Contemporary Ergonomics 2009: Proceedings of the International Conference on Contemporary Ergonomics 2009

Contemporary Ergonomics 2009 Contemporary Ergonomics 2009 Editor Philip D. Bust Loughborough University © 2009 Tay...

Author: Philip D. Bust

229 downloads 1523 Views 4MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Contemporary Ergonomics 2009

Contemporary Ergonomics 2009 Editor

Philip D. Bust Loughborough University

© 2009 Taylor & Francis The Development of HSL’s Safety Climate Tool – A Revision of the Health and Safety Climate Survey Tool S. Sugden, M. Marshall, S. Binch & D. Bottomley © Crown Copyright 2008/HSL Reproduced with permission of the Controller of her Britannic Majesty’s Stationary Office. The views expressed are those of the Author and do not necessarily reflect the views or policy of the Controller or any government department. The Controller, any government and Taylor & Francis Ltd accept no responsibility for the accuracy of any recipe, formula instruction contained within this population. Typeset by Macmillan Publishing Solutions, Chennai, India Printed and bound in Great Britain by Antony Rowe (A CPI-group Company), Chippenham, Wiltshire All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Every effort has been made to ensure that the advice and information in this book is true and accurate at the time of going to press. However, neither the publisher nor the authors can accept any legal responsibility or liability for any errors or omissions that may be made. In the case of drug administration, any medical procedure or the use of technical equipment mentioned within this book, you are strongly advised to consult the manufacturer’s guidelines. ISBN 978-0-415-80433-2 (Pbk) ISBN 978-0-203-87251-2 (eBook)

Contents Preface

XI

PLENARY SPEAKERS Inspection: From goods through passengers to systems C.G. Drury Business ergonomics beyond health and safety: Work environments for employee productivity, creativity and innovation J. Dul

3

16

Occupational safety: Charting the future Y.I. Noy

24

Ergonomics and public health P. Buckle

33

Human factors and development of next generation collaborative engineering J.R. Wilson, H. Patel & M. Pettitt

39

ACCESSIBILITY A design ergonomics approach to accessibility and user needs in transport R. Marshall, D.E. Gyi, K. Case, J.M. Porter, R.E. Sims, S.J. Summerskill & P.M. Davis Development of tools for reducing the stress of using public transport P.M. Davis, R. Marshall, K. Case, J.M. Porter, D.E. Gyi, S.J. Summerskill & R.E. Sims Validation of the HADRIAN system with a train station design case study S.J. Summerskill, R. Marshall, D.E. Gyi, J.M. Porter, K. Case, R.E. Sims & P.M. Davis

51

61

70

COMPLEX SYSTEMS A software tool for evaluating the effects of culture on military operations A. Hodgson & C.E. Siemieniuch V

81

VI

Contents

Decision-making after the product-service shift and some implications for ergonomics E.M. Molloy, M.A. Sinclair & C.E. Siemieniuch Governance, agility and wisdom in the capability paradigm M.A. Sinclair, M.J.D. Henshaw, R.A. Haslam, C.E. Siemieniuch, J.L. Evans & E.M. Molloy

91

100

DESIGN The role of ergonomics in technology oriented projects; the intelligent zero emissions vehicle project A. Woodcock & J. Taylor

111

Research and policy development into scooter access onto London buses S. Thomas

122

HAZARDS Study of verbal communication between nuclear plant control room operators during abnormal situations A.N. Anokhin & N.V. Pleshakova …It’s the way that you do it: Ensuring rule compliance J. Berman, P. Ackroyd & P. Leach

135

145

HEALTH ERGONOMICS AND PATIENT SAFETY Public requirements for patient-held records J. Binnersley, A. Woodcock, P. Kyriacou & L.M. Wallace

159

Systems analysis for infection control in acute hospitals P.E. Waterson

165

Portable pods: Design for unscheduled (urgent) care in the community S. Hignett, A. Jones & J. Benger

174

Using usability testing as a benchmarking tool – A case study on medical ventilators Y. Liu & A. Osvalder

182

Contents

Does the older workforce with high work demands need more recovery from work J.J. Devereux & L.W. Rydstedt

VII

189

HUMAN ADVISORY The role of the human in future systems – Considerations and concerns M.S. Young

199

Automation and technology in 21st century work and life S. Sharples

208

Vigilance and human supervisory control – A potted history J.M. Noyes

218

HUMAN ERROR/ACCIDENTS Unpacking the blunt end – Changing our view of organizational accidents B.M. Sherwood Jones Lessons from the past to give to the future: Learning from incidents S. Evans The development of HSL’s safety climate tool – A revision of the health and safety climate survey tool C. Sugden, M. Marshall, S. Binch & D. Bottomley

229

236

245

HUMAN FACTORS INTEGRATION Reporting location and environmental features when using metal detector or probe in a simplified mine detection task R.J. Houghton, C. Baber & J.F. Knight Predictive Operational Performance (PrOPer) model K.T. Smith Distributed situation awareness: Applications and implications in defence P.M. Salmon, N.A. Stanton, G.H. Walker & D.P. Jenkins

255

263

272

VIII

Contents

Human factors models of mini unmanned aerial systems in network-enabled capability C. Baber, M. Grandt & R.J. Houghton Human factors past and present W.I. Hamilton

282

291

Ergonomics challenges for digitisation of mission planning systems N.A. Stanton, G.H. Walker, D.P. Jenkins & P.M. Salmon

300

Human-centered design focus in systems engineering analysis: Human view methodology R.J. Smillie & H.A.H. Handley

310

Submarine safety and spatial awareness: The subsafe games-based training system R.J. Stone & A. Caird-Daley

320

Criteria for assessing social and organisational impacts of emerging technologies P. Sirett, R. Braithwaite, R. Cousens, J. Marsh & M. Mason

332

INSPECTION Animating the inanimate: Visual inspection from the medical to the security domains A. Gale

347

Inspecting for safety hazards K. Woodcock

354

Recent advances in technology to improve inspection training A. Gramopadhye, P. Stringfellow & S. Sadasivan

363

Using body discomfort correlations to assist in ergonomic inspections of computer users P. Stringfellow & A. Gramopadhye

373

One year later: How screener performance improves in X-ray luggage search with computer-based training A. Schwaninger & A.W.J. Wales

381

Contents

IX

METHODS AND TOOLS The hexagon-spindle model for ergonomics R. Benedyk & A. Woodcock

393

PRIMARY INDUSTRIES A biomechanics workload assessment of female milking parlour operatives M. Jakob, F. Liebers & S. Behrendt Swedish perspectives on ergonomics in the primary sector P. Lundqvist

405

414

ROAD ERGONOMICS A 50-driver naturalistic braking study: Overview and first results N. Gkikas, J.H. Richardson & J.R. Hill

423

Motorcycle ergonomics: Some key themes in research S. Robertson, A.W. Stedmon, P.D. Bust & D.M. Stedmon

432

Keeping it real or faking it: The trials and tribulations of real road studies and simulators in transport research A.W. Stedmon, M.S. Young & B. Hasseldine

442

SCHOOLS Pupil involvement in classroom (re)design: Participatory ergonomics in policy and practice A. Woodcock, J. Horton, O. den Besten, P. Kraft, M. Newman, P. Adey & M. Kinross Results from a post-occupancy evaluation of five primary schools M. Newman, A. Woodcock & P. Dunham

453

462

STRATEGIC ERGONOMICS Repositioning human factors/ergonomics (HF/E) – Raising high hazard industry understanding and commitment to HF/E D. Pennie, M. Wright & P. Mullins

473

X

Contents

Ergonomics for engineers P.J. Clarkson The trials and tribulations of trying to apply ergonomics in the real world M. Kenwright & D. Pennie

483

491

TRANSPORT A preliminary investigation of a low-cost method for prediciting the disruption of glances towards in-vehicle information systems A. Irune & G.E. Burnett

503

The effective management of user requirements and human factors issues C. Munro & S.R. Layton

513

WORKING PEOPLE Predicting the human response to an emergency G. Lawson, S. Sharples, S. Cobb & D. Clarke

525

Human factors contributions to the design of a security promotion campaign: A case study A.E. Peron, B.F. Davies & S.R. Layton

533

Author Index

543

Preface Contemporary Ergonomics 2009 is the proceedings of the Annual Conference of the Ergonomics Society, held in April 2009, at the Royal College of Physicians, London, UK. The conference is a major international event for Ergonomists and Human Factors Specialists, and attracts contributions from around the world. This year marks the 60th anniversary of the Ergonomics Society. Papers are chosen by a selection panel from abstracts submitted in the autumn of the previous year and the selected papers are subject to a peer review process before they are published in Contemporary Ergonomics. Each author is responsible for the presentation of their paper. Details of the submission procedure may be obtained from the Ergonomics Society. The Ergonomics Society is the professional body for ergonomists and human factors specialists based in the United Kingdom. It also attracts members throughout the world and is affiliated to the International Ergonomics Association. It provides recognition of competence of its members through its Professional Register. For further details contact: The Ergonomics Society Elms Court Elms Grove Loughborough Leicestershire LE11 1RG UK Tel: (+44) 1509 234 904 Fax: (+44) 1509 235 666 Email: [email protected] Web page: http://www.ergonomics.org.uk

XI

PLENARY SPEAKERS

INSPECTION: FROM GOODS THROUGH PASSENGERS TO SYSTEMS Colin G. Drury University at Buffalo, SUNY, USA Ergonomics work on inspection started with field studies of people inspecting components repetitively, often in a line production environment. The original papers on the ergonomics of industrial inspection were often written by psychologists with a highly interpretive and contextual view of the process they were reporting on. Since that time (1950’s) studies moved to laboratory settings, with emphasis on mathematical modeling, although many quantitative field studies were also performed. Modern production processes deemphasize and even denigrate inspection as unproductive or counterproductive. Ergonomics studies of inspection began to expand into maintenance inspection, e.g. of aircraft or their components, into medical radiology and more recently into the airport security domain. With good qualitative and quantitative models of inspection available, we should now be in a position to tackle more abstract domains, applying what we have learned to inspection of organizations for safety, or complex paperwork systems for compliance. In these areas, what we know about inspection complements what we know about auditing processes and systems for compliance. There is a demand to better understand such inspection/audit activities, for example in government oversight of aviation maintenance or procedure compliance.

Introduction Inspection tasks have been a minor but continuing theme in human factors / ergonomics (HFE) practice and literature since the 1950’s. Many ergonomists have been asked to improve jobs and workplaces where items were inspected and have contributed to an evolving understanding of how people make repetitive decisions, often Yes/No decisions about goods and services. Most of our understanding of inspection tasks is based on these practical studies, often augmented by off-line studies to better model the effects of variables noted but uncontrolled in the field. Of course ergonomists were not the only ones studying inspection: there are long traditions in quality control and equipment/automation design that have very little (and that often derogatory) to say about the human in the inspection system. More modern strictures on quality have emphasized open-loop control of industrial processes rather than inspection as a way of ensuring that defective items are all but eliminated from the process. This has led to calls to end inspection a position 3

4

Colin G. Drury

significantly aided by automation of many measurement tools. A different class of inspection tasks is embodied in the continuing inspection of built systems to ensure that defects have not arisen and thus the system is still safe for public use. Here we include the inspection of railway tracks for a range of conditions that could compromise passenger safety, and parallel inspections of civil aircraft structures, or of passenger bags being brought into the civil aviation system. Without too much of a stretch, this class can include medical inspection, e.g. radiography or endoscopy, designed to ensure that the human body is still working as expected. In parallel to the quality control literature, there is an equally well-developed literature on auditing, although again this rarely makes explicit mention of the human auditor. There have been audit programs developed for ergonomics / human factors but little consideration of auditing as inspection of systems rather than of objects. Finally we consider the potential application of findings on human and automated inspection to audit tasks by comparing their functional similarity to ask whether we can improve auditing processes by treating them as inspection of systems. The genesis of this extension came from a request by G. Salvendy for the author to combine his previous Handbook chapters on Human Factors and Automation in Test and Inspection (Drury, 2001a) and Human Factors Audit (Drury, 2001b) into a single chapter Human Factors and Ergonomics Audits (Drury, 2006). The obvious response to this request was to devise a framework that would cover both traditions and lead to new applications. This paper is the result.

Inspection of product Inspection has been a traditional part of the whole production process with written roots dating back almost 5000 years to ancient Sumaria. In the epic Gilgamesh, the narrator invites the reader to examine the quality of the walls of Uruk, built by the King: “Take hold of the threshold stone–it dates from ancient times! Go close to the Eanna Temple, the residence of Ishtar, such as no later king or man ever equaled! Go up on the wall of Uruk and walk around, examine its foundation, inspect its brickwork thoroughly. Is not (even the core of ) the brick structure made of kiln-fired brick, and did not the Seven Sages themselves lay out its plans?” (Gilgamesh, Tablet 1) The essence of the examining function is all there already. Bodily senses (look, take hold of ) are used to compare the existing item (wall) with some implied or actual standard (e.g. kiln-dried brick). In traditional craft skills the functions of inspection were part of the overall job of the craftsperson, no different in essence from other task elements. With the advent of the industrial revolution, and its development of F. W. Taylor’s “Scientific

Inspection: From goods through passengers to systems

5

Management”, the breakdown of jobs into separate tasks performed by separate people led to specialized inspectors. Taylor himself (Taylor, 1911 Chapter 2) studied the inspection of ball bearings in bicycle manufacture, concentrating (as might be expected) on speed and throughput rather than accuracy, although he did report increases in overall quality as well as quantity after the job was thoroughly studied and rest breaks introduced. Also early in the 20th century the concepts of statistical quality control were being introduced, where specified accuracy was paramount and neither speed nor worker well-being was an explicit consideration. The linking of accuracy, throughput and well-being in industrial inspection tasks awaited the insights of psychologists (human factors, ergonomics) after World War II. In the 1950’s and 1960’s studies of industrial inspection were undertaken in the UK and USA for a variety of products from electronic wiring (Jamison, 1966), through ball bearings (Belbin, 1963) to newly-minted coins (Fox, 1964). They combined detailed observations and interviews with measurements of defect detection performance. Performance was generally found to be poorer than expected (e.g. Table 1 of Gallwey, 1998) and often quite variable between inspectors and from day to day. These early studies also found some of the factors affecting human performance in inspection tasks, such as the presence of comparison standards for wool grading (McKenzie, 1958). Early work by Jamieson (1966), Thomas and Seaborne (1961) and McKenzie (1958) established that social pressures are an important part of the inspector’s job, and that inspectors can at times change their behavior and performance in response to such pressures. These field studies have continued throughout the study of industrial inspection, e.g. with knitting machine needles (Drury and Sheehan, 1969), and a whole series of electronics components (Harris and Chaney, 1969). Interesting new ideas were brought in by Lloyd et al., 2000), using customer ratings of the importance of defects to devise lighting systems for car bodies that emphasized those most important to the customer. Industrial inspection studies continue (e.g. Dalton and Drury, 2004 on sheet steel inspection). Good reviews of many published studies and the factors found to affect inspection performance are found in Megaw (1979) and Gallwey (1998). Parallel to this field (and field-based) work, there has been a tradition of modeling of inspection behavior. At its simplest, this has involved a classification of the subactivities of inspection, typically in the form of a function or task description. A typical list is adapted from Drury (2006) below but many others are available: Setup, where the inspection system is prepared for use. Needed tools, equipment and supplies are procured and calibrated, procedures are available to aid the inspector, and the inspector has been trained to perform the task correctly. Present, where the inspector and the entity to be inspected come together so that inspection can take place. Search is typically a sequential serial process during which the whole item to be inspected is brought piece-by-piece under scrutiny, stopping when an indication is found or further search is considered unnecessary. Decision is where the indication located by search is judged against a standard to determine whether it is a true defect. Response, where the action chosen in Decision must be taken correctly.

6

Colin G. Drury

In the 1960’s and 1970’s the Signal Detection Theory (SDT) model became popular with inspection researchers, partly because both the simplest form of SDT and the simplest form of inspection have outcomes expressible as 2 × 2 contingency tables (e.g. Drury and Addison, 1973). The TSD model of detecting a signal of known characteristics within random noise was at least plausible for inspection, although a number of authors questioned the applicability of TSD’s assumptions when applied to inspection (e.g. Buck, 1975). In parallel to this TSD modeling, a number of papers used models of visual search and visual detection for inspection tasks (e.g. papers in Morris and Horne, 1960). This modeling theme was taken up in inspection of medical chest X-ray images (e.g. Kundell and LaFollette, 1972), see later. Visual search models assume that detection of a target only takes place after it is fixated by the eyes. They also assume some process by which successive eye fixations are located on the item searched, either as a random process, a systematic process or a systematic process with fallible memory for prior fixation locations (e.g. Morawski et al., 1980; Arani et al., 1984). In fact both the search process (e.g. visual search) and the decision process (e.g. TSD) are probably important in industrial inspection tasks. Both would imply some form of speed-accuracy trade-off, as was seen by Drury (1973) and later used in a generic inspection model (Drury, 1975). However, while this flow of field studies and models of inspection tasks was taking place, the whole idea of industrial inspection was being challenged by newer manufacturing techniques and philosophies.

The end of product inspection? With the introduction of more tightly coupled production systems, error cannot be tolerated without widespread consequences, so that control of quality at source is a key component of advanced manufacturing systems such as Flexible Manufacturing Systems (FMS) cells. The corresponding push towards Lean manufacturing has also made early control of errors a priority. The concept of controlling key variances at their source has arisen through the various iterations of quality movements, but is also a key feature of Socio-technical systems design. One major effect of this change in production architecture has been to renounce to a large extent post-production inspection and replace it by in-process inspection. Indeed, the third of Deming’s (1982) fourteen points was: Eliminate the need for mass inspection as the way of life to achieve quality by building quality into the product in the first place. Require statistical evidence of built in quality in both manufacturing and purchasing functions. With the ultra-low defect rates now being demanded and achieved, even highly effective sorting inspection leaves the error rate too high. If we require defective rates in parts per million (∼ 10−6 ), then even a hit rate on inspection of 99% (error rate ∼ 10−2 ) will not give the necessary result unless the production defective


7

rate is ∼ 10−4 . Even this level may be difficult with human inspectors as they tend to be less effective at defect detection as the overall quality improves (e.g. Panjawani and Drury, 2003). Thus we look instead for necessary precursors to error rather than error. If we know the mean of a process has shifted by a given amount, then we can deduce that defects will be more common, even if we find no defects. This is the principle of in-process quality control, often using control charts. Where does this leave inspection? Is it to be eliminated completely, to be used on rare occasions when things go wrong, or must it move on beyond the manufacturing domain? In part the last of these alternatives is nearest the truth, mainly because of automation and information technology in manufacturing.

Inspection in other domains: Food, medical images, safety and security If we consider inspection ergonomics moving beyond its original industrial quality control domain, we can treat the reduced importance of this original domain as a cause or as an irrelevance. Inspection ergonomics had to evolve, but it was doing so well before the manufacturing revolution. One obvious area was in medical imaging inspection from the original Kundel and LaFollette (1972) beginnings to sophisticated use of models and tackling of new problems (Chen et al., 2008; Nodine et al., 2002). Another area that received ergonomics attention well before JIT was farm product inspection, e.g. of chicken carcasses (Chapman and Sinclair, 1973) or sexing of day-old poultry (Biederman and Schiffar, 1987). This activity still continues in all food industries but has received little recent attention beyond schemes for automation. There is a need for inspection to ensure airworthiness of civil and military aircraft, where inspection detects cracks, corrosion, loose fasteners etc. so that they can be repaired before they pose a threat to safe operations. Human Factors issues in civil aircraft maintenance inspection has received considerable attention following the fuselage damage to an airliner in 1988 for which the cause was found to be inadequate inspection performance (e.g. Drury et al., 1990). Since then studies have involved equipment design, training (e.g. Gramopadhye et al., 1987), job aids (Wenner et al., 2003) and good practices guides (Drury, 2003). A major growth area of inspection research has been in aviation security. This ergonomics research started many years ago (Funke, 1979) but expanded greatly after US terrorist attacks were widely perceived as resulting from inspection failures. Although all security screening tasks appear to fit the task description given above (Drury, 2002), most ergonomic attention has been given to X-ray screening of carry-on or checked baggage (e.g. Gale et al., 2000; McCarley et al., 2004; Hsaio et al., 2008). Models current in inspection, e.g. TSD, Drury Search Plus Decision model, visual search models and visual perception models, have all found applicability in this domain. Other areas of inspection that are increasing are single-stream recycling, where each object type has to be sorted (e.g. Associated Press, 2008), fire safety inspection of buildings and inspection of food facilities for compliance with local regulations. None have received ergonomics attention, but should be amenable to

8

Colin G. Drury

the observation, analysis and modeling techniques derived from studies of industrial inspection. In fact, the inspection of organizations for safety or compliance represents perhaps the next domain for inspection research. The manufacturing changes over the past two decades which heralded the demise of product inspection have also had the effect of moving organizations and safety professional away from reliance on post incident measures (e.g. injuries, equipment damage) towards pre-incident measures. In these, we are looking for valid indications of precursors to the rare events that accidents and injuries should become in a well-run organization. This field of examining organizations for compliance to standards already has its own techniques and tradition, under the rubric of auditing.

The auditing tradition When we audit an entity, we perform an examination of it. Dictionaries typically emphasize official examinations of (financial) accounts, reflecting the accounting origin of the term. Accounting texts go further, e.g. “. . . testing and checking the records of an enterprise to be certain that acceptable policies and practices have been consistently followed” (Carson and Carlson, 1977, p. 2). As with inspecting, auditing is mentioned in antiquity, at least in current translations. “. . . . . . he who does not pay the fine annually shall owe ten times the sum, which the treasurer of the goddess shall exact; and if he fails in doing so, let him be answerable and give an account of the money at his audit.” (Plato: Laws Book VI) Returning to the definition of an audit, the emphasis is on checking, acceptable policies, and consistency, to provide a fair representation of the business for use by third parties. A typical audit by a Certified Public Accountant would comprise the following steps (from Koli, 1994): Diagnostic Investigation: description of the business and high lighting of areas requiring increased care and high risk. Test for Transaction: trace samples of transactions grouped by major area and evaluate. Test of Balances: analyze content. Formation of Opinion: communicate judgment in an audit report. Such a procedure can also form a logical basis for safety audits. The first step chooses the areas of study, the second samples the system, the third analyses these samples while the final step produces an audit report. These are one way to define the broad issues in a safety audit design: How to sample the system: How many samples and how these are distributed across the system?


9

What to sample: What specific factors (valid and reliable precursors of safety problems) are to be measured? How to evaluate the sample: What standards, good practices or principles to use for comparison? How to communicate the results: Techniques for summarizing the findings, how far can separate findings be combined? Given this long auditing tradition, and its known applicability to ergonomics / human factors audits (e.g. Drury, 2001b), is there a structural similarity between inspection and auditing from which we can gain insight into the audit process?

Auditing as inspection of systems The simplest structural congruence between inspection and safety audit tasks is based on the task list of inspection given above: Setup implies the design of a sensitive, valid and reliable audit tool, and the training of auditors in its use. Present, implies a sampling plan for which parts of the organization to audit, and access to the sampling points chosen. Search is both the sequential following of the sampling plan and the active search of each area samples to locate possible indications. Decision is the judgment of the indication located by search against auditing standards to determine whether it is a true precursor of a safely problem. Response is the preparation and presentation of a clear audit report to those who commissioned the audit. Note that these map quite simply onto the four points derived from auditing: Diagnostic Investigation comprises the Setup task, Test for Transaction comprises the Present and Search tasks, Test of Balances is Decision while Formation of Opinion is the Response task. The list of how the four audit steps can be applied to safety audits merely rounds out this correspondence. This structural similarity can help the audit process by suggesting similar issues found in inspection tasks that could also occur in safety audits. All of these issues have been found in field studies of inspection and in the models used to better understand that function. A non-comprehensive list of issues is presented below in approximately the order of the generic inspection task list. Validity is the extent to which a methodology measures the phenomenon of interest. Content validity is perhaps the simplest but least convincing measure. If each of the items of our measurement device displays the correct content, then validity is established. Theoretically, if we could list all of the possible measures of a phenomenon, content validity would describe how well our measurement device samples these possible measures. In practice, it is assessed by having experts in the field judge each item for how well its content represents the phenomenon studied. Concurrent (or predictive) validity has the most immediate practical impact. It measures empirically how well the output of the measurement device correlates with

10

Colin G. Drury

the phenomenon of interest. Of course, we must have an independent measure of the phenomenon of interest, which raises difficulties. Construct validity is concerned with inferences made from scores, evaluated by considering all empirical evidence and models. Thus a model may predict that one of the variables being measured should have a particular relationship to another variable not in the measurement device. Confirming this relationship empirically would help validate the particular construct underlying our measured variable. The issue of validity of ergonomics audits has been addressed by many authors (e.g. Keyserling et al., 1993; Drury 2001b)) Validity for product inspection is typically less of a problem as the defects the inspection system is designed to prevent are usually based on customer input (e.g. Lloyd et al., 2000). However, even for products, validity can be compromised by the tendency for management to add indications to the list, so that in one example, inspectors were expected to detect over 80 different defects on the moving sheet steel (Dalton and Drury, 2004). Training has been seen as an important factor in inspection performance since the days of Belbin (1963). Clearly an untrained inspector is practically worthless, but the typical on-the-job training still found in many places is also highly inefficient. For example in X-ray security inspection (e.g. Schwaninger and Hofer, 2004) and in product inspection (Kleiner and Drury, 1992) a training scheme designed around psychological principles produced large performance improvements for even experience inspectors. For audit programs, the issue of providing effective training for auditors has not been addressed by ergonomists, although clearly safety auditors must be trained effectively. In industry there is typically a great respect for auditors with experience, but as noted above, training and experience may have different effects on performance. Accessibility is not typically an issue in product inspection but is a serious one for aviation maintenance inspection (Mozrall et al., 2000), where constricted spaces force awkward postures on inspectors of airframes and engines. A number of studies (Mozrall et al., 2000; Hsaio et al., 2008) have shown minimal effects of working posture on inspection performance, but postural discomfort is still suspected to be an ergonomic cause of poor performance in general. For audits, the equivalent issue is less one of posture than of permission to access. If an organization expects an audit then it can take steps to put on a good showing for the auditors. For this reason, most safety audits should be (are are) unannounced. It is still possible for organizations to hide problems (c.f. security inspection for threats) and to deny access to certain parts of buildings, so that auditors need to be aware of such potential biases in drawing up a sampling plan and in following logical indications when performing the audit. Access and Search overlap somewhat in safety audits, as they do for any other procedural inspection task. Speed-accuracy trade-off (SATO) has already been well-documented for product inspection and is clearly an issue for safety audits. Most of any SATO effect in product inspection appears to arise from the Search task (Drury, 1975). This results in an increase in probability of detection of a true defect and in probability of a false alarm as time taken per item increases (e.g. Ghylin et al., 2006, for security inspection). In audits the SATO issue has been characterized by Drury (2001b) as a balance between breadth, depth and application time. Ideally, an audit system would


11

be broad enough to cover any task in any industry, it would provide highly detailed analysis and recommendations, and would be applied rapidly. Unfortunately, the three variables of breadth, depth and application time are likely to trade off in a practical system. At the level of audit job aids, there are comprehensive ergonomics surveys such as the Position Analysis Questionnaire (McCormick, 1979) which take 2–3 hours to complete. Alternatively, there are simple single page checklists such as the Ergonomics-Working Position-Sitting Checklist (SHARE, 1990) that can be completed in a few minutes. The simplest job aid for any procedural task is the checklist. Safety inspectors in industry typically use a written checklist of several pages. Most checklists are used as memory aids for well-practiced tasks, so that they are structured as lists of commands, each of which is relatively terse: “switch both magnetos on: open fuel cock: prime engine for 5 seconds, etc.”. The user is expected to know the objects referenced and why each action is required. Checklists have their limitations, though. The cogent arguments put forward by Easterby (1967) provide a good early summary of these limitations in the context of design checklists and most are still valid today. Easterby quotes Miller (1967): “I still find that many people who should know better seem to expect magic from analytic and descriptive procedures. They expect that formats can be filled in by dunces and lead to inspired insights. . .We should find opportunity to exorcise this nonsense.” (Easterby, p 554) Easterby finds checklists can have a helpful structure, but often have vague questions, make non-specified assumptions, and lack quantitative detail. Checklists are seen as appropriate to aid operators (not ergonomists) in following procedural steps. Many checklists are developed, and many of these published, that contain design elements fully justifying such criticisms. Much research on checklists since that time has established a number of conclusions, all appropriate to safety audit checklists: 1. Checklist design: Use a “geographical” sequence of steps and good formatting/typography (Degani and Wiener; 1990), provide context information as well as instructions (Ockerman and Pritchett, 2000), use computer-based checklists as a more powerful alternative to paper-based ones (Larock and Drury, 2003) 2. People often use checklists in a manner not intended by designers, following their own sequence (Patel et al., 1993), using the object checked (an aircraft) for spatial cues (Pearl and Drury; 1995) or not using the checklist at all (Ockerman and Pritchett, 2000) Checklists are not the only issue in SATO for auditing as there is also the issue of time spent in the search phase of the task for any given design of audit procedure. Auditors, like inspectors, can be more diligent or less in their work. Decision Bias has been well-documented in inspection, typically in terms of the trade-off between missed defects and false alarms for product or security inspection. In auditing, there is a similar issue of potential bias in what is judged as an important

12

Colin G. Drury

finding and what is reported to the sponsor. Not all of the facts uncovered in an incident investigation are used in the report (Drury et al., 2003) and there is no reason to suspect that audits are immune to such selection biases. As Drury (2001b) notes, the design of the audit instrument can affect such biases by providing standards (Illumination must be >500 Lux) that can affect how the data is collected, rather than collecting the raw data (Record Illumination) and letting a computer program make the comparison to standards. Overall, a number of issues have been raised by this mapping of audits onto inspection tasks, all of which use inspection findings in the context of a safety audit. There are far more than can be pursued in future treatments, for example does a culture that encourages reporting and discussion of minor safety incidents (e.g. paper cuts) actually reduce the incidence of rare, large events?

References Arani, T., Karwan, M. H. and Drury, C. G. (1984). A Variable-Memory Model of Visual Search. Human Factors, 26, 63l–639. Associated Press (2008). Recycling innovation could end household ‘separation anxiety’ Wednesday April 30. Belbin, R. M. (1963). Inspection and Human Efficiency. Ergonomics in Industry Series, DSIR, London. Biederman, I. and Schiffar, M.M. (1987). Sexing day-old chicks: A case study and expert systems analysis of a difficult perceptual learning task. Journal of Experimental Psychology, Learning, Memory and Cognition, 13, 640–645. Buck, J. R. 1975, Dynamic visual inspection, in C. G. Drury and J. G. Fox (eds), Human Reliability in Quality Control (London: Taylor & Francis). Carson, A. B. and Carlson, A. E. (1977). Secretarial Accounting, (10th Edition) (Cincinnati, Ohio: South Western Publishing Company). Chapman, D. E. and Sinclair, M. A. (1975). Ergonomics in inspection tasks in the food industry. In C. G. Drury and J. G. Fox (Eds), Human Reliability in Quality Control, London: Taylor & Francis, 231–252. Chen, Y., Gale, A. and Scott, H. (2008) Anytime, Anywhere Mammographic Interpretation Training, Contemporary Ergonomics 2008, P. D. Bust (Ed.). Taylor & Francis, London, 375–380. Dalton, J. and Drury, C. G. (2004). Inspectors’ performance and understanding in sheet steel inspection. Occupational Ergonomics, 4, 51–65. Degani, A. and Wiener, E. L. (1990). Human Factors of Flight-Deck Checklists: The Normal Checklist, NASA Contractor Report 177548 (CA: Ames Research Center). Deming, W.E., (1982). Out of the Crisis, MIT, Cambridge, MA. Drury, C.G. (l973). The Effect of Speed of Working on Industrial Inspection Accuracy. Applied Ergonomics, 4, 2–7. Drury, C.G. (l975). Inspection of Sheet Materials – Model and Data. Human Factors, l7, 257–265.


13

Drury, C. G. (2001a). Human Factors and Automation in Test and Inspection, In G. Salvendy, Handbook of Industrial Engineering, Third Edition, Chapter 71, John Wiley & Sons, New York, 1887–1920. Drury, C. G. (2001b). Human Factors Audit, In G. Salvendy (ed), Handbook of Industrial Engineering. Third Edition, Chapter 42, John Wiley & Sons, New York, 1131–1155. Drury, C. G. (2002). A Unified Model of Security Inspection, Proc. Aviation Security Technical Symposium, Federal Aviation Administration, Atlantic City, NJ. Drury, C. G. (2003). Human factors and inspection: Best practices. Proceedings of the XVth Triennial Congress of the International Ergonomics Association, Ergonomics Society of Korea/Japan, Seoul/Korea, 584–587. Drury, C. G. (2006). Human Factors and Ergonomics Audits, In G. Salvendy (ed), Handbook of Human Factors and Ergonomics. Third Edition, Chapter 42, John Wiley & Sons, New York, 1106–1132. Drury, C.G. and Addison, J.L. (l973). An Industrial Study of the Effects of Feedback and Fault Density on Inspection Performance. Ergonomics, l6, l59–l69. Drury, C. G., Ma, J. and Woodcock, K. (2003). Do job aids help incident investigation? Contemporary Ergonomics, Proc. of the Ergonomics Society Annual Conference 2003, Edinburgh Scotland, April 15–17, 2003, Taylor & Francis, London. Drury, C. G., Prabhu, P. and Gramopadhye, A. (1990). Task analysis of aircraft inspection activities: methods and findings. Proceedings of the Human Factors Society 34th Annual Conference, Human Factors and Ergonomics Society, 1181–1185. Drury, C.G., and Sheehan, J.J. (l969). Ergonomic and Economic Factors in an Industrial Inspection Task. Int. Jnl. Prod. Res., 7, 333–34l. Easterby, R. S. (1967). Ergonomics checklists: an appraisal. Ergonomics 1967, 10, 548–556. Fox, J.G., 1964. The ergonomics of coin inspection. Quality Engineer 28, 165–169. Funke, D.J. (1979). Training: a significant factor in the effectiveness of airport baggage inspectors. Security Management, pp. 30–34. Gale, A.G., Mugglestone, M.D., Purdy, K.J., & McClumpha, A. (2000). Is airport baggage inspection just another medical image? Proceedings of the SPIE, 3981, 184–192. Gallwey, T. J. (1998). Evaluation and control of industrial inspection: Part II: The scientific basis for the guide International Journal of Industrial Ergonomics 22, 51–65. Ghylin, K.M., Drury, C.G., and Schwaninger, A. 2006, Two-component model of security inspection: application and findings, 16th World Congress of Ergonomics, IEA 2006, Maastricht, The Netherlands, July, 10–14. Gramopadhye, A. K., Drury, C. G. and Prabhu, P. (1997). Training for aircraft visual inspection, Human Factors and Ergonomics in Manufacturing, 3, 171–196. Harris, D. H., & Chaney, F. B. (1969). Human Factors in Quality Assurance. New York: Wiley.

14

Colin G. Drury

Hsiao, Y-L, Joseph, C., Joshi, S., Lapp, J., Pennathur, P. R, and Drury, C. G., (2008). Posture and performance: sitting vs. standing for security screening, Ergonomics, 51, 290–307. Jamieson, G. H. (1966), Inspection in the telecommunications industry: A field study of age and other performance variables. Ergonomics, 9, 297–303. Keyserling, W. M., Brouwer, M. and Silverstein, B. A. (1992). A checklist for evaluation ergonomic risk factors resulting from awkward postures of the legs, truck and neck, International Journal of Industrial Ergonomics, 9, 283–301. Kleiner, B. M., and Drury, C. G. (1992). Design and Evaluation of an Inspection Training Program. Applied Ergonomics, 24, 75–82. Koli, S. T. (1994). Ergonomic Audit for Non-Repetitive Task. Unpublished M.S. Thesis. State University of New York at Buffalo. Kundel, H.L., & La Follette, P.S. (1972). Visual search patterns and experience with radiological images. Radiology, 103, 523–528. Larock, B. and Drury, C. G. (2003). Repetitive inspection with checklists: design and performance. Proc. Ergonomics SocietyAnnual Conference 2003, Edinburgh Scotland, April 15–17, 2003. Lloyd, C. J., Boyce, P., Ferzacca, N., Eklund, N, and He, Y. (2000). Paint inspection lighting: Optimization of lamp width and spacing. Journal of the Illuminating Engineering Society, 28, 99–102. McCarley, J. S., Kramer, A.F., Wickens, C.D., Vidoni, E.D., & Boot, W.R. (2004). Visual Skills in Airport Security Screening. Psychological Science, 15, 302–306. McCormick, E. J. (1979). Job Analysis: Methods and Applications, (New York: AMACOM). McKenzie, R. M. (1958). On the accuracy of inspectors. Ergonomics, 1(3), 258–272. Megaw, E.D. (1979). Factors affecting visual inspection accuracy. Applied Ergonomics, 10, 27–32. Miller, R. B. (1967). Task taxonomy: science or art. In W. T. Singleton, R. S. Easterby and D. Whitfield (eds.), The Human Operator in Complex Systems. (London: Taylor and Francis). Morawski, T., Drury, C. G. and Karwan, M. H. (l980). Predicting Search Performance for Multiple Targets. Human Factors, 22, 707–718. Morris, A. and Horne, P. E. (eds) (1960), Visual Search Techniques (Washington, D.C.: National Research Council). Mozrall, J. R., Drury, C. G., Sharit, J. and Cerny, F. (2000). The effects of whole-body restriction on inspection performance. Ergonomics, 43, 1805–1823. Nodine, C.F., Mello-Thoms, C., Kundel, H.L., & Weinstein, S.P. (2002). Time course of perception and decision making during mammographic interpretation. American Journal of Roentgenology, 179, 917–923. Ockerman, J. J. and Pritchett, A. R. (2000a). Reducing over-reliance on taskguidance systems. In Proceedings of the Human Factors and Ergonomics Society 44th Annual Meeting, San Diego, CA.


15

Panjawani and Drury, C. G. (2003). Effective interventions in rare event inspection. Proceedings of the Human Factors and Ergonomics SocietyAnnual Meeting, Denver, CO, October 2003. Patel, S., Drury, C. G. and Prabhu, P. (1993). Design and usability evaluation of work control documentation, Proceedings of the Human Factors and Ergonomics Society 37th Annual Meeting, Seattle, WA, pp.1156–1160. Pearl, A. and Drury, C. G. (1995). Improving the reliability of maintenance checklists. Human Factors in Aviation Maintenance – Phase Four, Progress Report, DOT/FAA/AM-93/xx, (Springfield, VA: National Technical Information Service). Schwaninger, A. and Hofer, F. (2004). Evaluation of CBT for increasing threat detection performance in X-Ray Screening. In K. Morgan and M.J. Spector (Eds). The Internet. Society: Advances in Learning, Commerce & Security, (pp 147–156). WIT Press.SHARE (1990). Inspecting the Workplace, SHARE Information Booklet (Australia: Occupational Health and Safety Authority). Taylor, F. W. (1911). The Principles of Scientific Management, Harper & Brothers, New York. Thomas, L. F. and Seaborne, A. E. M. (1961). The sociotechnical context of industrial inspection, Occupational Psychology, 35, 36–43. Wenner, C. A., Drury, C. G. and Spencer, F. W. (2003). The impact of instructions on aircraft visual inspection performance: A first look at the overall results. Proceedings of the Human Factors and Ergonomics Society Annual Meeting, Denver, CO, October 2003.

BUSINESS ERGONOMICS BEYOND HEALTH AND SAFETY: WORK ENVIRONMENTS FOR EMPLOYEE PRODUCTIVITY, CREATIVITY AND INNOVATION Jan Dul Rotterdam School of Management, Erasmus University, The Netherlands The value of ergonomics for companies is beyond health and safety. With ergonomically designed work environments a company can reach a competitive advantage. Ergonomics can help to improve employee productivity (especially in organisations focusing on a cost reductions), and increase employee creativity (especially in organisations focusing on innovation). Hence, ergonomics can have a strategic value for business organisations. In this paper, models and examples are presented showing that ergonomics can create business value. The consequences for ergonomics researchers and practitioners are mentioned.

Introduction According to the formal description of ergonomics, approved by the International Ergonomics Association, ergonomics deals with the design of products and processes in order to improve ‘human well-being’as well as ‘total system performance’. Improvement of human well-being can be considered as the ‘health, safety and comfort’ goal of ergonomics, which is important for the users of products and processes (consumers and workers). Similarly, improvement of overall total performance -although this term is rather vague-, can be considered as the ‘business’ goal of ergonomics, which is important for the management of an organization that develops, produces or uses products. This paper elaborates on the business dimension of ergonomics. Models and examples are presented which show how ergonomics can contribute to business goals. First I present a model how ‘product ergonomics’ and ‘production ergonomics’ can be linked to the common ‘value chain’ business model. Next, for production ergonomics, I will discuss to possible business topics for ergonomics: how to design work environments for human productivity, and how to design work environments for human creativity for innovation.

Business ergonomics In management, a business process is usually described as a chain of valueadding activities (Porter, 1985). The upper part of Figure 1 shows the 16

Business ergonomics beyond health and safety

Research

Product Process Purchasing development development

Production

17

Distribution

Product creation and realisation Product ergonomics Ergo-strategy Manager Ergo-innovation

Production ergonomics

Differentiation strategy with user-friendly products

Cost strategy with productive, human-friendly processes

Ergonomic Product development

Ergonomic process development

Ergonomic products

Ergonomic processes

Designer Ergo-system User

Figure 1.

Ergonomics as part of the value chain of regarding the ‘product’ and the ‘production’ (From Dul, 2003).

chain: ‘research, product development, process development, purchasing, production, and distribution’. The first part of the chain focuses on the ‘product’ and describes activities that must result in effective (in terms of sales) new or improved products or services. The second part of the chain focuses ‘production’and describes activities that must result in efficient (in terms of costs) realization of the product or service. Also in ergonomics a distinction has been made between ‘product ergonomics’ and ‘production ergonomics’. The lower part of Figure 1 suggests that product ergonomics can add additional value to the process of product creation (research, product development), and that production ergonomics can add additional value to the process of product realization (process development, purchasing, production, and distribution). With product ergonomics a competitive advantage can be achieved, by developing products that fit or exceed the functional and affective expectations of the user. For this strategic reason the management (the marketing manager, for example) can decide to use ergonomics. With production ergonomics, effective and efficient production systems can be realized using the capabilities of motivated, productive and creative employees. For this strategic reason the management (the operations manager, for example) can decide to use ergonomics. At the same time more acceptable working conditions regarding health and safety can be reached, which usually is a secondary argument for business ergonomics. In the remainder of this paper the focus will be on two business issues that are relevant to production ergonomics: productivity and innovation, by discussing how to design work environments for human productivity, and how to design work environments for human creativity for innovation.

18

Jan Dul

Work environments for productivity Since the industrial revolution, organizations focus on productivity: maximization of output at minimum cost. To support this objective, management principles and instruments have been developed and used in manufacturing and services companies all over the world. As a result, labour has been divided into small specialized tasks, processes have been standardized and subsequently mechanized and automated, and workers are specialized to do remaining tasks, many times resulting in repetitive and monotonic work with standard workplaces. However, when high levels of mechanization and automation have already been realized, further efforts to mechanize and automate the operations may not be as effective (ELA, 2004). In addition to ergonomics programmes that have been developed and implemented in order to reduce the negative side effects such work environments in terms 1 of health and safety and motivation, there is an increased attention for an ergonomics approach that combines productivity and health and safety. When ergonomics is applied, work environments can combine more efficient human movements, easier to assembly products, fewer rejected products etc., with less tiredness, better motivation, and improved health and safety. Ergonomically designed work environments may contribute to a reduction of sickness absenteeism and presenteeism. Presenteeism refers to those productivity losses occurring before an operator goes on sickness absence (Meerding et al., 2005). With ergonomically designed work environments fewer working days may be lost, costs to assist sick workers may be lower, return to work may be easier and quicker, there may be fewer cases of disability, lower personnel turnover, and fewer temporary workers to replace sick workers. In the ergonomics literature there is and increasing interest in productivity issues, and attempts have been made to provide empirical evidence of the productivity and costs effects of ergonomics programmes. For example, Koningsveld et al., 2005 analyzed 9 cases of ergonomics improvement projects, and found that the benefits from performance improvements vastly exceeded the benefits from reduced sickness absence alone. Several studies in the manufacturing industry in different countries have shown that with ergonomics productivity improvements can be realized (in combination with improved health and safety conditions). For example, Looze et al. (2003) analysed two cases in the Netherlands (one assembling office furniture, and the other assembling, testing and packing valves) report that with ergonomically improved technical and organizational processes (layout, task design, and work place design), productivity gains of 15–20% (and substantial reductions of time spent in risky body postures) can be achieved. Lee (2005), found in a Korean electric appliance company, that applying low cost ergonomics improved workplaces resulted in 10–30% productivity increase and US$17.01 million cost savings (and reduced number of musculoskeletal problems). Sen and Yeow (2003) studied the ergonomic redesign of electronic motherboard in a computer peripheral manufacturing factory in Malaysia, and found US $581,495/year cost savings and improved operators’ health and safety. These and other studies demonstrate that with ergonomically designed work environments considerable productivity gains can be achieved when applying ergonomics in the design of the production


19

environments and that companies have indeed implemented this approach successfully.

Example Rhijn et al. (2005) studied work environment improvements of an assembly process in a Dutch company Famostar Emergency Lighting BV that develops, assembles and sells emergency lighting. The company is market leader The Netherlands. It produces 200.000 systems each year and is rapidly growing. In the old assembly process, a batch of parts were laid down on a table, and assembled and packed manually by workers who walked along the table. Finished products were placed manually on a pallet for further transportation. Due to an increase of the product volume and due to lack of space, a project was started to redesign the assembling process. The batch-type of production was changed into a flow-type of production. Sitting workplaces were introduced, which allowed picking parts from boxes close to the body, and lifting equipment was introduced to reduce manual lifting. The project showed the following results: labour productivity in terms of average number of products per person per day increased 44%, the order lead time reduced 46%, the space requirement decreased 44%, and manual lifting load reduced from unacceptable according to the NIOSH guidelines, to acceptable. However, the arm posture worsened because the workers worked 12% longer with a 20–60 degrees elevated arm posture. Overall, the results show that the assembling process was more productive and more human-friendly, although some nuances were found. This example shows how ergonomics can contribute the company’s productivity and other (strategic) business goals.

Work environments for creativity and innovation While most above examples are from manufacturing, in particular the economies of the Western world largely depend on services and knowledge work. In a knowledge based and innovation driven competitive business environment, work environments designed for productivity may not be the right choice. A company that needs to compete on innovation needs its organizational members not only for reaching productivity goals, but also for developing novel and useful ideas for solving problems and developing new products, services, processes, systems, work methods, etc. The creative potential of the workforce needs to be stimulated and work environments need to be designed to support worker creativity. Hence, organizations need workers – both employees at any level in the organization, and managers (Ceylan et al., 2008) – that can produce novel and useful ideas. Creativity has become a fundamental resource for many western organizations to maintain competitive. Organizations that pay attention to worker creativity normally focus on the worker, not on the work environment (Ceylan and Dul 2007). These organizations select creative workers, and provide creativity training. However, generating creativity by selection and training of the creative workers may not be sufficient. Creative workers that are placed in traditional productivity driven work environments may not show the desired creative behaviour. Organizational work

20

Jan Dul

Organizational Environment (e.g. job design, leadership)

(e.g. personality, cognitive style)

Creativity

Novel and useful ideas

Innovation

Individual

Physical Environment (e.g. workplace design, building design)

Figure 2. The effect of individual factors, organizational environment and physical environment on creativity and innovation (Adapted from Dul and Ceylan, 2006). environments that are designed for productivity, having formal management structures, time constraints, strict regulations, daily similar tasks, etc., may even obstruct or inhibit worker creativity. Also physical work environments that are designed for productivity, e.g. having standardised workplaces, may not support creativity. Dul and Ceylan (2006) have suggested that the ergonomics field could contribute to ‘bottom up’ innovation by studying and developing work environments that can support creativity and innovation. Based on empirical evidence from different fields like organizational behaviour, environmental psychology, business administration architecture, etc. they developed a conceptual framework called the ‘Creativity Development model’ (CD model) (Figure 2) on the relationship between individual characteristics, organizational work environment characteristics, physical work environment characteristics and individual creativity and related innovation. In comparison to existing models from the management literature the CD model explicitly highlights the role of the physical environment to support creativity. Based on this model Dul et al. (2007) developed the Creativity Development Quick Scan (CDQS) which is a checklist for measuring the creativity potential of work environments. The checklist includes 21 factors that could foster or hinder creativity. Four factors relate to job design for supporting creativity (challenging job, team work, task rotation, autonomy in job), 5 factors relate to leadership for supporting creativity (coaching supervisor, time for thinking, creative goals, recognition of creative ideas, and incentives for creative results), 5 factors relate to interior design for supporting creativity (furniture, indoor plants/flowers, calming colours, inspiring colours, privacy) and 7 items factors relate to building design for supporting creativity (window view to natural elements, any window view, quantity


21

of light, daylight, indoor (physical) climate, sound, and smell). The checklist has to be filled in by the employee; for each factor the checklist measures on a 7-point scale the worker’s perceived climate for creativity. The CD Quick Scan has been applied in several studies, which resulted in a database that currently contains data from more than 1000 individuals with various professional backgrounds in 5 nations from different regions of the world. If individuals work in the same organisation the individual data can be aggregated to obtain the climate for creativity of an organization. Presently data are available from 30 companies. Many problems and solutions regarding work environments for creativity are very similar to the problems and solutions for work environments for comfort, health and safety. Hence, existing ergonomics knowledge could be readily applied. For other problems the required knowledge may not be readily available (yet) in the current ergonomics domain, although it may be available in other domains (organizational behaviour, business administration, environmental psychology, architecture, etc.). Yet for other problems further studies are needed to develop solutions.

Example A Dutch consultancy firm intended to stimulate entrepreneurship, curiosity and creativity of its consultants in order to enter new markets with existing competencies. First a CDQS analysis of the existing work environment in the company was performed, and the results were benchmarked against 4 other comparable organizations in the Netherlands. Tables 1 and 2 show the results of this analysis; the consultancy firm of interest is company 7. It turns out that the company scored relatively low for both the organizational climate for creativity and the physical climate for creativity. For the organizational climate for creativity, only ‘autonomy in the job’ scored higher than average. For the physical climate for creativity only ‘any window view’ scored higher than average. This comparison convinced the management that there was a need to make work environment improvements. First the results were discussed with the employees (the consultants), in order to get suggestions for improvements and prioritize them. With respect to the organizational work environment, there appeared to be a need for more teamwork (e.g. more consultants per project) for better exchanging information and ideas between consultants, and for realizing social support between consultants. There was a need for a more supportive and coaching rather than controlling leadership styles of the managers in order to stimulate creativity of the employees (e.g. by training of managers in leadership styles). With respect to the physical environment, several suggestions were made to change the workplace, which can be realized relatively easy (furniture, plants, colour). Currently, the suggestions are being considered by the management for implementation. One main result of this process was that on the basis of objective data, discussions started between management and consultants about the present situations and possibilities for improvements, similarly to any other participatory ergonomics approach.

22

Jan Dul

Table 1.

Organizational climate for creativity of 5 companies.

Company

2

4

5

6

7

All

Challenging Job Teamwork Task rotation Autonomy in job Coaching supervisor Time for thinking Creative goals Recognition of creative ideas Intensives of creative results Total Organisational

6.13 6.41 4.69 6.13 6.41 4.69 6.18 5.76 4.18 5.62

4.59 5.22 3.24 6.00 5.00 4.24 4.47 4.71 4.12 4.62

5.01 4.68 3.43 5.62 3.51 3.59 3.65 3.76 3.58 4.09

5.43 5.57 4.71 5.93 5.36 4.14 4.71 5.43 4.79 5.12

4.75 4.08 3.67 6.25 3.25 3.50 3.58 3.67 3.75 4.06

5.18 5.19 3.95 5.99 4.71 4.03 4.52 4.67 4.08 4.70

Table 2.

Physical climate for creativity of 5 companies.

Company

2

4

5

6

7

All

Furniture Indoor plants/flowers Calming colours Inspiring colours Privacy Window view to nature Any window view Quantity of light Daylight Indoor (physical) climate Sound Smell Total Physical

4.53 5.18 4.71 3.47 3.76 3.59 5.35 5.82 5.88 4.59 4.18 3.88 4.58

4.88 3.82 3.59 3.00 5.06 4.00 5.06 5.18 5.65 4.00 4.24 3.82 4.43

3.86 3.38 3.00 2.59 4.32 4.15 5.11 5.14 5.38 3.64 4.32 3.71 4.05

4.50 3.93 3.14 2.71 3.57 3.64 3.54 4.71 4.43 4.21 3.50 3.86 3.81

3.67 2.42 3.00 2.42 3.33 3.17 5.09 3.67 2.42 3.00 2.42 3.33 3.16

4.29 3.75 3.49 2.84 4.01 3.71 4.83 4.90 4.75 3.89 3.73 3.72 3.99

This example shows how ergonomics can contribute a company’s creativity and innovation policy.

Conclusion The above models and examples show that the field of ergonomics can add value beyond health and safety. Ergonomics can contribute to a company’s competitive advantage by providing approaches and solutions to improve labour productivity, or to improve the climate for creativity and bottom up innovation. However, most organizations are not aware of such value of ergonomics, and the same may apply to many ergonomists. Therefore in order to further exploit this added value of ergonomics in the future, ergonomists need to develop their skills and knowledge to link ergonomics to business strategies and goals, and to communicate this towards the management and business community (Dul and Neumann, in press).


23

References Ceylan, C. and Dul, J. 2007. Expert opinions on human resources practices for creativity stimulating work environments for innovation in Turkey, the Netherlands, and Brazil. Paper presented at the 9th International HRM Conference on Changes in Society, Changes in Organizations, and the Changing Role of HRM: Managing International Human Resources in a Complex World, Tallinn, Estonia, June 12–15. Ceylan, C., Dul, J. and Aytac, S. 2008. Can the office environment stimulate a manager’s creativity? Human Factors and Ergonomics in Manufacturing 18(6), 589–602. Dul, J. 2003. ‘De mens is de maat van alle dingen’. Over mensgericht ontwerpen van producten en processen. (‘Man is the measure of all things’. On human centrered design of products and processes). Inaugural address. Erasmus Research Institute of Management, Rotterdam, The Netherlands. Dul, J. and Ceylan, C. 2006. Enhancing organizational creativity from an ergonomics perspective: The Creativity Development model. Paper presented at the 16th World Congress on Ergonomics (IEA 2006), Maastricht, The Netherlands, July 10–14. Dul, J., Ceylan, C. and Hendriks, H. 2007. A practical instrument to measure the creativity potential of the work environment. In: Proceedings of the 10th European conference on Creativity and Innovation, Copenhagen, Denmark. Dul, J. and Neumann, W. P. (in press) ‘Ergonomics Contributions to Company Strategies’, Applied Ergonomics. ELA (2004). The success factor people in distribution centers, European Logistics Association/Kurt Salmon Associates, Brussels. Koningsveld, E. A. P., Dul, J., Van Rhijn, G. W. and Vink, P. (2005). Enhancing the impact of ergonomics interventions. Ergonomics, 48(5), pp. 559–580. Lee, K. S. (2005) ‘Ergonomics in total quality management: How can we sell ergonomics to management?’ Ergonomics, 48(5), pp. 547–558. Looze, M.P. de, Van Rhijn, J.W., Deursen, J. van, Tuinzaad, G.H. and Reijneveld, C.N. (2003). A participatory and integrative approach to improve productivity and ergonomics in assembly. Production Planning & Control 14(2): 174–181. Meerding, W. J., IJzelenberg, W., Koopmanschap, M. A., Severens, J. L. and Burdorf, A. (2005). Health problems lead to considerable productivity loss at work among workers with high physical load jobs. Journal of Clinical Epidemiology, 58(5), pp. 517–523. Porter, M.E. 1985. Competitive advantage: creating and sustaining superior performance. Free Press, New York, USA. Rhijn, J.W. van, Looze, M.P. de, Tuinzaad, G.H., Groenesteijn, L., Groot, M.D. de, Vink, P. 2005. Changing from batch to flow assembly in the production of emergency lighting devices. International Journal of Production Research 43(17), 3687–3701. Sen, R. N. and Yeow, P. H. P. (2003) ‘Cost effectiveness of ergonomic redesign of electronic motherboard’, Applied Ergonomics, 34(5), pp. 453–463.

OCCUPATIONAL SAFETY: CHARTING THE FUTURE Y. Ian Noy Liberty Mutual Research Institute for Safety, USA The relentless human drive to overcome our inherent limitations and amplify our physical and mental capabilities is evident from examining the historical evolution of work systems. This paper spotlights selected milestones in the evolution of human work, illustrating the role that ergonomics innovations have had in shaping human civilization. With this as backdrop we discuss the relatively recent emergence of occupational safety as a societal priority. Yet, despite the enormous magnitude of the global burden of work-related injuries, progress has been slow. We argue that meaningful future progress in occupational safety will need to embrace the principles of resilience engineering and be integrated with modern approaches to management systems.

Introduction As the Ergonomics Society prepares to celebrate its 60th Anniversary it is fitting to reflect on the past achievements of the discipline and contemplate the prospects of ergonomics playing a more meaningful role in shaping the society of the future. The theme of the annual conference, “Contemporary Ergonomics”, challenges us to examine the current state of the discipline/profession in relation to important global issues. In this paper we examine the burden of occupational injuries as a major public health concern facing both industrialized and industrializing societies. In his book The Substance of Civilization: Materials and Human History from the Stone Age to the Age of Silicon, Stephen Sass (1999) traces how the discovery of new materials has shaped events throughout human history, enabling technological innovations and ushering in new eras. There is a hidden, yet a far deeper dimension running throughout the book that tells an even more compelling story. Read from the perspective that the discovery of new materials is but the outcome of the natural human drive to amplify our physical and mental power, what emerges is a compelling account of the evolution of ergonomics from an artform dealing with relatively simple machines to a science dealing with complex, dynamic systems. Indeed, the powerful story between the lines of this book provides a fascinating tour of the evolution of work systems. We begin with a brief description of selected milestones in the history of work systems, based loosely on Sass’ work, and with this as backdrop we discuss the emergence of occupational safety as a societal priority. From a high level perspective, workplace safety is a rather recent concern that demands far greater attention than it has received thus far. In particular, we argue 24

Occupational safety: Charting the future

25

that future progress in occupational safety will need to embrace the principles of resilience engineering and be integrated with modern approaches to management systems.

Evolution of work systems: From art to science Many associate the emergence of the science of work, with Frederick Taylor’s application of scientific principles at the turn of the 20th century to improve productivity. Actually, references to the science of work appeared 50 years earlier in Jastrebowski’s 1857 treatise entitled “An outline of ergonomics, or the science of work, based upon truths drawn from the Science of Nature”. Lexicon aside, it is evident that some form of work science was already developed thousands of years ago. It may not have benefitted from the application of the scientific method, but it was tried, tested and codified. How else could massive projects such as the pyramids of ancient Egypt be organized and carried out? How else could ingenious aqueducts that span hundreds of miles across the Roman Empire be built with great precision? Other examples of monumental projects include the Great Wall of China that extends over 6,400 km (4,000 mi) and was built in various stages starting in the 6th century BCE. These massive projects, and many more like them, required sophisticated methods to plan and organize, train workers and manage the work. Moreover, it is clear that some form of ergonomics knowledge was relatively advanced even 2500 years ago. For example, archaeological discoveries relating to ancient Greece provide abundant evidence of the extensive application of human-centred principles in the design of temples, theatres, tools and household implements, and methods relating to construction and medicine (Marmaras 1999). It is of particular interest that in 400 B.C.E., Hippocrates, the father of Western medicine, formulated a set of guidelines for surgical work in a hospital. His writings indicated that the design of the workplace, the use of tools, lighting and posture were important factors affecting the surgeon’s performance. These guidelines embodied ergonomics knowledge, though they may not have been scientifically derived.

Pre-historic work systems The fact of the matter is that the human drive to innovate and control our environment actually has its origins millions of years ago and coincides with the very foundations of society as we know it. The role and power of human-centred technology is a common thread in the evolution of human society. Early humans were by necessity constantly searching for ways to extend their capacity to survive, and ultimately to prevail. They fashioned simple tools from stone to dig up roots, crack open nuts, hunt game and to cut hides. Anthropologists have uncovered the earliest evidence of stone implements in the Olduvai Gorge in Tanzania, dating back more than two million years. This was the dawn of ergonomics as an artform. In shaping the hand ax for more complex tasks Homo habilis was applying ergonomic principles to improve biomechanical power. He had no formal

26

Y. Ian Noy

understanding of physics, biomechanics or properties of materials – but the end product was an ergonomic innovation nonetheless. It is particularly fascinating that the first appearance of stone tools coincided with the separation of the genus Homo from Australopithecus, which eventually became extinct. The aptitude for ergonomics as characterized by the intelligent and creative use of resources to support the human enterprise gave Homo a decided advantage. Homo survived for other reasons, including the sharing of responsibility for gathering wild plants and hunting game (early application of macroergonomics). Interestingly, other species, notably chimpanzees, exhibit intelligent use of tools and organized behaviour. But, humans evolved at a much faster rate. According to the renowned biologist, Christopher Wills, unlike other species that evolve in response to changes in the environment, humans evolve in response to changes that we ourselves make in our environment (Wills, 1998).

Agriculture-based society The emergence of Homo sapiens some 35,000 years ago led to further technological innovations, yet these modern humans continued to live a nomadic hunter-gatherer lifestyle for several thousands more years. The transition to a farmer-merchant economy first occurred in the Middle East some 10,000 years ago and was made possible by the discovery of clay and how to fire it. This ergonomic breakthrough facilitated the cooking and storage of liquids and grains. Moreover, by providing the means for transporting food and other products, clay vessels enabled trade with neighbouring Mesopotamia and Anatolia for needed raw materials and metals. Archaeological discoveries in Jericho suggest that its inhabitants learned to cultivate wheat and store food products. The advent of agriculture allowed them to establish permanent settlement. Two thousand years later, the technological spotlight shifted to southern Babylon. It is here five to six thousand years ago, that written language was developed for keeping inventory of farm goods, keeping records of trade, and maintaining property deeds. In the same way that stone tools extended human physical capabilities, written language extended human memory resources, and established a reliable basis for shared knowledge. Parenthetically, ancient written media also included stone. The most famous example is the Code of Hammurabi, carved in stone in 2200 B.C.E. This code is of some interest to ergonomics because it contains the first known set of design standards, and established the penalties of defective design. For example, according to the code, if a house should fall and cause the death of the owner, the builder shall be put to death. Further specialization of tools and instruments to suit the particular task at hand awaited the discovery of metals, such as copper and bronze. New tools such as saw blades and drills were developed that proved as important economically as chariots and swords were politically. Iron appeared later, being far more difficult to extract than bronze. In some regions, steelmaking was developed to improve the functional characteristics of iron. In The Substance of Civilization Stephen Sass recounts how the iron swords of the Gauls bent during their battles with the Roman legions,


27

which were armed with Toledo steel blades (Sass, 1999). The Gauls apparently had to straighten their blades across their knee after each blow, a deadly ergonomics disadvantage. But, the Roman Empire, despite its technological prowess, eventually fell, leaving the Western world in chaos. The quality of goods and production processes actually declined during the fifth and sixth centuries. The Spread of Islam in the seventh and eighth centuries stimulated renewed interest in science and technology, and it helped to introduce Eastern ideas into Western culture, ideas that were instrumental to the continued evolution of ergonomics. Printing and paper, both developed in China about a thousand years earlier, spread westward along the Silk Road in the ninth and thirteenth centuries, respectively. The ability to reproduce hundreds of books with identical text and diagrams was crucial to the rise of modern science and technology. Another Chinese innovation to be carried westward was gunpowder. It was decisive in the Muslim resistance against the Crusaders, leading to the development of cannons and the need to improve the quality and production of iron, which was to become the engine of the Industrial Revolution. The Renaissance marked the transition from medieval to modern times. The rediscovery of Greek and Roman literature led to the emergence of the humanist movement in the 14th century, and an interest is scientific knowledge. Humanism, the belief that each individual has significance within society, was yet a further expression of the silent, but powerful, human-centred ideal. By the end of the Renaissance, another idea emerged – it was profitable to turn out products for common people. A mass market was opening, and the new economy was being fueled by private enterprises in countries that encouraged it.

Industrialization: The precursor of ergonomics as a science Together with coal, iron made possible the industrialization of the modern world. The Industrial Revolution marked the transition from a stable agriculturalcommercial society to a modern industrial society relying on complex machinery rather than tools, thus opening a new chapter in the relationship between people and technology. Whereas in England, machines were used to improve quality, in America, machines were used to improve productivity. In both instances, however, there was a need to overcome existing human-system limitations. Dramatic changes in the social and economic fabric of society took place as inventions and technological innovations created the factory system and greater economic specialization. Things began to be mass produced for unknown users – users who had different needs, were of variable size and had different preferences. The challenge then, as it is now, was to design things that would be functional and appealing to a wide market. This gave rise to an interest in individual differences and its application to design, a topic central to ergonomics science. We also note the emergence of explicit human-centred concepts in the design of machines, such as the spinning jenny. The industrial revolution also gave rise to the modern institution of labour. It led to greater intellectual interest in the work activity of ordinary people. LaMettrie’s controversial book, L’homme Machine, published in 1748, concerned the relationship

28

Y. Ian Noy

between humans and machines. By the 19th Century, labour was viewed by scientists, social and political reformers as an immense reservoir of energy that needed to be managed (conserved) so as to optimize productivity (Rabinbach, 1990). The advent of the Industrial Revolution in the mid-18th Century ushered in what we know as the science of work. The application of the scientific method to work system design was at first an attempt to maximize human output, but later reoriented to address overall system performance and worker wellbeing. The period following the Industrial Revolution was marked by an explosion of technological innovations, which continue today at an accelerated rate, leading to the recent emergence of systems engineering as an approach to address the growing complexity of socio-technical systems. More recently, the construct of joint cognitive systems underscores the fact that humans and machines must be viewed as an integrated whole whereby humans act via the machine interface to affect change rather than on the interface (Hollnagel and Woods, 2005). This draws attention to the importance of humans and machines working cooperatively to control outcome in an unpredictable environment, rather than viewing them as separate units. We posit that this rather profound shift in paradigm, from human-machine interaction to human system integration, is central to thinking about new approaches to address the issue of occupational safety.

Occupational safety: Past and present Inventive work systems that had been developed to carry out the construction of the ancient pyramids or the Great Wall of China must have involved elaborate engineering and management. However, it is clear that worker safety was not a prime consideration in the design of work. For example, it is estimated that between one and two million workers died during the construction of the Great Wall of China. Much of the labour force in ancient Egypt and the Roman Empire comprised slaves who suffered unimaginable afflictions from their work, as they did from their oppressive masters. In fact, apart from survival, the primary motivation for ergonomics innovations throughout most of human history was the drive for political or economic superiority. Social objectives were secondary, if they mattered at all. Concern over occupational health and safety emerged during the Industrial Revolution. There was already some interest in this as evidenced by Ramazzini’s book in 1717, The Diseases of Workers, linking occupational hazards to work factors including repetitive tasks, posture and stress (Ramazzini, 1717). It took over half a century for the adoption of legislation on working conditions. Perhaps the earliest Act in the UK, The Health and Morals of Apprentices Act of 1802, applied to orphan apprentices in the textile industry. It stated that no children under the age of 9 were to be apprenticed and the working day was to be limited to 12 hours with no night work. Gradually, legislation was extended to other industries and smaller establishments. To be sure, since Ramazzini’s work, considerable progress has been made in understanding the aetiology of work-related disorders and developing interventions to control the risks in the workplace. Yet, the situation today remains quite dismal.


29

$ Billions Overexertion Fall on same level

$12.4 25.7%

Fall to lower level Bodily reaction Struck by object Struck against object Highway incident $6.4 13.3%

Caught in/compressed $5.3 10.8%

$4.8 10.0%

Repetitive motion $4.3 8.9%

Assaults/ violent acts $2.5 5.1%

$2.4 4.9%

$2.1 4.4%

$2.0 4.0%

$0.4 0.9%

Figure 1. Top ten causes and direct costs of workplace injuries in 2006. The WHO and ILO estimate that about 2.0 million work-related deaths occurred in 2000. While our knowledge of the global burden of workplace injury is woefully incomplete, it is widely acknowledged that it is enormous and unacceptable. Even in the developed world, the incidence of disabling injuries is unknown or understated. In the U.S.A., where reasonable data exist, the direct cost of the most serious occupational injuries in 2006 is estimated to be $48.6 B (Liberty Mutual Research Institute for Safety, 2008). The burden of serious1 workplace injury in the U.S.A. has been reported annually since 1998 in the Liberty Mutual Workplace Safety Index (LMWSI), which combines information from Liberty Mutual, the Federal Bureau of Labor Statistics and the National Academy of Social Insurance. These data have been influential in drawing attention to workplace injury as a public health and safety priority, and they permit prioritizing research efforts in areas that can have the most impact. Figure 1 shows the 10 leading causes of workplace injuries for 2006, as reported in the LMWSI. As shown, the leading cause of injuries is overexertion, accounting for over 25% of the burden, followed by falls on same level and falls to lower level, which together account for a further 25% of the burden. During the 8 years that the WSI has been published, the overall burden has not changed appreciably. While there have been differential changes in the relative contribution of the leading causes over the period 1998 to 2006, as depicted in Figure 2, the overall annual burden at nearly $50B remained unacceptably high. In sum, despite scientific progress in understanding the mechanisms underlying work-related injuries, efforts to develop effective countermeasures have met with limited success. We need to redouble our efforts to address this major public health issue. 1

Serious injury is defined as an injury resulting in six or more days away from work

30

Y. Ian Noy Overexertion Fall on same level Fall to lower level Bodily reaction Struck by object Struck against object Highway incident Caught in/compressed by Repetitive motion Assaults & violent acts

⫺40.0

⫺30.0

Figure 2.

⫺20.0

⫺10.0

0.0

10.0

20.0

30.0

Percent real growth change 1998–2006 by category.

Occupational safety: Where to from here? Comprehensive and meaningful improvement in occupational safety demands a change in our fundamental approach to the problem. Although considerable work remains to understand the physiological, biomechanical, behavioural and physical mechanisms underlying work-related injury, a wealth of knowledge exists to inform engineering solutions and workplace controls. It is increasingly evident that one major impediment to progress relates to the relatively underdeveloped knowledge in areas related to individual and organizational behaviour. In the remaining sections we briefly touch on topics that we believe should be vigorously pursued and developed, such as situated safety, resilience engineering and resilience management. The common theme is that safety must be regarded as an integral part of the whole work system, rather than as a separable property.

Situated safety Situated safety emphasizes the need to take into account critical situational factors such as target workforce, work organization, culture, language, infrastructure, and other socio-technical characteristics in developing an understanding of work-related injuries and prevention strategies. All too often research protocols and instruments are borrowed and applied in settings that differ in one or more fundamental ways from the settings for which they were developed or demonstrated to be valid. Similarly, interventions are implemented in situations where they have not been shown to be effective. It is essential to understand the limitations to generalizability of research protocols and interventions that appear in the literature.


31

Related to this is the notion that safety should be viewed in relation to a dynamic situation, rather than as a probabilistic outcome associated with exposure to static risks. Humans are goal oriented and remarkably adaptive. They often work under conditions of uncertainty or subject to unforeseen disturbances. As a result, risk must be assessed in relation to the context of the operation. This requires knowledge of intrinsic and extrinsic situational factors that may be time varying and diverse, which makes them very difficult to study and measure. The case-crossover methodology (Lombardi et al., 2003) developed within the field of injury epidemiology attempts to identify the transient conditions that existed at the time of injury. Clearly, it will be important to improve our ability to capture and analyze situational data.

Safety through resilience engineering Conventional approaches to safety employ a public health model. This entails a sequence of activities designed to identify the risks, isolate the causes, study the underlying mechanisms, determine the most effective means of prevention, implement and test effectiveness. This approach is inadequate inasmuch as it is reactive and relies on hindsight. Unfortunately, there will always be new paths to failures. Rather than erecting reactive barriers and defences, the preferred approach is to design work systems that are resilient. The key to resilience engineering is building capacity to recognize, adapt to and recover from variations, surprises and failures. This is especially important for complex, dynamic systems. Hollnagel et al. (2006) summarize the current state of the science underlying this approach and provide a comprehensive description of its various aspects. They believe that “safety depends on providing workers and managers with information about changing vulnerabilities and the ability to develop new means for meeting these”. New methods and tools are beginning to emerge that can be used to engineer the system to allow workers to cope with complexity under pressure.

Safety through resilience management Business excellence models are used extensively in Europe, America, Asia and Australia to promote and reward quality management. The common element across the various models is the pursuit of continuous improvement of goods, services and processes through programs that focus on customer results and empower and motivate employees (Noy, 2000). From an ergonomics standpoint, it is noteworthy that the various models of business excellence consistently focus on people, both as customers and employees. Safety in the organization’s quality management system is viewed as an integral part of business excellence rather than an afterthought, measured at the end of the month to inform corrective measures. It is planned for and monitored on an on-going basis in real time. Resilience management recognizes the value of integrating safety and ergonomics in strategic and operational objectives. It comprises a comprehensive set of processes for anticipating, detecting, reacting and adapting to situations as they arise. The structure and processes of an enterprise should be analyzed to determine the extent to which safety considerations are systematically incorporated in

32

Y. Ian Noy

performance and quality management and constitute an integral part of business decisions.

Conclusions Resilience engineering and resilience management systems are complementary in the sense that the former relates to the design of a work system whereas the latter relates to its operation. Together they represent shifting paradigms – away from regarding safety as an isolated metric of system outcome to an integral objective of system design and operation. Work systems have evolved over millennia to a point where current systems thinking can support emerging constructs such as resilience engineering and management. Developing methods for applying these constructs represents the principal challenge for future progress in occupational safety. Until these new paradigms can be turned into practical models and tools perhaps an understanding of these concepts will stimulate designers, engineers and managers to think about occupational safety in a different, more beneficial way.

References Hollnagel, E., Woods, D.D. 2005, Joint Cognitive Systems: Foundations of Cognitive Systems Engineering, (CRC Press) Hollnagel, E., Woods, D.D. and Leveson, N. 2006, Resilience Engineering: Concepts and Precepts, (Ashgate) Liberty Mutual Research Institute for Safety. 2008, From Research to Reality, Volume 11(3) www.libertymutual.com/research Lombardi, D.A., Sorock, G.S., Hauser, R., Nasca, P.C, Eisen, E.A., Herrick, R.F., Mittleman, M.A. (2003). Temporal factors and the prevalence of transient exposures at the time of an occupational traumatic hand injury, J Occup Environ Med; 45: 832–840. Marmaras, N., Poulakakis, G. and Papakostopoulos, V. 1999, Ergonomic design in ancient Greece, Applied Ergonomics, Volume 30, Issue 4. Noy, Y.I., 2000. The Need for Integrating Human-Centred Principles in Quality Policy, keynote paper, II APERGO Conference on Ergonomics in Quality Management, Costa da Caparica. Rabinbach, A. 1990, The Human Motor: Energy, Fatigue, and the Origins of Modernity (Basic Books) Ramazinni, B., 1717 and 1940, In W. Wright (trans.): The Disease of Workers, (Chicago: University of Chicago Press) Wills, C. 1998, Children of Prometheus: The Accelerating Pace of Human Evolution, (Basic Books) Sass, S.L. 1999, The Substance of Civilization: Materials and Human History from the Stone Age to the Age of Silicon, (Arcade Publishing)

ERGONOMICS AND PUBLIC HEALTH Peter Buckle Robens Centre for Public Health, Post Graduate Medical School, Faculty of Health and Medical Sciences, University of Surrey, Guildford GU2 7TE The disciplines of public health and ergonomics share scientific concepts and often address similar health issues. However, the potential for a significantly greater engagement with public health priorities now exists. Problems to be tackled include the ageing population (and workforce), obesity and patient safety. Priorities may be set using public health methods (e.g. epidemiological trends, accident data, sickness records) but an ergonomics system approach will be required to address the complex settings in which public health must be delivered. This paper seeks to discuss how ergonomics and public health priorities might be mapped to identify suitable areas for research and knowledge application.

Introduction In this paper I have sought to set out an agenda that might identify and then strengthen the links between public health and ergonomics. It will also seek to show how new areas for research and professional application might be identified. Finally, it will identify and emphasise the importance of a number of current public health concerns. Public health has been defined as ‘the science and art of preventing disease, prolonging life and promoting health through the organised efforts and informed choices of society, organisations, public and private, communities and individuals.’ (Wanless, 2004). The breadth of this definition may be usefully set alongside the IEA (2000) definition of ergonomics ‘Ergonomics (or human factors) is the scientific discipline concerned with the understanding of interactions among humans and other elements of a system, and the profession that applies theory, principles, data, and methods to design in order to optimise human well-being and overall system performance.’ Both definitions stress the need to take a systems perspective. They also share the need to apply evidence based, scientific approaches when addressing systems based problems. More specifically the impact of both disciplines on protecting the health and well-being of individuals, is readily apparent. Despite these similarities there has been little consistent effort exerted by either discipline to develop potential synergies. 33

34

Peter Buckle

Public health in practice The approach to public health promoted by the USA’s Centre for Disease Control (CDC, 2008) can serve as a simple introduction. They describe a conceptual cycle, from assessment, through policy development to assurance of public health. This cycle can be segmented as follows: • • • • • • •

Monitor health status to identify community health problems Diagnose and investigate health problems and health hazards in the community Inform, educate, and empower people about health issues Mobilize community partnerships to identify and solve health problems Develop policies and plans that support individual and community health efforts Enforce laws and regulations that protect health and ensure safety Link people to needed personal health services and assure the provision of health care when otherwise unavailable • Assure a competent public health and personal health care workforce • Evaluate effectiveness, accessibility, and quality of personal and populationbased health services In addition, there is a recognition of the need to maintain active research across all of these segments to gain fresh insights to problems and to seek new solutions, against an ever changing system in which in all public health is delivered. The steps described in this public health cycle follow closely those of ergonomics practice. Similarities and natural synergies are readily identifiable. To examine more closely the relationship between public health and ergonomics we might consider topical issues. In England, the Chief Medical Officer’s annual report (2008) describes how each of the nine public health regions is prioritising health agendas. A simplified list of topics includes: • • • • • • • • • •

Mental health (and mental health and wellbeing in employment) Patient safety and acquired infections Driving down heart attack rates Uptake of screening services Health effects of climate change (e.g. heat stress) Inequality and deprivation Floods and recovery from floods (so called ‘slow burn’ emergencies) Response to emergencies (e.g. Avian flu) Childrens’ health and wellbeing Reducing childhood injuries (including falls, transport accidents, poisoning, smoke and fire, drowning) • Road deaths Each topic might be explored from an ergonomics perspective, and even simple reflection shows that most have important dimensions that can only be tackled using ergonomics and human factors based approaches. This leads logically to the challenge of how to systematically consider each from an ergonomics perspective and thus enable topics to be extracted and prioritised for research and knowledge application.

Ergonomics and public health

35

How and where might ergonomics influence public health? To address this question it seemed appropriate to start with a well established model of ergonomics and consider how public health concerns might then be mapped to it. Moray’s (2000) model was selected as one helpful perspective of the ergonomics approach. The model has been simplified to enable each level of the system to be explored (i.e. individual, team, management/organisation, legal and regulatory rules, sociopolitical). This has then be used to consider how each level of the system might be affected with any given public health challenge (e.g. obesity, ageing workforce).

Obesity In the following example I have chosen to use the example of obesity. The Department of Health have assessed what the levels of obesity in England may be in 2010 if current trends in prevalence continue. In fact, according to the Department by 2050 60% of men and 50% of women could be clinically obese. Despite this trend, it is a topic that has, to date, received little attention from ergonomists. A recent letter published in the journal Applied Ergonomics (Williams and Forde, 2008) highlights some of immediate factors of concern relating to the increase in obesity with in the UK. The authors also identifies the paucity of ergonomics research that is considering obesity in the context of work. In the following example (see table 1) I have focused on the work context of the modern office. A similar approach might have been used to consider other work contexts, for example jobs involving manual handling, sedentary professional drivers or those with a predominantly standing requirement such as often found in the retail industry or on production lines.

Ageing workforce Another major demographic change of direct impact on ergonomics is that of the ageing workforce (CEC 2003, 2004). Buckle et al. (2008a & b) are among a number of researchers attempting to identify what the issues are that will arise from the ageing workforce (largely resultant from the ‘baby boom’ years) and the change in balance of younger and older workers in workplace teams. A second and perhaps more critical factor is likely to be a shift towards more of the workforce staying in employment beyond the traditional retirement age of 65. Little is known of their ergonomics ‘metrics’ as, to date, relatively small numbers who choose to work beyond this age and those that do make this choice are usually highly selected, and a survivor population. We have yet to see what the impact will be when many more workers find it necessary to work beyond 65 because they have, for example, inadequate pension provision. Huge challenges lie ahead for ergonomists and public health professionals if cohorts of 65 plus, 70 plus and potentially 75 plus workers are to be adequately protected and helped to remain healthy, productive members of the workforce. A failure to adequately address the ergonomics issues may have

36

Peter Buckle

Table 1.

Example of an analysis of Public health and Ergonomics.

Public health demographic concern: Rise in Obesity Context: Modern office System Levels

Potential changes

Individual characteristics Physical Increased body mass and factors concomitant anthropometric changes. Probable change in physiological fitness

Examples of Ergonomic concerns

Current ergonomics standards for equipment and environments may be inadequate. Current data no longer valid for design purposes. Current equipment may be inadequate or subject to failure. Poor work organisation may encourage sedentary work throughout the whole day. This may exacerbate other physiological disorders associated with obesity. Work-rest scheduling for obese workers may need to be adjusted to allow sufficient worker recovery

Psychological factors

Stress related to public/ colleagues perception of obesity (including harassment and obesity). Impact of inappropriate work systems on sense of well-being and performance.

Motivation and well-being at work for obese workforces requires co-ordination with public health agenda, particularly in regard of education and support.

Team

Need to ‘carry’ and ‘protect’ member of team who may be more vulnerable

General understanding of teamwork and sharing of work load is poor and even less knowledge is available regarding work design and obesity issues.

Organisation

Level of knowledge, skills to assess needs

Development of appropriate needs assessment methods and training advice to be embedded in organisation. Need to develop social and business cases. Management training needs to address obesity.

Legal and regulatory rules

Existing guidance not applicable

Need to generate new data on appropriate norms and ranges. Need to validate new guidance.

Sociopolitical

Economic (benefits) pressure to keep all workers at work and maintain a healthy workforce

Informing policy makers of the benefits of ergonomics in preventing ill-health and accidents in the workplace. Promoting work system design principles

Ergonomics and public health

37

substantial implications for work, employers, and long term sickness absence and accident rates. Some other emerging public health concerns that ergonomists might choose to investigate include: • Climate change and impact on health (e.g. slow burn disasters such as flooding, human behaviour with respect to water scarcity, heat stress in design of workplaces, transport and homes, energy use and behaviour). • Public health and care of older people (e.g. design of care homes, delivery of care in community settings, patient safety, telecare design, assisted living design, slipping, tripping and falling accidents). • Transport and public health (access to public transport with an ageing population, demographic, design of transport for changed anthropometric demographic) • Nutrition and public health (need to establish sustainable behaviours, user friendly information, access to good nutrition (perhaps integrated with workplace and transport design) • Infection and public health (design of hospital systems, and other care environments, to minimize hospital acquired infections; user friendly infection control measures. • Public health and emergent threats (ergonomics in the design of bioterrorism defence systems, human factors in flu pandemics). The potential list is clearly much longer but the scope for the ergonomics discipline to be more closely engaged with the public agenda should be evident. Without the contribution of ergonomics thinking and methods the public health threats may not be tackled effectively.

Conclusions Ergonomics has, hitherto, paid little systematic and sustained attention to the public health agenda. It has much to offer, particularly from a systems methodology. Simple mapping of public health concerns to the ergonomics perspective can identify needs and help in prioritising ergonomics applications. Further, public health priorities are often based on sound epidemiological and other health trend data that provide a helpful base against which ergonomics interventions might usefully be judged. Those responsible for the education and training of professional ergonomists might consider further how opportunities for shared learning alongside public health professionals might be achieved or how a better understanding of public health concepts and priorities could be delivered. Ergonomics research and practice might seek also to influence public health through a greater sharing of research and application methods and findings. Publishing in public health journals and presenting at public health conferences are helpful options. The opportunity to showcase ergonomics within a public health context is an important one and the agenda is extensive and ever changing.

38

Peter Buckle

References Buckle P, Woods V, Oztug O, and Stubbs D 2008a Understanding workplace design for older workers: a case study In ‘Contemporary Ergonomics 2008’ Ed. Bust, P Publ. Taylor and Francis ISBN 978-0-415-46575-5 pp3–7. Buckle P, Woods V, Oztug O, and Stubbs D 2008b Understanding workplace design for older workers. Final report to Strategic Promotion of Ageing Research Capacity (SPARC) http://www.sparc.ac.uk/projects_older_worker.asp#l1. CDC 2008 The Essential Public Health Services http://www.cdc.gov/od/ocphp/ nphpsp/essentialphservices.htm accessed 10/10/08. Commission of the European Communities 2003 Sec 429 The Stockholm and Barcelona targets: Increasing employment older workers and delaying the exit from the labour market. Commission of the European Communities 2004 COM 146 final. Increasing the employment of older workers and delaying the exit from the labour market. Dept of Health 2008 On the State of Public Health: Annual Report of the Chief Medical Officer 2007. Moray N 2000 Culture, politics and ergonomics. Ergonomics, 43, 858–868. Wanless, D 2004 Securing Good Health for the Whole Population London HM Treasury. WHO 2008 Closing the gap in a generation: Health equity through action on the social determinants of health ISBN 978 924 1563703. Williams, N and Forde M 2008 Ergonomics and obesity. Letter to the editor, Appl Erg 40, 148–9.

HUMAN FACTORS AND DEVELOPMENT OF NEXT GENERATION COLLABORATIVE ENGINEERING John R. Wilson, Harshada Patel & Michael Pettitt Human Factors Research Group, University of Nottingham, UK Collaborative working, and especially that which takes place distributed in time and space, presents particular challenges to human factors. This paper draws from a large European Integrated Project, CoSpaces, which is building collaborative engineering systems in VR/AR, mobile and location aware computing. We discuss the development of descriptive models of collaboration and the use of use cases and scenario construction within user requirements analysis.

Introduction The changes we are seeing in the ways that work systems are designed and organised and in how people work have great implications for ergonomics approaches, methods, models and techniques. A notable example of this is the trend towards collaborative and distributed collaborative working, and the consequent need to determine how ergonomists will examine the strategies and performance of collaborative work teams, how they will apply these methods, and any emergent guidance for design. Advances made to help facilitate the work of distributed collaborative teams have focused on introducing new technologies into the workplace, and thus much research has concentrated on the design of computer and communication systems and their respective interfaces, but we must be careful not to neglect the design of operating procedures, job and team structures and notions related to performance such as distributed team situation awareness, mental workload and mental models. This paper concerns two interacting aspects of the human factors work on a major European project (CoSpaces) in collaborative work environments. First is the development of user requirements, feeding these through into system specifications. This is of interest for three reasons: the need to ensure that requirements embrace both system usability and the support of collaboration; the focus on a large “messy” distributed work system rather than the usual ergonomics of a single interface or product development (Wilson et al, 2003); and the dearth of literature which makes explicit how to define, construct and implement user scenarios, use cases and systems requirements in the context of collaborative systems. The second aspect is the development of a descriptive model of collaboration and collaborative working. For a notion which has been explored thoroughly within the world of computer supported cooperative work (CSCW) for many years we were surprised by the lack of a universal understanding of what collaboration comprises and therefore how to measure it. Our model has linked with project work such as: the generation of user requirements, development of evaluation frameworks and profiling tools for collaboration readiness. 39

40

J.R. Wilson, H. Patel & M. Pettitt

This paper draws together the work carried out on the collaboration model and that on user requirements, within the context of CoSpaces. Difficulties faced of general interest included integrating the work of human factors with the software engineering teams and disparate views regarding models within this context. There are differing views on the degree of formality required in such models and the extent to which they should be descriptive rather than predictive. This is of course an old debate within and around human factors, for instance found within discussions of mental models, and shows no signs of going away.

CoSpaces and relevant human factors We have carried out work over years in a variety of settings and for a variety of purposes, all with the theme of distributed working. This distribution is across functional specialisms amongst co-located work groups – for instance in control rooms (ATC, rail etc.) and manufacturing shop floors (electronics, automotive), and is also temporal and spatial, within mobile or distributed workspaces (e.g. in international aerospace manufacturing). This paper is based primarily on current work carried out in a large European Commission funded Integrated Project, CoSpaces, in which there are three workspaces – co-located, distributed and mobile. CoSpaces has industrial, research and business partners from 12 European countries, developing innovative collaborative working solutions that are responsive to industrial needs. The project explores how advanced technologies (virtual, augmented, tele-immersive, mobile, and context aware) can be deployed to create collaborative workspaces for supporting product design, assembly/construction and maintenance activities in the aerospace, automotive and construction sectors. Early on we structured the involvement of the human factors specialists into the overall project in such a way as to be as supportive as possible of the technical development teams, acting as the representatives of the users and translators of their requirements. Another role was to ensure that the needs for collaboration (and usability) were not swamped in the rush to produce advanced architectures and exciting tools and those resultant systems could be implemented to benefit engineering and business performance. Figure 1 illustrates the human factors contribution – of course a too neat and optimistic process but something against which to assess the project.

Model of collaborative work Approaches to modeling One of the critical early tasks in CoSpaces was to develop collaborative models to inform the development and implementation of CoSpaces systems. However, clarity was needed on what we intended by models in this context, since the term is used quite validly in many different ways by different people. This is perhaps not so surprising when we consider that a model is merely a simplified representation of the relevant properties of an entity produced for a specific (often explanatory) purpose.

Next generation collaborative engineering: Human factors and development

41

Active distributed development space Design guidance

Visionary scenarios Partner involvement Collaboration model

Current practice scenarios

CoScope profiling

CoSpaces scenarios

Use cases

Requirements Functional specification

Evaluation framework

Development

External company involvement Evaluation

Figure 1.

CoSpaces route to functional specification and development.

The human factors team wanted a model for explanation – to give a better understanding of what is collaboration and collaborative working, and the attributes which influence and form part of collaborative work. Such a model would allow recognition of the degree and type of collaborative work currently taking place within the user partners, help to establish strengths and weaknesses, provide a framework within which to define user requirements, and subsequently inform change management strategies and evaluation. On the other hand the developers and technologists (but interestingly not the user partners) wanted to initiate development of models that could be implemented within CoSpaces systems development, primarily to be used computationally, however crudely, at the very least to allow some form of built-in user profiling and possibly to support development of intelligent agency within the workspaces. These formal models are only likely to represent simple and predictable processes, and unable to model dynamically changing, complex, and unpredictable organisational processes (Robin et al, 2007). We have concentrated upon a descriptive model, which in any case would be required to form the basis of more formal models.

Collaboration and collaborative working models Research into collaboration spans a number of disparate fields like organisational/social psychology, CSCW, management science, engineering and human factors, and domains including education, healthcare, etc.; thus existing definitions of collaboration are numerous, varied and often domain specific. However a number of common themes emerge, i.e. collaboration involves two or more people engaged in interaction, and that these people are working together towards common goals. Other associated attributes which emerge are: trust, communication, coordinating activities, shared resources, shared rules/norms, shared vision and responsibility, and shared skills. These attributes have been incorporated into our model of collaborative work. Collaboration tasks, processes, and related technology support have been classified according to various theories and frameworks, including: the group task

42


circumplex, task technology fit theory, media richness theory and coordination theory (Weiseth et al, 2006). Such classification schemes are an important focus for CSCW research, and benefit from an understanding of collaboration. Many models have been proposed seeking to structure the factors influencing collaboration (Girard and Robin, 2006; Hawryszkeiwycz, 2006; Harvey and Koubek, 2000; Warner et al, 2003 and Weiseth et al, 2006), Of the few existing collaboration models there is a tendency towards simplicity or to focus on only a small part of collaborative work. Other researchers have focused on developing models to aid evaluation of collaborative work, e.g. Neale et al (2004), who presented a model for evaluating activity awareness, in which they defined the underlying processes of collaboration as being communication, coordination, and work coupling. Some researchers have attempted to isolate the core actions and interactions that form part of collaborative work. Pinelle et al (2003) suggest that these ‘mechanics of collaboration’ are independent of organisational culture, individual differences, or the type of task being conducted, and include: communication with team members, gathering information, negotiating access to shared tools and transferring artefacts to others. However, this approach does not explicitly consider high level collaboration activities such as planning, decision making and exploration. A number of researchers have attempted to model different aspects of team collaboration (e.g. Rogers and Ellis, 1994, McNeese et al, 2000). There appear to be few empirical studies which have examined collaborative work and those that have been reported have tended to use a questionnaire and/or interview approach (exceptions are the work of Vande Moere et al, 2008; Détienne et al, 2006). At the root of problems associated with evaluating collaborative systems is that we do not have a universally agreed understanding of what it means to collaborate, and thus what enhances or detracts from effective collaborative work. Such understanding is needed to determine methods to measure and predict the outcomes of collaborative work and make recommendations on enhancement and design.

The CoSpaces collaboration model We have isolated the main factors forming and influencing collaborative work through a literature review across many domains, embedded work with industrial organisations, interviews, workshops and expert brainstorming sessions, experimental studies and our own user requirements elicitation work with CoSpaces partners. The factors forming our current working representation of collaborative work are summarised in Table 1. As with many models, representation is critical. Our first attempt used three main building blocks of collaboration: collaboration environment, collaboration process and collaboration support (see Figure 2, left). However it was felt this representation implicitly suggested a process flow and an arbitrary hierarchy. We could not be confident of the implied ‘weightings’, nor of any causative influences or other factor relationships implied by the structure. The second representation shows factors in embedded elliptical layers (Figure 2, centre), illustrating the close relationship between the factors of collaboration: actors, groups, process, tasks, support and context. In contrast to the first


Table 1.

43

Factors (main and sub-factors) of collaboration and collaborative work.

Individuals

Teams

Interaction Processes

Skills

Roles

Communication Type

Wellbeing Composition Psychological Common ground factors Shared awareness and knowledge Relationships Group processes

Coordination Decision making Learning Error management

Tasks

Support

Context

Knowledge Culture management Structure Teambuilding Environment Demands Training Business climate Resources Organisational structure Networks

Tools

Figure 2. Different representations of the CoSpaces Collaboration Model (note: figure to show representations, not detailed content).

version, this format does not suggest specific weights, representing all factors as interdependent and part of an overall context. We include more sub-factors in this second version. The third version is presented as a web of factors (Figure 2, right), illustrating the mutually dependent relationship between the factors. Individuals and teams are central to the process of collaboration, engaged in intraand inter-group collaboration. They are involved in interaction processes which are required to collaborate on performing tasks, which in CoSpaces form part of their design and engineering work. Providing support is essential for ensuring that collaborative work is effective and efficient, and that individuals and groups have access to the resources required to perform their tasks, and meet their goals and needs. Context forms the final segment of the web, usually dictating the actors and groups, tasks, support that is needed/provided, and will impact on the actual process of collaboration itself. Sub-factors are associated with these main factors within the web, and factors which ‘overarch’ the main factors (relevant across individuals, teams, process, tasks, support and context) are shown outside the web. We are currently exploring relationships between selected factors in laboratory and field settings, for instance testing any relationship between trust, culture and

44


performance, or between familiarity, communications, shared awareness, common ground and performance.

Requirements, scenarios, use cases and functional specifications CoSpaces presents several challenges for a requirements elicitation process. First, the project is focused on three industrial sectors, with four user partners, who are themselves groupings of a number of smaller partners. This is in stark contrast to the more normal development process that considers either a bespoke system for a single organisation or user/user group, or a general application for a wider consumer market. Second, the project intends to develop a system that supports collaborative working in three contexts: co-located, distributed and mobile. Within each user partner different combinations or instances of these collaborative workspaces may or may not be required. Third, the project has not started from a tabla rasa but has grown from a previous ‘road-mapping’ European project, FutureWorkspaces. Some of the CoSpaces partners (user and developer) were involved in this project, and therefore bring a certain amount of fore-knowledge with them. Others are newcomers and consequently the level of familiarity regarding the vision of the project, the ways in which members work together, and with the technologies being developed, varies between project partners. Fourth – particularly for the consortium newcomers – the nature of the collaborative technologies involved, and the underpinning virtual and distributive technologies within them, mean that potential uses, and more importantly potential work organisation and work system structures that might result from their use, can be unclear. As a result of these special challenges the user requirements gathering: was based upon an identification and prioritisation process which is structured, clear and traceable; was as far as possible user-centred and beyond this user-partnered; resulted in prioritised, verifiable requirements, which can be translated into practical specifications. The seven stage process is: Stage 1 – Raising enthusiasm and creativity through the use of visionary scenarios, stories, and technology demonstrations. CoSpaces began by demonstrating the potential vision for collaborative workspaces to establish as clearly as possible a common starting point. Technical setups were presented allowing user partners to identify commonalities and potential uses within their organisational settings. With such advanced virtual and distributed technology, traditional user requirements identification may not work; users’ needs should be neither too “Star Wars” (unrealistic visions) nor too earth-bound in today’s technology. Ambient interfaces, context-aware functionality etc. may be poorly understood by end users, who are unlikely to be aware of tangible benefits, likely frequency of use, nor potential problems. Conceptual scenarios were potentially powerful tools for later requirements elicitation but, early on, demonstrations of related systems were useful for stimulating discussion, enabling users to more clearly define what they want or could possibly use and to make more informed decisions, leading to better communication and collaboration between the users and the designers/developers (Carroll, 1997).


45

Stage 2 – First involvement and visits to user companies, in order to foster relations with user partners, gain an initial understanding of the collaborative contexts to be further investigated, and establish a full list of stakeholders. Stage 3 – Development of current practice scenarios for collaborative engineering. Beynon et al (2005) have differentiated between: user stories, defined as real world experiences, ideas, anecdotes and knowledge, activity and context produced in any medium; conceptual scenarios, which report common elements across stories and are perhaps equivalent to the essential use cases of Constantine and Lockwood (1999); concrete scenarios, of which several may be drawn up for one project; and, use cases (see Stage 5). The value of scenarios, for CoSpaces is that they provided a means to an end to elicit user requirements in a form subsequently usable for system specifications, and system development. Scenarios allowed user requirements analysts to explore the context, user needs, and requirements in a vocabulary acceptable to stakeholders, whilst also communicating the priorities of those stakeholders to developers. We created rich-detail, tabulated descriptions of current practices, focussing on task and process level detail, organisational goals, stakeholder information, key decision stages, the types of communication that occur and the organisational, social and cultural issues surrounding collaborative behaviour, as a gateway developing user requirements within context and opening a dialogue between researchers and users for discussion of the potential applications. Stage 4 – Detailed development of current practice scenarios through semistructured interviews. Building on work in the previous stage, and conducted with real end users, human factors researchers were able to follow up in more detail the information gathered in the scenarios. Interviews enabled: development of all scenarios to comparable levels of detail; discussion of where technology might be used to improve collaborative practice; and the opportunity to observe current practice first hand, focussing particularly on the key activities, communications, collaborative relationships and decision making processes involved in the described task(s). Researchers were able to identify problems that could be solved by introducing new technology or through organisational changes. In accordance with the fundamental philosophy of socio-technical system design, both types of need emerged from the interview process. For example, a common theme was that, whilst current tools used were not especially advanced – requiring significant set-up costs and not dealing well with large data files, and without the functionality to share CAD models – these were not in themselves enough to explain poor collaboration performance. A combination of technology solutions and organisational approaches was required, e.g. policies and incentives to encourage the use of collaborative tools. Stage 5 – Generation and agreement of use cases. Researchers and users worked together to produce specific use cases for new technology within the scenario settings (e.g. Cockburn, 2001). Use cases describe the interaction with a system yet to be designed in terms that are easily understood, concentrating on what people try to achieve, rather than how they achieve it. In CoSpaces use cases were textual, in structured stages. Stage 6 – Detailed user requirements production and prioritisation. Requirements were elicited for each step through a use case, rooting them firmly within the

46


use case and scenario environment (Maiden and Mavin, 2006). The aim was to articulate as requirement statements the precise functionality required to support the behaviour and activity described in each case. Requirements were then compiled into one document, synthesised, combined and categorised; by researchers working with user partners, and prioritised according to importance (high/low) and timescale (short-term/long-term) by both development and user partners. Stage 7 (and beyond) – System specification production and negotiation. The final stage involved discussion primarily between research and development partners, in order to clarify the meaning and interpretation of requirements, and determine technical feasibility. This process allowed developers to better understand the underlying needs behind the requirements and suggest initial groupings and technical solutions.

Next steps and conclusions Naturally, a good user-centred design process does not begin and end with requirements elicitation. Particularly for collaborative systems, where the ‘task’ of collaboration is often vague, transitory and difficult to specify, we need to ensure close contact with users throughout the technical development into real systems. In sympathy with this, CoSpaces is following a Living Labs approach, whereby we aim to test and develop throughout the project, involving users as much as feasible, and carrying out iterative development at user sites where practical. The Living Labs notion is not without its critics, but our approach is twofold: development of a digital community infrastructure, especially integrating CWE development across small suppliers in the motor industry; and methodological, akin to the “digitised observation home”. We have termed the second an Active Distributed Development Space (ADDS), through it bringing research into reality rather than reality into research and using the notion of design partners within a tradition of participatory and user-centred design (see Wilson et al, 2008). We are currently exploring how the collaboration model could be used to measure (quantitatively or qualitatively) the collaboration which takes place in a company by applying weightings to each factor in the model to represent how people currently work and how they would like to work in the future. This work has the ultimate intention of helping industrial organisations improve their collaborative performance across tangible (e.g. improved product development efficiency) and implicit (e.g. greater employee satisfaction) dimensions.

Acknowledgements This paper is based on work carried out within the EU Integrated Project CoSpaces.

References Beynon, D. Turner, P. and Turner, S. 2005, Designing Interactive Systems (Pearson Education, Essex)


47

Carroll, J.M. 1997, Scenario-Based Design. In: M. Helander, T.K. Landauer, P. Prabhu (eds.) Handbook of Human-Computer Interaction (Elsevier, Amsterdam) Cockburn, A. 2001, Writing Effective Use Cases (Addison-Wesley, London) Constantine, L.L. and Lockwood, L.A.D. 1999, Software for Use (Addison-Wesley, Harlow) Détienne, F., Burkhardt, J.-M. & Barcellini, F. 2006, Methodological principles to analyse distant and asynchronous collaborative design, Symposium on Methodological principles for analysing and assessing collaborative design, International Ergonomics Association Conference, Maastricht, The Netherlands, 10 to 14 of July 2006 Girard, P., and Robin, V. 2006, Analysis of collaboration for project design management, Computers in Industry, 57, 817–826 Harvey, C.M., and Koubek, R.J. 2000, Cognitive, Social and Environmental Attributes of distributed Engineering Collaboration: A Review and Proposed Model of Collaboration, Human Factors and Ergonomics in Manufacturing, 10(4), 369–393 Hawryszkiewycz, I. 2006, A framework for raising collaboration levels on the internet. Lecture notes in Computer Science, 4082/2006 Maiden, N. and Mavin, A. 2006, WHERE: Scenario Modelling, In The IET Seminar on: Answering the Six Questions About Requirements (The IET, London) McNeese, M.D., Rentsch, J.R., and Perusich, K. 2000, Modeling, measuring and mediating teamwork: The use of fuzzy cognitive maps and team member schema. Proceedings IEEE International Conference on Systems, Man and Cybernetics, (Institute of Electrical and Electronic Engineers, New York), 1081–1086 Neale, D.C., Carroll, J.M., and Rosson, M.B. 2004, Evaluating computersupported cooperative work: models and frameworks. Proceedings of CSCW ’04, November 6–10, 2004, (ACM Press, USA), 112–121 Pinelle, D., Gutwin, C., and Greenberg, S. 2003, Task analysis for groupware usability evaluation: modelling shared-workspace tasks with the mechanics of collaboration, ACM Transactions on Computer-Human Interaction, 10(4), 281–311 Robin, V., Rose, B., and Girard, P. 2007, Modelling collaborative knowledge to support engineering design project manager, Computers in Industry, 58(2), 188–198 Rogers, Y. and Ellis, J. 1994, Distributed cognition: an alternative framework for analyzing and explaining collaborative working, Journal of Information Technology, 9(2), 119–128 Vande Moere A., Dong A. and Clayden J. 2008, Visualising the Social Dynamics of Team Collaboration. International Journal of CoCreation in Design and the Arts (CoDesign), Taylor & Francis, pp. 151–171. Warner, N.W., Letsky, M., and Cowen, M. 2003, Structural model of team collaboration, Office of Naval Research, Human Systems Department, Arlington, VA, unpublished manuscript, http://www.au.af.mil/au/awc/awcgate/navy/model_of_ team_collab.doc

48


Weiseth, P.E., Munkvold, B.E., Tvedte, B. and Larsen, S. 2006, The wheel of collaboration tools: A typology for analysis within a holistic framework, Proceedings of the 2006 Conference on Computer Supported Cooperative Work, Banff, Canada, (ACM Press, USA), 239–248 Wilson, J.R., Jackson, S. and Nichols, S. 2003, Cognitive work investigation and design in practice: the influence of social context and social work artefacts. In: E. Hollnagel (ed) Cognitive Task Design, (CRC Press, USA) 83–98 Wilson, J., Patel, H., Pettitt, M. and Saikayasit, R. 2008, ADDS: Living Labs for collaborative engineering, in J. Schumacher and V.P. Nitamo (eds.) European Living Labs: a new approach for human centric regional innovation, (Wissenchaftlicher Verlag, Berlin)

ACCESSIBILITY

A DESIGN ERGONOMICS APPROACH TO ACCESSIBILITY AND USER NEEDS IN TRANSPORT R. Marshall1 , D.E. Gyi3 , K. Case2 , J.M. Porter1 , R.E. Sims1 , S.J. Summerskill1 & P.M. Davis1 1

Department of Design and Technology, Department of Mechanical and Manufacturing Engineering, 3 Department of Human Sciences, Loughborough University, Loughborough, UK

2

This paper describes research carried out into the area of accessibility and ‘design for all’. The Accessibility and User Needs in Transport (AUNT-SUE) project was initiated to develop and test sustainable policies and practice that would deliver effective socially inclusive design and operation in transport and the public realm. Loughborough University’s role in the project focuses on the provision of data on users that is accessible, valid, and applicable and a means of utilising the data to assess the accessibility of designs during the early stages of development. These needs have led to the development of our inclusive design tool called HADRIAN. Data were collected on 100 people the majority of whom are older or have some form of impairment. These data include size, shape, capability, preferences and experiences with a range of daily activities and transport related tasks. These are partnered with a simple task analysis system. The system supports the construction of a task linked to a CAD model of a design to be evaluated. The task is then carried out by the virtual individuals in the database. Accessibility issues are reported by the system allowing excluded people to be investigated. Thus HADRIAN supports designers and ergonomists in attempting to ‘design for all’ by fostering empathy with the intended users, meeting their data needs through an accessible and applicable database and providing a means of gaining some of the feedback possible with a real user trial at a much earlier stage in the design process.

Introduction Research has identified that there is a clear and well established need for all those involved in the design of products, services and environments to take an inclusive approach, and avoid discriminating or disadvantaging users based on their size, shape, age, abilities, needs, or aspirations (Coleman et al., 2003). This need is driven by a number of factors: the ageing population (WHO, 2008), increasing legislation 51

52

R. Marshall et al.

(DDA, 2005), and a strong moral case that design should not be embodying a philosophy that prevents certain people from using, and enjoying the outcomes that it produces. The persuasiveness of these drivers together with the perceived support for inclusive design amongst the design community would suggest that evidence of inclusively designed outcomes would be common place. However this is often not the case with many designs still primarily catering for the younger and able-bodied. We believe that there is, therefore, a need for a new approach in order to effectively support designers when attempting to ‘design for all’ that must take into account a number of important issues: • Tools are often data driven. If the data is unavailable, inaccurate, misleading, unapplicable or just difficult to apply then often the tool is of limited value. • Even with accurate, applicable data, tools provide an additional layer of activity and often require some expertise. If a design team lacks the expertise in a particular discipline tools may be used to mitigate against this. Assuming tools are easy to use it can often be the case that the tool is used without a full understanding of the results or advice provided by the tool. • Tools are rarely good at addressing the softer, cognitive or emotional issues within design. Even if the tool manages this element of design well, it is not straightforward what needs to be done. This paper describes our current research as part of the AUNT-SUE project to provide better support for designers in their efforts to design for all. AUNT-SUE (Accessibility and User Needs in Transport for Sustainable Urban Environments, Porter et al., 2006) is a consortium of UK academic institutions including London Metropolitan University, University College London and Loughborough University, together with local councils and other public and private bodies such as Camden Council, Hertfordshire Council, and the RNIB. The consortium’s aim is to produce methodologies for sustainable policies and practices that will deliver effective socially inclusive design and operation of transport and is funded as part of the EPSRC’s SUE programme.

Our approach to inclusive design Our current approach to inclusive design was driven by an early survey of 50 designers which aimed to identify the current situation in designing when taking into account the needs of older and disabled people (Gyi et al., 2000). The results highlighted that available data tends to be ‘patchy’ and rarely in sufficient detail to enable design professionals to make more informed decisions. In addition, existing data tools are not in a format or language that designers can access and relate to easily. Finally it was noted that the majority of designers used at least one computer aided design package. These findings have also been identified amongst a broader range of concerns for businesses in a more recent study by Goodman et al. (2006). From this we identified that there is a clear need to provide ergonomics data in a highly visual form and to integrate design for all philosophy into existing good

A design ergonomics approach to accessibility and user needs in transport

53

practice, such as the use of CAD and other computer based design tools such as SAMMIE (System for Aiding Man Machine Interaction Evaluation, Porter et al., 2004). Human modelling systems such as SAMMIE allow ergonomics issues to be explored in a CAD environment during concept design. The benefits of being able to explore issues such as fit, posture, reach and vision at an early stage of the design are invaluable in achieving products that fully accommodate users. There are many issues when trying to apply ergonomics data to design and the same is true of ergonomics tools such as human modelling systems. Sources of anthropometric data often have limitations on how representative they are for any given design exercise. Concerns with the age of the data, access to data on appropriate Nationalities, the support for designing from 5th–95th %iles, the lack of task specificity, and so on. These concerns are compounded when the vast majority of the existing data do not relate to older or disabled people. In addition, the successful use of such tools is often coloured or constrained by the need for ‘expert’ users. For ergonomists to truly support widespread design practice, we need to develop and communicate information and methods that meet the needs of designers themselves. In response to these issues it was decided to develop an inclusive design tool. This tool is called HADRIAN (Human Anthropometric Data Requirements Investigation and Analysis, Marshall et al., 2004) and works together with SAMMIE to target the first two bullet points outlined in the introduction, namely: the provision of relevant, accessible and holistic information on people of a broad range of size, shape, and ability and a means of utilizing the available information to assess the inclusiveness of a proposed design. Together the tools support ‘playing’ with ideas at an early stage in the design process and provide feedback as to the impact of those ideas on users. In principle, the tools take the model of fitting trials, or user trials that are performed with real people and real products and move that into a virtual world where the costs and complexities of the real world are avoided and yet ‘some’ of the valuable feedback is available in a much more timely manner.

Design relevant data on people HADRIAN comprises a database of physical and behavioural data on 102 people covering a broad range of ages and abilities. The age of people in the database ranges from 18 to 89 years with 46 people who are over 60. Within the sample 59 people have some form of impairment including: limb loss, asthma, blood conditions, cerebral palsy, epilepsy, head injuries, non specific knee problems, multiple sclerosis, arthritis, vision and hearing impairments, heart problems, paraplegia, Parkinson’s disease, stroke, and dyslexia, amongst others. Of the 43 able bodied people 20 were aged over 60 and had undiagnosed or minor impairments associated with being older. In addition, the sample also contains people who reported difficulties with using some forms of transport impaired by issues such as pushchair use. The sample is clearly not representative of the more general population. To address the lack of data in this area and to directly support designers in inclusive

54

R. Marshall et al.

Figure 1.

Reach envelopes define the boundary of reach capability for an individual.

design it was a deliberate decision to skew the data towards those who are older and/or with some form of impairment. Whilst the sample is biased it has been carefully selected to cover as broad a range as possible for every measure recorded. The data consist of 26 Anthropometric body measures and 18 joint angle ranges of motion (see Table 1 for a summary of all the available data). For those people who use a wheelchair four wheelchair measures were also taken. A somatotype value is used to record body shape (Carter and Heath, 1990) in addition to two digital photographs taken from the front and side of the subject. Reach range is recorded in the form of an envelope of coordinate data centred upon the shoulder. These coordinates describe a surface mesh representing the boundary of reach capability for the subject’s arm (Figure 1). In addition to traditional anthropometric measures the data also contain a whole body scan. Using a [TC]2 whole body scanning system, subjects have been scanned to capture their body form. Body scan data is not available for all of the people in the database due to a number of limitations with the system. In particular, wheelchair users could not be scanned in their own wheelchair as the chair would interfere with the scanning process. In order to address issues with the applicability of existing data sources it was decided that the database should contain functional data in addition to anthropometry. Based upon a survey of user needs, conducted with 50 older and disabled people (Oliver et al., 2001) three areas were identified for investigation: a range of generic kitchen based tasks, a range of seating scenarios, and ingress and egress from a range of public transport types. Taking a pragmatic approach the data collection focused on tasks that were sufficiently specific to be relevant to design needs, yet generically applicable so that we were not designing a kitchen design tool, or creating a system that required data on every possible task situation in order to be useful. Using a kitchen rig consisting of wall and floor units together with a mock up of an oven, participants were asked to pick and place a range of one handed and two handed loads to various locations. A second rig was used to simulate entering and exiting from UK rail, coach and bus vehicles with a representative range of step heights and

1. Anthropometry (mm) Stature Weight Arm length Upper arm length Elbow-to-shoulder (link) Wrist-to-elbow (link) Abdominal depth (standing) Abdominal depth (sitting) Thigh depth (standing) Thigh depth (sitting) Knee-to-hip (link) Ankle-to-knee (link) Ankle height Foot length Sitting height Sitting shoulder height Hip-to-shoulder (link) Chest height Chest depth Head height Eye-to-top-of-head Buttock-knee length Knee height Shoulder breadth Hip breadth Hand length Hand grip length Wheelchair length Wheelchair height Wheelchair width Wheelchair seat height 6. Task capability (encoded postures for each task plus task videos) 4 pick & place tasks (high shelf, worksurface, oven, low shelf) with 3 load types (cup, bag, tray) each set to maximum comfortable weight, 1 or 2 hands as appropriate. Seating: 2 designs – high & hard, low & soft; restricted access to single side (bus), both sides (toilet cubicle), no restriction.

5. Whole body scan (VRML file)

4. Somatotype (3 digit number)

3. Reach range (∼100 coordinates mm) Functional reach volume generated by dominant arm/hand

9. Background Age Nationality Occupation / work history Handedness Disability Front and side photographs

8. Transport questionnaire (question and answer transcripts and videos) Transport use (frequency etc) Issues with transport usage (problems, assistance required etc) Issues with lifts, steps, escalators Issues with environment (personal safety etc) Issues with signage and timetables Local issues

7. Additional capability Bending to touch toes Getting up from lying down Reaching to tie shoelaces Twisting upper body to left and right Peg test (dexterity) Grip strength Vision

Ingress / egress: step up / step down from maximum comfortable step height, two handle types, maximum of 4 handle locations

Summary of data in HADRIAN database.

2. Joint constraints (deg) Shoulder extension/flexion Shoulder abduction/adduction Upper arm extension/flexion Upper arm abduction/adduction Upper arm medial/lateral rotation Elbow extension/flexion Elbow pronation/supination Wrist extension/flexion Wrist abduction/adduction

Table 1.

A design ergonomics approach to accessibility and user needs in transport 55

56

R. Marshall et al.

Figure 2.

Data in the database is structured around individual people.

handle locations. Where possible the data collected reflects real-world application. Thus, comfort maximums were recorded to reflect what the subject would be likely to attempt in reality where absolute maximums would not normally be used. All task activity was recorded from two positions via digital video camera and subsequently used to provide video clips of tasks being performed. Task data stored within the database includes a success or a failure for each task element. In addition, the data also records how a task was performed in terms of coded task behaviour. AUNT-SUE has also seen an expansion of the database beyond the physical into cognitive, emotional and sensory data associated with travel. Whilst these are often complex they are also the source of some of the most fundamental issues in transport accessibility. Using a questionnaire on transport activities, data were captured for each individual’s ability to deal with tasks such as using trains, buses, trams, London-style taxi cabs and minicab taxis; as well as issues surrounding travelling with luggage; the types and frequency of journeys made; problems using stairs, lifts or escalators; route planning, dealing with crowds, understanding signs and other public information, and perceptions of crime and personal safety. Thus the questionnaire provides information concerning issues that may arise at any point during the whole journey process. A key feature of the database is how the data are presented (Figure 2). To address concerns with designers being presented with faceless numbers, with the lack of visualization in current tools, and in the complexities of multivariate accommodation, the database is effectively a catalogue of individuals. Data are not broken down into categories of individual measures but are instead maintained as a set associated with a single person. Thus the user can browse through the people in the database see a picture of that individual and explore the data about that person. This approach fosters empathy between the designer and the people who they are


57

designing for, and attempts to minimize the dehumanizing effects of percentiles and of the virtual environment in which the design is being created.

A means of applying the data to design problems To address the utilization of the data to assess the inclusiveness of a proposed design, HADRIAN also includes a simple task analysis system. HADRIAN focuses on an integrated approach to support the designer in defining a task description associated with using their design, subsequent analysis of the task performed with the design, and final result reporting and analysis feedback. This approach then aids the designer in the evaluation of a specific design, establishing a form of semiautomated virtual fitting trial. This fitting trial can then be performed many times using the individuals in the database as virtual subjects. The analysis system was designed to facilitate a means of describing how a product would be used. ‘Product’is taken in its broadest context and may include any object that requires physical interaction. The system allows a complete task such as “purchase ticket from ticket machine” to be broken down into recognizable elements such as “select ticket type” or rather its interpretation as “reach to touch-screen”. The dynamic process of performing a task is broken down into static ‘frames’ associated with these elements. SAMMIE is then used to model the elements of these static frames including a posture for the human model, a target object, and an environment. The system then responds to each individual task element in a chain, performing physical interactions, and assessing the success of each element towards completing the task. The actual process of describing a task element involves selecting a task command such as Look, Reach, View, Move and so on. The command then needs a target that will be represented within the model to be analysed. This target will be the focus of the task element and form a verb + noun pairing such as Look at Screen, Reach to Card Slot. The task definition is then completed through the specification of task parameters. This may include an acceptable viewing distance, a particular grip type to adopt, or what limb a reach should be performed with. In an attempt to make the analysis as realistic as possible the system encodes the task behaviour recorded in the database. This behaviour is then replicated when an individual in the database is asked to perform a task of a similar nature. Thus, HADRIAN synthesizes how the task is performed by each individual in the database as the task progresses. Whilst a full treatment of the analysis model is beyond the scope of this paper, various factors are taken into consideration during this process including the dominant hand of the individual, the current orientation of the individual, the nearest limb to the target, any impairments to the non dominant limb, the details of the next task element, and so on. Irrespective of the underlying complexity of the process, the designer sees individuals in the database attempting to perform the task as defined. Each individual will have their own size, shape, capability and behaviour represented in the human model. The individuals will all perform the task slightly differently based on the

58

R. Marshall et al.

process outlined above. The outcome of the analysis is then reported to the designer in terms of a percentage excluded. The percentage is based upon those individuals in the database and so if 5 of the 100 people fail to achieve the task then 5% are reported excluded. The designer can then explore those who failed, identify exactly what individuals experienced problems, with what part of the task and why. The designer may now explore the affects of changes to the design or even the task. Because the process is occurring during an early stage of the design, possibilities can be rapidly explored and an optimum quickly identified.

Ongoing developments HADRIAN’s development has been targeted at addressing discrete design problems or products. However, the concept of inclusive transport is not solely related to any single piece of design, rather it concerns a network or system of designed elements. These designs could include everything from a flight of steps to a train carriage and yet they all form potential barriers to travel. This network is part of the transport infrastructure, combining a number of directly related, and indirectly related design problems that must be addressed holistically if inclusive transport is taken in the context of the ‘journey’. To succeed in providing truly accessible transport we must be able to ensure that a door-to-door journey for example, from home to the doctor, or from the bank to the theatre is possible at every stage. Two tools are being developed to address this need; the Inclusive Journey Planner and the Journey Stress Calculator. The inclusive Journey Planner addresses the needs of end users by providing improved information on accessibility along the whole length of a journey. The system then allows much more informed decision making on route and transport mode choice before the journey is undertaken. The Journey Stress Calculator supports the needs of transport professionals by allowing HADRIAN style analyses of a whole journey. This enables transport professional and practitioners to identify which people are most likely to avoid a journey, compare the accessibility of different routes and ensure spending is targeted at removing stressors that cause greatest exclusion (Davis et al., 2009). HADRIAN is also going through a process of validation (Summerskill et al., 2009). Validation will consist of three phases that will compare the findings from: HADRIAN used by a designer, SAMMIE used by an experienced ergonomist, and a real user trial conducted by an experienced ergonomist. All three assessments will use people from the HADRIAN database in real or virtual form to assess a number of existing transport related designs. The results will be used to refine HADRIAN.

Conclusion HADRIAN has been developed to support designers and ergonomists in developing inclusive products addressing limitations with current data and tools. The


59

HADRIAN database provides extensive data on 102 individuals including their size, shape, ability, behaviour, and range of other information on their background and specific transport related experiences. This data can then be employed through a simple task analysis system to assess the inclusiveness of a design. This predictive process greatly increases the chance that designers will identify issues that users may experience with a design long before the design is complete. HADRIAN is currently undergoing validation to assess its performance and to ensure that it is useable by the people it was designed to support. The database and analysis system will be made available to designers to support their efforts and help establish inclusive design as an integrated part of design practice.

References Carter, J.E. and Heath, B.H., 1990. Somatotyping Development and Applications, Cambridge: Cambridge University Press. Coleman, R., Lebbon, C., Clarkson, J. and Keates S., 2003. From Margins to Mainstream, In: J. Clarkson et al. eds, Inclusive Design, design for the whole population, London: Springer-Verlag, pp 1–25. Davis, P., Marshall, R., Case, K., Porter, J.M., Summerskill, S.J., Gyi, D.E. and Sims, R.E., 2009. Reducing the stress of using public transport. In: Contemporary Ergonomics 2009, proceedings of the Annual Conference of the Ergonomics Society. London, UK, April 2009. DDA., 2005. Disability Discrimination Act 2005 [online]. Office of Public Sector Information. Available from: http://www.opsi.gov.uk/acts/acts2005/ukpga_ 20050013_en_1 [Accessed 01/09/2008]. Goodman, J., Dong, H., Langdon, P.M. and Clarkson, P.J., 2006. Industry’s response to inclusive design: a survey of current awareness and perceptions. In: P.D. Bust ed. Contemporary Ergonomics 2006, proceedings of the Annual Conference of the Ergonomics Society. Cambridge, UK, April 2006, pp. 368–372. Gyi, D.E., Porter, J.M. and Case, K., 2000. Design practice and ‘designing for all’, Proceedings of the IEA 2000/HFES Congress, Human Factors and Ergonomics Society, San Diego, California, USA, August 2000. Marshall, R., Case, K., Porter, J.M., Sims, R.E. and Gyi, D.E., 2004. Using HADRIAN for Eliciting Virtual User Feedback in ’Design for All’, Journal of Engineering Manufacture; Proceedings of the Institution of Mechanical Engineers, Part B, 218(9), 1st September 2004, 1203–1210. Martin, J., Meltzer, H. and Elliot, D., 1994. OPCS surveys of disability in Great Britain: The prevalence of disability among adults, London: HMSO. Oliver, R.E., Gyi, D.E., Porter, J.M., Marshall, R. and Case, K., 2001. A Survey of the Design Needs of Older and Disabled People, In: M.A. Hanson ed. Contemporary Ergonomics 2001 proceedings of the Annual Conference of the Ergonomics Society. Cirencester: Taylor & Francis, 2001, pp 365–370. Porter, J.M., Case, K., Marshall, R., Gyi, D.E. and Sims, R.E., 2006. Developing the HADRIAN inclusive design tool to provide a personal journey planner,

60

R. Marshall et al.

In: P.D. Bust ed. Contemporary Ergonomics 2006, proceedings of the Annual Conference of the Ergonomics Society. Cambridge, UK, April 2006, pp. pp 465–469. Porter, J.M., Marshall, R., Freer, M. and Case, K., 2004. SAMMIE: a computer aided ergonomics design tool. In: N.J. Delleman, C.M. Haslegrave, and D.B. Chaffin eds. Working Postures and Movements – tools for evaluation and engineering. Boca Raton: CRC Press LLC, 454–462. Summerskill, S.J., Marshall, R., Gyi, D.E., Porter, J.M., Case, K., Sims, R.E. and Davis, P., 2009. Validation of the HADRIAN System with a Train Station Design Case Study. In: Contemporary Ergonomics 2009, proceedings of the Annual Conference of the Ergonomics Society. London, UK, April 2009. WHO, 2008. WHO Ageing [online]. World Health Organisation. Available from: http://www.who.int/topics/ageing/en/ [Accessed 01/09/2008].

DEVELOPMENT OF TOOLS FOR REDUCING THE STRESS OF USING PUBLIC TRANSPORT P.M. Davis1 , R. Marshall1 , K. Case2 , J.M. Porter1 , D.E. Gyi3 , S.J. Summerskill1 & R.E. Sims1 1

Department of Design & Technology, Wolfson School of Mechanical & Manufacturing Engineering 3 Department of Human Sciences, Loughborough University, UK 2

Loughborough University’s initial role in the Accessibility and User Needs in Transport (AUNT-SUE) project led to the development of a database that describes 100 people. This drives a 3D human modelling tool called HADRIAN that allows designers and ergonomists to assess discrete interaction points in a journey. The need for both the public and practitioners to assess the accessibility of whole journeys has led to the development of two further tools. The Inclusive Journey Planner is based on the premise that reducing stress in anticipation of a journey will increase inclusion. It will show how existing internet journey planners could be improved to provide richer and more personalised information so that users are prepared for potential barriers and avoid needlessly stressful journeys. The Journey Stress Calculator provide a novel way for transport professionals to investigate transport accessibility improvements. It models the stress that the 100 HADRIAN participants would experience as a result of multivariate ‘stressors’ including having to climb steps or cross a road, navigating a crowded environment, stand on a bus, or purchase tickets.

Introduction AUNT-SUE A multidisciplinary team at Loughborough University has been part of the EPRSC funded Accessibility and User Needs in Transport (AUNT-SUE) project since its inception in 2004. AUNT-SUE was initiated to develop and test sustainable policies and practice that would deliver effective, socially inclusive design and operation in transport and the public realm. Now in its second phase; AUNT-SUE is producing tools that may be used by transport planners, designers of urban infrastructure and ergonomists to ensure that future developments improve accessibility. Tools being developed by AUNT-SUE include AMELIA, which provides guidance on improving practical accessibility based evidence collected in a geographical information system. It enables planners to compare possible solutions that increase access to 61

62

P.M. Davis et al.

important locations in a region. The Street Design Index also uses geographically collected data, but uses measures that are associated with perception of personal safety within a street environment to enable planners to pinpoint problem areas through which pedestrians will be unwilling to travel.

HADRIAN Loughborough University’s role in the first phase of AUNT-SUE led to the development of a database that describes 100 real people in terms of their size, shape, capabilities, behaviours, and experiences with transport and transport related tasks. As an alternative to using national population data; this database is intended to represent variation in humans and as such includes 59 people with many different forms of disability. The first implementation of this database is a design tool called HADRIAN. Based on the well established SAMMIE human modelling system, HADRIAN will allows designers and ergonomists to carry out virtual fitting trials using models of people in the database. In addition, a user-friendly version of the database including additional information, photographs and video clips will be available to help practitioners understand the behaviour of their subjects.

Whole journey assessment HADRIAN is suited to assessment of interaction points in a journey such as ticket barriers, lifts and vehicle ingress. However, there is also a need to assess the accessibility of the whole journey (Porter et al., 2006). From the perspective of individuals wishing to use public transport public; whole journey assessment is needed in order to find a journey that is best suited to their capabilities and prepare for anything that might make their journey difficult. From the perspective of practitioners; whole journey assessment is needed to compare journeys or problems within a journey in order to prioritise improvements. Two further tools are being developed to meet these needs. The Inclusive Journey Planner will build on the functionality of existing internet journey planners to assist individual users. The Journey Stress Calculator will provide practitioners with a new way to understand and assess whole journey accessibility by virtually sending each of the 100 HADRIAN Participants on any given journey. The challenge with both of these tools is taking into account not only features of the journey that physically prohibit travel, but also those that make participation difficult or unattractive. This challenge is being approached by considering the fundamental role that psychological stress plays in excluding individuals.

Stress and social exclusion The difficulty with assessing whole journey accessibility is that access is not simply achieved by removing ‘hard’ barriers that practically obstruct specific groups of disabled people. The current focus on social inclusion emphasises the need to ensure that people who are disabled, elderly, young, female, minority ethnic or low

Development of tools for reducing the stress of using public transport

63

socioeconomic status are not excluded. It is not always straight forward to identify the barriers that exclude these people. Many people cite unreliable service as a reason not to use public transport. This suggests that people can also be excluded by uncertainty due to a lack of knowledge or control over aspects of a journey. Within the AUNT-SUE project; the Street Design Index assesses ‘soft’ barriers such as lack of natural surveillance or vandalism to identify urban areas which people may be deterred from travelling through. While it is clear that these soft barriers are affect people psychologically; it is also possible to consider the psychological impact of hard barriers. Doing so means that all barriers can be considered in the same terms. Psychological stress theory provides the language for this.

Stress is a unifying concept The accepted model for psychological stress is the transaction between stressor and response (Cassidy, 1999). While major life events such as bereavement may be considered as stressors, daily hassles have been found to be better predictors of psychological symptoms and health. Such stressors may be described as acute, ambient or anticipatory. A hard barrier such as steps up into a vehicle will be an acute stressor for a person who has difficulty climbing steps. Soft barriers such noise or vandalism are ambient stressors because they can make a person feel at risk without presenting an immediate problem. Uncertainties such as not knowing what time a bus will arrive or which way to walk are anticipatory stressors because the person predicts that they may become late or lost. Considering all barriers as stressors allows comparison between practical, perceived and predicted problems with the stress response being the unifying measure of severity.

Stress is unfair The widely advocated social model of disability defines disabled people as those who face unequal difficulties in order to accomplish the same objectives as able people. While there is no reason to suggest that disabled people have a lower capacity to cope with stress; by his definition, disabled people will be exposed to more stressors as they participate in society. Getting on bus is more stressful if it is physically difficult. Wayfinding is more stressful if you have difficulty reading signs. Variation in the way people experience stress is not confined to people with medical disabilities. People of lowest socioeconomic status show greater signs of distress (Mirowsky & Ross, 1989). This has been attributed to greater exposure to severe stressors and fewer resources to aid coping in everyday life. Response to stress also varies with age; young adults show greater anxiety attributed to the stress of taking on new responsibilities, whilst older adults show greater depression attributed stress associated with deteriorating health. Other social patterns of distress relate to having children and the nature of employment. Such variation in people’s exposure to stressors and capacity to cope with stress explains why certain groups of people are more likely to find using public transport stressful and therefore be socially excluded.

64

P.M. Davis et al.

Stress excludes people Although a hard barrier, such as inability to walk up stairs, would stop some people travelling, there are many people who choose not to make public transport journeys for less clear-cut reasons. Such non-participation can be explained as a stress transaction as follows: A person anticipates that they could fail to complete a journey. This is anticipatory stressor and causes emotional distress. The person chooses to respond by flight from the activity. The person behaves in a way that enables them to avoid the journey. As such, anticipatory stress, based on predicted problems, has excluded the person. This means that it is not only the most severe stressors that need to be assessed. All stressors encountered in a journey, even if they are minor or unlikely, can contribute to the anticipatory stress that excludes people from travel. Therefore any stressor that can be eliminated from the public transport system is an opportunity to increase inclusion.

Stressor elimination is universally beneficial Reducing stress is not only important in terms of reducing exclusion. The elimination of stressors would benefit the wellbeing of all those who travel. When a person encounters a stressor that they cannot take direct action against (by fight or flight) then they will remain stressed. In activities such as travel where a person can encounter multitude of stressors that are beyond their control this stress will accumulate. Repeated or prolonged stress leads to deterioration of a person’s physical and psychological health. This is recognised in transport with Costa et al. (1988, quoted in Cassidy 1999, p. 50) finding increased psychological illness in commuters. As an approach to improving public transport accessibility; stressor elimination is attractive because it takes into account the full range of problems that can contribute to exclusion, acknowledges variation in people and has the potential to make using public transport a healthy and uplifting experience for all.

The Inclusive Journey Planner The Inclusive Journey Planner is intended to play a significant role in reducing the anticipatory stress that causes people to avoid travelling. It will do this by enabling users to find a route that will cause them least stress and give them the information they need so that they can assess whether the whole journey is accessible and reduce uncertainties as they go on the journey. While the Inclusive Journey Planner must be suitable for all, it should particularly benefit people with any of the difficulties shown in Table 1. The Inclusive Journey Planner will build on the functionality of existing internet journey planners such as Transport Direct, Traveline and the Transport for London


Table 1.

65

Inclusion of users with all kinds of disadvantage.

Circumstantial

Physical

Cognitive and Sensory

Inexperienced With Young Children Awkward Luggage Limited Money Limited Time No Car etc.

Limited Movement Wheelchair User Low Dexterity Low Strength Low Stamina Slow Walking etc.

Language Problems Learning Difficulties Hearing Impaired Vision Impaired Dyslexic etc.

(TfL) Journey Planner. Preliminary research was carried out to investigate how current UK journey planners provide travel information and how they might be improved upon. Significant variation in interface quality and level of information provided was found with the TfL Journey Planner outstanding in terms of the provision of accessibility options and information. User trials and structured interviews were then carried out with 14 people, including five with various disabilities. The trials picked up on usability issues and causes of dissatisfaction with the TfL journey planner. The interviews then explored opportunities to improve functionality and established some additional options and information that people would find beneficial. This process resulted in more than 60 recommendations that shall be implemented in the design of the Inclusive Journey Planner. The majority of these are encompassed by four main concepts: Personal Profiles, Genuine Journey Choice, Rich Journey Plans, and Better Walking Routes.

Personal Profiles Many existing journey planners have ‘Advanced Options’ which, for example, enable the user to specify preferred modes of transport and set a maximum walking time. However, most participants in the trials, including those who would benefit most from using these options, ignored them with the result that three disabled people found journeys that they were physically incapable of doing. Many of the options that are provided were found to be unnecessary, whilst new options such as change time allowance and choosing to avoid crowded or dark locations were popular. Personal Profiles should include only a limited range of useful settings so that users can choose preferences quickly before they proceed to entering their journey search. These would then be saved to ensure that people are given journeys that are suit to their needs every time without needing to re-enter information.

Genuine Journey Choice Current journey planners present a limited range of journey options to the user. These may be repetitions of the same route or a small number of similar routes at different times. Given that the user has specified a time of travel this is not the best way to provide choice.

66

P.M. Davis et al.

Figure 1.

Figure 2.

Personal Profile concept design.

Genuine Journey Choice concept design.

A Genuine Journey Choice system would group repetitions of the same journey as into a single journey option. The system would then present alternative routes that suit the user’s profile and enable the user to sort and filter these according to various criteria, including cost and walking time, so that they can always find their best possible way to make the journey.

Rich Journey Plans The TfL Journey Planner provides additional details beyond travel times, locations and vehicles. For many people this information, including service updates and the presence of steps, lifts and escalators, can be very useful and further information such as vehicle access, warnings about crowding and even weather predictions would be welcome. The challenge is to design such Rich Journey Plans without overloading the screen or the user so that people get the information they need to travel with confidence.

Better Walking Routes The final concept for the Inclusive Journey Planner goes beyond transport links to help people navigate through the urban environment. Whilst many existing journey


Figure 3.

67

Rich Journey Plan concept design.

planners provide route maps for walking between stops and stations, these are not tailored to the needs of the user. Providing Better Walking Routes means giving people accessible routes and information such as locations of pedestrian crossings, toilets, cash machines, steps and lifts. This may not only be done with conventional maps, but also text descriptions and local photographs.

Development, testing and outcomes The Inclusive Journey Planner will be developed as a web-based prototype with limited functionality to enable testing. Initial trials will be carried out to identify any usability problems before the final design is reached and tested against existing journey planners. The prototype will then be made available as an exemplar for future journey planners along with the results of the trials and guidance documentation. In addition to this openness, AUNT-SUE is actively seeking to assist transport organisations in the design of journey planners that reduce stress and promote inclusion.

The Journey Stress Calculator Given suitable options and the right information, individuals can make good judgements about whether they are able to go on a journey. For transport professionals however, it is very difficult to take into account the variation in the preferences and abilities of individuals when assessing the accessibility of different journeys. The Journey Stress Calculator is intended to solve this problem. It will make use of information about the 100 HADRIAN participants in order to model the level of stress that each individual is likely to experience as a result of encountering acute, ambient and anticipatory stressors in a journey.

68

P.M. Davis et al.

Figure 4.

Journey Stress Calculator showing variation in stress at different journey stages.

The stress calculator concept The stress calculator contains a library of stressors that may be encountered in journeys. For each of these stressors every participant has been assigned a stress response based on their ability to cope with that stressor on a scale where 0 = unaffected, 1 is mildly effected, 2 = moderately effected and 4 = significantly effected. The practitioner then determines the occurrence of each stressor at each stage in the process. For each individual, the total journey stress is simply the sum of all their stress responses. total_ journey_stress = (stressor_occurencen × stress_responsen ) The Journey Stress Calculator currently includes the responses of participants to 58 standard journey stressors. By specifying the combination of stressors that are present in the walking, interchange and vehicle stages, any given public transport journey can be uniquely modelled.

Tool development Development of the Journey Stress Calculator will follow three main strands. The first is the simplification of the journey inputting process. Ideally it would be able to take output of a journey planner directly as an input; however, practitioners will need to use their knowledge of the journey to determine the likelihood of some of the stressors. Inputting this information correctly requires a good user interface. The second strand is to develop different outputs. The Journey Stress Calculator can be used to judge whole journey accessibility and compare alternative journeys, but it can also provide in-depth information about who is excluded, why people


69

are excluded and identify particular stages in the journey that would benefit from investment. This means that results need to be presented in ways that suit different objectives. Thirdly, testing is needed to ensure that the results provide a realistic approximation of the actual stress that people experience in a journey. The way that stress is calculated may need to be modified to adjust the impact of different types of stressor. Once development is complete the Journey Stress Calculator should prove to be a practical tool and promote the understanding of exclusion in terms of the different levels of psychological stress that people experience.

References Cassidy, T. 1999, Stress, Cognition and Health, 1. (Routledge, London) Costa, G., Pickup, L., Di-Martino, V. 1988, A further stress factor for working people: evidence from the European Community, 1. A review. (International Archives of Occupational and Environmental Health), 60(5), 371–6. Cox, T. 1981, Stress, (MacMillan, London) Mirowsky, R. and Ross, C.E. 1989, Social causes of psychological distress, (Aldine de Gruyter, New York) Porter, J.M., Case, K., Marshall, R., Gyi, D.E. and Sims, R. 2006, Developing the HADRIAN inclusive design tool to provide a personal journey planner, Contemporary Ergonomics 2006 (Taylor & Francis, Cambridge), 470–4.

VALIDATION OF THE HADRIAN SYSTEM WITH A TRAIN STATION DESIGN CASE STUDY S.J. Summerskill1 , R. Marshall1 , D.E. Gyi3 , J.M. Porter1 , K. Case2 , R.E. Sims1 & P.M. Davis1 1

Department of Design and Technology, Department of Mechanical and Manufacturing Engineering, 3 Department of Human Sciences, Loughborough University, Loughborough, UK

2

The HADRIAN (Human Anthropometric Data Requirements Investigation & Analysis) human modelling system is under development at Loughborough University as part of the EPSRC funded AUNTSUE (Accessibility and User Needs in Transport for Sustainable Urban Environments) project. The HADRIAN system aims to foster a ‘design for all’ ethos by allowing ergonomists and designers to see the effects of different kinds of disability on the physical capabilities of elderly and disabled people. This system is based upon the long established SAMMIE system (System for Aiding Man Machine Interaction Evaluation), and uses data collected from 102 elderly and disabled individuals (Joint range of motion and anthropometry, ability to use steps and stairs, lifts escalators etc.). The HADRIAN system allows three dimensional CAD data of new products to be imported, with subsequent analysis using all of 102 sample members. The 102 sample includes a stature range of 1st% UK female to 99th%ile UK male, and also includes a range of disabilities that have been assessed using scales from Martin et al. (1994). In this way the needs of people with specific conditions, such as arthritis, can be demonstrated and where possible, design accommodation can be improved. This paper describes the validation activity that is underway with the HADRIAN system. The validation reflects the transport focus of the AUNT SUE project by using HADRIAN to analyse the user interaction points that people encounter when using the Docklands Light Railway in London. This includes the use of ticket machines, the use of the train station infrastructure such as lifts and steps and stairs, and the use of ATMs to obtain cash. The validation is being performed by comparing the predicted results from HADRIAN and the abilities of users when performing real life tasks such as retrieving a ticket from a machine, or pressing a floor button when in a lift.

Introduction Human modelling systems (HMS) such as JACK, RAMSIS, SAMMIE (Porter et al. 2004), and SAFEWORK are used in the design of vehicles, manufacturing environments and workstations. These systems use Computer Aided Design 70

Validation of the HADRIAN system with a train station design case study

71

Figure 1. The use of SAMMIE in the design of an automobile interior (2007). software to represent the size and shape variability of humans in simulations of environments such as car interiors (see Figure 1). The ability to simulate how people of different sizes and nationalities are accommodated by a product removes the need for costly early physical prototypes. If used correctly within a design process that includes later prototypes that verify the HMS analysis results, HMS can be highly cost effective. Currently HMS systems support design activity with a focus on able bodied people. The aging population in the UK (WHO 2008) and a greater awareness of the needs of disabled people (Disability Discrimination Act (DDA 2005)), have raised the prospect of using HMS to simulate the effects of disability, supporting the design of more inclusive products. Using human modelling systems to represent the effects of disability does raise issues in terms of the expertise of the end user. The designers and engineers that use HMS in the product design process generally have little experience of the effects of disability or the coping strategies used by disabled people. Current HMS systems have no methods of representing the coping strategies used by disabled people, such as sliding a kettle along a work surface instead of carrying it. This paper describes the validation of a new HMS system that has been designed to combine anthropometric, joint range of motion and behavioural data for a sample of disabled people. The aim of the new system is to foster a greater awareness of the effects of disability amongst designers and engineers, whilst providing a tool that supports a design process resulting in greater accommodation of the needs of elderly and disabled people. The test bed for the new system (HADRIAN) is the long established SAMMIE system (Marshall et al. 2004).

The HADRIAN system A key feature of the HADRIAN system is that the process of evaluating a product is automated, removing the product designer or engineer from key stages of the HMS

72

S.J. Summerskill et al.

process that require knowledge of the behaviour of disabled people. The HADRIAN system allows a product analysis to be performed on the basis of a task description provided by the software user. For example, if a ticket machine is to be evaluated, the user would import a CAD model of the ticket machine into the HADRIAN system and build a list of tasks to be performed. Example tasks for a ticket machine include the use of a control to select on screen options, and depositing money into the coin slot. The HADRIAN system can perform these interactions for all of the sample members built into it, using data on the coping strategies of each individual to automatically perform tasks such as the positioning of the virtual user, to allow the best reach to the various controls built into the product. The system can then identify which users were unable to complete certain task stages based upon their ability to reach and view the interaction points of a product. This allows design changes to be identified that can increase the accommodation of the product, such as lowering a control or changing a screen angle. The behavioural data that is used in HADRIAN was captured from the video recording of kitchen based activities of daily living performed by each participant. The system also contains demographic information such as the type of disability, age, sex and occupation.

The HADRIAN sample of users HADRIAN is based upon data collected from a sample of 102 people, the majority of whom were registered as disabled, or had age related impaired mobility. The sample of elderly and disabled users participated in the following data collection activities; anthropometric data, joint range of motion data, reach range data, completion of a questionnaire detailing the use of different modes of public transport, and the collection of baseline data on the ability of the participants to perform kitchen based activities of daily living. Within the sample of 102 people 59 people have some form of impairment including: limb loss, asthma, blood conditions, cerebral palsy, epilepsy, head injuries, multiple sclerosis, arthritis, vision and hearing impairments, heart problems, paraplegia, Parkinson’s disease, stroke, and dyslexia, amongst others. Of the 43 able bodied people 20 were aged over 60 and had undiagnosed or minor impairments associated with being older. The remaining participants provide baseline information on the capabilities of non disabled people. All of the sample members included in HADRIAN were capable of living independently. Each subject was assessed using a modified version of the OPCS sample frame (Martin et al., 1994) to allow a comparison of the severity of disability exhibited by the HADRIAN sample to prevalence and severity of disability in the UK.

Anthropometric data used in the HADRIAN subject simulations The following anthropometric measures were collected from each participant. Stature, Arm length, Upper arm length, Elbow-shoulder, Abdominal depth, Thigh depth, Knee-hip length, Ankle-knee length, Ankle height, Foot length, Sitting


73

Figure 2. The postures adopted by participants when performing kitchen based tasks. height, Sitting shoulder height, hip-shoulder length, Chest height, Chest depth, Head height, Eye-top of head, Buttock-knee length, knee height, Shoulder breadth, Hip breadth, Hand length and Grip length.

Joint constraint data used in the HADRIAN subject simulations The following joint constraint data measures were collected from each participant; shoulder extension, shoulder flexion, shoulder abduction, shoulder adduction, arm extension, arm flexion, arm abduction, arm adduction, arm medial rotation, arm lateral rotation, elbow extension, elbow flexion, elbow pronation, elbow supanation, wrist extension, wrist flexion, wrist abduction, and wrist adduction.

Data collected on positioning and posture The prototype version of HADRIAN containes automation data based upon the kitchen tasks that were performed in the user trials. The participants were asked to move a variety of objects onto a high shelf, a work surface, and into cupboards and shelves of standard kitchen units. This process was video recorded to allow the postures that were adopted to be coded (see Figure 2). Table 1 shows the positioning and postural data that were captured for both ambulant and wheelchair using participants. These coded data were used to inform the behavioural aspects of the HADRIAN task automation. A more detailed description of the HADRIAN system can be found in Marshall et al. (2009).

The validation of the HADRIAN system The HADRIAN validation process aims to verify and potentially improve the data that drives the automation of the product assessment process. This is done by comparing the results of the analysis of the accessibility of three travel interaction

74


Table 1. The coding information used to classify the postures exhibited by the HADRIAN sample members during the kitchen tasks. Postures and orientations to be coded Orientation of the user to the kitchen cupboards Arm used for the tasks The posture of the legs during the kitchen tasks (ambulant participants only) Back twist Back bend Shoulder Head orientation

Coding criteria for use in HADRIAN Face on, side on, angled approach Left or right Straight, bent 1 (knee angle 170–120◦ ), bent 2 (knee angle 119–40◦ ), crouch (knee angle 39–0◦ ), left kneel, right kneel, full kneel, sitting Left or right >10◦ Upright (0–10◦ ), lean (11–45◦ ), bend (46◦ +) Relaxed, extended Yaw (neutral, left, right +/− >10◦ ) Pitch (neutral, forward back +/− >10◦ ) Tilt (Neutral, left, right, +/− >10◦ )

points, namely, a ticket machine, a lift at a railway station and an ATM machine. The selection of travel interaction points reflects the aim of the AUNT SUE project, which is to reduce social exclusion by supporting increased accessibility to the travel infrastructure by elderly and disabled people. Three different analysis techniques are used to assess the interaction points. Firstly an expert user of HMS, with extensive experience of the coping strategies used by disabled people, performed an analysis using HADRIAN data in SAMMIE. The SAMMIE system allowed the expert HMS user to simulate the anthropometry and joint range of motion only, with the experience of the expert user driving the behavioural aspects of the product analysis. The second analysis technique involves the use of the HADRIAN system to perform an automated analysis. Finally, a selection of the HADRIAN sample are asked to perform the same product interactions that were performed by their virtual counterparts with the real products. This has the potential to illustrate unexpected coping strategies or task failures that were not anticipated in the virtual user trials performed in SAMMIE and HADRIAN. Any new information gained regarding user behaviour can then be applied to refine the automation of HADRIAN for specific tasks.

Validation sample members Ten HADRIAN subjects will participate in the validation process. The subjects selected were two people over 65 with arthritis (male and female), an ambulant disabled female with cerebral palsy who uses a wheeled walking frame, a male mobility scooter user with mobility problems caused by a brain tumour, a powered wheelchair user with limited strength in the right arm due to a stroke, and five manual wheelchair users with a range of upper body mobility. These participants exhibit physical limitations in joint range of motion and reaching ability that affect their ability to interact with products, which can be compared between the three analysis techniques.


75

Figure 3. The analysis of an NCR ATM design with a paraplegic participant with good upper body mobility.

Validation progress The analysis of the lift, ticket machine and ATM has been performed by the expert user with the sample of ten HADRIAN participants. Figures 3, 4 and 5 show the analysis in progress for the ATM, ticket machine and lift respectively.

The analysis of two NCR ATM designs The cash machine manufacturer and project collaborator NCR has provided two cash machine designs for the validation process. This includes real ATM fascias for an analysis using the HADRIAN sample members, and CAD data to allow a virtual analysis using HADRIAN and SAMMIE as shown in Figure 3. The ATM illustrated in Figure 3 is a new design that is currently being installed world wide. An earlier model has also been provided by NCR. The analysis of the ATM involves different mounting heights for the ATM, the range of which is found internationally. Each HADRIAN participant will be asked to perform typical product interaction stages (insert a credit card, select a cash withdrawal by using a screen button, collect the cash, deposit a cheque in an envelope) which have been reproduced using the virtual analysis techniques.

The analysis of the ticket machine used by the Docklands Light Railway (DLR) in London The analysis of the ticket machines used by the DLR will be performed using a sample of participants based in the London area. These participants have been

76


Figure 4. The analysis of the ticket machine used in DLR stations in London with a participant that has limited mobility due to a stroke.

matched to members of the original sample in terms of their type and severity of disability. This allows an analysis of the general applicability of the design recommendations made using the original HADRIAN sample members, when compared to other people with similar conditions. For example, the participant shown in Figure 4 exhibits limited shoulder rotation and arm strength due to a combination of a broken back suffered in a car accident and a subsequent stroke. This meant that the joint constraints applied to the human model in SAMMIE and HADRIAN indicated difficulty in reaching to the controls and coin slot incorporated in the design. This necessitated a near sideways approach by this wheelchair user which allowed the virtual participant to reach all of the interaction points, but also caused a poor viewing angle to the screen. The completed virtual analysis determined that the user would find it difficult to read the on screen information. The comparison between the virtual results and those of the real participants will be the basis for the validation HADRIAN data for this product. The virtual analysis has been completed in both the HADRIAN and SAMMIE systems, providing a range of design recommendations that would improve the accommodation of the device for use by elderly and disabled people, whilst not disadvantaging non disabled users. The next stage of the validation process will be the matched sample user trials in London, which will follow a scenario that includes the use of the ticket machine, accessing the platform level using the lift or a ramp that is over 100m long, boarding a train and then returning to the original station. The DLR station is due for refurbishment. The DLR management have supported the validation process by providing CAD data for ticket machines and access to the Greenwich station for measurement and assessment activities. The design recommendations produced by the validation process will be considered by the DLR management for inclusion in the refurbishment of the Greenwich DLR station.


77

Figure 5. The analysis of a lift found at the Greenwich Docklands Light Railway (DLR) station in London with a participant with mobility problems caused by a brain tumour.

The analysis of a lift used in the DLR train station in Greenwich, London The analysis of the lift used in the DLR train station will follow the matched sample approach used in the analysis of the DLR ticket machine. The virtual analysis of the lift has indicated a number of design recommendations. The virtual evaluation shown in Figure 5 indicated that a mobility scooter user would be unable to share the lift with other users due to space constraints. Also, the limited mobility of the user indicated difficulty in looking backwards when reversing the mobility scooter out of the lift (the user had removed the mirrors usually attached to the mobility scooter). Other HADRIAN sample members indicated that the limited width of the lift would make it difficult for people with limited upper body mobility to access the buttons for the alarm and holding the doors open, as these were located in the highest location on the control panel. The matched sample that will be asked to use this lift at the Greenwich DLR station will be observed to determine if these issues arise.

Conclusions The HADRIAN validation process has been designed to verify and improve the automation of the HMS analysis of products that are associated with the travel

78


infrastructure. The virtual analysis that has been completed has highlighted access issues that could hinder the use of public transport systems by people with disabilities and older people. The verification of these results with real users will demonstrate the effectiveness of the HADRIAN system for the analysis of tasks which are different to those upon which the automation is based, i.e. kitchen based tasks. Any differences in the assumed behaviour of the virtual HADRIAN sample members that are found in reality, will serve to fine tune the automation of the HADRIAN system. The dissemination of the HADRIAN dataset will take three forms. The first form will be a database that presents the information gathered to researchers and designers. The second form will be data which can be applied to all current HMS systems in terms of anthropometry and joint range of motion, and the third form will be HADRIAN itself, which will combine all of the data available with automation of the product analysis. The use of HADRIAN has the potential to support designers and ergonomists in their efforts to improve access for disabled people who have the capability to live independently, but who are excluded from public transport use by an infrastructure design process that does not currently take account of their needs.

References Disability Discrimination Act 2005 [online]. Office of Public Sector Information. Available from: http://www.opsi.gov.uk/acts/acts2005/ukpga_20050013_en_1 Marshall, R., Case, K., Porter, J.M., Sims, R.E. and Gyi, D.E., 2004. Using HADRIAN for Eliciting Virtual User Feedback in ‘Design for All’, Journal of Engineering Manufacture; Proceedings of the Institution of Mechanical Engineers, Part B, 218(9), 1st September 2004, 1203–1210. Marshall, R, Summerskill, S.J., Gyi, D.E., Porter, J.M., Case, K., Sims, R.E. and Davis, P., 2009. A design ergonomics approach to accessibility and user needs in transport. In: Contemporary Ergonomics 2009, proceedings of the Annual Conference of the Ergonomics Society. London, UK, April 2009. Martin, J., Meltzer, H. and Elliot, D., 1994. OPCS surveys of disability in Great Britain: The prevalence of disability among adults, London: HMSO. Porter, J.M., Marshall, R., Freer, M. and Case, K., 2004. SAMMIE: a computer aided ergonomics design tool. In: N.J. Delleman, C.M. Haslegrave, and D.B. Chaffin eds. Working Postures and Movements – tools for evaluation and engineering. Boca Raton: CRC Press LLC, 454–462. WHO, 2008. WHO Ageing [online]. World Health Organisation. Available from: http://www.who.int/topics/ageing/en/ [Accessed 01/09/2008].

COMPLEX SYSTEMS

A SOFTWARE TOOL FOR EVALUATING THE EFFECTS OF CULTURE ON MILITARY OPERATIONS A. Hodgson & C.E. Siemieniuch Loughborough University This paper presents the military rationale for developing an improved understanding of the effects of culture on the performance of military organisations and their partners. In particular the change from conventional high intensity warfare to asymmetric warfare increases the interactions between different cultures within multi-national military forces and in the local populations. A brief description of culture and cultural attributes is presented. A culture-based evaluation tool (the ‘Soft Factors Modelling Tool’), which has been developed by the authors, is described.

Introduction The paper reports on work funded by the UK Defence Technology Centre Systems Integration and Integrated Systems for Defence: Autonomous and SemiAutonomous Systems. The focus of the project is on the configuration of complex military systems (which will comprise various combinations and configurations of technical and non-technical [ie human] components). There is an increasing recognition of the potentially deleterious effects of incompatible individual and organisational cultures on complex systems and organizations. This applies in both military and civilian environments. In the case of the military, these effects are exacerbated by the increasing involvement of multinational forces and nongovernmental organizations as the world moves into an era of ‘asymmetric warfare’, explained below. A basic premise of the project is that these systems need to exhibit a range of desired behaviours commensurate with the environment within which they are operating, the tasks they have been set and the degree of autonomy desired. It is anticipated that the research will enable a greater understanding of the impact that different configurations of cultural attributes can have in facilitating or impeding systems in making, communicating and implementing decisions, including the requirements for organisational change.

Military rationale for research into a cultural assessment tool Major changes are occurring to the post Cold War environments in which Western armed forces are operating. These changes are described briefly in this section. 81

82

A. Hodgson & C.E. Siemieniuch

The increasing frequency of asymmetric warfare In military terms, the USA is now virtually unchallengeable in conventional high intensity warfare capability. However, this hegemony, by limiting the military options of would-be foes, has forced them to select alternative, ‘asymmetric’ or ‘complex irregular warfare’ approaches (International Institute for Strategic Studies, 2005). Western strategists have realized that the West will not only face asymmetric warfare threats from terrorists based in ‘failed states’, but also from developing countries which have already tested and evaluated asymmetric warfare capabilities (Schneider 2007). It is generally accepted that it will be necessary to project force wherever the perceived enemy is detected.

A need for changes in Western military forces The Western soldier will in future operate increasingly in vulnerable situations, where he or she is always effectively on the front line, facing surprise attack, hostage-taking and arbitrary changes in local alliances amongst armed groups who use the civilian population to shield themselves. It is important to reduce the risks that this soldier faces, wherever possible; this risk reduction will come about via more appropriate training, cultural awareness, improved weaponry and physical protection and improved information. Based on more than fifty years experience of anti-terrorist activities at home and abroad, the UK Ministry of Defence is well aware of many of the issues associated with low-level asymmetric warfare, and while much greater reliance must, in the future, be placed on unmanned systems for surveillance, communications, targeting and as weapon platforms, there will always be the need for ‘feet on the ground’ in future conflicts.

The project It is recognized within the SEAS DTC that personal culture and organizational culture have a significant effect on the effectiveness of military units. Therefore, the DTC is funding research projects covering cultural issues – the work of one such project, Impact of Different Cultural Attribute Sets on Semi/Autonomous System Decision Structures and Interfaces, is the main focus of this paper. One of the first tasks of the above research project was to carry out a major review of the literature related to culture and its effects, in particular with regard to the military (Siemieniuch & Meese 2006). The outputs of this survey covered twenty-four topics, including: • • • •

What is culture? National culture Professional culture Organizational culture

A software tool for evaluating the effects of culture on military operations

• • • •

83

Military culture Cross-cultural operations Transforming culture Cultural dimensions

These outputs provided an essential guide to the subsequent activities of the research project, in developing a cultural evaluation tool.

Military scenarios of interest There are five key areas where culture has a particular role to play; these are described below: 1. Increasingly powerful soldiers: Advanced autonomous and semi-autonomous systems place greater power than ever before in the hands of individual soldiers, and it is important that they make appropriate decisions in using this power. Individual, group and organisational culture play a significant part in such soldiers’ effectiveness. 2. Effects-based operations and force packaging: UK and US forces are moving towards ‘effects-based operations’, where performance measures are based on the achievement of objectives, not the provision of fire power. ‘Force packaging’ (modular, self-contained sets of units borrowed from the various commands for a given mission) is an important contributor to this approach. Force packages will be assembled at short notice from different service cultures to form a force appropriate for an unforeseen crisis. 3. Multinational and non-governmental organisations: Multinational forces are increasingly being deployed, typically working alongside non-governmental organisations (NGOs) and operating in conflict, peace-making and peacekeeping roles. However, the organisations operate within different cultural contexts and exhibit a range of cultural attributes which affect cohesion and increase the likelihood of disagreements and misunderstandings. 4. Local populations: Military forces are spending ever longer periods in foreign countries. To achieve ‘success’, they must interact in culturally sensitive ways with the local populations of the host countries in order to go from crisis to communal action. 5. The development of semi/autonomous systems: Designers make their own culturally-biased assumptions about user requirements and expectations. Organisational and military research shows that technology tends to be adopted by users for their own purposes, and this is heavily influenced by the extant organisational culture.

Cultural perspectives Culture is a major and underestimated factor in the performance of any complex system. Cultural effects can be examined at the level of the individual, the team,

84

A. Hodgson & C.E. Siemieniuch • Unique • Part genetic • Part learned, e.g. individual behaviour, values

INDIVIDUAL

COLLECTIVE

UNIVERSAL

Figure 1.

• Shared • Learned, e.g. national, professional & organizational culture • Common • Genetic • Pre-programmed, e.g. laughing, weeping, aggression

Hofstede’s perspectives on culture.

and organisation; however, culture also affects the design of systems (whether or not they contain humans). Hofstede’s (1984) definition of culture is regarded by the research team as a useful ‘working basis’: Culture is “the collective programming of the mind which distinguishes the members of one human group from another . . . includes systems of values; and values are among the building blocks of culture.” Hofstede (1991) provides three broad perspectives on culture, as depicted in Figure 1. The perspectives that are amenable to change are at the individual and the collective culture levels. Hofstede uses the term ‘cultural values’ where value is seen as ‘a broad tendency to prefer certain states of affairs over others’. In this project the term ‘cultural attributes’ has been used to describe a set of high level culturerelated features, the values of which can be used to describe individuals and systems (including technical systems).

National culture National culture is usually a product of heritage (religion, history, language, climate, population density, availability of resources, politics, etc.). The following can vary according to national culture and can therefore shape expectations and performance: • • • • •

Leadership styles (hierarchical vs. consultative) Superior – inferior relationships (accept vs. question decisions) Communication styles (direct and specific vs. indirect and non-specific) Emotional reaction (showing reaction, emotion or aggression vs. hiding reaction) Following vs. breaking rules


85

Professional culture Professional culture is usually manifested in its members by a sense of community and by the bonds of a common identity (Helmreich & Merritt, 1998). Attributes of professional culture can include: • • • • • • • • • • •

Members have specific expertise and a shared professional jargon Norms for behavior and common, binding ethical values Selectivity and competition for entry Prestige and status with badges or defining uniform Extensive training requirements Professional and gender stereotyping Status differentials Self-regulation Institutional and individual resistance to imposed change Reluctance to admit error and denial of ‘vulnerability’ Reduced awareness of personal limitations

Organizational culture There are two layers of organizational culture: • Formal, surface, visible structures e.g. uniforms, routines, documents, etc. • Informal, inner, invisible layers, e.g. values, beliefs, subconscious assumptions. • Organisational culture is more amenable to influence than professional or national culture and it is organisational culture that effectively channels the effects of the other two cultures into standard working practices. Organisational culture is specific; what works in one organization may not work in another, and may not work at a different branch of the same organization. • Factors thought to influence or engender organizational culture include: – Reduced awareness of personal limitations – Strong corporate identity (e.g. nature of the product and market branding) – Effective leadership – belief in the organisation’s mission and products – High morale and trust – confidence the organisation’s practices, management, communication and feedback – Cohesive team working and cooperation – Job security – Development & training – Degree of empowerment

Cultural attributes In contrast to much work on culture, some cultural attributes described in this section can be exhibited by technical elements such as intelligent software agents. Since systems comprise both technical and non-technical elements, the technical components of semi/autonomous systems will also need to demonstrate appropriate decision-making behaviours and an ability to perform in particular environments.

86


Attributable to human agents Individualism

Collectivism

Universalism

Particularism

Hunter-gatherer Power by achievement Mastery

Nurturer Power by status Fatalism

Attributable to human & technical agents Proactive Information integration Time synchronization High power distance High risk-taking

Orthododox Information analsis Time sequencing Low power distance Low risk-taking

Figure 2. Cultural attribute pairings. For brevity, only one of these pairs is described. Further details are provided in Johnson et al,. (2007).

Technical sub-systems or agents are designed by humans, and it is inevitable that positions on the cultural attribute pairings will, in many cases be inherent in the design of these technical subsystems, in particular where there is intelligence in the sub-system.

Overview of the chosen cultural attributes In this study Culture is currently assessed based on ten different pairings of cultural attributes. of which five can only be exhibited by the human agents in a system; the other five attributes can be exhibited by both human (i.e. non-technical) and technical (i.e. software-based) agents within a system, as indicated in Figure 2. The project researchers believe that there is an identifiable relationship between the configuration of cultural attributes exhibited by a system (comprising both human and technical agents and components) and the performance of that system in a particular environment. Cultural attributes can relate to the perceptions of a single agent (e.g. an individual or missile launcher), a group of sub-systems (a troop or a communications infrastructure) or an overall system (the army or the set of assets carrying out a mission). Each attribute pairing defines a range, anchored by a description of the likely beliefs, perceptions, etc., at each end. Individuals, groups or systems will select a position towards one end or the other. It should be remembered that the attribute ranges, of themselves, do not imply right or wrong, merely that the attribute value is more or less appropriate given a required level of performance in a particular environment.


87

Individualism vs. collectivism: This refers to the balance struck between individuals and groups. In individualistic societies, ties between individuals (other than immediate family members) are loose; each person takes responsibility for his or her actions; individualists tend to speak directly and factually, and are willing to argue and to question others’ views. Generally, individualists tend to exhibit a higher level of trust than collectivists. In collectivistic societies, individuals are integrated into closely knit groups, often in the form of extended families; in return for unquestioning loyalty, they gain the protection of their group. People from outside the group tend to be treated with suspicion. Collectivists try to avoid direct, confrontational approaches and find criticizing others difficult. Hierarchies tend to be rigid, and losing face is to be avoided at all costs.

The Soft Factors Modelling Tool The latest version of the Soft Factors Modelling Tool (SFMT) enables individuals and representatives of subsystems (teams, squads, platoons, etc.)to input estimates of their positions with regard to each of the cultural attribute pairings. These positions are converted into numeric values and feedback is presented to the user(s) graphically to show conflicts between cultural attributes at the various system levels. The tool will highlight: • Any similarities or conflicts in cultural attribute values that exist between the various assets • Areas where particular assets may show a cultural attribute value that will facilitate or inhibit a desired behavior or ability to operate in a selected environment The target end-users of the final version of the tool are mission planners who are required to put together a package (comprising human and technical components) to carry out an operational requirement in a particular environment. For example the tool can help answer the question ‘Is this particular configuration of military assets with this mission capable of demonstrating appropriate decision-making, information processing, communication, adaptive skills and behaviour in an environment where the command style is control free, authority is delegated, operational tempo is unpredictable and the battle space is ill-defined?’

Impact of cultural attributes on behaviour As indicated earlier, it is believed that exhibiting certain combinations of attribute values will facilitate or inhibit certain types of decision-making behaviour. For example if there is a required behaviour for processing information which included the ability to deal with ambiguity and contradictions in the information or to deal with incomplete information, it is evident that an agent or system demonstrating a position towards the analysis end of the analysis v synthesis spectrum would find this difficult. Within the SFMT the cultural attributes have been evaluated

88


Table 1.

Part of behaviours checklist.

Skill class

Desired behavior

2nd Lt. Smith Troop

Regiment

Communication & interaction skills

Say what you mean/mean what you say Convey meaning indirectly diplomatically Dispel conflict Ability to trust and be trusted Willingness to collaborate/ co-operate Transparency/openness

POS

POS

UNCLR

NEG

UNCLR

POS

POS POS POS

POS POS UNCLR

POS UNCLR POS

POS

POS

UNCLR

Process information rapidly

POS

POS

UNCLR

Deal with ambiguity Deal with complexity Deal with contradictions Deal with uncertainty Deal with incomplete information Objective analysis of technical data Prioritize information Sharing information

NEG UNCLR UNCLR UNCLR UNCLR

UNCLR POS UNCLR POS UNCLR

POS UNCLR POS UNCLR POS

UNCLR

POS

POS

UNCLR POS

POS POS

POS NEG

Information processing

against a set of desired behaviours and a simple scoring system developed Table 1 illustrates a section of a spreadsheet that lists on the left a range of skill classes and, within these, behaviours; note that some of these are mutually exclusive, some are complementary. Cultural attribute pairs are listed across the spreadsheet and scored, although only one of these pairings, individualism vs. collectivism, is shown in Table 1 due to space limitations. The third and fifth columns of Table 1 indicate the capabilities of collectivism and individualism to achieve the various behaviours. Differing combinations of behaviours would be required for effective decision-making in various combat, peacemaking and peacekeeping situations. The analysis engine evaluates the values input for the cultural attribute pairs against each of the classes of behaviour of Table 1, using a simple traffic light display system. Note that, for this paper, the green, amber and red colours in Tables 1 have been replaced with the words POS (positive), UNCLR (unclear) and NEG (negative).

Impact of cultural attribute values on the ability to perform in particular environments Military environments vary considerably in terms of climate, terrain, command structures etc. The cultural context, within which military operations take place, also varies.


Table 2.

89

Extract from sample analysis results for behavior.

Environment class

Class characteristics

Command structure and style

Centralised Structure POS Decentralised Structure NEG Authoritative/interventionist Collaborative/control free POS Formal communication structure Informal communication structure Strong directive leadership Consensus based leadership POS

Function/authority/ Stove-piped function skills distribution distribution Dispersed function distribution Centralised authority over actions Delegated authority over actions Specialist skill set available Multi-skilling predominates

Collectivism

No obvious tendency

Individualism NEG POS

UNCLR

UNCLR

POS

UNCLR

POS

NEG

NEG

POS

POS

NEG

NEG

POS

As stated earlier, agents or systems exhibiting certain positions on the cultural attribute pairings will inhibit or facilitate agent or system performance in a given environmental context exhibiting certain characteristics. Another table, similar to Table 1 captures the environmental characteristics.

Presentation of SFMT results A final table provides the results of the analysis, after comparisons between the Tables above, providing an assessment of the goodness of fit between the proposed team and the environment – see Table 2 below. It has the same kind of structure, with a traffic light system to show distinct incompatibilities. The user may then decide whether there are too many reds and ambers that the plan must be reconsidered, or to go ahead anyway, or adopt some other alternative.

Further work Following completion of the current version of the SFMT, it will be evaluated by a wide range of military and other scenarios, prior to the development of a final version and recommendations for additional work. Many issues are still to be resolved, for example: • What are the optimum sets of cultural factors and behavioural/skill classes that achieve comprehensiveness without confusing overlaps? • Are the environmental factors too environment-specific to be transferable across environments?

90


• Who should input the soft factor values? • Could this tool be used to good effect in civilian scenarios?

References Helmreich, R.L. and Merritt, A.C. 1998. Culture at Work in Aviation and Medicine. Aldershot, UK, Ashgate. Hofstede, G. 1984. Culture’s Consequences: International Differences in WorkRelated Values. Beverley Hills, Sage. Hofstede, G. 1991. Cultures and organisations. London, McGraw-Hill. International Institute for Strategic Studies. 2005. Complex Irregular Warfare: The Face of Contemporary Conflict. In Military Balance, 105(1): 411–420. Johnson, P., Siemieniuch, C.E. and Woodhead, M.A. 2007. Description of 1st Prototype Version of Soft Factors Modelling Tool, Technical Report, SEICRP-0605, SEIC, Dept. of Electronic & Electrical Engineering, Loughborough University, UK. Schneider, W. 2007. Asymmetric Military Aspirations and Capabilities of the People’s Liberation Army of the People’s Republic of China. US-China Economic and Security Commission, Washington DC, March 2007. Can be downloaded from: http://www.uscc.gov/hearings/2007hearings/transcripts/mar_29_ 30/schneider.pdf Siemieniuch C.E. & Meese N.C. 2006. Technical Deliverable Report SEIC-RP520, SER006, SEIC, Dept. of Electronic & Electrical Engineering, Lougborough University, Leics, UK.

DECISION-MAKING AFTER THE PRODUCT-SERVICE SHIFT AND SOME IMPLICATIONS FOR ERGONOMICS E.M. Molloy, M.A. Sinclair & C.E. Siemieniuch Loughborough University, UK This paper introduces a Decision-Making System Framework developed in the UK EPSRC-funded Grand Challenge, entitled ‘Knowledge and Information Management – through life’ (KIM), for the evaluation of decision-making processes and systems. The development of the framework is outlined, along with a brief outline of the validation work. The use of the framework will be discussed, with reminders that the goal of this framework is not to provide the ‘right’ answer, which is improbable, if not impossible, to predict when dealing with such extended and complex systems.

Introduction A significant challenge, particularly for engineering organisations, is the ‘productservice shift’, whereby organisations move from delivering a product to the provision of through-life, sustained capability support (Oliva and Kallenberg 2003; MOD 2005; Anon 2008). This affects companies in their internal organisation and requires changes to the processes that they employ; particularly for knowledge and information management. Furthermore, the problems will get steadily worse as systems evolve from standalone status to components of systems-of-systems, with all the issues associated with this shift. While this happens, the making of complex decisions will remain a human and organisational issue. This is the context of this paper. The paper outlines the background of decision-making systems (DMS) and complexity before outlining the development and subsequent validation of a DMF (Decision Making system Framework) and supporting process. It builds on work initially presented at the ErgonomicsAnnual Conference 2008 (Molloy et al., 2008).

Organisational decision-making While there has been a considerable literature on Decision Support Systems, there appears to be a dearth on the characterisation and evaluation of DMSs that may use DSSs; Simon (1997) is one example. This is a pity; from an ergonomics perspective; 91

92

E.M. Molloy, M.A. Sinclair & C.E. Siemieniuch

as Rasmussen (1997; 2000) has said, “A closer look at the recent major accidents show that they are not caused by a stochastic coincidence of failures and errors, but by a systemic migration of organisational behaviour towards the boundaries of safe operation. Major accidents are the side-effect of decisions made by several decision-makers, in different organisations, and at different points in time, all doing their best to be effective locally.” One of the underlying causes of this is complexity: “A system exhibits complexity if it is composed of many integrated entities of heterogeneous parts, which act in a coordinated way and whose behaviour is typically nonlinear.” (Rycroft and Kash 1999). Some organisational characteristics which give rise to this are (Gregg 1996): • • • • • • •

many agents many kinds of agents operating in parallel with a degree of behavioural autonomy non-linear interactions multiple steady states for the system multiple steady states for human agents

As Gregg says, not many of these are needed for the system to exhibit emergent behaviour. A personalised view of this (Vaill 1998) is: “Furthermore, there was the problem of the intrusiveness of events: things did not occur one at a time; no competence could be practiced in pristine singularity. Instead, at any moment I was flowing with the multiple, disjointed time streams of the various projects in which I was involved.. . . Everything was interactive. . . . I simply had to learn to understand myself in a spatio-temporal field of relationships, flowing and shifting. It was a field of multiple players, each of whom had his or her own schedules, expectations of the rates at which things ought to proceed, and resistances to being sidetracked by other people’s temporal perceptions and priorities.” This is the context of DMS; seldom is it discussed in the literature, but it is the problem that we chose to research, because of its relevance to practitioners of systems ergonomics.

Decision-making systems and decision support systems It is well-known that decision-making is affected by external environmental and commercial pressures, and by internal effects and pressures. Many of these, such as established processes, structures, and incentivisation policies, are determined by the company’s overall strategy. It is important that strategy is developed and implemented in an arena of good governance, otherwise the endeavour to make good decisions will be vitiated and detrimental emergent behaviour will be a nearcertainty.

Decision-making after the product-service shift and some implications

93

Decision-Making Systems (DMS) Included within the term DMS as it is used within this paper are: • Agents – software or human based, who are involved with decisions • Activities – the decision-making activities which enable decisions to be made • Infrastructure and technology – to facilitate decision-making, including computer based support • Knowledge and Information. Its distribution across the organization (the ‘knowledge configuration’) determines the effectiveness of decisions made. DMS will of course be affected by time (for example, time available in which to make the decision, time by when the decision must be made, time when the effect of the decision is realized and expected duration of the decision) and the style of decision-making.

Decision Support Systems (DSS) In this research, DSS are considered to form part of the overall DMS. In much of the literature (Silver, 1991; Finlay, 1989) DSS are considered to include only computer based tools. However, this research widens that definition to include any form of support or guidance, which may be computer based or from a more human source, such as Communities of Practice (CoP) (Coakes and Clarke, 2006; Wenger, 2000).

Development of a decision-making systems framework The aim of the DMS framework and the associated process of use is to assist organisations in configuring their DMS to help them to cope with the risks and opportunities of long life, complex, engineered projects and systems. A top level view of the framework is shown in figure 1; lack of space precludes more detail. The framework has been developed from meta-analysis of accident and incident reports, pilot studies and industrial based case studies, all of which embrace complexity in some way. These are outlined below.

Literature studies The incidents meta-analysed were predominantly accident cases where a wealth of information and documentation is available (e.g. Enron, Piper Alpha). These studies investigated the decision processes and issues leading up to the various incidents, not the causes nor the blame. The analyses identified commonalities between the incidents; for example, people not available to do the work, poor communication, complacency, etc. A categorization of these issues and commonalities allowed initial identification of the attributes of DMS (e.g. agents, activities, infrastructure, knowledge and information) and the organisational aspects which could be affected by the DMS (e.g. internal and external variables, cultures and level of decision-making) and provided the development of the initial framework.

94


Pilot studies A number of pilot studies were carried out, to aid development of the framework and to develop methodology for the subsequent studies. Two student-based studies were investigated. Both were year-long design projects specifically mimicking organisational and technical aspects of full-scale engineering projects. These were followed by two field studies, one on the design and build of an award-winning innovative educational building, and an engineering study, evaluating a research and engineering centre. All of the studies looked at the identification of key decision points and the mechanisms of decision-making and decision support, leading to further development of the framework (and its associated system). Some key initial findings were: (1) not all decisions are consciously made, creating a rich environment for miscommunication and emergent behaviour; (2) corporate and top level strategy is not always flowed down or instantiated in day to day activities, which can impact on decision rationale; and (3) small scale organisational success with stakeholder involvement does not always scale up and can have unexpected outcomes, meaning the right processes may not work as expected if applied for misguided reasons.

Case studies This research involves two industrial case studies, in addition to the pilot studies outlined above. In both the studies, a qualitative approach to data gathering was used. Preliminary work involved analysis of existing documentation followed by a combination of email communication, interviews and workshops. The studies took place within a multinational aerospace organization experiencing the issues of the product-service transition. Both focused on the critical process of deciding whether or not to bid for a contract, exploring identified problems at different levels of activity decomposition. Data were gathered on activities and roles, information flows and decision points, along with process and problem perception and other contextual details. An outline of these studies is included in the table below.

The DMS framework A representation is given in Figure 1. The row headings are component parts of a DMS. The column headings are aspects of the overall system or system of systems, which could affect or be affected by the DMS. These are outlined below.

Components of the DMS The following are the constituent parts of a DMS: • Agents: DM (decision-making) agents may be human or software based, for example, who are involved in the DM process. • Activities: DM activities are those activities necessary for a decision to be made.


Table 1.

95

Outline of industrial case studies that helped to develop the framework.

Study

Problem statement

Overview of key findings

Industrial study 1

A stage of the bid process had become congested with mandated activities and reviews, leaving little time to complete all necessary work to progress the bid. Bids could be small, filling niches, or billion-dollar, ‘bet the company’ size. The study explored whether all the activities were necessary.

Industry study 2

There was concern about the quality and consistency of a particular pre-bid technical review. There seemed to be a lack of understanding of the training requirements for those involved in the review; one obvious knock-on effect being an incorrect assessment of the technical issues in bids. This could be very unfortunate in big bids.

• A change of focus has occurred with contracts written on service-based ideas with a more delicate balance of capability and profit. For the bid process, this represents new risk and its analysis is not as mature as it could be to deal with this. • The change has encouraged overlap in review activities between different functions • It is not enough to adapt existing processes to deal with these new contracts – new process must be developed and employed • Not all decisions are consciously made, especially during status review. • For the new service-based contracts there was uncertainty about what should be reviewed, with concern that some things are not reviewed at all, leading to unknown risk. • Purpose and importance of activities was not clear, which has issues for bid tailoring and consistency of output. • These things need to be addressed before appropriate training can be developed.

• Infrastructure and technology: That which provides support for the DM process. • Knowledge and information: DM knowledge and information is that which is necessary for a decision to be made. The DM knowledge and information flows through the infrastructure and technology, around the DM activities to the agents to allow decisions to be made.

Impacting variables on the DMS These include: • • • •

Internal variables: Contextual issues (e.g. stage of lifecycle) Environmental variables: External influences (e.g. legislation, partners) Organisational variables: Structure, empowerment, risk – taking, etc. Level of DM: Strategic, tactical or operational.

96


Figure 1. Top-level snapshot of the framework, illustrating a DMS audit. Numbered circles are issues. For example, issue 2 is the lack of personnel in charge of specific parts of the process. Linked circles show multiple categorizations of the same issue. Note that some combinations of void cells can imply that the audit was insufficient.

Using the DMS framework The aim of the DMS is to assist the audit of any DMS within an organization and to aid its dynamic reconfiguration for the extended lifecycles of the product-service shift. Figure 1 is a top-level view of the prototype DMS Framework. Issues identified in an audit may be oriented in this framework, according to the component of the DMS and impacting variable concerned. Mapping of issues allows ‘quickview’ comparisons of both clusters and of voids to highlight areas for investigation. This surfaces issues and allows stakeholders to commence discussions to enact improvements; it is essentially a tool to initiate and direct discussion. It does not present ready-made solutions. Work is in progress on the application and use of the framework and its databases and techniques. Detailed use scenarios are being developed. It is currently envisaged that will be possible to apply the tool both for prospective and retrospective work. On the negative side, there are some issues that the framework does not currently address. These include training and workload. However, as noted above, the framework has been used to investigate problems in these areas. Furthermore, the use of the DMS framework and associated processes relies heavily on honest identification of issues and subsequent categorization. Given that few people are able to understand more than a small part of their own organization, and given the power


97

relationships and politics to be found in any organization, these are not easy goals to achieve. Finally, an upgrading process will be made available for the framework, to allow it to be tailored for a given organization and its ever-changing context and to provide updates as new information becomes available.

Validation A final case study was used as validation for this work. The aim was to validate both the tool and the process for its use. The study took place in a manufacturing organisation that provided promotional products for retailers, a very different industry to that which formed the focus of the primary studies for this work. During the time of the study, the organisation was transitioning to lean manufacturing techniques after a commercial ‘Near Death Experience’ ((OSTEMS) 1998). The problems they were encountering included: • Management of a transient workforce • Loss of knowledge through loss of key personnel • Issues arising from changing customer requirements. The DMS framework was used alongside more established techniques (such as the Role Matrix Technique (Callan, Siemieniuch et al., 2005) and IDEF activity modelling ((IDEF) 1993) and has been shown to complement such techniques, providing extra contextual information from a decision making point of view and helping to identify some potential route causes for issues ascertained elsewhere. The tool was validated insofar as: • It was useful for the company • It was able to reveal a number of relevant issues that had not been considered The study also highlighted the expert level of knowledge required to use the prototype version of the tool. A more usable form will need to be developed before widespread use.

Commentary The framework does not provide ‘right’answers, but informs organisations about the characteristics of their DMS to enable them to be resilient in unexpected situations. With new contract and project types, there is a requirement for new processes, which may require more (different) people and resources and will require wider knowledge. These processes must be accepted and integrated into the organisational processes if they are to be effective. Considerations from agility show that a paradigm shift is required. This includes the hard bits of the organization to change – culture, trust, leadership etc.

98


There will also need to be a significant upgrading of the infrastructure that joins the community of organizations and people that deliver and use the capability. This must at least accompany, if not be in the lead, of the changes outlined above. With the growth of ‘through-life capability’contracts comes a shift in responsibility. Suppliers’ responsibility to provide a through-life service means the decisions they make are now long-term. This calls for better decision support, with better identification and assessment of risk (technical and business). These form part of the new processes and must be easy to use, if not implicit within regular working practices. If a decision is not consciously made, the outcome is neither known nor accounted for. This can lead to unanticipated behaviour (emergence), increasing risks and missed opportunities. With this interlinking of issues, a multi-disciplinary approach is a necessity in any such contract. The collaboration requirements mean that whole supply chains must work in partnership. The multi-disciplinary approach will cross organisational boundaries, calling for a better alignment between different organisational processes, as noted in the industrial studies. Without full information flows, trust and partnerships are essential. By focusing attention away from the facility at the point of handover to the services supported by that facility over the longer term, the current changeover to product-service projects provides new challenges for decision-making. With service typically being consumed or used as it is produced, it is probable that each instantiation will be different. It becomes fundamental for a service organization to ensure that it is resilient enough to cope with a very changeable environment. All of the above emphasises the key message of this paper: effective and appropriate, embedded decision support is essential for successful service organizations, and tools for this task must be flexible enough to withstand the manifold challenges of complexity. This framework tool is intended to help organisations to achieve effective and appropriate, embedded decision support.

Acknowledgements The work presented herein was undertaken in the Knowledge and Information Management (KIM) Through-Life Grand Challenge Project (www.kimproject.org) funded primarily by the Engineering and Physical Sciences Research Council (EPSRC Grant Number EP/C534220/1), the Economic and Social Research Council (ESRC Grant Number RES-331-27-0006.

References (IDEF) (1993). Integration Definition For Function Modeling (IDEF0). Gaithersburg, MD 20899-3460, National Institute of Standards & Technology. (OSTEMS), A. (1998). Understanding Best Practice for key elements of the New Product Introduction Process, DTI/ESRC Business Processes Resource Centre, University of Warwick, Coventry, UK.


99

Anon (2008). Implementing Systems Engineering in Defence. D. S. Engineering, UK Ministry of Defence. Callan, K., C. E. Siemieniuch, et al. (2005). A case study example of the role matrix technique. INCOSE 2005, Rochester NY. Coakes, E. and Clarke, S. (2006) Encyclopaedia of communities of practice in information and knowledge management, London, Idea Group. Finlay, P. (1989) Introducing decision support systems, Oxford, NCC Blackwell. Gregg, D. (1996). Emerging challenges in business and manufacturing decision support. The Science of Business Process Analysis, ESRC Business Process Resource Centre, University of Warwick, Coventry, UK, ESRC Business Process Resource Centre. MOD (2005). Defence Industrial Strategy. M. o. defence, Her Majesty’s Stationery Office: 143. Molloy, E.M., Siemieniuch, C.E. and Sinclair, M.A. (2008) “A new decisionmaking systems framework”, in Contemporary Ergonomics 2008, Bust, P. (ed), Taylor & Francis: 55–60. Oliva, R. and R. Kallenberg (2003). “Managing the transtion from products to services.” International Journal of Service Industry Management 14(2): 160–172. Rasmussen, J. (1997). “Risk management in a dynamic society: a modelling problem.” Safety Science 27(2/3): 183–213. Rasmussen, J. (2000). “Ergonomics.” Ergonomics 43(7): 869–879. Rycroft, R. W. and D. E. Kash (1999). The complexity challenge. London, Pinter. Silver, M.S. (1991) Systems that support decision makers: description and analysis, Chichester, John Wiley & Sons. Simon, H. A. (1997). “The future of information systems.” Annals of Operations Research, 71: 3–14. Vaill, P. B. (1998). The unspeakable texture of process wisdom. Organizational wisdom and executive ocurage. S. Srivastva and D. L. Cooperrider. San Francisco, Jossey-Bass: 25–39. Wenger, E. (2000), “Communities of practice and social learning systems”, Organisation articles, 7(2): 225–246.

GOVERNANCE, AGILITY AND WISDOM IN THE CAPABILITY PARADIGM M.A. Sinclair, M.J.D. Henshaw, R.A. Haslam, C.E. Siemieniuch, J.L. Evans & E.M. Molloy Loughborough University BAE SYSTEMS As a consequence of an investigation into Engineering Governance, an exploration of Agility within the realm of national defence, and the burgeoning understanding of Complexity within systems-ofsystems, the usual conclusion is supported: it’s the people that are the most important asset. The paper discusses some of the structural aspects for ensuring Through-Life Capability Management in changing environments, and some tools to enable managers and others to plan these structures

Introduction The paper presents an argument based on the assimilation of a number of different projects, all carried out in manufacturing industry, covering engineering governance (Nendick, Hassan et al., 2006; Nendick, Siemieniuch et al., 2006), supply chains (Siemieniuch and Sinclair 1999; Siemieniuch and Sinclair 2000; Siemieniuch and Sinclair 2002), knowledge management (Siemieniuch and Sinclair 1999; Siemieniuch and Sinclair 2004), and organisational preservation (Siemieniuch and Sinclair 2002; Siemieniuch and Sinclair 2003; Siemieniuch and Sinclair 2004), and decision support (Molloy, Siemieniuch et al., 2007; Molloy, Siemieniuch et al., 2008). Aspects of these have been assimilated into a discussion of the effects of complexity, the growth of systems-of-systems, and the role of Human Factors in their long-term survivability (Sinclair 2007). This paper extends this assimilation to include governance, agility, and the importance of wisdom in organisations.

The scenario The current shift in the manufacturing paradigm for the capital goods industry away from selling a product and towards a longer-term capability that delivers an effect (“we no longer sell you computers at a one-off cost; we sell you an fully-serviced IT infrastructure, for an annual fee, that frees you to concentrate on your business.”), necessitates a significant shift in the way that supplier companies organise their operations. The scenario is shown in Fig 1; a supply chain to an end-customer (e.g. providing jet engines to an airline, on a ‘power by the hour’ basis, so that the airline can rely on a given level of readiness in its fleet, and has no need to resource and manage this. For both the sides this results in a smooth, predictable, long-term cash flow). Azarenko, Roy et al. (2008) provides an interesting discussion of this approach. 100

Governance, agility and wisdom in the capability paradigm

101

Figure 1. Illustration of a supply chain. At right is a customer, utilising the assets provided by the suppliers. At left is the constellation of suppliers supporting these assets through life. OEM = Original Eqpt Manufacturer. Cap Gp = a capability grouping of assets to achieve an effect. For the companies in the supply chain, there is a requirement to extend the operational envelope; it is not enough to design and deliver the product to the customer’s specification; now, the supply chain must anticipate the usage of the product and the support it will require, in conjunction with other products, in a changing environment, for a long time ahead, and reorganise itself accordingly. Hence, as Ashby’s Law indicates (Ashby 1956), the supply chain must become more complex to handle this increased operational complexity. There is also an increased requirement of information transfer; unfortunately, for reasons of security and commercial confidentiality, while data transfer may increase significantly, information and knowledge transfer rates may not. Considered as a multi-agent environment, Fig 1 has the following characteristics (Gregg 1996): • An evolving, uncertain, environment (client, suppliers, etc.) • Many agents, of different kinds, with lots of connections • Communicating in parallel • Evolving agents, with many steady states • Interactions between agents across organisational boundaries • Interactions between different goals within an agent • Interactions between agents with different goals • Language/culture differences • Restricted time (deadlines, interruptions, etc.) As Gregg pointed out, if just a few of these are present, then the system’s behaviour is no longer predictable; emergent, probably unwanted, behaviour is possible. It is with governance that one may ameliorate this emergence.

Engineering Governance Governance is concerned with three questions: • Are we doing the right things? • Are we doing those things right? • How do we know this?

102

M.A. Sinclair et al.

A top-level view of engineering governance covers four perspectives: • • • •

Enterprise – risk management (‘ensure an appropriate market presence’) Performance – delivery of capability for business benefit (‘make a profit’) Asset quality – delivery to specification (‘get paid’) Legal/social – meet legal/social requirements (‘stay out of jail, and be respected’)

Humphrey (1989) said it well: “. . . a disciplined environment empowers the intellect, whereas the regimented environment marginalises it.”, and this should be the goal, while meeting the top level perspectives. It requires both the freedom for people to exercise their capabilitiers whilst fulfilling the needs of the organisation; discipline implies freedom within constraint, and the organisation, while creating an environment in which to prosper, should also address the issues of disciplinary transgression; there must be a justice system provided as well as a reward system to complement the organisation’s processes (Reason 2001). The processes themselves require documentation and ownership to ensure their long-term integrity; bearing in mind Humphrey’s encomium above, it becomes evident that the principles and precepts of socio-technical theory are applicable here (Cherns 1976; Davis 1982; Eason 1988). Tools exist to map processes (e.g. IDEF0, UML/, SysML, RADS); others to map both role relationships and cultural characteristics in relation to processes are in somewhat shorter supply, examples being the Role Matrix Technique (Callan, Siemieniuch et al., 2006), Social Network Analysis and the Cultural Values Modelling Tool (Siemieniuch and Sinclair 2006; Hodgson and Siemieniuch 2008), FOCUS’ and the Competing Values Framework. Because of the long-term nature of capability provision, governance must necessarily include knowledge lifecycle management and its propagation within a regime of constant personnel turnover (Siemieniuch and Sinclair 2007). A particular issue in governance is that when audits, reviews, and other governance assessments are carried out, with whatever metrics are deemed appropriate at the time, it is tempting to compare against benchmarks arising from past performance. Unfortunately, past performance is not a good guide to future demands; it is appropriate to discuss agility at this point before returning to governance, the reason being that the demands of agility may render past benchmarks irrelevant. One particular aspect of agility which demands good governance is the fact that if an organisations would gain in agility, there is a pre-requisite that it be wellgoverned. This is because any permanent gain in agility must stem from some degree of re-organisation of the enterprise.

Organisational agility It is in the speed of response and the adequacy of the response to altered circumstances (whether planned or unplanned) where competitive advantage may lie; in the commercial world, one might consider the travails of Microsoft in producing the Vista operating system compared to Apple which produced two operating systems and also changed platforms in the same time frame. Organisational agility arises from both robustness and resilience. Furthermore, there are different classes of agility, so some clarity of terms is required at this


103

point. Organisational robustness is the result of pre-planning; risk analysis followed by risk amelioration in the design, development and subsequent upgrading of the capability system that is instantiated. It ensures that we have available the necessary assets that comprise the capability together with the plans to redeploy these with minimum effort and adjustment to address the risks we have foreseen, thus being able to maintain organizational and system integrity. Organisational resilience includes the capability to adjust to unexpected events as well. As above, we assume that system integrity will be maintained, but it will be an adapted form of integrity. The implication is that our capability assets will be rearranged in real time into new configurations, and some will be used in novel ways, to deliver the goals. Resilience represents second- and third-loop control, with a consequent requirement for wisdom, knowledge and wide-ranging, timely information for its success. Organisational agility addresses both the extent of change and the rate of change to the new circumstances pertaining in the organisation’s context. Necessarily, this includes support for delivered capability as well. In this respect, a thoughtful paper from a defence perspective to be presented at an INCOSE conference discusses four levels of agility with regard to time as shown below (Mackley 2008): • • • •

Tactical Operational Strategic Lifecycle

respond within a day or so respond between a week and a year or two respond within several years respond over a longer period.

On a Tactical timescale one adapts with what one has, essentially by re-allocating resources and changing the patterns by which they interact. These changes may be robust, because they were foreseen and alternative plans for reconfiguration exist, or they may be ad hoc, because of an emergent surprise. As one goes down the scale, greater change is possible, but more slowly, and at greater cost (i.e. we can demonstrate limited agility very quickly, but more extensive agility beyond this will only happen with more time). This is because the assets themselves can now be changed, as can their interaction capabilities – we can adapt, alter, replace, extend, innovate, etc. This now involves the supply chain, reaching back a distance both in organisation, co-ordination and in time. Foresight is the key issue here; if the supply chain that delivers the assets has looked ahead sufficiently acutely, then the assets will have most if not all the resilience built in (i.e. we will have ensured robustness), allowing us to be Tactical, rather than Operational or Strategic in our response. Hindsight has to do with the reactivity of the supply chain given the emergence of a problem; again there is the opportunity of fast reaction, but the organisations involved must be geared towards this and must have long-term access to relevant information and understanding – from the suppliers’ perspective they need ‘reach-forward’ fairly extensively into the customers’ information environment. And, equally importantly, there must be a considerable level of trust among the suppliers involved in rearranging any assets. Clearly, governance must address these issues of change as well; agility is a never-ending exercise. It is also evident that wisdom has an important part in this,

104


particularly where resilience is required. As indicated above, resilience implies something of a step into the unknown; we must understand the situation which is developing in front of us, and must respond well and quickly to it.

The relevance of organisational wisdom The argument developed to this point is that the shift to a service paradigm for capability introduces greater complexity into the supply chain, requiring a more complex structure to address it. Furthermore, the long-term nature of the capability shift means both that the more complex structure will have to adapt to changing circumstances, and to required increments in agility, with continuous, perhaps far-reaching effects of knowledge loss due to personnel turnover. While good governance will ameliorate the likelihood and effects of emergent behaviour, and knowledge management techniques will overcome most of the knowledge loss issues, there is a role for wisdom in economising on the effort to govern and to manage knowledge; uniquely, wisdom also addresses effectively the ‘unknown unknowns’ to which a capability service is exposed. Wisdom thus provides a third strategy for organisational benefit, and we address this now. It is somewhat staggering to discover that organisational wisdom is a topic that has been virtually ignored for over a century by the soft sciences (Ardelt 2005). While there has been recent interest in the wisdom of individuals and social groups, there appears to be very few documents which discuss wisdom in organisations (Hammer. 2002 does this very well). Wisdom combines an ethical perpective with perception, knowledge, experience and communication. Because experience implies the passage of time, wisdom tends to be associated with age, but the link is not strong – and could be made weaker by the use of simulation and gaming techniques. Some characteristics of wisdom: • Wisdom is acknowledged; it cannot be claimed • Self-serving behaviour is never considered wise. Ethical behaviour usually is. • Wise people receive the same inputs as the rest of us; what distinguishes them is this combination: they see bigger patterns in these inputs, seek more rich corroboration, understand better the alternative decisions and their outcomes, and deliver solutions which are better at that time. • People may be ‘wise’ in regard to some aspects of the environment, and ordinary in relation to other aspects. Hence, ‘wisdom’ should include ‘in relation to X’. • Wisdom is not free of errors; in a changing world, wise performance at one time may be silly, irrelevant or wrong at another. • Wisdom is usually in short supply. Hence, it needs management. Industry’s typical perspective , as explicated by Hammer, is as follows: • It’s needed at the top. • Identify your talented people; create a career path to give them experience (and make them wise).


105

• Improve the rest; wisdom works best within a knowledgeable environment. • Create a mentoring function, with those at or near retirement as the mentors. • Retain your wise people. However, in relation to Fig 1, and coupled with the issues of governance and agility outlined above, arguably this is not enough. The orthodox approach is positive, but not focussed on actual needs, processes and the changing environment; we need wisdom at the top of the organisation – but where else as well? The capability paradigm implies that wisdom will need to be distributed horizontally as well as vertically, and especially where the capability is delivered and realised. In turn, this implies that career paths may have to be lateral instead of always-upwards; there may need to be ‘gates’ such as ‘You cannot be appointed to this position until you have performed X; you cannot perform X until you have experienced Y. . .’, with governance metrics, reward systems and a culture that appreciates the worth of wisdom. If this argument is accepted, then it follows that the Human Resource Management function in organisations has a much larger role to perform. We offer Fig. 2 as

Figure 2. A roadmap to illustrate progress towards a wise organisation.

106


a roadmap for further progress, for organisations to achieve ‘fitness for functions’ (Jenks 2004; Dixon, Brown et al. 2007).

Summary The paper has outlined some of the consequences of a shift to a capability perspective for the capital goods marketplace. Most of the issues stem from the problems of complexity, allied to the long-term nature of capability support. It is argued that governance is a key issue in ameliorating the effects of complexity, but that governance is not a simple issue, given the organisations must consider agility as well. Furthermore, wisdom is at least as important as governance is ensuring that organisations are fit for purpose in a capability environment. It is the people that matter in this, and their integration with the technology is a key issue.

References Ardelt, M. (2005). Foreword. A handbook of wisdom. R. J. Sternberg and J. Jordan. New Your, Cambridge University Press: xi–xvii. Ashby, W. R. (1956). An introduction to cybernetics. London, Chapman & Hall. Azarenko, A., R. Roy, et al. (2008). “Technical product-service systems: some implications of innovation within the Machine Tool industry” Journal of Manufacturing Technology Management in press. Callan, K., C. E. Siemieniuch, et al. (2006). “A case study example of the Role Matrix Technique.” International Journal of Project Management 24(6): 506–515. Cherns, A. B. (1976). “Principles of socio-technical design.” Human Relations 29: 783–792. Davis, L. E. (1982). Organisational Design. Handbook of Human Factors. G. Salvendy. New York, Wiley: 433–452. Dixon, K., S. Brown, et al. (2007). Integrating the intelligent enterprise. 17th Annual Symposium of the International Council on Systems Engineering, San Diego. Eason, K. D. (1988). Information technology and organisational change. London, Taylor & Francis. Gregg, D. (1996). Emerging challenges in business and manufacturing decision support. The Science of Business Process Analysis, ESRC Business Process Resource Centre, University of Warwick, Coventry, UK, ESRC Business Process Resource Centre. Hammer, M. (2002). The Getting and Keeping Of Wisdom – Inter-Generational Knowledge Transfer in a Changing Public Service. Ottawa, Research Directorate, Public Service Commission of Canada. Hodgson, A. and C. E. Siemieniuch (2008). A software tool for evaluating the effects of culture on military operations. AACC-08: Adaptive Agents in Cultural Contexts. Arlington, VA, American Association for Artificial Intelligence.


107

Humphrey, W. S. (1989). Managing the software process. Reading, MA, AddisonWesley. Jenks, C. L. (2004). “The well-being of social systems.” Systems Research and Behavioural Science 21: 209–217. Mackley, T. (2008). Concepts of agility in Network Enabled Capability. The NECTISE Conference. Leeds University, UK, BAE Systems. Mackley, T. (2008). The role of lifecycle systems in the through-life engineering of system solutions M. A. Sinclair. Cranfield University School of Engineering: Conference paper yet to be publshed. Molloy, E.-M., C. E. Siemieniuch, et al. (2007). The Effects of Complexity on the Product-Service Shift. International Conference on Complex Systems. Boston, MA. Molloy, E.-M., C. E. Siemieniuch, et al. (2008). Developing a DecisionMaking Systems Analysis Framework. KIM Project Conference. Reading University, UK. Nendick, J. V., M. Hassan, et al. (2006). ‘Good Engineering Governance’ – an issue for Ergonomists. IEA2006: International Ergonomics Association 16th Congress. Maastricht, International Ergonomics Association. Nendick, J. V., C. E. Siemieniuch, et al. (2006). Ergonomic aspects of ‘good engineering governance’ in the design process. Contemporary Ergonomics. P. Bust. London, Taylor & Francis: 62–66. Reason, J. (2001). Managing the risks of organisational accidents, Ashgate. Siemieniuch, C. E. and M. A. Sinclair (1999). “Organisational aspects of knowledge lifecycle management in manufacturing.” International Journal of Human-Computer Studies 51: 517–547. Siemieniuch, C. E. and M. A. Sinclair (2000). “Implications of the supply chain for role definitions in concurrent engineering.” International Journal of Human Factors and Ergonomics in Manufacturing 10(3): 251–272. Siemieniuch, C. E. and M. A. Sinclair (2002). “On complexity, process ownership and organisational learning in manufacturing organisations, from an ergonomics perspective.” Applied Ergonomics 33(5): 449–462. Siemieniuch, C. E. and M. A. Sinclair (2002). The role of the Process Owner in avoiding the ‘drift to disaster’ in processes. 2nd International Conference on Occupational Risk Prevention, Maspalomas, Gran Canaria. Siemieniuch, C. E. and M. A. Sinclair (2003). Changes in organisational roles when disaster strikes, and the lessons that can be learnt from this, in the manufacturing domain. Human factors of decision making in complex systems, Dunblane, Scotland, University of Abertay, Dundee. Siemieniuch, C. E. and M. A. Sinclair (2004). Long-Cycle, Distributed Situation Awareness and the Avoidance of Disasters. Human Performance, Situation Awareness and Automation: Current Research and Trends. D. A. Vincenzi, M. Mouloua and P. A. Hancock. Mahwah, NJ, Lawrence Erlbaum Associates. 1: 103–106. Siemieniuch, C. E. and M. A. Sinclair (2004). “Organisational Readiness for Knowledge Management.” International Journal of Operations & Production Management 24(1): 79–98.

108


Siemieniuch, C. E. and M. A. Sinclair (2006). Impact of cultural attributes on decision structures and interfaces. 11th ICCRTS Coalition Command and Control in the Networked Era, Cambridge, UK, US DOD. Siemieniuch, C. E. and M. A. Sinclair (2007). “Using corporate governance to enhance ‘long-term situation awareness’ and assist in the avoidance of organisation-induced disasters.” Applied Ergonomics 39(2): 229–240. Sinclair, M. A. (2007). “Ergonomics issues in future systems.” Ergonomics 50(12): 1957–1986.

DESIGN

THE ROLE OF ERGONOMICS IN TECHNOLOGY ORIENTED PROJECTS; THE INTELLIGENT ZERO EMISSIONS VEHICLE PROJECT A. Woodcock & J. Taylor The Design and Ergonomics Applied Research Group, Coventry School of Art and Design, Coventry University, UK Social, political, economic, ecological and technological factors will bring radical changes to future transport systems. The Intelligent Zero Emissions Vehicle (IZEV) project concerned the design and installation of intelligent, wireless computing, and hydrogen fuel cell technology in a taxi and narrow boat. Such telemetry will enable fleet owners, drivers and passengers to pull down real time information pertinent, concerning, for example, route and tourist information, fuel efficiency and location information. This will enable travelers to make informed, intelligent decisions about their travel behaviour. The creation of such demonstrators provides opportunities for stakeholders to understand how technological developments will influence their everyday lives and to provide input future scenarios. Their feedback can help designers better understand requirements, modify their designs and consider ways of presenting their ideas for greater public awareness and acceptance. This paper describes both the end user involvement which occurred in the four month IZEV project and reflects on the opportunities such projects offer for enquiries into ergonomic issues.

Introduction ‘It is widely recognised that the world has no more than twenty years to meet the urgent challenges of climate change and oil depletion.’ Harris (2007). Approximately a third of Britain’s CO2 emissions are from transport. These are arguably the most intractable forms of CO2 emission and certainly the ones rising fastest. The West Midlands faces the problems of many urban conglomerations in the UK and has recently been awarded £26 million to tackle transport related problems. An array of solutions are possible to alleviate traffic related such as congestion charges, toll roads, pedestrianisation, vehicle taxes, speed limits and bumps, and the design of energy efficient vehicles and fuels. The Intelligent Vehicle Demonstrator Programme, funded by Advantage West Midlands under the Science City initiative, sought to address transportation problems by using in-vehicle technology to enhance transportation networks within a City Centre environment. To this end a number of applications were designed and 111

112

A. Woodcock & J. Taylor

installed in a car and narrow boat to support remote tracking, monitoring and fuel usage, as well as ‘pushing’ interactive content to drivers and passengers through a dedicated touch screen. The rationale for delivering more information on vehicle performance, routes and driver behaviour is that the passenger, driver or fleet owner can use this information to make intelligent decisions about their journeys. The immediate questions this gave rise to related to whether the technology was needed, or offered services over and above those already available, how such systems would be used, and whether they had been designed appropriately.

The vehicles The narrow boat Canals are an under utilised resource. ‘One of the most energy efficient means of moving goods is by canal and the threats of global warming and oil depletion are resulting in a resurgence of interest in this means of transportation.’ Harris (2007). The zero-emission hydrogen hybrid narrow boat (Figure 1a) was developed by the University of Birmingham to demonstrate how magnet and fuel cell technologies can be used to power inland waterways craft more efficiently than carbon based alternatives. Green hydrogen is considered a clean fuel as it has a minimum impact on the environment and can reduce the levels of carbon dioxide and other greenhouse gas emissions.

The microcab The Microcab (Figure 1b) was developed in 2001. It has been the subject of investment, iterative design and review by Jostins and others (e.g. Atkinson, 2007). It has evolved into a series of silent, small urban vehicles operated by scooter-like handlebar and lever controls, with zero emissions, suitable for use as a taxi or light freight carrier. It is powered by hydrogen fuel cells from which the only emission is water vapour. Weighing 250 kg, it is an ultra light, urban vehicle with a maximum seed of 30 mph and a range of 100 miles.

Figure 1.

1a) The narrow boat and 1b) Microcab.

The role of ergonomics in technology oriented projects

113

These development vehicles formed the carriers for the intelligent systems, which were designed to track the vehicles use, monitor fuel use and provide information to vehicle occupants. It was expected that the incorporation of such technology will increase the efficiency, effectiveness and enjoyment of vehicle use. The following section provides details of the technology.

The intelligent systems These consisted of four parts 1) Remote, (online) live tracking of vehicles – overlaid onto Microsoft Virtual Earth maps. This enables vehicles to be tracked to street level and their movements displayed in real time (Figure 2), 2) Remote tracking of vehicle journeys, fuel, engine temperatures etc showing how the vehicle is being driven (Figure 3), 3) In-vehicle touch screen to support individual email accounts, web cams, hands free voice over IP Phone calls, GPS, MP4 and internet access, 4) Location based content ‘push’ to vehicle touch screens (to include parking information and what’s on guides) The industrial partners, Shoothill and RDM expect that such systems will be used, for example, by transport firms to monitor the current location and expected arrival times of vehicles, to monitor traffic flows, to plot the quickest and most costeffective routes, and show individual vehicle performance. Given that it is now possible to collect a vast amount of information from vehicle sensors and send this to different users, one of the main challenges involves understanding what information is needed for which user, at what time and how this can be presented. For example, currently, research technicians are interested in gathering fuel and performance information such as stack temperature, voltage, current etc; drivers may be interested in fuel consumption given the shortage of re-fuelling stations. To explore this, a portal was developed (see Figure 3) which

Figure 2.

Snap shot of vehicle location and journey details.

114

Figure 3.


Representation of system showing real time route and fuel usage.

displays real time information. The portal developed (Figures 2 and 3) provides a way of communicating the potential of new technology to ‘non-technical’ visitors in innovative ways, by showcasing car routes and a live camera feed from the car.

The role of ergonomics Similar to many funded projects, IZEV was technology led. Its main objective was to produce a series of demonstrators, exemplifying how technology could be used to enhance transport systems of the future. In such projects there may be little scope for ergonomic investigations – the time frame is short (in this case 4 months), user trials are dependent on technology being installed and tested prior to the end of the project, and many of the underlying issues, such as those relating to interface design and task interactions have been ‘addressed’ by the developers. Lessons learnt from this and previous projects (e.g. Woodcock and Scrivener, 1997) indicate that detailed user trials (as opposed to those which occur in prototype development) can only occur in a follow on project, with funding organized to ensure that this is not simply a requirement but is a fundamental part of exploitation and commercialization. As far as possible, studies were undertaken to understand the opportunities the new technology would provide and how it would be received. Additionally two technology trials were conducted with early versions of the systems in both the narrow boat and the Microcab to discover usability and acceptance issues. These are both related in the following section. However, because such systems are only ‘working prototypes’, which suffer because of interoperability problems and limited functionality, potential users are often disappointed with their experiences.


115

This is problematic. We need to involve end users in usability testing and discussing the wider implications of planned technology usage as early as possible. However, the systems they experience at such early stages may be far removed from what the software developers are planning or imagine they are showing. This may bias and frustrate users and reduce opportunities for the full discussion such systems require. The authors firmly believe that part of their responsibility in such projects is about increasing public awareness and providing opportunities for public engagement. In the IZEV project, as we were not able to use, test or demonstrate fully working prototypes, we developed persona and usage scenarios to understand how the technology might be used. Mindful of the need to raise greater awareness of the underlying rationale and technology, these were developed with ProMedia, along with more technical material, into ‘an educational’ DVD for councils, teachers, end users, wider stakeholders and future investors to understand the potential of the research.

User groups and usage of intelligent, zero emission vehicles Through a state of the art review, qualitative investigations with project partners, interviews with members of the public and different stakeholders, issues relating to the user groups and usage of the technology were discovered. It was predicted that there could be three main user groups for the small road vehicles (whether these acted as a light van or a Microcab) – the passenger, driver and fleet owner. Three main user groups were distinguished for the narrow boat – the freight or fleet company, owner occupier and holiday maker. Persona were developed for each of these groups. 25 interviews were conducted with taxi drivers in Coventry in order to understand the way they work, their experience of and attitudes towards technology. Current systems provide taxi drivers with satellite-navigation, radio, meter and radio. Surprisingly, not all taxis had a meter and only a few had satellite-navigation. All drivers were happy with their current systems, the information which was provided and the way it was presented. The radios were problematic at busy times when it was difficult to interact with the controller. The fleet owners similarly expressed satisfaction with current technology. Concerns were expressed over touch screens (and to a certain extent satellitenavigation) as these could lead to vandalism and theft. Drivers already felt unsafe at night when picking up fares from in and around the city centre, as passengers are often drunk and show antisocial behaviour. Only three drivers had had cc-tv installed in their cabs as part of a pilot system to address driver safety. They would not contemplate installing further systems until security issues had been resolved. Additionally drivers were worried about taking their eyes off the road when they used such systems To summarise, 14 of the drivers (just over half) were interested in the system, 6 would not consider it because of the security issues, and the remaining 5 felt they were not computer literate enough to use the described system and felt it would

116


complicate their job instead of making it easier. For the drivers that said they would consider one, they would like: 1. Whilst stationary – some form of entertainment (mostly internet), details of points of interest such as the location of the nearest petrol station. 2. Whilst going to pick up a passenger – Drivers would like to be able to use GPS to show directions to the fare, to use the system to communicate with the office to accept jobs and get addresses. 3. Whilst driving to their destination – Most drivers would like route information and to be able to calculate and charge the passenger for the journey. 4. Whilst returning to base – Use the system to look for more jobs near their current location. Approximately 50% of the drivers said that they did not record information about their fuel consumption, either because they were not interested in this, did not understand the concept or because it was not yet mandatory. 12 drivers recorded information to work out the cost of fuel in comparison to their earnings. Vehicle emission rates are inspected once a year by the Council. If a vehicle has too high a level it is taken off the road until its emissions are within regulations. This can lead to loss of earnings. Therefore, if drivers had a way of knowing this in advance, they would be interested in monitoring their emissions. In summary, from discussions and questionnaires with taxi drivers and passengers, found the following barriers to use: • For taxi cab owners – cost of the system, replication of facilities already on the market (e.g. way finding, cc-tv), over complication and security were major issues. Additionally, the drivers were not overly concerned with reducing their vehicle emissions as a way of controlling pollution. • For taxi passengers (who would be able to use services in the vehicles) – payment of services, replication personal facilities (eg phone and computing), the need for certain types of information on a short journey were barriers to use. It was questionable whether passengers on short journeys around town would make use of the services provided. • If the Microcab was going to be used as a light van – major concerns included the availability of fuel and safety issues related to the silent running of the vehicle (also see Hinchliffe 2008). However, fleets of such vehicles will be welcomed by councils and organizations wishing to reduce their carbon footprint. It may be concluded from this that the ‘taxi industry ‘needs to be persuaded of the benefits of such systems. One way of doing this will be to provide fully working demonstrators, however these may ‘skip over’ the issues relating to usability and cognitive overload which need to be addressed through laboratory based studies.

Barge owners A questionnaire was sent to barge owners (19 responses) and comments received from a web forum relating to the use of the proposed technology on barges. The


117

responses indicated a need for up to date information before and during the journey relating to locks, turning points, mooring stations, distances, boat yards, stoppages, winding and refuelling points and information which would help in journey planning. The route planner should provide the opportunity to plan the route on how far you wished to travel and provide a list of possible points of interest. It should also show the time, distance, fuel consumption and the number of obstructions (locks etc). Some barge users already use radios and wireless internet to gain updates on conditions and waterways that could affect their journey such as weather reports, engineering points, hazards and flood warnings. They might accept additional technology if it provided them with useful, usable information. Additionally respondents felt that ‘push technology’ could be used to provide information regarding local amenities such as shops, restaurants, waste disposal, wi-fi and communication points, public transport, roads adjacent to or near to the canal, emergency services such as medical support, veterinary services and police. Canal boats pass through, or are moored in high risk vandal areas, any form of route tracking or surveillance would therefore be reassuring to owners and travelers. In terms of on board performance, whether the barge ran on low emission fuel or not, respondents were interested in receiving information on how the boat was running such as engine performance, engine revs, water temperature, oil pressure, battery capacity, speed and currents. Although the responses indicated a level of support for the technology, some negativity was expressed. This fell in to three categories; • worries that it was merely replacing systems already in place, e.g. “What would be available that isn’t already available via Nicholson’s and Google maps?”,“My laptop is a GPS, depth sounder, VHF, weather tracker and AIS receiver”, • cost of such systems in relation to information that already freely (or cheaply) available and meets needs), “There’s already a lot of information available in printed and on-line form”, • lastly, as predicted, some barge owners were anti technology, “Personally I like to get away from computers when I am on the water.” Additionally respondents were split about having the information they required sent to their phones or on an onboard computer system. Some respondents would like information sent to their phones as these were portable and easy to access while on the move. Others would like to check progress and updates on an onboard computer system. The respondents who were negative, felt that the information was already available on internet sites and through books and did not want expensive equipment on their boats, commenting that if a book got wet it would not be ruined or expensive to replace but that a computer or mobile would be expensive and easier to damage or break while on a canal. If a system is going to be made available to barge users it needs to be carefully tailored to the usage context and specifically provide the functionality that is needed. This might be worth investment as two additional markets were found during the research. Firstly, the holiday maker – who would be interested in novelty, points of

118


interest and have the time to read tourist information about entertainment facilities in towns as well as details about the journey. Although this type of information will also be made available in taxis, it may not be used owing to the time of the journey. Secondly fleet owners (not just of holiday boats), but those who are interested in using canals to transport freight. Remote sensor systems could monitor fuel usage, security and progress.

Technology trials Although full technology trials were planned using working prototypes in both the narrow boat and the Microcab. The required systems were not sufficiently integrated with the vehicles to enable these to take place. Instead two preliminary technology trials were conducted on both vehicles using a subset of system functionality. 10 participants aged between 20–64 years evaluated the integrated driver’s systems in the taxi. A task analysis was undertaken to ensure that a representative range of tasks were selected. Anthropometrics was also considered as the system would be placed in the drivers work space envelope. Participants were video recorded using the system in a stationary Microcab, they were encouraged to verbalise as they completed the tasks and completed a post test questionnaire relating to ease of use of the technology. The following findings emerged: 1. In relation to anthropometrics, users had to stretch too far to use the system which was attached to the windscreen. When these results are compared with anthropometric data for the average forward grip reach of British males and females, the screen was found to be positioned too far away to be used comfortably from the normal driving position by even a 50 percentile population. All the participants either had to stretch to use the system or sit forward in their seats. Operation was obstructed by the switch to turn on the window wipers which was knocked several times when reaching past the steering wheel to use the screen, resulting in turning on the wipers. 2. Although the interface did not consist of novel features, and the ‘drivers’ had no other tasks to attend to, they found the interface confusing when selecting postcodes, areas and web pages 3. 50% of participants commented on the small size of the screen and the difficulties this caused for navigation and key pad use. Participants suggested a bigger screen to help navigation and to making internet access worthwhile. 4. Information loading speed was unacceptably slow 5. Most participants were concerned about operating the system whilst driving as they would have to reach over the steering wheel and take their eyes off the road to complete the tasks. They would have provided voice activation and instruction 6. The advanced prediction features of the GPS system caused problems in relation to automatic letter and number predictions for the postcodes and data acceptance. All but one participant saw value in the integrated system and would use one if it was available – although this would depend on cost and the length of time spent on the


119

road. The other participant commented “I do not think it was anything particularly new”. Additionally one participant felt that such a system would invite theft.

The technology in the narrow boat As the system installed in the boat had not, at the time of the trials been customized to a water based transport system, only the researchers were invited to provide their comments during a system walkthrough. Although they were enthusiastic about the potential of such a system, most of their comments related to the lack of functionality that specifically met their needs and the fact that the system was not programmed for travel on waterways.

Future potential The Intelligent Zero Emissions Vehicle project linked together hydrogen fuelled vehicles with intelligent systems in close to market exemplars. Such ‘Studies on the performance of the boat {and the car} will establish the viability of hydrogen for energy storage and as a fuel. We wanted to improve the science and engineering in this field by creating a real working example of this type of transport application and to enhance the public’s understanding and acceptance of hydrogen.’ Harris (2007). Taken together the technology provided represents what is feasible in the near future, with suitable levels of investment. From an ergonomics perspective the questions which still need to be resolved include whether the users of transport systems want or will need such systems in the future. Some respondents were ‘anti-technology’; others felt that they are not ‘clever enough’to understand the systems, were content with their current provision and did not see the benefits of integration. Although cost was mentioned, technology is more affordable, and is not so much a barrier. Importantly people are happy with current systems and view integration as adding complexity. However it is likely that if they have mastered and adopted current technology they may move seamlessly to the next generation of devices. Ubiquitous systems are threatening. The ‘big brother’ complex was not raised as much as anticipated – in fact those worried about crime wanted more surveillance, not less. However, this might have been overlooked because very few of the respondents were exposed to the full working system – they did not appreciate that every detail of one’s journey and transport behaviour could be monitored, or what might be done with this information. For respondents, the more complex, or powerful and visible a system, the more likely it was to attract criminal interest. Security measures therefore need to be highlighted in the design to deter would-be thieves and vandals and reassure purchasers. Lastly, even the limited technology trials revealed standard usability problems such as design of feedback and controls, positioning of equipment. In the rush to expand technological horizons, basic usability issues can be neglected. This means that the results and systems developed from such projects must be subjected to a full battery of user testing prior to launch.

120


Summary The Intelligent Zero Emissions Vehicle project has demonstrated hydrogen and wireless technologies in vehicles based at the University of Birmingham for intelligent vehicle management and emissions free operation. Wireless technology may be used to offer an enhanced user experience (furnished through a touch screen monitor) and remote tracking online which supports real time monitoring. The development of such vehicles will enable (for example): • Fleet managers to track vehicles remotely (view location, routes, petrol usage etc.), • Allow parents to ‘live’ track online when the school bus will arrive outside the house or to view the location of the nearest taxi through any internet connection, • Provide wireless cc-TV for taxi drivers, • Deliver free vehicle based internet phone calls within a city centre environment, • ‘Push’ interactive content to vehicles (either to support drivers with traffic and parking information or enhance a passenger’s journey through tourist and what’s on information), • Support local authorities in tracking traffic flows. The inclusion of ergonomists in such projects is most welcome. Ergonomists should be providing advice on the requirements, design and evaluation of future systems, especially ones which will increase the amount of information and tasks people have to manage. However, there has to be real engagement with ergonomists, representative end users and stakeholders at both earlier and later stages than were possible in this project. Time and resources should be allowed for this at the front end of projects. This needs to include full and meaningful dialogue with developers and service providers. Additionally the role of ergonomics needs to be expanded to provide opportunities for stakeholders to consider the wider societal costs and benefits, and usage contexts of such systems.

Acknowledgements The project was sponsored by Advantage West Midlands Science City and was a collaboration between Coventry University and the University of Birmingham, in conjunction with industrial partners, RDM, BT PLC, Shoothill Ltd and ProMedia.

References Atkinson, P. 2006, Can a small taxi be accessible? Notes on the development of the Microcab. In P. Bust (ed) Contemporary Ergonomics, 475–482, Taylor and Francis: London. Jostins, J. 2008, Micro:cab, accessed 29/08/08 at http://www.microcab.co.uk/ index.htm


121

Harris, R. 2007, Hydrogen Hybrid Canal Boat: Clean and Silent Propulsion for the Inland Waterways accessed on 29/08/08 at http://www.newscentre.bham.ac.uk/ press/2007/09/Hydrogen_Barge_Launch_21_9_07.shtml Hinchliffe, M. 2008, Electic cars ‘silent killers’ accessed on 29/08/08 at http://www.news.com.au/couriermail/story/0,23739,24224490-5010760,00.html Woodcock, A. and Scrivener, S.A.R.1997, User requirements and networked applications. In S.A. Robertson (ed) Contemporary Ergonomics, 516–521, Taylor and Francis: London.

RESEARCH AND POLICY DEVELOPMENT INTO SCOOTER ACCESS ONTO LONDON BUSES S. Thomas CCD Design & Ergonomics Ltd Access to the London Transport network is growing more important with the increasing use of motorised scooters by people with a range of mobility problems. TFL are concerned that the current policy on mobility carriage transportation on board buses is unclear and does not give enough clarity on which types, specifically scooters, that may access their bus fleet. Previous research cited in Department for Transport’s: Carriage of Mobility Scooters on Public Transport (2007) recommended that only scooters of a comparable size to conventional wheelchairs be allowed access onto public transport. CCD Design & Ergonomics ran and led the research programme and produced a final report detailing the findings.

Introduction Being committed to developing an accessible transport system for all, Transport for London (TfL) have ensured that their bus fleet is now completely accessible to wheelchair users through ramp access and specified wheelchair spaces on every bus in service in the Capital. With the increasing use of motorised scooters by people with a range of mobility problems it is now being argued, by some, that scooter access to the transport network is important to allow those with mobility impairments to be able to travel freely, and that this should form part of any accessible transport service, alongside wheelchair access. TfL are concerned that their Current Conditions of Carriage for buses do not give bus drivers or the general public enough clarity about the types of scooters which might be able to access their bus fleet. TfL therefore wished to undertake research to inform the development of a clear and consistent policy towards the access of mobility scooters onto the bus fleet for all the relevant stakeholders, including London bus companies, their drivers, users of mobility scooters, wheelchairs and the wider public. TfL envisaged that such research would establish which scooter models were suitable for access to the bus fleet, identify the key stakeholders and ensure that the conclusions could be communicated clearly to them. Previous research by the Department forTransport: Carriage of Mobility Scooters on Public Transport (2007) recommended that only scooters of a comparable size

122

Research and policy development into scooter access onto London buses

123

to conventional wheelchairs be allowed access onto public transport. It also identified further issues with the carriage of scooters that required clarification and which TfL wished to investigate by commissioning research specific to London Bus Services. CCD is a human factors and design consultancy specialising in the usable design of public transport vehicles, systems and infrastructure. CCD supported their research by drawing upon the facilities and capabilities of partners Ricability and MIRA, in order to develop a project team offering countless benefits. CCD ran and led the research programme and produced a final report detailing the findings. Ricability supported the scooter access user-trials design by CCD, specifically providing and organising access to their established panel of mobility impaired trials participants and securing suitable models of scooters.

Figure 1. Typical Scooter Models: Wispa, Sun-runner, Go, Go. Ricability has established contacts within scooter distributors to facilitate the borrowing of test scooters. Ricability’s panel of mobility impaired people includes scooter users and these formed a test group for user trials. To facilitate easy access for the user group to testing facilities, Intertek laboratories in Milton Keynes, with whom Ricability traditionally collaborate on research, were employed. CCD are very experienced in the design and build of bespoke mock-up and/or test rigs. Our approach to assessing the accessibility of various bus designs by scooters was to design and build an adjustable test rig, which could be easily and quickly re-configured to represent different bus layouts. The evaluations included: • Objective assessment of scooter access (and stability issues) to buses via ramp and manoeuvrability in mock-up bus vestibule spaces • Manoeuvrability of different scooter types into, around and out of current DfT compliant wheelchair spaces within the different bus configurations • Usability differences between scooter types (wheel configuration, seat shape/type and position, accessorised etc) and possible safety implications within current DfT compliant wheelchair spaces in buses

124

S. Thomas

• Scooter user opinions/experiences of access to bus types (i.e. subjective opinion of acceptability, risk and comfort) The project was also supported by MIRA, who provided computer-based simulation facilities to investigate the dynamic stability of suitable scooter models to quantify the risk of their tipping and/or moving during sharp turns or sudden braking, both in forward and rear facing positions and in comparison to wheelchairs that are already accepted for.

The brief The issues TfL wished to investigate included: • Good practice, including policies and procedures or codes of practice from other bus companies, and good practice in terms of communicating and disseminating revised conditions of carriage to the general public. • The application of the DfT wheelchair size recommendations to the London bus fleets and the ability of the current range of scooters to comply. • The manoeuvrability of different scooter models with respect to the designs of the different vehicles in the London bus fleets and the different locations of wheelchair spaces. • The health and safety risks associated with each model including the stability of different models (3 and 4 wheel models) with reference to the designs of the different vehicles and the locations of wheelchair spaces, and the risks associated with unsecured mobility vehicles when buses are in transit. • The issues faced by wheelchair and scooter users when travelling by bus and how new policy may address those issues, i.e. when is it appropriate to ask travellers to move from a scooter to a seat during transit? Does bus design make access difficult? • The views of London bus companies and their experiences so far in meeting/not meeting the needs of scooter users. • The implications of the findings to future design of compliant scooters. The research project produced the following outputs: • Understanding of the size and model restrictions in terms of the access of scooters onto the London buses. • Clear policy direction in terms of security and risk relating to different models of scooter when on board. • Guidance for manufacturers and retailers concerning the suitability of different scooters (design/size) for the bus service. • Understanding, via consultation, of the needs and views of users of scooters, buses and drivers/bus companies, which will inform the development of a clear and useable policy on the access of scooters. • Examples of good practice in communicating policy guidelines to the public and bus companies/drivers, including use of endorsements/kite-marks, etc.


125

Methodology Research & survey work CCD undertook a literature and internet based search for examples of good practice, policies, procedures and codes of practice from other bus companies, both in the UK and abroad. CCD also used local contacts in USA, Canada, Australia and Europe to ascertain and review the policies and codes of practice being implemented by bus companies and in particular what means they have used in disseminating their conditions of carriage. Additionally, CCD contacted usability/mobility groups, discussion forums and websites in order to get user feedback on their understanding of terms of carriage, an understanding of where the gaps in their knowledge are, issues they have with both the terms and how they are made available and views on which means of dissemination they find to be most effective. CCD also surveyed the London bus fleet, building on data and measurements developed for previous studies into push chair access, to determine the variation in the range of space and facility provision, particularly in terms of door access, vestibule size, wheelchair space (i.e. size, layout, position in bus and provision of rest) as well as other potential impediments to manoeuvre such as seating, stanchions, poles etc. The range of vehicles currently operated includes double deck, single deck and articulated buses all with low floor capability and ramped access. While there are a fairly large number of bus models in operation, these tend to be based on a limited number of chassis types with bodies from a number of coach builders.

Stakeholder consultation TfL currently contract some 18 bus companies to provide services in and around London. To get the broadest coverage and exchange of views CCD held a number of interactive workshops at their London offices with representatives from each company. The workshops were aimed at understanding the needs and views of the bus companies, their experiences so far in meeting or not meeting the needs of scooter users, their views on operational issues (such as access protocols, delay to service, issues with the possible use of scooter restraints, checking/monitoring proper procedure by the scooter/wheelchair user etc) and safety risks and their methods and associated experiences of disseminating codes of practice (both to their staff and to the public). Each workshop involved invited company stakeholders responsible for policy and operation. Some 40 bus drivers, from various bus operating companies, were also interviewed on site during their shift breaks. CCD ran a series of user workshops across London to undertake a structured review of the needs and views of scooter and wheelchair users in relation to bus access, in particular to understand their requirements for access, their experiences with accessing buses and with how bus companies operate their policies, their views

126

S. Thomas

on the availability, clarity and dissemination of terms of carriage and their views on issues such as the provision and use of wheelchair/scooter restraints and transfer to a seat during transit. CCD directly contacted a number of relevant user groups (such as Transport for All and Shopmobility) with the aim of gaining wider access to user views with a questionnaire survey. CCD also completed user interviews at the Mobility Road Show (12th to 14th June 2008). The show is the largest annual mobility and lifestyle event in the world and attracts thousands of people with mobility problems from all ages who take the chance to assess products and services available to help them. The show offered unrivalled access to the widest range of interested and motivated potential mobility impaired users of bus services.

Access and manoeuvrability testing CCD assessed the size and model restrictions on scooter access and the ability of various scooter models to comply with DfT recommendations by a series of user trials using various models of scooter on a mock-up bus test rig. CCD specified an adjustable bus rig capable of being re-configured to represent the different examples from the London fleet including a ramp, doorway, vestibule area and wheelchair bay as well as any poles and stanchions etc. that might affect scooter access. The aims of the trial were as follows: • to identify which models of scooter can be manoeuvred by real users onto a bus and into a wheelchair space • what aspects of bus and/or scooter design most impact on or restrict manoeuvre/ access (especially scooter turning circle and wheel base and relative positions and design of bus structures) • what elements of access might pose risks to scooter or other bus users (i.e. internal manoeuvre, sudden speed increase at the top of the ramp, stability on or leaving the ramp etc) • whether the manoeuvrability of different models of scooter within the bus or their mode of entry limits them to facing forward or rearward when the bus is in transit CCD believed that this last issue is of some significance, since previous research and policy in both Europe and Canada suggests that scooters and wheelchairs provide safer seated position in transit when facing the rear of the bus. Indeed the majority of wheelchair spaces provided in buses are designed for rearward facing occupation. CCD believed that it was important to understand whether some models of scooter that might otherwise fit in terms of size, may not be capable of being manoeuvred into a wheelchair space to face rearward.

Health & safety assessment As part of the stakeholder consultation workshops (as described above) CCD ran hazard identification and hazard and operability sessions (HAZID and HAZOP)


127

Figure 2. Test Rig.

to identify and quantify perceived health & safety risks associated with the carriage and access of scooters on buses and, where possible, identify options for risk mitigation and design improvement. This provided a detailed understanding of the level of risk and concerns of safety from the perspectives of both bus companies, their drivers and of the service users. As part of the user trials of scooter, access and manoeuvrability potential health and safety risks associated with each model (3 and 4 wheel) that meets TfL’s wheelchair space requirements were identified, especially in relation to stability

Figure 3.

Mock up.

128

S. Thomas

during ramp access and subsequent internal manoeuvre and in relation to potential interaction with other bus users. Of particular interest to TfL (and as highlighted in previous research) is the potential risk of tipping, particularly for 3 wheel scooters, when the bus is in transit. CCD investigated dynamic stability and potential issues with sudden swerves or braking by computer-based simulation. CCD employed MIRA’s Simulation and Safety Engineering department to make a series of simulation tests with their MADYMO 3D system of those scooter models which meet TfL requirements, each with occupants of various masses (e.g. 5th, 50th and 95th percentile). The occupied scooter models were subjected to a series of defined accelerations from cornering, swerving, braking and accelerating from which the potential tipping or other types of response from the scooter and occupant were determined. Importantly MIRA’s assessments were compared with responses between forward facing and rearward facing positions, in order to determine the validity of current practice and research which suggests that rearward facing positions for scooters are safer. Following the simulation assessment, CCD made a design based assessment of the various buses in operation with a view to commenting on or developing mitigation to those risks identified, such as tipping prevention devices and/or alterative types of restraint that might be required. The aim was be to determine what, if any, measures might feasibly be employed (in terms of adaptation or process) to enable the carriage of models of scooter that present too high a risk of tipping, but might otherwise meet TFL’s requirements.

Results Research & survey work Mobility Road show The most interesting findings from the road show were from non-London transport users and the reasons why they chose not to use their scooters on public transport and buses. The following were a few of the reasons: inaccessibility, they owed a disability vehicle, crowding issues, confusion on policy, previous bad experience.

Focus groups and forums See ‘Stakeholder Consultation’ below.

Survey of London bus fleet In total, eight models of bus were identified by TFL as those with the most limited interior space in terms of the wheelchair allowance, infrastructure such as poles, etc. These models were then photographed, measured, assessed and ranked in terms of space limitation, the worst five going through to pilot stage.


129

Stakeholder consultation Managers and Drivers. Both managers and drivers were of the opinion that the current situation is that scooters are allowed onto buses at the discretion of the driver in the absence of any firm guidelines; and that there also seems to be a lack of guidance on what happens if a wheelchair/scooter needs to board and the bay is occupied by a wheelchair, child buggies, which may or may not be collapsible, etc. In any event, the result could be confrontational with the driver. There are a multitude of motorised scooters of different sizes and configurations (3 wheel, 4 wheel etc) and the suitability of allowing the whole range is unlikely. There appears to be three main options, all of which will not be welcomed by some Stakeholders: • Disallow all scooters • Allow all scooters • Allow “compliant” scooters by some form of “badge” scheme. Drivers felt that to allow scooters onto a bus, they would have to present minimal additional disruption when compared to a wheelchair. In general a wheelchair can enter the bus by the ramp powered by the occupant or an assistant. It can then turn within its own length, reverse into the bay, rest against the backboard and apply a brake to stop movement during motion of the bus. Exit is a direct drive out. On entrance, if the bus is unable to get close to the kerb then it is likely that the wheelchair can still get off the kerb and access and exit the ramp from road level. Drivers would therefore suggest that to allow access to scooters, then certain criteria need to be met to ensure that they are comparable to a wheelchair for both safety and boarding times. Users. It was evident from the focus groups and forums that the majority of users interviewed had, at least, attempted to use their mobility scooter on public transport. Those who had not would certainly like to be able to as they felt it would enable them to be more independent. A few of these users had attempted but were refused by drivers for reasons such as too many people currently on board, not being allowed to carry them, the model in question was too big and so on. Drivers were often being quite rude and users stated that they are made to feel that they are a burden to the public. Of those who had and do use their scooters on public transport, the majority felt that it gave them more freedom. The majority of users also claimed that they had travelled in the forward facing position and not rearward as specified in the manual wheelchair regulations. Users felt that by travelling forwards it enabled them to establish their position and right to travel on the bus, they didn’t have people staring at them, they were able to see where they were going and (in the absence of an I-bus) could see when to get off. Users felt strongly that there was no need to travel in the rearward facing position, since the seat pad was not designed for a scooter. Users felt that should the policy be changed, it would not be possible for all models of scooter to travel on the bus and those models that could, would need to

130

S. Thomas

be identified somehow to users and drivers. The issue of prioritisation would also need to be addressed in terms of buggy, wheelchair users and the elderly. There would also need to be re-instruction to bus drivers as many users felt that they were ignorant to their needs and to all bus users.

Conclusions • It is clear that scooter users would generally be keen to travel on London buses and would benefit from doing so. Usage is currently limited because users do not believe they are allowed on buses, because they do not know that buses might be accessible to their scooters or because drivers refuse to take them, for various reasons. • The user trials have demonstrated that a number of scooter types can access to some buses, although clearly not all models are compatible with all buses. • Accessibility for scooters is affected by a range of factors. Just because a scooter’s physical dimensions fit within the standard wheelchair space does not mean it can be carried. Their manoeuvrability, the shape and position of their wheels and handle bars, their floor clearance and so on interact with the access ramp, available space and various structures in the bus in order to limit access. Additionally, time taken to access/egress is likely to prove an important additional limiting factor. Some scooters require a lot of manoeuvring to access the wheelchair space (typically multi-point turns) which takes a considerable time. Clearly, in trying to operate a timetable, TfL may be forced to exclude some scooter models simply because they take so long to enter or exit buses that they would seriously affect the ability to run a timely service. • The policy for the carriage of scooters will have to clearly identify those scooters and buses that are compatible. Due to the range of factors influencing accessibility, it is likely that TfL will need to test future models and keep a register of those that are suitable for carriage, updating it as new models of scooter and bus evolve. • The most significant limitation to access in terms of bus design is the position stanchions/handrails around the wheelchair space. Opportunities for removing, re-designing or re-positioning these should be considered in order to improve accessibility, especially in reducing the time it takes to manoeuvre in and out of the wheelchair space. • Rearward facing travel (for the sake of safety) is unlikely to be feasible and if made a policy requirement would seriously limit the range of scooters that could be carried. Few scooters are sufficiently manoeuvrable to turn around inside the bus in order to reverse into the space, and given both the limited width of the ramp, limited capability of scooters and the nature of users (many suffer from restricted joint mobility and cannot easily turn around) it is not feasible or safe to expect users to reverse on to the bus. Additionally, it is likely that many users would be capable of transferring from their scooter to a bus seat (the bus must remain stationary). It is also thought unreasonable that scooter access be restricted on this basis given that wheelchair users often (if not typically) travel


•

• •

•

131

facing forward, despite policy to the contrary. It would seem unfair not to allow scooter users to face the same risks as wheelchair users choose to. In addition to defining the range of scooters that can be carried on certain buses, the carriage policy will need to clearly and unambiguously explain the conditions of carriage, safety issues and limitations. For it to be successful, the policy needs the widest possible dissemination, including at least relevant websites, scooter manufacturers, mobility advisory bodies and assessment centres, mobility forums and groups and so on. It will also be important to reach the widest possible audience, especially the non-scooter using public, who also need to be made aware of the policy if it is going to be effective and accepted. Drivers need to be properly and carefully trained on the policy itself and how to improve access through judicious choice of ramp deployment, announcements to passengers to make way for scooters if the bus is crowded, etc. Drivers need an easy way of identifying compatible scooters. The simplest method will be badge or flag on the scooter to allow the driver to see clearly that a given model is compatible with their bus. For this to work it is likely that TfL will have to employ some form of registration scheme whereby users apply for the identifying mark declaring the model of scooter they use. There are likely to be some operational problems that will need resolving if the policy is to operate smoothly. Drivers will require some means of indicating to waiting scooter users that the wheelchair space is already legitimately occupied (e.g. an external indicator light on the bus or external PA facility).

References Department of Transport, 2006, Carriage of Mobility Scooters on Public Transport Feasibility Study. http://www.dft.gov.uk/transportforyou/access/tipws/cmspt/ (at 21 October 2008). Department for Transport, 2000, Public Service Vehicle Accessibility Regulations (PSVAR) 2000 – Guidance.

HAZARDS

STUDY OF VERBAL COMMUNICATION BETWEEN NUCLEAR PLANT CONTROL ROOM OPERATORS DURING ABNORMAL SITUATIONS A.N. Anokhin & N.V. Pleshakova Obninsk State Technical University for Nuclear Power Engineering, Obninsk, Russia Communication plays an important role in the joint mental activity of Main Control Room (MCR) team members at a Nuclear Power Plant (NPP). Today a study of communication is considered valuable not only as a means for investigating information exchange, but also as a tool to improve personnel reliability and for reducing human errors. In the present work, the authors performed a detailed analysis of the communication which takes place when operators are dealing with an emergency situation at NPP. Based on interviewing the operators, the analysis of emergency procedures, and by conducting of a series of experiments, the following outcomes were obtained: six types of communication were identified, the amount and relative proportions of these various types communication were estimated, the factors influencing the communication were revealed and assessed, and recommendations of ways as to improve communication within the MCR were formulated.

Introduction Communication is defined as the exchange of information while preparing for, or performing work, during NPP control (Barnes et al., 2001). Purposes of communication include maintaining a common understanding between team members and thereby increasing operator reliability. Numerous researches have considered communication and its influence on the effectiveness of teamwork. Studies by Jentsch et al, (1995) and Johannesen et al, (1994) found that teams, who were more effective in their performance, used more standard communication and vocalized more situation awareness observation relative to changing important process parameters and alarms. Communication can result in a unified and accurate team situation model, that, in turn, aids collaborative action planning and is then needed to adjust strategies or to develop new ones to deal with the situation (Ma et al., 2006). Schraagen and Rasker (2001) studied the role of communication in the development of the shared mental models of the process situation. Their results showed that communication is not so beneficial in routine situations, when team members know exactly what is going on. In novel situations, however, communication is necessary for operators to respond to environmental cues, to explain to each other why previous strategies will not work in the current, novel situation and then it 135

136

A.N. Anokhin & N.V. Pleshakova

permits co-operative determination of new strategies and the prediction of future states. Increasing the level of automation can influence and improve communication between team members (Johannesen et al., 1994). However, new control room designs also can impose challenges on team working. Norros and Savioja (2004), noted, that transition to ‘soft control’ can diminish the horizon of observation and hinder information exchange. One of the methods for measuring team communication is to categorize communication types and then quantify communication in accordance with the categorization. Schraagen and Rasker (2001), for instance, considered and estimated different communication categories such as information exchange, determining strategies and performance monitoring. Barnes et al (2001) studied the communication errors and showed that a verification feedback serves an important error-checking function in the communication process. Through feedback the similarity of the meanings given to the message by the sender and understood by the receiver can be verified. An example of feedback in verbal communication is when the receiver ‘repeats back’ the message and the sender either agrees with the receiver’s repeat back or corrects it. Today, communication more and more plays an additional role in that it is used as a means for increasing the reliability of human operations and particularly, to avoid operator errors. After the series of accidents at NPPs during the 1970s and 80s the emergency management procedures tend to be more strictly formalized and structured. In order to implement the next step in a procedural sequence, the unit shift supervisor has to read the required step’s wording out loud. Then, the appropriate operator (reactor operator or turbine operator) executes this step and reports back about any obtained results. Such a protocol, contributes to increasing the reliability of operator actions. However, this kind of communication is superimposed on the regular exchange of information between MCR operators, field operators, plant specialists and power grid dispatchers. This requirement thus increases the frequency and intensity of communication traffic which can overload the information transmission between operators. As a rule, the emergency procedures are used under stress conditions and under time pressure. In such situations, the communication in itself may be subject to errors and the operators are forced to find ways for optimizating the message exchange. This paper is devoted to the analysis of the communication which takes place between NPP MCR operators when managing abnormal events. The paper considers the categorization and the structural models of various types of communication acts, the estimation of relative proportions of the various communication types, the analysis of factors preventing communication, and a description of methods used by operators in order to facilitate communication.

Methodology The study comprised three stages, namely: interviews with operators, analysis of emergency procedures, and an experimental study. The reference NPP was a Pressurized Water Reactor (PWR-1000) unit generating 1000 megawatts of electrical

Study of verbal communication between nuclear plant control room operators

137

power. The experienced MCR shift personnel were interviewed at the first stage of the analysis. Most of operators emphasized that they felt a lack of time when required to read the wording of a procedure step by step, instead of immediately implementing the required actions in a rapidly evolving emergency situation. Then the relevant emergency procedures were studied. Two broad kinds of procedures are used at NPPs, namely ‘symptom-oriented’procedures and ‘event-oriented’ procedures. Procedures of both types are represented in a two-column tabular format. The third stage, the experimental study, consisted of observation of MCR teams while managing an emergency scenario presented using a full scope NPP training simulator. The scenario reproduced the superimposition of two faults – a leakage of water from the primary loop to the secondary loop inside the steam generator and a leakage of water from a primary loop into the containment. Four MCR teams took part in the experimental study. Each team consisted of operators who have had some years of work experience in the MCR. Each MCR team consists of four persons, namely – reactor operator, turbine operator, shift supervisor of the reactor plant and the unit shift supervisor. Two teams used the symptom-oriented procedure in order to overcome the situation, and two teams used the event-oriented procedure. The actions performed by operators were recorded by video camera, and all relevant process parameters, manipulation of controls and technological events (alarm signals, switching of equipment) were registered by the simulator computer. A hierarchical task analysis and timeline analysis were carried out. All verbal messages from the MCR operators (160–180 messages in each simulator run) were categorized and their expected duration was estimated. The following work has been undertaken in the course of the study: • the main types of communication were revealed, categorized and described; • relative proportions of the various communication types were recorded. • the performance shaping factors and the main problems were considered and analyzed.

Types of communication acts MCR operators communicate both with each other and with outside personnel. Among themselves, operators communicate verbally, sometimes accompanying their messages with gestures. Outside participants are field operators, shift supervisors of various departments, the NPP shift supervisor, NPP senior managers, and the dispatchers of various systems – power grid, fire brigade, etc. MCR operators communicate with outside personnel by means of telephone or over the public address system. An elementary component of communication is an act of information transfer from one sender to one or more receivers. In the more complex case, communication takes place in the context of some action. Such a communication includes two or more information transfer acts (for example, question-answer). Figure 1 presents a structural flowchart for the act of information transfer. In the simplest case, such an

138


Figure 1. A structural flowchart illustrating the act of information transfer from sender to receiver.

act includes just one activity element, namely: the sender speaks the information. In the more complex case, the receiver provides feedback and either confirms adequate perception of the information, or asks for repetition or to clarification of the information. The sender repeats the message or provides some additional data. After the exchange, the act may be completed or continued until the transferred information is satisfactorily perceived by the receiver. Information transfer acts can be categorized into three generalized types depending on the purpose and content of the information to be transferred, namely: 1) Query – The sender asks the receiver for certain information; 2) Report – The sender informs the receiver about certain events; 3) Command – The sender instructs the receiver to perform certain actions. The communication act is a combination of the above mentioned three types of information transfer act. Figure 2 illustrates the six various types of communication, between NPP MCR operators, that were revealed during the study. (a) An Information report consists of one act of information transfer to one or more receivers. Usually an information report is initiated by a reactor or turbine operator (or field operators through communication facilities). As a rule, such a report contains information about incoming alarms, the deviation of process parameters and any change in technological equipment state. (b) A Query-report includes two information transfer acts. In most cases, the MCR shift supervisor asks operators for necessary information or the MCR operators issue a query to field operators. A query can by expressed either as an interrogative sentence (for example, ‘what is the current concentration of boric acid?’) or as an instruction to perform an inspection action (for example, ‘Check the current concentration of boric acid‘). A Query implies an obligatory reply.


139

Figure 2. Flowchart diagram for the six types of communication: (a) information report, (b) query-report, (c) discussion, (d) report about action performed without command, (e) command-action, (f) request to perform action.

(c) A Discussion takes place when the receiver and sender exchange information and try to find a root cause of situation. During such mutual discussion, members of the MCR team usually clarify the current situation or make a decision about any necessary joint action. (d) A Report about action performed without command takes place after the reactor or turbine operator has independently performed a certain technological operation. The purpose of this type of communication is to maintain the situational awareness of other team members. (e) A Command-action. This type of communication commences with an instruction to perform a certain technological operation. Usually, the unit supervisor is the initiator and sender of the command, whereas the reactor or turbine operator is the receiver. In some cases, the command can be initiated by an MCR operators and will be addressed to the field operator. After the command has been issued, the receiver performs the prescribed operation (usually this is a motor action – such as switching on a pump, controlling a valve, etc.) and then reports back as to the outcome. In some cases, the report may be omitted. (f) A Request to perform an action is initiated by operators when they need an approval of this action from the plant or unit supervisor. After the unit supervisor has received the query, he appraises the situation and then either approves or rejects the action. Upon receiving approval, the operator completes the action, after which he can then report on the results. In total, 529 acts of communication were identified during four run-throughs of the abnormal scenario and their total duration was 2 hours and 17 minutes. Communications took up about half of this total time. The average duration of one communication act was 7.7 sec. Table 1 shows the quantitative characteristics of the different

140


Table 1.

Proportion of the various types of communication.

Sort of communication

Relative proportion of the communication type, %

Relative time spent on this type of communication, %

Average duration, sec

Information report Query-report Discussion Report about action performed without command Command-action Request to perform action

38 23.5 5 3.5

16 28.5 12 4.5

3.3 9.4 18.4 10.1

24 6

34 5

10.9 6.8

communication types. The table shows that the majority of the communication time was occupied by information reports, query-reports and command-actions.

Factors which influenced communication During the running of the scenarios and the analysis of the video records, it was revealed that a considerable proportion of information transfer acts included requests to clarify or repeat information (see Figure 1 for the structure of each act). A High level of noise within the MCR was identified as one of the main causes. During abnormal situations, the noise is produced by activating audible alarm signals and the extremely intensive conversations between the operators (particularly between MCR operators and external personnel). Each alarm is accompanied by an audible signal such as a loud ring or a siren. Though the operators try to avoid talking when alarms are sounding, the signals often occur after the communication act has started. In the case of a rapidly developing situation, the frequency of alarms is very high and operators have insufficient time for the exchange of messages in the extremely short breaks between consecutive alarm signals. The frequency of communication and the frequency of audible alarm signals during the four emergency scenario runs are shown in Figure 3. In most situations there is a low correlation between the two curves (the Pearson correlation coefficient is 0.3–0.7). This is to be expected because the operators must report about the deviation of process parameters and any change in equipment operational mode immediately after activation of the appropriate alarm signals. Moreover, when the situation is developing rapidly, the operators make a joint effort to find an explanation for the occurring events and to develop a suitable strategy of actions. In order to be heard, the operator is forced to raise his voice and to shout above both the sound of alarm signals and over any talking between other team members. Another cause of excessive noise in the MCR can be the presence of additional personnel such as the safety engineer, shift supervisors from various departments, the nuclear physicist, etc., who may all be required in certain situations. Additional personnel can arrive at the MCR both during the abnormal situation or in the


141

Figure 3. Frequency of communication and audible alarm signals when dealing with the emergency scenarios (——– – Frequency of communication acts, - - - - - – Frequency of audible alarm signals).

case of scheduled changes to the unit operational mode (for example, raising the reactor to criticality). The additional personnel may assist the MCR team members to solve problems and may partially relieve the reactor and turbine operators of their high workload. However, these external specialists can increase and overload communications and inevitably raise the level of noise in the MCR. The second, essential factor preventing fast and accurate communication is the occurrence of several communication acts together. Superimposition is the appearance of two or more communication acts concurrently. Most often superimposition takes place when several, different senders try to transfer information to one receiver. For instance: • when the state of the unit equipment is changing very rapidly and both operators (reactor and turbine) report main events to the unit shift supervisor at the same time; • when the shift supervisor issues a command to an operator to perform a certain action while this operator is already busy talking with a field operator. In the case of the superimposition of communication acts, the receivers often asked the sender to repeat the message. The same situation was observed when an audible alarm was superimposed on a communication.

142


A third essential factor influencing communication is the information content of the transferred message, query and command. In accordance with the existing rules for the implementation of emergency procedures, the unit shift supervisor has to read out loud the whole of each sequential step of a procedure, following which, the appropriate operator (reactor or turbine) must carry it out.. It had been revealed as a result of a previous ergonomics assessment of procedures, that most of the procedure steps, as formulated by ‘written language’, were much more ponderous and verbose when compared with colloquial language. The wording of a procedural step typically contains complex, technical equipment identifiers, such as 4RT11,12,13,14S11, which can be awkward to perceive verbally although they can be read without difficulty. When interviewed, the operators said that, under the conditions of time pressure and in a rapidly expanding emergency situation, reading all instructional steps aloud and verbatim, actually retards the operator’s response to the situation. Analysis of video records revealed that when the personnel were in a hurry they cut out the wording of some steps and omitted unnecessary and redundant words. When operators mentioned some equipment items and actions, they often used professional slang expressions which have been established as a result of longterm, joint working within a team. Such a use of slang enables the reader to save time and to reduce the duration of reading of procedural steps, on average, of three times. This assessment was obtained by measuring (using the video recordings) the duration of each communication act, including reading of the instruction and then comparing the measured duration with the expected duration. In order to estimate the expected duration, each instruction was read aloud verbatim five times. Each time the reading duration was measured to an accuracy of 0.1 sec and the average time was calculated. In a number of cases, the unit supervisor himself performed the operation of checking of a process parameter, rather than reading out the relevant step of procedure and then seeking the value of this parameter from the operator. This type of task substitution reduces the duration of step performance by, on average, a factor of five to six. At the same time, the reduction of wording and the usage of slang can lead to ambiguous interpretation of the content by the receiver. Use of slang to refer to technological equipment may force operators to perform a mental conversion of information because all control panels and documentation use only “official” labels and identification. Obviously, this increases the cognitive workload on operators and creates the preconditions for mistakes. However, it should be noted that, during the conducted experiments, no such errors were observed.

Additional findings and recommendations Along with the analysis of communication, the way in which operators restored the power unit from transient, abnormal, state to normal (stabilized) mode was also studied. An evident relationship was revealed between the time spent by MCR personnel for communication with each other and the overall time spent to normalize


143

the situation at the unit. The more time operators spend in communication, and the more carefully and accurately they transmitted information between each other, then the more quickly, smoothly and effectively they achieved the normal state of the unit. This conclusion is in agreement with similar results reported by Jentsch et al (1995). This dependency is predictable, since the communication enables the team to maintain awareness about process events and about the actions performed by all team members. Nevertheless, if the amount of communication exceeds a certain critical limit then operators may lose the time required to perform the necessary actions. It was also noted that senders often call the receiver by name (or name and patronymic, as is customary in Russian) at the beginning of an information transfer act. This immediately stimulates the recipient and concentrates their attention, thus allows them to avoid superimposition of messages and reduces the need for clarification questions. Another way to improve the efficiency and reliability of communication is by read back, when the receiver confirms that he/she has received and interpreted the information in the correct way. Most often, the receiver simply repeats back the received information. Undoubtedly joint planning and decision-making regarding the required followup actions can be very useful. Joint cognitive work can often lead to more effective teamwork. However, it is not always so. For example, in one conducted experiment, an operator incorrectly identified the initial event and then persuaded the unit supervisor to follow an inappropriate procedure. The following recommendations, which are aimed at improving communication have been formulated as a result of the analysis of findings. 1. There is a need to adjust not only the volume, but the tone of audible emergency and warning alarm signals in the MCR. It is important to ‘divorce’ the tone of the audible signals from the frequency and tone of the human voice. 2. The wording of procedural steps were more easily perceived when rewritten particularly when complex linguistic constructions were replaced. 3. It would be useful to unify and standardize the use of slang terminology in order to develop an unambiguous description of certain actions and technological equipment. 4. Provide the unit supervisor with contextual technological information in which the content depends on the current conditions and procedural step. This would particularly reduce the number of ‘query-report’ acts and would lead to a general reduction in communication load. 5. In order to avoid the loss of important information in case of superimposition of communication acts, it is necessary to concentrate the attention of the receiver to high priority messages (for example, by using the keyword ‘important’).

References Barnes, V., Haagensen, B. and O’Hara, J. 2001, The human performance evaluation process: a resource for reviewing the identification and resolution of human performance Problems (NUREG/CR-6751), (U.S. NRC, Washington, DC)

144


Jentsch, F., Sellin-Wolters, S., Bowers, C. A. and Salas, E. 1995, Crew coordination behaviors as predictors of problem detection and decision making times. In Proceedings of the HFES 39th Annual Meeting, (HFES, Santa Monica, CA), 1350–1353 Johannesen, L.J., Cook, R.I. and Woods, D.D. 1994, Cooperative communications in dynamic fault management. In Proceedings of HFES 38th Annual Meeting, (HFES, Santa Monica, CA), 225–229 Ma R., Kaber, D.B., Jones, J.M. and Starkey, R.L. 2006, Team situation awareness in nuclear power plant process control. In Proceedings of 5th ANS International Topical Meeting NPIC&HMIT 2006, (ANS-Omnipress, Washington DC), 459–462 Norros, L. and Savioja, P. 2004, Usability evaluation of complex systems. A literature review (STUK-YTO-TR 204), (STUK, Helsinki) Schraagen, J.M. and Rasker, P.C. 2001, Communication in command and control teams. In Proceedings of 6th Command and Control Research and Technology Symposia, (CCRP, U.S. Department of Defense, Washington, DC)

. . . IT’S THE WAY THAT YOU DO IT: ENSURING RULE COMPLIANCE J. Berman, P. Ackroyd & P. Leach Greenstreet Berman Ltd, Reading RG1 4QS, UK Ensuring compliance with procedures and process is of great importance not only for high-hazard organisations, but also in other sectors such as Finance. Whilst much is known about the causes of noncompliance, such as inadvertent actions, misunderstandings or deliberate violations, non-compliance remains a challenge. The challenge for industry is to understand the potential causes of non-compliance and to identify appropriate measures to defend against them. The solutions frequently lie in areas of conventional ergonomics – task, interface and procedure design, and operator competence. And yet organisations frequently find difficulty in seeing a clear way forward. The importance of an ergonomics approach is highlighted, but as part of a coherent strategy for managing rule compliance, including supervision and leadership skills. It also emphasises the contribution this makes to creating a resilient organization.

Introduction What do the ‘rogue traders’ at Société Générale and Barings Bank, and the explosions at the BP Texas City refinery and the Buncefield oil storage depot in Hertfordshire have in common? They were all either caused, or aggravated, by failure to follow procedures. Ensuring compliance with procedures and process is of great importance for all organisations – whether traditional ‘high-hazard’ or economic high-consequence. Much guidance has been written concerning noncompliance, ranging from explicit guidance (e.g. HSE, 1995) through to general guidance on ergonomics, procedure design, etc. and hence it could reasonably be assumed that the problem would now be properly controlled. The reality is that many organisations fail to ensure compliance with those processes and procedures that they believe are critical to their organisation’s commercial or safety performance. This begs the question – why is it apparently obvious (albeit with hindsight) what should be done, but apparently so difficult to do it? This paper discusses how the issue can be manifest, and the challenge of identifying and implementing effective solutions. Whilst many solutions lie within the domain of ergonomics, selection of the appropriate approach requires an understanding of the goals that underpin the observed behaviour. The paper considers the nature of non-compliance, the importance of the concept of goals, and the idea of a contract between the individual and organisation. This leads to consideration of the manner in which non-compliant behaviours are manifest (rather than focussing 145

146

J. Berman, P. Ackroyd & P. Leach

solely on the actual breach of process). The paper then considers what some of the solutions might look like, and approaches to proactive improvement. This is then discussed in the context of resilience, and the manner in which organisations can continue to move forward.

What is the issue? It’s important to understand the nature and extent of non-compliance – what we mean by the term ‘non-compliance’ and hence what we are trying to address. It’s easy to consider a set of behaviours as representing a common set of deficiencies, and hence being amenable to a single set of solutions, when in practice they encompass a wide variety of shortfalls and challenges. Compliance is concerned with the willingness and ability to follow agreed processes and procedures. It is important that the agreed processes, whether or not they are formally described in procedures, are acknowledged both by the organisation and by those charged with undertaking the tasks and activities, to be the required manner by which the desired outcomes are achieved. It would therefore appear that ‘non-compliance’ should be self-evident – the failure to undertake tasks in accordance with the agreed processes. Unfortunately, the reality is often less clear. What is the agreed process? Is the observed action a failure to follow the process or a legitimate deviation? What is the person attempting to achieve? What is the difference between non-compliance and a simple slip or mistake? What is the role of poor operator knowledge? The cause of non-compliance cannot be directly observed – only inferred from the context. Non-compliance is concerned with intended actions which deviate from an agreed process or procedure for a given task (i.e. it excludes simple errors). Three attributes are important: • Whether the deviation is deliberate – this distinguishes between active noncompliance (‘I choose not to follow a particular instruction’) and passive non-compliance (‘I was not aware that what I was doing was a deviation from the accepted procedure’). Passive non-compliance is particularly amenable to ‘ergonomics’ solutions. • The ‘correct’process is agreed – there is a contract between the individual and the organisation in respect of undertaking a task in a particular manner. The notion of contract implies obligations on both sides. Compliance must be feasible (all of the elements of ergonomics that contribute towards successful task performance), and the organisation must ensure that these elements are maintained in an effective manner. • Shared individual and organisational goals – compliance is concerned with voluntary behaviour by individuals, and hence it is important to understand the extent to which the individual’s goals align with the organisation’s. There will always be a disparity, but the greater the disparity, the more likely it is that non-compliance will occur. Understanding the disparity helps to identify where ergonomics can support solutions, and tends to be a key element in determining effective compliance management.

. . . It’s the way that you do it: Ensuring rule compliance

147

Given these attributes, there is a set of interrelated factors that can lead to non-compliance: • Culture: there may be attitudes within the organisation that affect the perceived importance of the implied compliance contract, or the extent to which it applies. • Goals: there needs to be sufficient consistency between the organisation’s goals and those of the individual. • Interfaces: the provision of information to the operator, and how the operator can execute the tasks affects their willingness to undertake the tasks in a given manner. • Procedures: the appropriate approach to undertaking the task can be agreed only if it is suitably described and communicated. Furthermore, the agreement requires more than merely being informed of the required approach – it must also be agreed as being feasible, viable, effective and required. • Training: the organisation must ensure that the people charged with undertaking the task are suitably equipped to do so in terms of knowledge, competence, and understanding of the context in which it is required. • Job design: the demands of the job must be reasonable and realistic. If an organisation is to address non-compliance effectively, it needs to consider how all of these issues come together, and to address them holistically and with the cooperation of those most directly affected – the ones who are apparently failing to comply. McDonald (2006) cites an example of a study of aircraft maintenance where one third of the technicians reported that the procedures do not present the best or safest approach and hence are not followed. What is notable about this is not only the proportion of staff who fail to comply with procedures in a highly regulated environment, but also the lack of surprise expressed by other in the industry when advised of this – it implies not only that non-compliance may be endemic, but that it is, in effect, condoned. This is not to suggest that the industry is unsafe – its record is very good – but that there appear to be many factors contributing to the choice of behaviours exhibited by staff who are apparently expected to exhibit high compliance. It should be noted, of course, that there may be very good reasons for the procedures as they are which are merely not apparent to the maintenance technicians. A concern is the apparent acceptance of a ‘pick-and-mix’approach as to which procedures require compliance. We have observed similar phenomena in such sectors as the nuclear industry where safety assessments have elicited such comments as “yes – that’s what the procedure says but actually we don’t do it that way”.

How is it manifest? At one level this may seem an odd question – it is surely obvious when staff fail to comply with a procedure. In practice, the obvious indicators are merely that there is a problem – but not what are the causes of the observed problem, and hence there is confusion as to effective solutions. An ergonomics approach (considering the entire work context and environment) would suggest that looking beyond the immediate issue is potentially beneficial, particularly in supporting a proactive approach to

148


avoiding non-compliance. The question therefore is not “how are non-compliant acts manifest”, but rather “how are the organisational behaviours manifest which indicate a vulnerability to non-compliance”? • Culture: How does the organisation ‘reward’ different behaviours? What standards are demonstrated? What resources are available to revise work practices? • Goals: How does the organisation communicate its goals, and monitor (and support) the goals of its employees? • Interfaces: What evidence of interface shortfalls is sought within the organisation? How are interface concerns raised and addressed? What is the incidence of ‘special’ tools and job-aids? What is the evidence of ‘DIY’ modifications? • Procedures: What level of training is being implemented to compensate for potential procedural deficiencies? How is procedural guidance presented to the users? How are changes to procedures promulgated? • Training: How is training assessed? Have training syllabi kept pace with developments in the task and equipment? • Job Aids: Are there simple job aids that could be introduced? Is there reliance on ad-hoc job aids to compensate for poor task design? If the nature and cause of non-compliance is properly understood, then the set of potential solutions starts to emerge. Non-compliance is not inevitably associated with failure – many instances will occur without harm arising, and potentially with what appear to be positive outcomes such as reduced time or resources needed, such that the behaviour might be perceived as acceptable. It may be only after some time that a failure arises and the organisation focuses on those behaviours. It is therefore not sufficient to rely only on conventional indicators of performance. The organisation has an obligation to provide suitable monitoring arrangements – and this is part of the ‘contract’ between the organisation and staff. One set of indicators of non-compliance, ironically, might include evidence of apparent good performance. Why is a task now being completed in half the time? Why with significantly reduced cost? Of course this is not to say that any instance of good performance is indicative of an underlying violation. However, it reinforces the importance of investigating and learning from good performance. This also is part of the contract. If the good performance arises from a valid and appropriate alternative arrangement, then the lessons should be exploited. Alternatively, if it arises from an inappropriate approach – perhaps a failure to undertake an important safety check due to a perception that it is unnecessary – then addressing that deviation before it leads to an incident is clearly sensible. The individual instance may provide a leading indicator. It may also provide information about diverging goals.

What do the solutions look like? There are no generic solutions. There are many reasons for non-compliance, and hence each situation will demand different approaches. All too often the


149

Make compliance easier/remove problems Apply discipline/ sanctions/ rewards

Make rule better/correct

Involve staff in rules and implementation

Set clear standards Improved compliance Introduce a behavioural modification approach

Educate

Make rule and compliance important

Make non-compliance difficult/impossible Supervise/ monitor

Figure 1.

Potential solutions.

organisation’s reaction is to pursue training solutions, and perhaps procedural enhancements (assuming that they do not merely adopt a disciplinary approach). Whilst these might be of value, there are potentially many other solutions that might be relevant in certain circumstances. Figure 1 is derived from a toolkit developed for the rail industry (Berman et al., 2005), and indicates the range of potential solutions that could be considered. A key element of the toolkit is support for solution selection and implementation. It is rare that only one solution would be adopted, particularly when addressing compliance generically rather than merely seeking to prevent the recurrence of a specific instance. Furthermore, solutions that work in some instances may not be appropriate in others. For example, setting clear standards where the non-compliance was due to confusion about what the organisation expected (i.e. goal differences), would not be the optimal solution for non-compliance arising due to poor task design making compliance extremely difficult. A user of the tool from a chemical manufacturing plant noted that the generic approach allowed him to identify a non-compliance as unintentional, and that it would be most effectively addressed through a behavioural approach, whereas his original perception was that it was due to deliberate deviation. It is also important to emphasise that this is a solution set – it’s not a set of causes. One powerful cause is a misunderstanding of the demands of the prevailing situation, leading to incorrect rule selection. This can arise for many reasons, including

150


poor situation awareness (e.g. Banbury and Tremblay, 2004), or the effects of recognition-primed decision-making where the decision-maker uses probabilistic reasoning based on past experience (e.g. Klein, 1993). The importance of the solution-set is that it provides potential means of overcoming the consequences of these effects. However, other approaches to improved management of situation awareness and decision-making remain valuable, and this underlines the importance of a coherent approach to compliance management. Additionally, whilst improving the rule may improve compliance, it is also important to consider why the rule was not optimal. It is inevitable that there will be ‘bad’ rules – systems and processes change and evolve. The challenge arises when the organisation makes it difficult for a rule to be improved. This will tend to lead to non-compliance rather than identification and improvement. Thus, a proactive solution will focus also on improving the change management arrangements. It may further wish to consider means of proactive review of procedures to identify challenges (such as might arise through a HAZOP-style evaluation process). In one instance, an organisation was concerned about staff failure to wear highvisibility vests. This arose in part due to inconvenience and was eventually overcome by making the overalls high-visibility and hence removing the need for the vests. By making compliance easier, the need for the ‘rule’ was removed.

What would help? A number of challenges have been noted in terms of recognising and understanding the nature of a given compliance problem, the ‘contract’ between the organisation and the individual and how that contract affects behaviours and goals, the range of potential solutions (which span the gamut of ergonomics), only a subset of which will be relevant to a given compliance problem, and understanding how to implement them. These challenges are no different from those facing any ergonomics intervention and should be amenable to a similar ergonomics approach: • • • •

Recognise the problem (be clear about what is happening) Analyse the problem (understand the activity, constraints and goals) seeing the solution (considering a set of potential solutions) devising an effective implementation strategy (it is likely to involve interventions that affect attitudes and behaviours)

Guidance tends to be beneficial in supporting selection, tailoring, implementation and monitoring of solutions. A further requirement is to help the organisation not only to be able to correct identified shortfalls that could lead to non-compliance, but also to become more resilient to the influence of those factors. In other words, the optimal approach would reduce the likelihood of future shortfalls arising. The rail industry software tool (op. cit) was developed to support organisations that wished to understand and manage safety critical rule compliance. Unsurprisingly, the tool itself is generic (although the examples are predominantly drawn from the rail industry). In developing the tool, much time was spent working with a range of rail industry organisations (Train Operating Companies, infrastructure


151

maintenance companies, etc) to understand the range of contexts and potential users of the tool, and the questions the tool was to support. What emerged was clarity of the elements that could give rise to non-compliance and the importance of identifying which one(s) were relevant in a given context (and typically there would be more than one, and therefore also more than one preferred solution). It was also apparent that the tool was likely to be of most use to those managers and supervisors who were best placed to address the issues. The tool was constructed as a set of questions that would encourage the user to explore the evidence for potential factors that could affect compliance, and then to consider various means of addressing those factors. The tool linked the causes and potential solutions, together with examples of how such solutions might be implemented. This was also found to be a key element of the tool, as it is important that the set of selected solutions are tailored to the specific demands of the organisation. In other words, the tool assisted users to identify their own solutions, rather than presenting ‘off-the-shelf ’ solutions. It provides examples. The tool was empirically derived, and its validation has tended to be pragmatic – organisations have found it has led them towards solutions that might not have been immediately apparent to them, but which appear to have led to reduced noncompliance. In piloting the tool, it was clear that there was a need to be able to communicate with senior management concerning the nature of the issues and the potential ‘solution space’ in order that appropriate resources would be made available. However, it was also clear that selection of appropriate solutions required cooperative effort between managers and staff at all levels, and the tool needed to be able to facilitate this communication. It is important not to imply a single solution. The questions arising were also important. Whereas many managers might initially make use of the tool to allow them to understand and resolve non-compliances identified through incident reporting and investigation, it became apparent that this was often merely the means of engaging with those managers, and that once familiar with the approach they were equally interested in proactive assessment of potential non-compliances – perhaps arising from planned changes to working practices. This prompted the development of a specific part of the tool that supported proactive assessments.

Integrating solutions Implementing effective solutions requires a coherent and wide-ranging approach. It’s not sufficient simply to address a particular instance of non-compliance – nor to focus on one aspect, such as poor task design, or poor procedures. There are two distinct areas where solutions may lie, and typically only one is considered. Depending on the nature of the non-compliance, there will be a tendency to look for solutions either within conventional ergonomics (task design, interfaces, procedures, training, workload, etc), or within organisational factors (supervision, culture, leadership, communications, sanctions and rewards, etc). It is not our intention to imply that ergonomics excludes organisational factors – ergonomics can be considered to embrace the latter – the distinction is drawn to highlight the perception

152


that can exist within organisations. All too often there is an attempt to resolve non-compliance through improved procedures or training, without considering the importance of addressing the organisational factors. Attention to conventional ergonomics is necessary but not sufficient. Effective management of compliance requires a balance of both of these aspects – it is rarely the case that an organisation has deficiencies in only one of these areas. This returns to the issue of the ‘contract’ between staff and organisation. The organisation has an obligation to ensure that the ergonomics issues are addressed such that compliance is made as easy as possible, and non-compliance is made both difficult and non-rewarding. This permits the member of staff to fulfil their part of the contract which is to operate systems in a compliant manner. At the same time, the organisation also has an obligation to support the member of staff through provision of effective supervision etc., which in turn can reveal goal disparities. Without a clear and recognised balance between the two elements, there is the potential for the solutions to fail. Furthermore, it is attention to the organisational issues that will provide both a clear demonstration to staff that non-compliance is unacceptable to the organisation (and that it is prepared to commit resources to reducing it), and also a clear demonstration that the organisation is not merely paying lip service to improving compliance but instead is establishing processes that will provide high quality information on which to act. Senior staff and line managers create the perceived priorities within an organisation and, often unwittingly, place an undue priority on commercial or operational goals thereby undermining safety. Staff may perceive that it is acceptable, or even required, to take risks to achieve the company’s goals. Language, allocation of resource and what managers show interest in can all create this position. Staff may feel that it is not appropriate or possible to raise safety concerns. This is an area where senior managers need to consider their own behaviours carefully. It is vital that they connect directly with the workforce to find out what is really happening, what concerns staff have, and to consider the nature of the culture within the organisation. Positive environments where success is celebrated can inadvertently stifle concerns. Senior managers either not wanting to listen to ‘bad news’ or appearing to challenge such news aggressively will inevitably prevent key information being raised and acted on. Middle managers become a filter that either blocks or distorts information from the workforce. Good news is quickly passed upwards whilst bad news may be reduced, downplayed or blocked. Thus, effective management of compliance requires attention to attitudes and culture issues, in order to maximise the benefits from attention to more conventional ergonomics issues such as procedures, training and task design.

Resilience No discussion of compliance would be complete without considering organisational resilience. It’s important that organisations not only address the current factors that


153

might encourage non-compliance, but that they also ensure they are able to prevent the recurrence of those or similar factors, or the gradual erosion of the effectiveness of controls. Many organisations rely on procedural controls to ensure effective management arrangements – and these more than any are frequently at risk of degradation due to non-compliance. Many events from finance (Société Générale, Barings Bank) or conventional high-hazard industries (Longford gas explosion, Royal Commission, 1999; Grayrigg train derailment, RAIB, 2008) have the failure of procedural controls as a significant causal factor. Amalberti (2006) presents a revision of Reason’s resilience model as a cascade of resiliences, where a series of sets of prevention, detection, recovery and mitigation are provided to prevent failures from propagating into accidents. Control of compliance should be considered in the same manner, whereby arrangements are implemented and maintained to prevent not only the identified non-compliances, but also potential future or undetected ones. It is the ability to develop arrangements that remain robust over time that is of particular importance, and why this paper has emphasised goals and solution selection rather than the solutions themselves. It is also important to consider how those arrangements can degrade over time precisely because of the factors that give rise to non-compliance. This is noted by the present authors when considering organisational drift (Berman and Ackroyd, 2006). Attention to improving compliance is a key defence against drift, and a key factor in supporting resilience. Furthermore, the defences against non-compliance are themselves powerful means of enhancing resilience. It was noted above that the proactive assessment element of the toolkit was seen by users to be of significant value, such as for the assessment of planned changes. This is particularly relevant to organisational drift and resilience. One of the challenges with respect to drift is compensating behaviours – where staff adopt a ‘can-do’ attitude to overcome emerging challenges. The difficulty arising from this is that those compensating behaviours can mask the potential degradation in resilience. They are frequently indicated by deviations from agreed procedures, but by virtue of their nature are often not immediately apparent to the organisation. An example we have seen is where a change to a procedure gave rise to resource constraints that meant a particular task check became difficult to undertake. However, because the check was of a high-reliability system and had never revealed a fault, staff were not concerned that it was no longer undertaken. But the informal nature of this compensation meant that the organisation still believed that the check was being made, and claimed benefit from it in their safety assessment.

The way forward Our experience is that successful compliance management is a multi-faceted process. In particular, it requires both a proactive and reactive approach to identifying potential non-compliance, and then a balanced programme of solutions driven out from an understanding of goals, that cover both the design of tasks (in the broadest

154


sense) and how the organisation goes about its business. The key elements of compliance management appear to be: • • • • • • • • • •

Understanding individual goals Identification of clear and realistic goals, targets and benchmarks Focus on competence and supervision Consideration of incentives and sanctions (both intentional and unintentional) Provision of resources and equipment (including all aspects of task and job design) Application of consistent standards and procedures Effective investigation of concerns, successes, incidents and near-misses Effective arrangements for workforce involvement and change management Provision of effective review Accountability

There is no single ‘best’ method for improving rule compliance. However, the rewards of doing so are significant, in terms of improving organisational resilience and performance. Management of rule compliance is central to the domain of ergonomics, but must start from an understanding of the disparity between the goals of the individual and the organisation, which are then impacted by organisational issues, workload and ergonomics. By employing competent people, it is inevitable that there will be constant pressure towards non-compliance (albeit without malicious intent). This therefore requires constant attention to the management of non-compliance if the organisation is to survive. Most staff comply with arrangements if they can recognise the value and importance of doing so. All too often that value and importance is inadvertently hidden. Ergonomics is central to compliance management – but only if a broad perspective is maintained.

References Amalberti, R. 2006, Optimum System Safety and Optimum System Resilience. In: E. Hollnagel, D. Woods and N. Leveson (eds) Resilience Engineering. Aldershot: Ashgate Banbury, S. and Tremblay, S. 2004, Situation Awareness: A Cognitive Approach. Aldershot: Ashgate Berman, J. and Ackroyd, P. 2006, Organisational Drift – a challenge for enduring safety performance. Proceedings of Hazards XIX, Rugby: IChemE Berman, J., Ackroyd, P., Mills, A. and Davies, T. 2005, Management Toolkits – Solutions for Rule Compliance. In: J. Wilson, etc (eds). People and Rail Systems: Human Factors at the Heart of the Railway. Aldershot: Ashgate HFRG, 1995, Improving Compliance with Safety Procedures. London: HSE Klein, G. 1993 A recognition-primed decision (RPD) model of rapid decision making. In: G. Klein, J. Orasanu, R. Calderwood and C. Zsambok (eds) Decision Making in Action. New York: Ablex


155

McDonald, N. 2006, Organisational Resilience and Industrial Risk. In: E. Hollnagel, D. Woods and N. Leveson (eds) Resilience Engineering. Aldershot: Ashgate. RAIB, 2008, Derailment at Grayrigg 23 Feb 2007. RAIB Report 20/2008. Dept for Transport Royal Commission, 1999, The Esso Longford Gas Plant Accident: Report of the Longford Royal Commission, State of Victoria

HEALTH ERGONOMICS AND PATIENT SAFETY

PUBLIC REQUIREMENTS FOR PATIENT-HELD RECORDS J. Binnersley1 , A. Woodcock1 , P. Kyriacou2 & L.M. Wallace1 1

Coventry University, UK 2 City University, UK

Electronic patient-held record devices enable people to carry around information about their health. They need to be easily accessed in emergency situations and should provide enough information to enable front line staff to deliver effective care. Although it has been suggested that patients taking responsibility for their records could improve patient safety, little is known about attitudes towards this. In taking a user centred approach to the development of such a system, surveys were conducted with the public and health professionals. The survey of 258 members of the public showed that 85% thought patient-held records would be useful. The types of information that most people wanted to carry were name, allergies, long term conditions, current medication and blood group. A smart card design was preferred by 62% of the participants and 68% thought it should be provided free of charge.

Introduction In many areas of health care, professionals rely on patients or their relatives to provide information about medical history. Accurate information is vital for the safety and effectiveness of clinical decision making but many people struggle to provide this (for example because they are unconscious, have forgotten, or do not know or understand what information to provide). Although it has been suggested that patients taking responsibility for their records could improve patient safety (Hall, 2007), little is known about the UK public’s attitudes towards this or what the design and requirements of such a system would be. The study reported here is part of a wider Engineering and Physical Sciences Research Council funded project with City University to develop a patient held health record device which takes into account the requirements of the public and health care professionals. It will act independently of health service initiatives, being kept, carried and updated by the patient. This paper focuses on the first stage of the project, a survey undertaken to understand attitudes towards, and requirements for, such a device. The information will be used by product and software designers to create solutions which meet user requirements and acceptance criteria. Previous studies, such as Ross et al. (2005) have pointed to the interest of patients in having certain types of information available for privileged access. However, Pyper et al. (2004) highlighted concerns that unauthorised people might gain access 159

160

J. Binnersley et al.

to electronic records. These, coupled with worries about surveillance might mean that regardless of the benefits of such a system, it may not be accepted by the UK public. Additionally, there has been a lack of consensus about the types of information that should be entered, access rights, the design of the device, and input and display requirements. Electronic patient held record systems are currently used in many countries and many of these use smart cards. They are used as combined ID/health cards in Malaysia and Germany (Frontier Strategies, 2001). In the USA, devices being used include USB or memory sticks such as MEDeCARD and dog tags for the military (Burnham, 2000). In the UK a pilot study was conducted between 1989–1992 in Exmouth, where13,000 smart cards were used by patients (NHS Management Executive, 1990). Patients were able to restrict the information contained on the cards, but provision was made to include health care professional identifiers, ethnic group, language spoken, religion, medication details, allergies/drug sensitivities, donor status, emergency information, symptoms, examination findings, laboratory information, tests and measurements, diagnostic information, employment details and administrative data. Access to the information was available to health professionals. The main result of the trial was that the patients liked the concept of an electronic patient-held card and also suggested that more information should be included on the cards. The study was not continued, with the availability of technology at the time limiting its wider feasibility. The survey outlined below was undertaken with the aim of collecting information from the UK public about support for patient-held records, potential barriers to their use, the sort of information that should be held on a patient held health record, access rights and the user requirements for a device.

Method In order to determine the questions to include in the survey five focus groups were held, consisting of a total of twenty-five people. These were members of the public, designers, IT specialists, health professionals and people with chronic illnesses. The full survey was conducted in pharmacies in London and the Midlands. Pharmacies were selected to include participants from a range of ethnic backgrounds, geographical locations and social classes. Customers over the age of 16 were invited to participate in the study and were given verbal and written information. Only those who signed the consent form were included. Some participants with poor eyesight and levels of English literacy were assisted in completing the questionnaire and 258 questionnaires were returned.

Results Participant characteristics The age range of the participants was 17–89 years, with a mean age of 45 years. 43% of the participants were male, and approximately two thirds were of white

Public requirements for patient-held records

161

European origin. The largest occupational group was white collar professionals (27%). 61% lived in the Midlands and 39% in London. 26% considered themselves to have a long term health problem and 13% had already experienced being given part of their health records to look after. The most common form of these had been paper records, including letters and repeat prescriptions. The most commonly used paper records were maternity records and the most frequently cited advantage of keeping records was being able to share information with health professionals, stated by eight participants. The most frequently listed disadvantage was the possibility of losing them, mentioned by six participants.

Support for patient-held records 85% of the participants said they would find a patient-held record device useful. The most common circumstance where participants thought a patient-held record would be useful was if they were too ill to give information to a health professional. The most common concern was that unauthorised people would gain access to the information and 64% of the participants were worried about this. Participants were provided with a list of items and asked whether they thought these should be included on a patient-held device. The answers given in the table below represent positive responses for each item. The results show that the items most participants thought should be on patientheld records were current medication, name, allergies, blood group and long-term conditions. Few participants could think of other items that they would want included but nine specifically stated that they would not want their address on a device. Most participants thought that all health professionals should be able to see these items. 81% of the participants stated they agreed or strongly agreed that health professionals should be able to access different levels of information according to their profession. Seven participants specifically stated that they thought pharmacists should be able to access some information and 27 participants stated that non-medical personnel should not be able to see the information.

Table 1.

Information to be included on a patient-held record device.

Item Current medication Name Allergies / intolerances Blood Group Long-term conditions Next of kin Age Previous major health problems NHS number Address

Percentage agreement 92 92 92 91 89 88 86 75 75 69

Item Disabilities Dietary information Living will and donor information Implants Carer contact Religion Information about social services care Non-prescribed medication Ethnicity

Percentage agreement 63 59 55 54 50 42 39 34 34

162

J. Binnersley et al. 70 60 50 %

40 30 20 10 0 Smart card Key fob

Jewellery

Others

Device Computer memory linked to a mobile stick phone

Figure 1. Answers to question about preferred form of device.

When asked to indicate whether they agreed or disagreed with statements regarding their attitudes towards patient-held record devices, the majority of participants thought it would be good to use a patient-held record device (77%), that people who were important to them would approve of this behaviour (64%), that they would find such a device easy to use (82%) and would intend to use one if such devices were available (76%). Most participants wanted themselves or their doctor to be able to update the information. The majority (85%) wanted to be able to update information at their doctor‘s surgery and to use a keyboard form of entry (71%). When asked what type of device they would be most likely to use, the most popular was smart card, ticked by 62% of the participants who answered the question. Figure one illustrates the device preferences. When asked how much they would be willing to pay for such a device, 157 (68%) of the participants indicated that they thought it should be provided free of charge. Only seven people were willing to pay over £20.

Discussion The present survey replicates the findings of Ross (2005) who found that shared medical records were almost universally endorsed across a broad range of ethnic and socioeconomic groups. Pyper et al. found a high level of support for electronic records, but also concerns about unauthorised people gaining access to the data. Designers will need to consider using personal access codes in order to keep the data as secure as possible. They will need to address what happens when the card is stolen, misplaced or inaccurate information is entered and the consequences of this. Most members of the public would like a wide range of information held on the device, with the most commonly cited being current medication, name, allergies,

Public requirements for patient-held records

163

blood group and long term conditions. The NHS Management Executive study (1990) found that patients tended to want more information to be included after they had become used to having the device. However, this may not be the case if the onus falls on the patient to enter and regularly update such information. This will be a major consideration in the design stage, as ease of use will significantly affect regularity of use. Liaw’s (1993) study found that most participants preferred a credit card or walletsized patient-held record device. Similarly, in the current study, most participants preferred a smart card design. Advantages of this over other design forms are portability, familiarity and cost. A ‘card’ format may also be more acceptable as an input device (e.g. as opposed to a piece of jewellery) as it can be accommodated more readily in existing readers. The card itself can also be a carrier of information (e.g. a picture of the patient, life threatening illnesses and allergies). The design and the relationship between this and the information carried in the software is being addressed through development of a prototype. The survey had several limitations. Some potential participants declined to take part because they did not see the advantage of such a system for them or because English was not their native language. The study did not include consideration of how patient-held records could be managed for children or adults who would be unable to take responsibility for their own records. None of the participants in this study were under the age of sixteen. It is likely that these groups may derive greater benefit from a patient-held record system. Lastly, some participants found it difficult to imagine such a device (and system) without seeing a prototype. Future work could therefore address increasing the public‘s confidence in the security of databases and awareness of the benefits of patient held health records. One way of achieving this will be through the development of working prototypes which will enable members of the public to achieve a greater understanding of the costs and benefits of such a system. For example, costs in terms of data entry, remembering to carry the card and update it. The team are also mindful that only patients have been considered. Without the acceptance and use by health care professionals, a patient held health records system will be redundant. Therefore a similar survey is currently being conducted with this group.

Acknowledgements The research has been supported by an Engineering and Physical Sciences Research Council grant number EP/F00323411 28/02/2007. Lloydspharmacy permitted data collection on their premises.

References Frontier Strategies, 2001, Smart cards for smart customers (Frontier Strategies, Canada) [online] Burnham, T. 2000, Dog Tag Health Card; Soldiers Dec 2000

164

J. Binnersley et al.

Hall, J. 2007, What roles for the patient in patient safety research? In conference proceedings Patient Safety Research: Shaping the European Agenda, 24–26 September 2007, Porto, Portugal [online] Liaw, S.T. 1993, Patient and general practitioner perceptions of patient-held records. Family Practice 10, (4) 406–415 NHS Management Executive. 1990, The Care Card: Evaluation of the Exmouth project. (HMSO, London) Pyper, C., Amery, J., Watson, M. and Crook, C. 2004, Patients‘ experiences when accessing their on-line electronic patient records in primary care. British Journal of General Practice 54, (498) 38–43 Ross, S.E., Todd, J., Moore, L.A., Wittevrongel, L. and Lin, C.T. 2005, Expectations of patients and physicians regarding patient-accessible medical records. Journal of Medical Internet Resources 7, (2) 1–32

SYSTEMS ANALYSIS FOR INFECTION CONTROL IN ACUTE HOSPITALS P.E. Waterson Department of Human Sciences, Loughborough University, Loughborough, LE11 3TU, UK This paper makes the case for applying a systems perspective to the analysis of hospital-based infection outbreaks. Most of the research that has been conducted on behavioural aspects of infection control has focused on explanations at an individual level of analysis (e.g., interventions to improve hand washing). The infections outbreaks at the Maidstone and Tunbridge Wells NHS Trust are analysed in detail using an established framework for risk management. The paper further outlines the human and organisational issues raised by the analysis and provides a means through which these aspects of infection can be highlighted as part of a future research agenda within systems ergonomics.

Introduction Within the last few years the subject of hospital infection control has become the subject of much media attention (e.g., BBC Panorama, 2008). A number of high profile hospital outbreaks within the UK involving bacterium such as Clostridium difficile (C. diff.) and MRSA (Methicillin-resistant Staphylococcus aureus) and the number of mortalities resulting from these outbreaks, has made infection control into a central priority for the UK NHS and other health care systems worldwide (Allegranzi et al., 2007). Much of the debate so far has concentrated on improving hygiene within hospitals (e.g., hand washing), very little research has been conducted on the wider behavioural, social and organisational factors that may also determine infection control outbreaks (Griffiths, Renz and Rafferty, 2008). The intention of the present study is to consider the potential of adopting a systems ergonomic perspective towards hospital-based infections. More specifically, the paper describes the advantages to be gained from applying existing systems analysis techniques to infection outbreaks and using these to draw out both lessons learnt as well as strategies for improvement. In order to demonstrate the value of such an approach the paper focuses on a specific case study namely, the Maidstone and Tunbridge Wells NHS Trust outbreaks which occurred between 2005–2007 (Healthcare Commission, 2007). 165

166

P.E. Waterson

Systems analysis and systems ergonomics Hospitals represent a good example of a complex large-scale sociotechnical system involving a large diversity of professions spanning a range of roles and specialisms as well as technologies that range from the latest eHealth applications (e.g., electronic records) to more established aspects such as physical design components (e.g., wards and buildings). Within systems ergonomics, Rasmussen (1997) has provided a modelling framework for understanding the dynamic interaction between levels within large-scale sociotechnical systems. Vicente and Christoffersen (2006) have used the modelling framework to identify the lessons learnt from the May 2000 outbreak of E. coli which occurred in Walkerton, Canada. Their analysis used a graphical representation of the contributing factors that led up to the Walkerton outbreak. These ranged from decisions made at governmental levels (e.g., privatisation initiatives), the action of actors within the system (e.g., failures to take water samples), as well as equipment failures (e.g., shallow water wells). In the next section the main events leading up to the outbreaks at Maidstone and Tunbridge Wells are outlined. These are used to suggest ways in which the outbreak could be analysed in more detail, and modelled using the Rasmussen and Vicente frameworks.

The Clostridium difficile outbreaks in the Maidstone and Tunbridge Wells NHS Trust Background to the outbreaks and timeline During the period between April 2004 and September 2006 an estimated 90 people died at the Maidstone and Tunbridge Wells NHS Trust as a result of becoming infected with the Clostridium difficile (C. diff.) bacteria (Healthcare Commission, 2007, p. 5). C. diff. is the major cause of serious bacterial infectious diarrhoea acquired in hospitals in the UK and is particularly resistant to drying, chemical disinfectants and alcohol. The main events at the Maidstone and Tunbridge Wells NHS Trust are summarised in table 1.

Contributory factors leading up to the outbreaks The Healthcare Commission (2007, HC, 2007) report identified a number of factors that contributed to the outbreaks that occurred with the Trust. These can be summarised in terms of five main themes: the role played by external organisations; management of the trust; clinical management on the hospital wards; the role played by the infection control team; and, equipment and hygiene factors.

The role of external organisations Within the report both the setting of government-led targets and financial pressures on NHS Trusts are mentioned as background, contributory factors that had an impact on the day-to-day operation of the Maidstone and Tunbridge Wells Trust. In particular, the report mentions the need for Trust board members and managers

Systems analysis for infection control in acute hospitals

Table 1. Time period

167

Summary timeline of infection outbreaks.

Event

April 2000 2001/2–2005 October 2005– September 2006 Before 2005

Trust established following merger between two other local NHS Trusts High turnover of senior managers and period of organizational stability More than 500 patients developed the infection, 60 patients estimated to have died due to C. difficile infection. Trust has a high level of infection with C. difficile but no one in the trust or local health authority was aware of this Autumn 2005 Number of patients infected doubles. Approximately 150 patients affected, a number of whom died as a result of the infection. (first outbreak) April–Sept. 2006 258 patients in total affected Beginning 2006 Number of newly infected patients declines. April 2006 Trust recognizes it has a major outbreak and reports this to strategic health authority and health protection unit. (second outbreak) April 2007 Healthcare Commission finds unacceptable examples of the use of contaminated equipment October 2007 Healthcare Commission Report published.

to meet targets for the use of beds. Higher bed occupancy meant that there was less time for the cleaning and a higher probability of transmission of infection between patients (HC, 2007, pp. 69–70). The need to meet financial targets such as spending on equipment and buildings also placed pressure on the Trust to cut back in areas that impacted upon infection control such as financing for new buildings and isolation areas. Infection control within the UK NHS is regulated by a number of bodies including the Health Protection Agency (HPA). The remit of the HPA is to provide advice and support to NHS, local authorities and other agencies with regard to public health issues. The creation of the HPA in April 2005 coincided with the first outbreak at the Trust. One part of the HPA, the health protection unit (HPU), was set up in order to support organisations in their management of infections. The report highlights that this caused some confusion within the Trust at the time of the outbreaks, as the expectation was that the HPU could give provide guidance covering the supervision and monitoring of infection control. The HPU did not have close involvement with the Trust and generally worked in a reactive way, responding to concerns as they arose (HC, 2007, p. 8). Similar problems were encountered within the much larger Strategic Health Authority (SHA) who are responsible for implementing government policy and fiscal control within regions of the UK.

Management of the Trust The report describes a catalogue of problems and failures associated with the management of the trust at the time of the outbreaks. In terms of clinical risks and incidents, management strategy in general “had been fragmentary and poorly understood” (HC, 2007, p. 77). The reports from an internal group set up within

168

P.E. Waterson

the Trust in order to analyse complaints, claims and incidents highlight, amongst others, the following issues: the unsatisfactory nature of some “escalation” areas (areas temporarily set up to deal with infected patients); the impact that the accident and emergency (A&E) target had on the quality of care; poor quality handover and transfer to wards from A&E; concerns about staffing levels, and, bank staff managing wards on some shifts. The style of leadership within the Trust and the overall management culture are criticised in the report. Many staff described the leadership of the chief executive as being “autocratic” or “dictatorial” (HC, 2007 p. 91). The report concluded that the person appointed as director of infection prevention and control had “no real understanding of the role at the outset” (HC, 2007, p. 5). Turnover of managers and directors was also high. Finally, the trust’s management of staffing is criticised heavily within the report in several places. The number of nurses working on wards had fallen since the period 2002/03 and at the same time the number of beds had also reduced. In 2006/07 the number of nurse per bed was 1.52, the same number as in 2003/04 (HC, 2007, p. 82). Trust managers had not carried out a comprehensive review of staffing levels or a determination of minimum staffing levels.

Clinical management on the hospital wards A review of the case notes of 50 patients who had died having had C. diff. found that in 80% of the cases, at least one element of the clinical management, or monitoring of C. diff at ward level was unsatisfactory (HC, 2007, p. 4). A number of elements were mentioned, including: infrequent reviews of patients by doctors; lack of systematic monitoring as to whether or not a patient was recovering from C. diff.; and, failure to change antibiotic treatment when a patient failed to respond to the initial treatment (HC, 2007, p. 4). Delays in starting treatment occurred on the wards, mostly because there was a delay in sending samples for analysis (HC, 1007, p. 33). There was also little evidence that once C. diff. had been diagnosed, that patients were monitored for severe signs of the infection (HC, 2007, p. 34). In other cases, it was clear that diagnoses were either not considered or had been missed. In 34% of the cases reviewed, medical records did not indicate that a regular review of C. diff had taken place (HC, 2007, p. 38). The management of fluids and nutrition on the wards was also inconsistent. In 36% of the cases there was evidence of poor fluid management and in 34% nutritional needs had not been assessed or managed (HC, 2007, p. 38).

The infection control team The role played by the infection control team within the trust was a complex one and one made difficult by problems relating to accountability, the amount of resources available to them and their ability to function as a team. The arrangements for accountability were not clear (HC, 2007, p. 54) and it was not clear who was responsible for the team. Infection control nurses were accountable to the director of nursing, however, the pathology manager held the budget for these nurses, but did not consider that he had any management responsibly for infection control. Not


169

until September 2006 did the trust take steps to clarify the management of the team (HC, 2007, p. 51).

Equipment and hygiene Hygiene practices within the trust and the state of hospital buildings contributed a great deal to the outbreaks. Wards, bathrooms and commodes were not clean and patients had in some cases to share equipment (e.g., Zimmer frames) which were not cleaned before use (HC, 2007, p. 4). The infection control team were keen to isolate patients once they had been identified as C. diff. cases, however the scarcity of side rooms made this difficult. As a result many patients before and after the outbreaks were kept on open wards. The design of buildings and their age meant that many wards did not have sufficient space for storage or the provision of hand basins in utility rooms. The buildings in the trust were generally old or in a poor state of repair and when they were first opened did not have adequate cleaning and laundry services (HC, 2007, p. 6).

Analysing the outbreaks: A systems perspective The outbreaks which occurred within the Maidstone and Tunbridge Wells Trust represent the combined impact of a complex set of factors extending over several years. In common with most examples of accidents, disasters or large-scale adverse events, the outbreaks are best interpreted as arising through the combination of a number of interrelated systemic factors and influences (Turner, 1978; Reason, 1995). Figure 1 attempts to use some of the elements of Rasmussen’s (1997) risk management framework in order to further analyse the outbreaks. In order to illustrate the framework as it applies to the outbreaks, a small sample of the contributory factors are used to link together some of the system components. In the following sections, the various levels and the boundaries between them, are examined in more detail.

Government, regulatory bodies and trust governance At the very highest level of the system it is difficult to isolate the role played by government-set targets as a contributory factor leading to the outbreaks. Targets placed many individuals, particularly those at trust board and management levels under a great deal of pressure. This pressure in itself may have led them to make poor decisions, and in some cases to prioritise bed occupancy rates at the expense of the risk of an infection outbreak. Previous research on the influence that targets have on management decision-making in health care tends to be equivocal. Bean and Hood (2006) for example, show that the impact of satisfying a specific target (e.g., hospital waiting times) has not been analysed in terms of how this influences other related services (e.g., quality of care). Others have suggested that health care targets represent: “tin openers rather than dials … they do not give answers but prompt further investigation and inquiry, and by themselves provide an incomplete and inaccurate answer” (Carter, Klein and Day, 1995). Within the trust it is likely

170

Figure 1.

P.E. Waterson

Mapping of infection outbreak influencing factors with system levels and boundaries.

that targets exerted considerable pressure on the system as a whole and this pressure filtered down various levels of the system. It is possible that the drive to comply with these targets increased the likelihood of an adverse event or set of events taking place at some stage within the trust. Poor communication, confusion of responsibilities and accountabilities between and within the various regulatory bodies delayed the time in which they could react to the outbreaks. A separate report by the Healthcare Commission (2008) examined the underlying causes of serious failures in NHS health care providers and identified large-scale organisational processes such as mergers and poor change management procedures as common factors. Within the wider literature on disasters (e.g., Perrow, 1999) the nature of organisational linkages and structures are also widely acknowledged to be significant explanatory factors.

Hospital management Within the hospital the actions of senior managers were identified as significantly contributing to the failure to prevent and deal with the outbreaks. The link between management, human resource management (HRM) practices and work performance outcomes has been investigated in detail in the last few years. Wood and Wall


171

(2002) for example, reviewed the evidence that suggests there is a link between highinvolvement HRM practices and employee productivity. High involvement HRM practices typically include empowering employees to make their own decisions and the presence of self-managed teams. The review showed that there these types of practices in organisations do tend to increase levels of employee productivity. Similar effects have been shown between HRM practices and measurements of safety outcomes (e.g., number of adverse events). In general, there is strong evidence to suggest that aspects of management behaviour partially shape and determine the culture of safety within organisations (e.g., Zohar, 2000). Within health care specifically, West et al. (2002) carried out a large-scale survey of the relationship between HRM practices and general in-hospital mortality. The survey showed some aspect of high involvement HRM were associated with lower mortality rates after adjustment for patient and hospital characteristics. Aside from the way in which senior managers behaved at the trust, the questions still remains as to why they ignored, or at least failed to realise the seriousness of the outbreaks and their consequences. Many of the managers interviewed in the original Healthcare Commission report reported that they were aware of how serious the situation had become within the trust, but were powerless to do anything about it. One possible explanation is what Vaughan (1996) in her study of the Challenger shuttle disaster termed the “normalization of deviance”, namely that managers over time began to accept and take for granted the level of infection risk within the Trust. Only after the level of risk built up to a point where it could not be controlled, did they begin to realise the gravity of the situation.

Clinical management and equipment and buildings Understaffing and general lack of resources together played a part in the outbreaks. Staffing ratios and levels of staff morale almost certainly contributed to the problem of containing the spread of infection on the wards. In general, the research literature provides some evidence that lower levels of staffing increase the likelihood of infections occurring. Hugonnet et al. (2004) (cited in Griffiths et al., 2008) examined the numbers of nursing staff and staff downsizing relative to infection levels. The researchers found an inverse relationship between staff downsizing and the rate of hospital-based infection. Curiously, little research has been conducted on the impact of job satisfaction/morale on hospital infection levels, however, work in other domains (e.g., manufacturing and service industries) suggests that lower levels of satisfaction are clearly linked to lower levels of job performance (e.g., Parker, 2007). It might be conjectured that the behaviour of clinicians and other health care professionals within the trust shares similarities with those of senior managers and trust board managers. Many individuals at ward level were aware of the levels of poor hygiene and inadequate patient monitoring practices, but saw no way to improve the situation. Weick and Sutcliffe (2003) analysed data from the Bristol Royal Infirmary Report (2001) and concluded that hospital staff became locked into particular lines of action or behaviour where they “search for confirmation that they are doing what they should be doing” (p. 73). These so-called “cultures of

172

P.E. Waterson

entrapment” inhibit an organisation’s ability to break out of patterns of behaviour that over time can lead to adverse outcomes. In the case of the trust they may provide some means with which to explain shared boundary spanning behaviours between levels within the hospital subsystem (figure 1).

Ways forward and conclusions The analysis presented in the paper has shown that there are advantages in analysing hospital-based infection outbreaks from a system’s perspective. Many of the issues that have been discussed have not been researched within the patient safety literature in much depth, particularly organisational phenomena (Waterson, 2008). It is still the case that most research within the area of infection control has concentrated on individual levels of analysis (e.g., behavioural interventions in the form of hand washing campaigns). The paper has only touched upon some of the behavioural issues involved with a system as complex as hospital-based healthcare. Future research should examine causal links between system levels in a more systematic manner and address the issue of influences at the meso-level of analysis, that is linkages and relationships between individual (macro-) and organisational (macro-) system levels (Karsh and Brown, in press). Further empirical studies which focus on behavioural aspects of infection control which go beyond individual levels of analysis are needed. In addition, there appears some scope for modelling the relationships between system components and comparing these models against hospital-based data (e.g., Brailsford and Schmidt, 2003).

References Allegranzi, B. Storr, J., Dziekan, G., Leotsakos, A., Donaldson, L. and Pittet, D. 2007, The first global patient safety challenge “Clean Care is Safer Care” from launch to current progress and achievements. Journal of Hospital Infection, 65 (supplement 2): 15. Bean, G. and Hood, C. 2006, Have targets improved performance in the English NHS? British Medical Journal, 32, 419–422. BBC Panorama 2008, How Safe is Your Hospital? (Broadcast on 27th April, 2008), (http://news.bbc.co.uk/1/hi/programmes/panorama/default.stm, accessed 28th April, 2008) Brailsford, S. and Schmidt, B. 2003, Towards incorporating human behaviour in models of health care systems: An approach using discrete event simulation. European Journal of Operational Research, 150, 19–31. Carter, M., Klein, R. and Day, P. 1995, How Organisations Measure Success: The use of Performance Indicators in Government. (London: Routledge). Department of Health 2001, Learning from Bristol. The Report of the Public Inquiry into children’s heart surgery at the Bristol Royal Infirmary 1984–1995. (London: HMSO). Griffiths, P., Renz, A. and Rafferty, A.M. 2008, The impact of organisation and management factors on infection control in hospitals: a scoping


173

review. Kings College, London. (http://www.kcl.ac.uk/content/1/c6/04/08/39/ Infectioncontrolreport.pdf, accessed 15th September, 2008). Healthcare Commission 2007, Investigation into outbreaks of Clostridium difficile at Maidstone and Tunbridge Wells NHS Trust. (http://www. healthcarecommission. org. uk/_db/_documents/Maidstone_and_Tunbridge_ Wells_investigation_report_ Oct_2007.pdf (accessed 8th April 2008). Healthcare Commission 2008, Learning from Investigations (http://www. healthcarecommission.org.uk/_db/_documents/Learning_from_investigations. pdf (accessed 17th September 2008). Huggonet, S., Harbath, S., Sax, H., Duncan, R.A. and Pitter, D. 2004, Nursing resources: a major determinant of nosocomial infection? Current Opinion in Infections Diseases, 17, 4, 329. Karsh, B-T. and Brown, R. in press, Macroergonomics and patient safety: the impact of levels on theory, measurement, analysis and intervention in medical error research. Applied Ergonomics. Parker, S.K. 2007, Job satisfaction. In S. Rogelberg, (Ed.) The Encyclopaedia of Industrial and Organizational Psychology. (New York: Sage). Perrow, C. 1999, Normal Accidents. (Princeton: Princeton University Press). Rasmussen, J. 1997, Risk management in a dynamic society: a modelling problem. Safety Science, 27, 183–213. Reason, J. 1995, A systems approach to organizational error. Ergonomics, 38, 1708–1721. Turner, B.A. 1978, Man-Made Disasters. (London: Wykeham Publications). Vaughan, D. 1996, The Challenger Launch Decision: Risky Technology, Culture and Deviance at NASA. (Chicago: Chicago University Press). Vicente, K.J. and Christoffersen, K. 2006, The Walkerton E. Coli outbreak: a test of Rasmussen’s framework for risk management in a dynamic society. Theoretical Issues in Ergonomics Science, 7, 2, 93–112. Waterson, P.E. 2008, Taking stock of the systems approach to patient safety. In S. Hignett, B. Norris, K. Catchpole, A. Hutchinson and S. Tapley (Eds.), Improving Patient Safety. (Loughborough: Ergonomics Society). Weick, K.E. and Sutcliffe, K.M. 2003, Hospitals as cultures of re-enactment: a re-analysis of the Bristol Royal Infirmary. California Management Review, 45, 2, 73–84. West, M.A., Borrill, C., Dawson, J., Scully, J., Carter, M., Anelay, S., Patterson, M. and Waring, J. 2002, The link between the management of employees and patient mortality in acute hospitals. International Journal of Human Resource Management, 13, 8, 1299–1310. Wood, S. and Wall, T.D. 2002, Human resource management and business performance. In P.B. Warr (Ed.), Psychology at Work (5th edition). (Harmondsworth: Penguin Books). Zohar, D. 2000, A group-level model of safety climate: Testing the effect of group climate on microaccidents in manufacturing jobs. Journal ofApplied Psychology, 85, 587–96.

PORTABLE PODS: DESIGN FOR UNSCHEDULED (URGENT) CARE IN THE COMMUNITY S. Hignett1 , A. Jones1 & J. Benger2 1

Healthcare Ergonomics and Patient Safety Unit, Department of Human Sciences, Loughborough University, UK 2 Academic Department of Emergency Care, University of the West of England, Bristol, UK In the last five years UK government policy has moved towards increasing the provision of unscheduled (urgent) care in the community. To meet these changes the emergency ambulance service is shifting from an organization designed to convey patients to hospital to a professional group capable of assessing urgency and delivering the appropriate treatment to the patient; ‘providing the right response, first time, in time’. This paper explores modular design for supporting (portable) technology and draws on examples from emergency care practitioner services (Ambulance Trusts), out-of-hours GP services and Minor Injuries Units (Primary CareTrusts), and Emergency Departments (Acute Trusts) to investigate the workplace layout and clinical activities, including the use of equipment and consumables.

Introduction In May 2004 the Department of Health (DH) commissioned a strategic review of National Health Services (NHS) ambulance services in England, focusing on how the ambulance service could shift from providing resuscitation, trauma and acute care towards “Taking healthcare to the patient: transforming ambulance services in the community” (DH, 2005a). The strategic review set out a five-year vision to: • Improve the speed and quality of call handling, ‘hear and treat’. • Increase the range of mobile healthcare services for patients requiring urgent care, ‘see and treat’. • Increase the range of other services such as diagnostics and health promotion. • Improve the quality and speed of service to patients with emergency care needs. In 2007/08 there were 7.2 million emergency and urgent ambulance calls in England, with 81% (5.9 million) resulting in an emergency response and 4.26 million patient journeys from 12 ambulance Trusts (NHS Information Centre, 2008). Before 1st April 2007 emergency and urgent calls were prioritized and classified into four categories: • A, immediately life threatening (8 minute response); • B, serious but not life threatening (19 minute response);

174

Portable pods

175

• C, not immediately serious or life threatening (locally agreed response time standards); • Urgent, usually in response to a GP/midwife/health care professional request for transportation via a separate phone line (NHS Information Centre, 2008). After 1st April 2007 the urgent calls have been included in the emergency call system to ensure that all patients are triaged within the same response time target framework (DH, 2005a). Urgent calls are usually classified as a category C, accounting for 40.4% of all calls in 2007/08. The response standard was further changed from 1 April 2008 (‘Call Connect’), with the clock start changed to the time of the call connection to the ambulance control room rather than when key information had been obtained from the 999 caller, with the aim of increasing survival rates for cardiac arrest patients (DH Ambulance Policy, 2008). Urgent Care is defined as ‘the range of responses that health and care services provide to people who require – or who perceive the need for – urgent advice, care, treatment or diagnosis. People using services and carers should expect 24/7 consistent and vigorous assessment of the urgency of their care need and an appropriate and prompt response to that need’ (DH, 2005b). In 2007 the Royal College of General Practitioners commented that the organization of urgent and unscheduled care services needed to improve to address the fragmentation of care services including duplication of provision, wasted resources and unnecessary handoffs between providers. They suggested the adoption of a ‘whole systems approach’ to ensure integration of services and the creation of virtual urgent care centres for different providers to work together (Lakhani et al., 2007). Emergency and Urgent Care Networks (EUCNs) were set up to co-ordinate and organize a diverse group of primary, secondary, out of hours (ambulance) and social services. The care board structure included a range of stakeholders from acute, primary care, ambulance, NHS Direct, mental health (crisis teams), social care, and patient representation. Some EUCNs also included police, passenger transport executive, local authority, public health, health promotion, community pharmacy (Turner et al., 2008). Emergency and urgent care services are faced with the triple hurdle of delivering a service that is more responsive, more resource efficient, and that also employs the latest medical technologies (Phillips and Caldwell, 2007). In 1999 the role of Emergency Care Practitioner (ECP) was developed in the Ambulance Service. This role aimed to raise the clinical skills of paramedics, ensuring that patients receive the right care at the right time, and in the right place (DH, 2004). Snooks et al., (2004) reviewed the early operational activities of ECPs and found no significant difference in the rate of conveyance, but that a longer time was spent with patients, generating a more in-depth assessment with more comprehensive clinical records. However by 2007 a different picture was emerging, with Mason et al. (2007) finding that ‘overall ECPs carried out fewer investigations, provided more treatments and were more likely to discharge patients home than usual providers’, suggesting that the increased clinical skills were achieving the goal of delivering the appropriate treatment to the patient. In order to achieve the vision of a modernized workforce that is able to provide a greater range of mobile urgent care, ambulance clinicians will need a wider range

176

S. Hignett, A. Jones & J. Benger

of competencies and underpinning knowledge whilst maintaining the vocational nature of their training (DH Ambulance Policy, 2008). Many Trusts are changing the profile of their workforce in line with the recommendations including consideration of the interface between people, processes and technology using a sociotechnical framework (DH, 2005a). However an initial review of the changed response/service model has identified potential risks including poor state of fleet and equipment and poor systems for integrating community responders (DH Ambulance Policy, 2008).

Project aim The vision for this project was to support the effective delivery of urgent care in the community by developing a system of configurable, rapidly exchangeable pods ready-loaded with equipment and consumables (portable pods) to suit various types of presenting complaints; attached to base vehicles (mobile pods) for delivery to the point of care at a variety of locations. It was proposed that pods could be exchanged, or remain with the vehicle; used as multiples for major incidents with a view to broadening their function beyond the current emergency mode of use. Figure 1 shows the conceptualization of the project, with a central focus on clinical activities (in the ergonomics work package) to: 1. Understand and identify current and future care activities in emergency departments, minor injury units, ambulances and out-of-hours GPs which could be delivered in the community. 2. Define the configurations for the portable and mobile pods. A previous paper (Jones et al., 2008a) described the process for stakeholder participation to identify and categorise clinical complaints for treatment and care in the home/community by ECPs as physical minor, physical uncertain, physical major, social, mental and elective. A set of 6 presenting complaints were selected for further study (JB): breathing difficulties (physical minor); chest pain (physical uncertain); lacerations (physical minor); falls (physical uncertain/social); neck pain

Figure 1.

Project overview.

Portable pods

Table 1.

177

Participants at 2007 workshops (n = 22).

Organisation

Participants (job title)

Acute

Emergency Department (ED) Registrar; ED Charge Nurse; Emergency Nurse Practitioner (2); Research Nurse GP; PCT Chair; Out-of = Hours Nurse Practitioner; MIU Nurse Practitioner (2); Medical Director (Deputy CEO); Strategic Lead Occupational Health Services; Nurse Advisors (2); Change Manager Assistant Director of Corporate Services; Fleet Manager; Service Improvement Manager; Back Care Advisor; Emergency Care Practitioner (3)

Primary Care

Ambulance

(physical minor); head injury (physical minor) (Jones et al., 2008b). This paper will describe the iterative process used to develop the design specification for portable pod through stakeholder workshops, responder bags audits, observations and design decision groups.

Method Data were collected using stakeholder workshops, audits and clinical observations. In 2007 two workshops were held with 22 stakeholders participating from acute, community and ambulance NHS Trusts (East Midlands and South West). They identified categories of complaints and lists of equipment and consumables for the portable treatment pods. In 2008 a further workshop was held with fleet (4), clinical (5), service (4) and health and safety managers (2) from 5 Ambulance Trusts to present the findings of the 2007 workshops, audits and observations. Data were collected as a series of semi-structured questions in individual workbooks, for example: • How might medical equipment used by the ambulance service be: 1. Better organised, laid out and accessed in vehicles? 2. Effectively rationalised to provide what is needed without carrying excessive or unnecessary stock? • How do you see the role of portable kit and diagnostics changing: 1. By 2015 2. By 2025? The written responses were entered into NVivo (Bazeley and Richards, 2000) for analysis. NVivo is a qualitative data management programme that support coding, searching and theorizing. An audit of portable equipment (used by Fast Responders/Paramedics) was carried out in 2002 (Redden, 2002). This was updated in 2008 for ECPs (Reynolds, 2008). Observational data were collected from 84 patients at EDs and Minor Injuries Units presenting with the 6 complaints identified from the 2007 workshops (Jones et al., 2008a). Data were recorded with Link Analysis (Jones et al., 2008b) with supplementary field notes for equipment and consumables requirements.

178

Table 2.


Observational data for equipment and consumables from EDs and MIUs.

Treatment groups

Equipment/Consumables

Minor wounds

Dressing pack, Irrigation fluid, forceps, scissors, tissue glue, steristrips, suture kit (sutures and instruments), Ear, nose and throat Tongue depressors, thermometer (tympanic), suction, auroscope Respiratory Oxygen, masks (including tracheotomy), stethoscope, pulse oxymeter, nebuliser, nasal and oral airways, peak flow meter, suction kit Blood monitoring Specimen bottles, phlebotomy kit, blood pressure cuff, cannula (various), giving sets, blood glucose testing strips, Eyes Irrigation fluid, eye pads Basic life support Defibrillator, oxygen, face mask, electrocardiogram (ECG) capability (computerized transmissible) Communications: Mobile telephone, tele-medicine capability, on-line decision support software Urinary Urinalysis kit, sample bottles, catheter equipment, incontinence pads General Gloves (sterile/non sterile), neck collar, bandages, gauzes, dressings, waste bins (sharps, clinical, domestic), sphygmomanometer, skin preparation wipes, apron, handwash facilities, tissues, syringes, needles, lubricant, magnifying glass, razor, tweezers, scissors, referral letter/x-ray/prescription (prescribing guidelines), patella hammer, ring cutter, safety glasses, helmet Drugs Glyceryl trinitrate (GTN) spray, aspirin, intravenous fluids, paracetamol, anti-inflammatories, antibiotics, local anaesthetics

Table 3.

Problems and suggested design solutions for the portable pod (workshop 2008).

Problems/Barriers Individual preferences resulting in duplication, different equipment (and quantities) and different locations Weight and size of bags for carrying (stairs, long distances) Infection control risks

Design ideas/solutions Standardisation of contents and layout Organise into clinical ‘jobs’ (cardiac, breathing difficulties) Miniaturise and modularise diagnostic equipment (reconnect to base station in vehicle, ‘wi fi’) Smaller bags (‘snatch/grab’ bags) Disposable bags for individual patients Equipment in safe, sterile, disposable containers

Results The 2007 workshops produced a list of equipment and consumables, grouped by clinical criteria. These were compared with the audits and the observational data to give a proposed generic list of equipment and consumables (table 2). The workshop in 2008 reviewed the findings. The analysis of the workbooks resulted in a number of thematic codes including problems/barriers and design ideas/solutions for the portable pod (table 3).

Portable pods

Figure 2.

179

Design idea for the portable pod.

These findings will be reviewed at Design Decision Groups (DDGs) to be held in November/December 2008 with operational staff (ECPs and Fast Response Paramedics) and management/staff representatives (Equipment Working Group) from EMAS. The DDG format will both challenge current practice (with word maps and round robin questionnaires) and support innovative re-design through mock-ups/prototypes (Wilson et al., 2005). Staff will be given an opportunity to prepare bags for specific care pathways. These will be reviewed and discussed in small group sessions. The first session will finish with a drawing exercise to create portable pods with improved functionality and usability. For the second session, prototypes (example in figure 2) based on the outputs from session one will be manufactured and used as the focus for the discussion and modification of the design specification.

Discussion The concept of modularisation has been called the goal of good design (Gershenson et al., 1999), with benefits including ease of product updating, increased product variety, and ease of design and testing. Modularisation has also been described as a ‘general systems concept’ (Schilling, 2000), with modular products as ‘systems of components that are ‘loosely coupled’. The challenge for the design of modular products/systems is to maximise flexibility, with staff being willing and able to distinguish between the performance, quality and value attributes of different components (e.g. equipment and consumables). Poland et al., (2005) questioned whether interventions (e.g. standardized equipment/consumables) would necessarily bring benefits. They suggested that professional work is ‘constructed as requiring the identification of best practices through careful and rigorous empirical evaluative research and ‘applying’ these as faithfully as possible’, however it is important to retain clinical autonomy in diagnosis and treatment so a degree of flexibility will be necessary to support individual variation (both clinical practitioner and patient). A modular design, including ‘grab bags’ may provide the optimal solution; this design approach will be further explored. A greater challenge will be the dissemination of the research findings. The intricacies of the UK healthcare market mean there is no central point of access for innovations and there is no central unit responsible for ensuring innovations are

180


considered or taken up (Phillips and Caldwell, 2007). A study commissioned by the Association of British Healthcare Industries (ABHI, 2002) found the behaviour of the NHS to be the biggest limiter to growth, including a limited capacity to innovate as a customer due to, for example, a lack of mechanisms for trialing/evaluating clinical and cost effectiveness; a lack of staff motivation to innovate due to day-today pressures and a blame culture that penalises innovation; a lack of dissemination of good practice between Trusts. We are involved with the National Ambulance Strategic Procurement Team so hope that this, together with professional and research dissemination will lead to the successful implementation of the portable pod.

Acknowledgements We would like to thank the EPSRC for funding the Smart Pods project and our partners at the Royal College of Art, Bath and Plymouth Universities. We would particularly like to acknowledge the contribution of the 6 participating NHS Trusts: University Hospitals Bristol NHS Foundation Trust (UHBT), University Hospitals Leicester Foundation NHS Trust (UHL), BrisDoc, Leicester County and Rutland Primary Care NHS Trust (LCRPCT), Great Western Ambulance Service NHS Trust (GWAS) and East Midlands Ambulance Service NHS Trust (EMAS).

References ABHI, 2002. A competitive analysis of the healthcare industry in the UK; CoMap II. An independent report commissioned by the Association of British Healthcare Industries. Bazeley, P., Richards, L. 2000, The NVivo®Qualitative Project Book, (Sage Publications, London) DH, 2004, The Emergency Care Practitioner report – Right skill, Right time, Right place, http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/ PublicationsPolicyAndGuidance/DH_4093086 (Accessed 1 April 2008) DH, 2005a, Taking Healthcare to the Patient. Transforming NHS Ambulance Services. http://www.dh.gov.uk/assetRoot/04/11/42/70/04114270.pdf (Accessed 2 Sept. 2005) DH, 2005b, Urgent Care: Direction of Travel. Consultation Document, (Department of Health, London) DH, 2008, Guidance on Pre-registration Education and Funding for Paramedics, (Department of Health, London) DH Ambulance Policy, 2008, Changing Times: sustaining long-term performance against ‘Call connect’ for NHS Ambulance Services, (Department of Health, London) Gershenson, J.K., Prasad, G.J., Allamneni, S. 1999, Modular Product Design: A Life-cycle View Journal of Integrated Design and Process Science, 3, 4. http://www.me.mtu.edu/∼jkgershe/lel/research/JIDPS.pdf (Accessed 26th August 2008)

Portable pods

181

Jones, A., Hignett, S., Benger, J. 2008a, Identifying current and future care activities in ambulances, emergency departments and primary care. Emergency Medicine Journal, 25 (Suppl):A3 Jones, A., Hignett, S., Benger, J. 2008b, A Comparison of Urgent Tasks and Workspace Design in Pre-hospital Care. In Hignett, S., Norris, B., Catchpole, K., Hutchinson, A., Tapley, S. (Eds.) 2008, From Safe Design to Safe Practice. Proceedings of the Improving Patient Safety conference. The Ergonomics Society. 16–18 July, Cambridge. 121–125 Lakhani, M., Fernandes, A., Archard, G. 2007 Urgent Care: a position statement from the Royal College of General Practitioners, ( Royal College of General Practitioners, London). Lattimer, V., Brailsford, S., Turnbull, J., Tarnaras, P., Smith, H., George, S., Gerard, K., Maslin-Prothero, S. 2004, Reviewing emergency care systems 1: insights from system dynamics modelling. Emergency Medicine Journal 21, 6, 685–691. Mason, S., O’Keeffe, C., Coleman, P., Edlin, R., Nicholl, J. 2007, Effectiveness of emergency care practitioners working within existing emergency service models of care. Emergency Medicine Journal 24, 239–243 NHS Information Centre, 2008, Ambulance Services, England, 2007–08, (The Information Centre, London) Phillips, W. and Caldwell, N. 2007, Supplying Innovative Healthcare Technologies into the UK Healthcare Sector: A literature review. Internal document. Centre for Research in Strategic Purchasing and Supply, University of Bath School of Management Poland, B., Lehoux, P., Holmes, D., Andrews, G. 2005, How place matters: unpacking technology and power in health and social care, Health and Social Care in the Community, 13, 2, 170–180 Redden, D. 2002, An Evaluation of Fast Response Vehicle Paramedic Bag Systems. Unpublished M.Sc. Dissertation. Dept of Human Sciences, Loughborough University Reynolds, R. 2008, Standardisation of Emergency Care Practitioners Equipment and Consumables. Unpublished B.Sc. Dissertation, Dept of Human Sciences, Loughborough University Schilling, M.A. 2000, Toward a general modular systems theory and its application to interfirm product modularity, Academy of Management Review, 25, 2, 314–334 Snooks, H., Kearsley, N., Dale, J., Halter, M., Redhead, J., Cheung, W.Y. 2004, Towards primary care for non-serious callers to the emergency ambulance service: results of ‘Treat and Refer’ protocols for ambulance crews, Quality and Safety in Healthcare, 13, 435–443 Turner, J., Nicholl, J., O’Caithain, A. 2008, A preliminary study of Emergency and Urgent Care Networks. http://www.dh.gov.uk/en/Consultations/ Responsestoconsultations/DH_080369 (Accessed 25th August 2008) Wilson, J.W., Haines, H., Morris, W. 2005, Participatory Ergonomics, in Wilson J.W. and Corlett, E.N. (Eds.) Evaluation of Human Work, Third Edition, (Taylor and Francis, London) 933–962

USING USABILITY TESTING AS A BENCHMARKING TOOL – A CASE STUDY ON MEDICAL VENTILATORS Y. Liu & A. Osvalder Division Design, Department of Product & Production Development, Chalmers University of Technology, Gothenburg, Sweden Medical ergonomics has become an important topic in the field of ergonomics due to the increasing deficiencies in medical device design. The aim of the present study was to use usability testing as a benchmarking tool of modern and complex ventilator machines in a real hospital setting to propose future redesign ideas for a target machine – SERVO-i. A usability study was carried out at the neonatal intensive care unit (ICU) with 6 expert nurses. During the tests, an artificial lung was used to simulate the real treatment context on a neonatal patient. A number of potential usability problems were detected on SERVO-i. Using Babylog and Stephanie as two reference machines, the strengths and weaknesses of SERVO-i were identified as well. Some valuable lessons were learnt from the study, which could be used as guidance for the choice of test subjects and control of tests in real life human-machine working environments.

Introduction In recent years, various medical accidents have been reported. However, the number of accidents, incidents and near mistakes with medical equipment is much larger than the number reported officially (Gaba et al., 1995). Reports of critical mishaps show that misuse and human error are more frequently involved than equipment failures (Hyman and Schlain, 1986; Craig and Wilson, 1981; Cooper et al., 1978). Concerning human errors in hospitals, Russell (1992) indicated that avoidable mistakes killed 100,000 patients a year. Human error in medicine, and the adverse events that might follow, are related to problems of psychology and engineering, rather than those of medicine (Senders, 1994). Medical ergonomics has become an important topic in the field of ergonomics. Medical devices play important and life-sustaining roles in the health care sector, especially in critical-care medical settings, such as Intensive Care Units (ICUs), Anesthesia Workstations, and Operating Rooms (ORs). Deficiencies in medical device design were increasingly identified and addressed in recent years. As is widely accepted, design of medical devices concerns not only the quality of the medical care delivery, but also the patients’ safety and health. Nowadays, usability and risk management in the medical field become an important topic of human factors research. Manufacturers have realized that safe and effective use of a medical device depend on the interaction between the operating environment, user 182

Using usability testing as a benchmarking tool

183

capabilities, stress levels, and device design, therefore usability issues have to be considered in the design process (Sawyer, 1996). Due to the academic research in usability engineering, the focus of risk analysis and risk management has already been shifted from technical aspects and human performance to human-machine interaction (HMI). How to improve safety and achieve user-friendly goals by detecting design deficiencies and reducing errors in the design process has increasingly drawn attention of designers and manufacturers. If design deficiencies or human errors could be detected during evaluations in the design process, numerous medical accidents or injuries could be avoided and more lives, money and labor could be saved. Ventilator machines are used as essential and critical medical devices in Anesthesia Workstations, ICUs, and ORs. How to improve the usability of ventilator machines attracts designers’ and ventilator manufacturers’ attention today, and thus how usability tests should be performed. SERVO-i has been an advanced ventilator machine used normally for adult patients. However, the manufacturer of SERVO-i chose it as a target machine for further development due to their future marketing strategy for neonate patients in neonatal ICUs at hospitals. The purpose of the present study was: (1) to detect potential usability problems of a modern complex ventilator machine (SERVO-i) for use in a neonatal ICU, and (2) to investigate the strengths and weaknesses of the target machine (SERVO-i) compared to two reference ventilator machines (Babylog & Stephanie). The aim of the study was to provide an example of using usability testing as a benchmarking tool of ventilator machines in a real hospital setting to propose future redesign ideas.

Methods and materials The usability study was carried out at the neonatal ICU in Queen Silvia Children’s Hospital in Gothenburg, Sweden. In the neonatal ICU, ventilator machines are frequently used on premature and neonate babies (0–3 months old), who require critical medical treatment. 6 expert nurses from the neonatal ICU, who had worked with 2 to 6 different brands of ventilator machines for 10 years at average, participated in the usability tests. All test participants were expert users on Babylog and Stephanie, but had never used SERVO-i before. The three ventilators had different types of interfaces. SERVO-i had a highly sensitive touch-screen interface, Babylog had a monochrome screen and a few hard buttons for menu choice, and Stephanie had a colorful screen and a large scroll knob for software menus (Figure 1). 6 significant task scenarios were selected with the help of a pediatrician specialist. The task scenarios were classified into two categories according to frequency of use, i.e. (I) Frequently used tasks and (II) Infrequently used tasks. Category (I) included Task 1 – Starting ventilation, Task 2 – Alarm handling, and Task 3 – Checking critical parameters. Category (II) included Task 4 – Trigger handling, Task 5 –Displaying ventilation process in diagrams, and Task 6 –Displaying alarm history.

184

Y. Liu & A. Osvalder

Figure 1. The three ventilators included in the usability study.

The usability tests were conducted according to a general test procedure proposed by, for instance, Nielsen (1993) and McClelland (1995). During the tests, an artificial lung was used to simulate the real treatment context on a neonatal patient. Prior to the usability tests, two pilot tests were carried out to verify the test procedure. At the beginning of the test session, each test subject was informed about the purpose of the test and given short instructions regarding the test procedure. Two test sessions were conducted with each test subject, one session on the reference machines (i.e. Babylog & Stephanie), the other session on the target machine (SERVO-i). Each test session took approximately one hour. During the tests, video cameras were used to record the whole test procedure; however, all the test subjects were notified that only their hands would be recorded, thus ensuring anonymity. During the tests, the number of errors was collected from each task scenario. At the end of each test session, an after-session interview was held with the test subjects to get their comments and suggestions on the interface design of the ventilators. A simple 10-item scale questionnaire, the System Usability Scale (SUS) questionnaire (Brooke, 1996), was also used to get each test subject’s overall satisfaction of each ventilator. At the end of the second test sessions, a psychometric technique called paired comparisons (Thurstone, 1927) was used in order to get the test subjects’ subjective assessment on the interface design of three ventilators. Pictures of the three ventilator interfaces were presented in pairs to each test subject. The test subjects had to select which interface they thought better, until all possible paired combinations had been used.

Results The results were based on the data analysis of the test subjects’ behavior and their verbal comments. A number of potential usability problems were detected on SERVO-i, such as (1) Sensitivity problem of the touch screen;


185

Number of failures

7 6 5 4

Babylog

3

Stephanie

2 1

Figure 2.

6

5

sk Ta

4

sk Ta

3

sk Ta

2

sk Ta

sk Ta

Ta

sk

1

0

Number of failures on Babylog and Stephanie.

(2) Inconsistency confirmation methods; (3) Difficult to get an overview of all functions; (4) Bad symbol of ventilation start button; (5) Bad information organization on the screen; (6) Bad readability of measured values; (7) Lack of ‘Help’ function; and (8) Terminology problem. Then number of failures during completing the frequently used tasks (i.e. Task 1, Task 2, and Task 3) and the infrequently used tasks (i.e. Task 4, Task 5, and Task 6), were collected and analyzed both for Babylog and Stephanie (Figure 2). In general, more errors were shown in the infrequent-used tasks for both machines. Although the test subjects were novice users on SERVO-i, the results of the paired comparison method implied that SERVO-i was most preferable due to better information clarity, information layout, adequate amount of relevant information, and ease of understanding. However, Babylog was most preferable due to its ease of use. Concerning overall satisfaction, a within-subjects’ Friedman test was conducted. No statistical significant difference was shown between SERVO-i and the two reference machines (p = 0.74). Nevertheless, higher ranking was found on SERVO-i by the test subjects who were first time users, which implied the test subjects’ positive attitude to SERVO-i. According to the after-session discussion, the test subjects commented freely on the pros and cons of the three ventilator machines, which provided useful information for future redesign. The strengths of SERVO-i included modern touch screen design, clear information display and value settings, ease of navigation, and logical menu structure. Sensitivity problem of the touch screen, bad navigation path to ‘Mode’ settings, bad readability and clarity of the ‘Trend’ display, were counted as weaknesses of SERVO-i. Although the strengths and weaknesses of the two reference machines were not the focus of this study, the results could be used to get an overall picture about competitors’ products and their competencies. For instance, the strengths of Babylog were counted as ease of use, simplicity of interface, and adequate amount of information on the display, while the strengths of Stephanie were ease of navigation, clear display, ease of use in parameter settings, and a good overall picture of all

186


information on one screen. The information could be used as useful implications for future redesign and marketing strategy.

Discussion In the present study, 6 expert nurses who had in average 10 years working experience with 2–6 different types of ventilator machines, interacted with a new ventilator machine (SERVO-i) as first time users. The results showed that the test subjects were very skilful in frequently used tasks and were almost frustrated on infrequently used tasks, when interacting with the familiarized machines that they were used to in their daily work. This phenomenon conforms to Nielsen’s (1993) opinion that even an expert user might be quite novice with respect to infrequently used tasks. In other words, expertise of an expert user might be task-bounded. The results posed a practical question concerning how to define expert users in the medical field. In the present study, the test subjects were actually task-experts rather than machine-experts. In the medical health care sector, time and expertise concern safety and health during the treatment. In case of emergency treatment, it is impossible for health carers or nurses to get time and look up the users’ menu about some infrequent-used functions of the medical device. Thereby, it is important that expert users of medical devices should be skilful in all available functions rather than partial functions of the devices. However, the after-session discussion revealed a special working organization factor at the neonatal ICU, that is, the medical doctors dominated in the work with ventilation treatment, while ventilator nurses were only subjected to the doctors’ permission during the treatment process. To some extent, such convention determines or affects constituent of nurses’ expertise. However, whether such affects should be counted as positive or negative is a question to the medical health care sector. During the usability evaluation, sensitivity problem of the touch screen was found as an important and typical problem on SERVO-i, which affected the test subjects’ performance in some ways, such as distinguishing and judging the interactive items and non-interactive items on the screen. In addition, the test subjects might have an inappropriate mental model concerning how to interact with some graphical items on the screen due to the sensitivity problem. Many of the test subjects thought that they should use the large adjustable knob rather than the touch screen to activate graphical items. This implied that the users’ mental models and performance were basically influenced by the features of the touch screen. Another typical usability problem detected was related to terminology. The results revealed that terminology differences between the ventilators confused the test subjects. Due to their previous accumulated experience and knowledge, expert users are always assumed to have better guessability (Jordan, 1998) to cope with the differences in the components and layout of a new or unfamiliar interface. However, the terminology difference hinders the effect of guessability. Consequently, the expert users might be confused and thereby hesitate to make a trial on further interactions. This posed a question to the anaesthesia and ventilation field: Is it possible to have common or standard terminologies on all ventilator products for


187

designers and manufacturers? Without doubt, standardization on terminology can reduce unnecessary confusion or errors in navigation. Since the test subjects used Babylog and Stephanie in their daily work, they were familiar with and used to work with these machines. SERVO-i was new and unknown to the test subjects. It is easy to hypothesize that the test subjects might be influenced by their experience or habits on the familiarized machines. However, the results of the subjective assessments did not prove this hypothesis. Instead, the results implied that expert users are not always biased or subjected to the constraints of their previous experience in subjective assessments. Some valuable lessons were learnt from the study, which could be used as guidance for choice of test subjects and control of tests in real life human-machine working environments. In the present study, the best way to perform usability tests was to make a test schedule in the morning, which depended on the available staff and how much activity there currently was at the neonatal ICU. This had to make because of the typical feature of ICUs at hospitals, i.e. an intensive and stressful working environment. Due to its strength in evaluating handling characteristics, usability test is suggested to be performed after the clinical trial during purchase stage (Liljegren and Osvalder, 2004). To perform usability studies at hospitals requires a lot of flexibility in time schedule and planning. It is hard to arrange a reliable schedule since everything is very much event driven in this type of working environment. Therefore, adequate negotiation with both high-level managers and nurses is necessary and important for performing a successful usability study in a hospital setting. In addition, it was found difficult to get a large number of test subjects in ICUs due to the stressful and event driven working environment. However, Nielsen (1993) indicated that 5 test subjects are enough in empirical usability evaluations in terms of reliability or accuracy level. With 5 test subjects, a 70% confidence probability of getting within ±15% of the true mean could be gained (Nielsen, 1993).

Conclusion The intention of the present study was to use usability testing as a benchmarking tool to evaluate the current design of SERVO-i in a real hospital setting, with the focus on investigating first time use experience by expert ventilator nurses. Based on the results, the following conclusions can be drawn: 1. The potential usability problems of SERVO-i for future improvement included: • sensitivity of touch screen • consistency of confirmation methods • overview of all functions • symbol of ‘Start’ button • information organization • readability of measured values • ‘Help’ function • visualization of interaction • terminology

188


2. SERVO-i was preferable to the two reference machines in terms of better information clarity, information layout, adequate amount of relevant information, and ease of understanding. 3. Referring to the two reference machines, the ‘ease of use’ aspect should be improved on SERVO-i. 4. The terminology difference between the same type of medical devices from different manufacturers might be an important factor affecting expert users’ performance in the interaction. 5. Expert users might not always be biased or subjected to the constraints of their previous use experience when giving subjective assessments on a new or unknown medical interface. 6. The present study provided a tentative example of how to carry out benchmark tests concerning usability aspects on medical equipments in a hospital setting.

References Brooke, J. 1996, SUS – A quick and dirty usability scale. In P.W. Jordan et al. (eds.) Usability Evaluation in Industry, (Taylor & Francis, London) Cooper, J., Newbower, R.S., Long, C.D. and McPeek, B. 1978, Preventable anesthesia mishaps: a study of human factors, Anesthesiology, 49, 399–406 Craig, J. and Wilson, M.E. 1981, A survey of anesthetic misadventures, Anesthesia, 36, 933–936 Gaba, D.M., Howard, S. and Small, S. 1995, SituationAwareness inAnesthesiology, Human factors, 37(1), 20–31 Hyman, W.A. and Schlain, L.A. 1986, User problems and medical device recalls, Medical Instrument, 20, 14–16 Jordan, P.W. 1998, An Introduction to Usability, (Taylor and Francis, London) Liljegren, E. and Osvalder, A.L. 2004, Cognitive engineering methods as usability evaluation tools for medical equipment. International Journal of Industrial Ergonomics 34(1), 49–62 McClelland, I. 1995, Product assessment and user trials. In J.R. Wilson and E.N. Corlett, (eds.) Evaluation of Human Work, (Taylor and Francis, London) Nielsen, J. 1993, Usability engineering. (Academic Press, San Diego) Russell, C. 1992, Human error: Avoidable mistakes kill 100,000 patients a year, Washington Post, Feb. 18, WH7 Sawyer, D. 1996, Do it by design: An introduction to human factors in medical devices. U.S. Department of Health and Human Services, Food and Drug Administration (Centre for Devices and Radiological Health) Senders, J.W. 1994, Medical devices, medical errors, and medical accidents. In M.S. Bogner (ed.) Human error in medicine, (Lawrence Erlbaum Associates Inc., Publishers, New Jersey), 159–177 Thurstone, L.L. 1927, A law of comparative judgment. Psychological Review, 34, 278–286

DOES THE OLDER WORKFORCE WITH HIGH WORK DEMANDS NEED MORE RECOVERY FROM WORK J.J. Devereux1 & L.W. Rydstedt2 1

University of Surrey, UK University West, Sweden

2

A follow-up study investigated how psychological and physical work demands affected older workers need for recovery at the end of a working day. A comparison of need for recovery results measured at time 1 and time 2, fifteen months apart, showed similar results. Psychologically demanding work was the stronger risk factor compared to high physical work demands. Older workers were more likely to have a significantly higher need for recovery from work after performing psychologically and physically demanding work compared to younger workers.

Introduction Approximately one third of the UK labour force will be age 50 or over by the year 2020. According to the Office of National Statistics, ageing of the workforce will be the most significant development in the labour market over the next 15 years. As ageing in the population increases and the need to work longer into later life takes effect in the workplace, it will become more important to understand the needs and physical and mental abilities of workers performing different kinds of work. Older workers might be at risk of overload from the physical and psychological demands of work. Besides working into older age, there is also the added complication of having the workforce adapt to different organisational work patterns including flexible working hours, the 24/7 work culture and irregular shifts. These changes in the workforce and the organisation of work may have serious implications for an increased risk of cardiovascular disease, subjective health complaints, musculoskeletal disorders and duration of sickness absence. Recent studies have used the concept of need for recovery from work to explore the temporary overload from the physical and psychological demands of work. It is viewed as an individual perception of the time required to recover from adverse working conditions that results in insufficient unwinding after exposure to stressful work characteristics and is a strong predictor of subsequent cardiovascular disease (van Amelsvoort et al., 2003), perceived job stress (Devereux & Rydstedt, 2008), subjective health complaints and duration of future sickness absence (Sluiter et al., 2003) In the study by Sluiter et al., (2003) the authors recommended that age be treated as an effect modifier between work demands and need for recovery after work. 189

190

J.J. Devereux & L.W. Rydstedt

The purpose of this study, using longitudinal data, was twofold. First, investigate the relationship between physical and psychological work demands and subjective need for recovery after work reported at two different time points 15 months apart. And secondly, determine whether age modified the risk relationship between the physical and psychological demands of work and subjective need for recovery after work. Ethical approval was given for this study by the University of Surrey Committee on Ethics.

Study population The respondents were all participating in a large longitudinal study on work stress and work-related musculoskeletal and mental disorders (Devereux et al., 2004). This comprised a baseline cross-sectional study of 8,000 workers of whom 3,139 were followed for 15 months (approx.). The cohort was drawn from 20 organizations across 11 industrial sectors in the U.K. The occupational groupings from the study were compared to the Standard Occupational Classification 2000 system using the 9 major groups. The study sample had an over-representation of professional workers and an under-representation of skilled trade, personal service, sales customer service and elementary workers. This was partly caused by the different response rates across different occupations. There was no age or gender difference between the study population of 3139 and the total population of 8000. Male and female workers within the age range 17–69 years were included in the study. Part-time workers and workers principally based long term within client organisations were excluded from the study sample. Exposure data, demographics and need for recovery were assessed using a questionnaire at baseline (time 1), 3056 workers had valid need for recovery and age scores. After attrition mainly due to turnover and downsizing in some of the organisations, the sample for the resurvey at time 2 consisted of 2091persons with valid need for recovery scores, about 68% of the participants at the baseline survey (time 1). The attrition did not affect the age and gender distribution of the sample.

Measures The ‘need for recovery’ scale consists of 11 dichotomous items concerning recuperation after one day of work. Examples of items are ’my job causes me to feel rather exhausted at the end of the working day’ and ’after a working day I am often too tired to start and activities’, and ’sometimes, I cannot optimally perform my job because of fatigue during the last part of the working day’. This scale has been evaluated against neuroendocrine activity and subjective health complaints (Sluiter et al., 2001). Furthermore, the test-retest reliability over a 2 year period in a stable work environment is good to excellent (Intraclass correlations 0.68–0.80) (de Croon et al., 2006). All items were coded so that higher scores meant more complaints, i.e. more need for recovery after work. Participants with missing data on any of the 11 items were excluded from analysis. The Cronbach’s alpha of the entire scale was 0.85. Median split was used to contrast employees with low and high levels of need

Does older workforce with high work demands need more recovery

191

for recovery after work and work content variables. All other measures have been reported previously (Devereux et al., 2004). Job demands (Chronbach alpha 0.66) was assessed using the modified Job Content Questionnaire comprising 4 items (working very fast, intensively, having enough time and difficulty combining tasks). Extrinsic effort was assessed using the Effort-Reward questionnaire. Extrinsic effort items were rated on a dichotomous (agree/disagree) scale. Extrinsic effort was measured by 6 items including the item concerning physical loading. Extrinsic effort without the physical demands item was also used, as a large proportion of the study population performed sedentary work. The physical demands item was also reported separately. The 5 item version of extrinsic effort had a Chronbach alpha 0.85. Extrinsic effort items included constant time pressures, interruptions and disturbances at work, job responsibility, pressured to work overtime and increasing demands of the job. The percentage of workers characterised into each exposure group across each need for recovery measure was first analysed. Each worker was classified into one of four age groups (17–29, 30–39, 40–49, 50–59). The age grouping was selected to allow for equal number within each group. The odds ratio and 95% confidence interval was then calculated for each potential risk factor and compared across each measure of need for recovery using the Mantel Haenszel Common Odds Ratio Summary statistic stratified across each of the 4 age groups. The stratum specific odds ratios and 95% confidence intervals were then calculated for each age stratum to investigate potential effect modification. Data were analysed using SPSS version 15 and Epi Info version 3.5.1.

Results There was a similar distribution of subjects across exposure groups at each time point (T1 and T2). The odds ratios for each risk factor across each time point was similar, as shown in table 1. The psychological demands were the strongest risk factors followed by physically demanding work and heavy lifting. Repetitive arm movement was also a risk factor but prolonged keyboard use was significantly weaker then the other physical demand variables and it is tentative as to whether the increase in the risk observed was a true effect. The risk from long working hours was similar to the risk for heavy lifting. There was a weaker odds ratio for late night or shift work but the results suggest that there was a true effect. Workers were approximately three times more likely to report a higher need recovery after work if exposed to psychologically demanding work, as measured using psychological job demands and extrinsic effort variables. Workers were approximately two times more likely to report a higher need for recovery after work when exposed to physically demanding job. Tasks requiring pushing and pulling, lifting loads greater than 16 kg, driving for more than half a day and performing almost continuous arm movement had odds ratios between 1.27 and 1.44. Gender did not increase the risk for need for recovery after work. Table 2 shows the risk from individual characteristics, work organisation and work content across each age stratum and whether the same effect was observed at

192


Table 1.

Risk factors for need for recovery at time 1 and time 2 adjusted for age.

Factor and exposure level

MH Common OR & 95% CI at T1

MH Common OR & 95% CI at T2

Individual characteristics Gender Men vs woman

1.13 (0.10–0.98)

1.08 (0.90–1.30)

Work organisation 42 or more hours worked per week Late or night shift vs day shift

1.79 (1.54–2.08) 1.28 (1.09–1.51)

1.67 (1.38–2.01) 1.46 (1.19–1.79)

Work content Psychological job demands Extrinsic effort Extrinsic effort without physical loading

3.10 (2.64–3.63) 3.22 (2.72–3.82) 3.14 (2.70–3.65)

2.91 (2.37–3.56) 3.14 (2.57–3.84) 2.57 (2.14–3.08)

1.94 (1.66–2.27) 1.43 (1.23–1.66) 1.30 (1.04–1.64) 1.27 (1.09–1.49) 1.68 (1.35–2.08) 1.33 (1.05–1.68) 1.19 (1.03–1.39)

1.98 (1.64–2.40) 1.41 (1.17–1.70) 1.26 (0.94–1.70) 1.27 (1.05–1.54) 1.44 (1.10–1.89) 1.33 (0.99–1.79) 1.07 (0.89–1.29)

1.50 (1.26–1.79)

1.44 (1.16–1.80)

Physical work demands Physically demanding job Task requiring pushing/pulling Lifting 6–15 kg Lifting 16–45 kg Lifting >45 kg Driving > half working day Using a keyboard for 4 or more hours a day Repetitive arm work Almost continuous arm movement vs Intermittent arm use and Regular arm movement with some pauses OR, Odds Ratio; CI, confidence interval

both time 1 and time 2 measurement of need for recovery after work. Table 3 shows the risk from physical work demands. The odds ratio for psychological job demands, extrinsic effort, driving, and tasks requiring pushing and pulling increased from the youngest to the oldest age group. The oldest age group had a much higher odds ratio compared to the youngest age group for the variables involving lifting greater than 16 kg and repetitive work. No such effect modification was observed for gender, working late or night shifts, lifting 6–15 kg, or using a keyboard. Prolonged working hours show an increased risk for the two older age groups compared to the two youngest age groups but this effect was not observed at both time 1 and time 2. The excess risk for the work content variables and the physical demand work variables was more than 30% in the older age group compared to the youngest age group.

Discussion Psychologically and physically demanding work increased the risk for need for recovery from work. Psychologically demanding work was the stronger risk factor

1.53 1.70

Work organisation 42 or more hours worked per week Late or night shift vs day shift 1.78 1.51 1.28

1.04 1.13

0.88

3.85 3.58 2.67

2.24 2.57

1.79

2.45 2.69 2.78

1.50 1.73

1.13

30–39

1.60 1.82 1.91

1.04 1.16

0.78

3.74 3.98 4.04

2.17 2.57

1.62

2.60 3.30 2.68

1.68 1.31

1.06

40–49

OR, Odds Ratio; CI, confidence interval X X This means that a: there is an increased risk in the oldest compared to the youngest group and, b: there is effect modification at both T1 and T2, defined by an excess risk of 30% or more.

2.62 2.32 1.85

1.25

Individual characterisitics Gender Men vs woman

Work content Psychological job demands Extrinsic effort Extrinsic effort without physical loading

17–29

OR (95% CI)

1.79 2.26 1.90

1.19 0.90

0.75

3.76 4.82 3.77

2.37 1.90

1.49

4.79 4.67 3.21

2.04 1.14

0.89

50–69

2.98 3.09 2.18

1.38 0.71

0.60

7.69 7.07 4.75

3.02 1.81

1.32

X X X

T1

Effect modification of age on the odds ratios for individual characteristics, work organisation and work content.


Table 2.

X X X

X

T2

Does older workforce with high work demands need more recovery 193

2.07 1.300 1.157 1.302 1.330 1.064 1.176 1.351

Physical work demands Physically demanding job Task requiring pushing/pulling Lifting 6–15 kg Lifting 16–45 kg Lifting >45 kg Driving > half working day Using a keyboard for 4or more hours a day Repetitive arm work 1.38 0.89 0.63 0.87 0.76 0.59 0.81 0.87

3.11 1.90 2.13 1.95 2.34 1.92 1.70 2.11

1.81 1.23 1.46 0.95 0.99 1.18 1.16 1.21

30–39 1.23 0.85 0.79 0.64 0.57 0.67 0.80 0.78

2.66 1.80 2.70 1.41 1.71 2.09 1.68 1.86

1.78 1.29 1.35 1.26 1.28 1.33 1.23 1.62

40–49

OR, Odds Ratio; CI, confidence interval X X This means that a: there is an increased risk in the oldest compared to the youngest group and, b: there is effect modification at both T1 and T2, defined by an excess risk of 30% or more.

17–29


OR (95% CI)

1.24 0.91 0.77 0.87 0.75 0.74 0.87 1.05

2.54 1.83 2.36 1.81 2.19 2.38 1.74 2.50

2.39 1.99 1.10 1.70 2.53 2.01 0.72 1.64

50–69

Table 3. Effect modification of age on the odds ratios for physical work demands.

1.63 1.35 0.60 1.14 1.47 1.08 0.47 1.06

3.51 2.92 2.02 2.53 4.35 3.75 1.09 2.54

X X X X X X

X X X X

T2 X X

T1

194 J.J. Devereux & L.W. Rydstedt

Does older workforce with high work demands need more recovery

195

compared to high physical work demands. Gender differences and prolonged keyboard work were not important factors. A comparison of the results measured at time 1 and time 2 shows similar effect sizes and so provides stronger support for the observed relationships. These results are supported by the cross-sectional results from Sluiter et al., (2003). They found that physical work demands were associated with need for recovery from work in construction, and hospital nurses. Our study also examined age as a potential effect modifier and found that older workers are more likely to have a higher need for recovery from work after performing psychologically and physically demanding work compared to younger workers. The excess risk for workers greater than 50 years old performing frequent lifting of loads >45 kg and experiencing high psychological demands was significant compared to the 40-49 year old age group. A high number of hours worked showed a marked increase in the likelihood for older workers at time 2 but not at time 1 so there is weaker evidence for this work organisation factor. The low response rate to the baseline questionnaire may have underestimated the effect size of work content variables, as a sub-sample of non-respondents indicated a higher time pressures compared to respondents (a potential response bias). Furthermore, the population may also be a survivor population in that older workers performing heavy lifting may have already left the study population before the baseline survey was conducted. This would make it harder to detect a true effect. Despite these limitations, an increased risk was observed for psychological and physical work indicating a true effect. There were no differences in age and gender in non-respondents confirmed by company records. Physical and psychological work demands were relatively stable over the follow-up period according to a re-survey of a cohort sub-sample. The majority of workers were unlikely to have changed from low to high exposure or vice versa. The need for recovery scale has shown to have good to excellent test-retest reliability under stable work conditions. No other study had examined age as an effect modifier across physical and psychosocial work demands so it is difficult to draw any firm conclusions. Further research is needed to explore this potential relationship as it is important for developing better health prevention strategies in the workplace. In this study a high need for recovery from work at time 1 was associated with taking more than 5 days off for health reasons in the last 12 months. (OR 1.62 95% CI 1.33–1.98) compared to workers taking no days off. Psychological job demands have been shown to be a determinant of long term mental strain in some occupational groups (Rydstedt et al., 2007) and the physiological effects of long term job strain can be detrimental to health (Rydstedt et al., 2008) Recovery is considered an important variable between adverse work demands, the development of work-related stress reactions and health problems in the long-term (Devereux & Rydstedt, 2008; Sluiter et al., 2003). It is important that the physical and psychological demands of work are assessed and controlled especially for older workers. Work organisations should consider

196


working together with older employees in their workforce to determine optimal workload levels so that the need for recovery from work is minimised. Furthermore, design changes to the work environment to support declining cognitive, perceptual and psychomotor abilities are recommended (Charness, 2008). Sluiter (2006) has suggested job specific functional tests for high demanding jobs in order to monitor individual functioning and facilitate appropriate interventions. We would like to thank the Health and Safety Executive for funding this study.

References Charness, N. 2008, Aging and human performance. Human Factors, 50, 548–555. de Croon, E. M., Sluiter, J. K., & Frings-Dresen, M. H. W. (2006). Psychometric properties of the Need for Recovery after work scale: test-retest reliability and sensitivity to detect change. Occupational and Environmental Medicine, 63, 202–206. Devereux, J. & Rydstedt, L. 2008, The relationship between psychosocial work factors and job stress: prospective data from the Stress and MSD Study. In L.Sznelwar, F. L. Mascia, & U. B. Montendo (Eds.), Human Factors in Organizational Design and Management-IX (IEA press, Santa Monica), 211–216. Devereux, J., Rydstedt, L., Kelly, V., Weston, P., & Buckle, P. 2004, The role of work stress and psychological factors in the development of musculoskeletal disorders: The Stress and MSD Study, (HSE Books, Sudbury). Rydstedt, L., Cropley, M., Devereux, J. J., & Michalianou, G. 2008, The relationship between long-term job strain and morning and evening saliva cortisol secretion among white-collar workers. Journal of Occupational Health Psychology, 13, 105–113. Rydstedt, L., Devereux, J. J., & Sverke, M. 2007, Comparing and combining the demand-control-support model and the effort-reward imbalance model to predict long-term mental strain. European Journal of Work and Organizational Psychology, 16, 261–278. Sluiter, J. K., de Croon, E. M., Meijman, T. F., & Frings-Dresen, M. H. W. 2003, Need for recovery from work related fatigue and its role in the development and prediction of subjective health complaints. Occupational and Environmental Medicine, 60, 621–670. Sluiter, J. K., Frings-Dresen, M. H. W., Van der Beek, A. J., & Meijman, T. F. 2001, The relation between work-induced neuroendocrine reactivity and recovery, subjective need for recovery, and health status. Journal of Psychosomatic Research, 50, 29–37. Sluiter, J. K. 2006, High-demand jobs: age-related diversity in work ability? Applied Ergonomics, 37, 429–440. van Amelsvoort, L. G. P. M., Kant, I. J., Bültmann, U., & Swaen, G. M. H. 2003, Need for recovery after work and the subsequent risk of cardiovascular disease in a working population. Occupational and Environmental Medicine, 60, i83–i87.

HUMAN ADVISORY

THE ROLE OF THE HUMAN IN FUTURE SYSTEMS – CONSIDERATIONS AND CONCERNS M.S. Young Human-Centred Design Institute, Brunel University, UK In 1990, Don Norman espoused his critical view of automation – namely, that its problems lie in being both too smart and yet not smart enough. Nearly 20 years on, we are no nearer to the ‘autopian’ vision of the future, although the ergonomics perspective on the problem has changed. In this paper, traditional frameworks and critiques of automation (allocation of function, ironies of automation) are summarised and contrasted against more contemporary thinking on the topic (support, cooperation and teamwork). While we wait for technology to catch up and for automation to be smart enough to be an effective member of a human-machine team, a new approach to automation is proposed based on principles of crew resource management. The argument is brought full circle by suggesting that ergonomics can take a lead on technology by specifying the capabilities of automation to fulfil the requirements of the team.

Automation – the ghost in the machine Technological progress is an inexorable tide – the rise of ubiquitous computing purports to make our lives safer and more efficient. From automatic protection systems in aircraft to artificial intelligence in our washing machines, the aims of such developments are very much in line with those of ergonomics – to improve safety, efficiency, and satisfaction. Nevertheless, the ergonomics literature on automation holds a much more conservative view, and we have known for some time now that to simply try to automate the human out of the loop does not provide the solutions that engineers crave. In the best-selling book ‘Jurassic Park’ (Crichton, 1980), the mathematician character Ian Malcolm criticises scientists for focusing too much on whether they can do something without stopping to consider whether they should. The same criticism has been levelled at designers of automation by some of the most influential ergonomists working in this field (e.g., Parasuraman, 1987; Wiener and Curry, 1980). Whilst the principle of avoiding technology merely for its own sake remains valid, it is fair to say that the current view in ergonomics is somewhat more mature in trying to understand how humans and automation can work together safely and effectively. It would be foolish to think we could stem technological progress – nor would we want to, lest Ian Malcolm accuse us of the kind of narrow-minded thinking he refers to as ‘thintelligence’. Humans and technology can and indeed should work together, and technological progress should be exploited – but in the 199

200

M.S. Young

right way. This paper considers the future for the ‘ghost in the machine’, arguing for a philosophical approach to the design of automated systems which allows humans and technology to coexist in a truly dualist system.

Promises and problems Before looking to the future, it is worth briefly reviewing some key developments in ergonomic thinking on automation. The classical perspective is captured in three seminal works. Firstly, Lisanne Bainbridge described the ‘ironies of automation’ in 1983. An initial irony lies in the designer’s view of the human operator as being unreliable or inefficient. Automation is popularly assumed to circumvent human error – by simply removing the human element in the system. What designers overlook, though, is that they are human too – and shortcomings in the system design can cause new problems. For instance, the new European Rail Traffic Management System (ERTMS) offers unprecedented levels of train protection, and to all intents and purposes will be an advance in performance and capacity on the rail network. However, its automatic protection systems are dependent on a set of data input by the driver at the start of a journey – thus moving the human error potential further back in the causal chain. Furthermore, Bainbridge argues, the operator is often left to do the tasks which the designer cannot solve through automation. But in order to take over control, a human operator must be practised at the task – which is impossible when the automation has been controlling it. Operator training cannot offer a solution, since it is impossible to train for the unforeseeable, so only general strategies may be learned. The irony is then in training the operator to follow instructions yet expecting them to provide intelligence in the system. Perhaps the final irony, Bainbridge notes, is that it is the most successful automated systems, with little need for human intervention, which require the greatest investment in operator training. James Reason (1990) expands on this with his ‘catch-22’ of human supervisory control, in that humans are only present in an automated system to deal with emergencies. They do this by drawing on stored knowledge of such, but with little opportunity to practice procedural responses – coupled with the unique nature of each emergency – means they have little knowledge or experience to draw upon. Reason concludes the catch-22 by stating that automation is most effective when it is least required, and vice-versa. That is to say, under normal operating conditions, automated systems can cope perfectly well. However, so do humans – thus begging the question of why automation has been implemented in the first place. In emergency situations, stressed humans can become overloaded and performance can deteriorate; it is under such circumstances when assistance would be most valuable. Ironically, it is in these situations when automation also surrenders, relying on the human to provide creativity and quell the emergency. Also in 1990, Donald Norman espoused a view that suggested it is not the presence of automation per se which is the problem, rather that the design of automated systems is inappropriate. The problem with automation is that it is at an intermediate level of intelligence – specifically providing an insufficient level of feedback. This

The role of the human in future systems

201

imposes more cognitive demands (the need to actively gather information about system state) and fewer perceptual-motor demands. Ironically, it is exactly these perceptual-motor tasks which humans are most skilful at, rather than the cognitive tasks which generally require greater mental demands. Thus the culprit is not necessarily the automation, rather the lack of feedback and interaction which keeps the human out of the loop. All of this leads Norman to a new irony: that it is not powerful enough. If automation were perfect (i.e., never fails), feedback would be unnecessary. However, the vision of a perfect automation system is still a long way off; the unique skills and flexibility of human operators will continue to prove crucial in critical situations. Take the example of United Airlines Flight 232, on 20th July 1989 (as reported by Faith, 1996). The DC-10 aircraft had suffered an explosive failure of its number two engine (the DC-10 has three engines, one on each wing and one mounted on the tail structure; it was the latter engine which failed), which in the process had destroyed the hydraulics to the control surfaces. Unable to fly the aircraft by conventional means, the crew enlisted the help of an off-duty pilot who happened to be on board, and they managed to fly to an airport at Sioux City, Iowa, purely by using the balance of the two remaining engines. The approach was going well considering such improvisation, but an unfortunate windshear on landing caused a catastrophic impact with the loss of 112 lives. However, the fact that there were 184 survivors can be put down to the ingenuity and skill of the flight crew. The common issue running through the critiques of Bainbridge, Norman and Reason is that automation changes the role of the human in the system from active control to passive monitoring – a task at which humans are particularly poor (Parasuraman, 1987). This out-of-the-loop performance problem can be manifest in vigilance failures (Molloy and Parasuraman, 1996) or difficulties recovering control from automation failure (Kaber and Endsley, 1997). Such problems have been attributed to mental workload (Young and Stanton, 2002), situation awareness (Endsley, 1995) and trust (Parasuraman and Riley, 1997). Although there may still be controversy surrounding the cause of automation-related complacency, there is no doubt that monitoring performance is impaired if the task is executed by automation instead of a human (Molloy and Parasuraman, 1996) – it is the qualitative change in the nature of the task that causes the problem. However, more recent thinking supplants this view, suggesting that whilst automation does change the nature of work, it actually adds a whole set of new demands. The problem of automation now is one of coordination.

Teamwork and taskwork In the last 10 years, there has been a shift in emphasis across the ergonomics literature regarding the interaction of humans and automation – and the associated solutions for the problems of automation. Rather than thinking in terms of the passive monitor, the human’s role is now seen as active manager of a set of resources, which happens to include the automated system (Dekker, 2004). The human is no longer out of the loop, just more distanced from it (Schutte, 1999). Automation does

202

M.S. Young

indeed qualitatively change the nature of the task, but this is by adding a whole new set of coordination demands (Parasuraman et al., 2000). Such a view is not wholly inconsistent with previous observations – one of the consequences of being a manager is that the demands increase at crucial times such as abnormal or emergency scenarios. But instead of this being due to vigilance monitoring issues, it is now felt that the problem arises from the increased demands of teamwork and taskwork. Teamwork is defined as “the ability of team members to work together, communicate effectively, anticipate and meet each other’s demands, and inspire confidence, resulting in a coordinated collective action” (Cannon-Bowers and Salas, 1997). In the teamwork literature, the demands of teamworking are divided into taskwork (those task-oriented activities which are necessary to actually get the job done) and teamwork (the coordination and communication activities, over and above taskwork, which are required to work together effectively). Thus in some cases, adding more members to the team can actually make overall performance less efficient, since the additional demands of teamwork outweigh the reductions in taskwork. Whilst the teamworking literature typically considers human-human teams, the principles of communication, cooperation and coordination apply across humanmachine teams when it comes to automation. Parasuraman and Riley (1997) defined automation as “the execution by a machine agent (usually a computer) of a function that was previously carried out by a human” (p. 231). Traditionally, this has been thought of as taking tasks piecemeal away from a single human operator, but it could equally be seen as the machines replacing human agents in the team. In essence, then, we are talking about delegation of tasks to another agent (cf. Dekker, 2004). But in doing so, we must ensure that the teamworking elements – the communication, cooperation, and coordination – remain intact. In the abnormal scenario, for instance, the human needs support from the automation in order to resolve the situation. The problem now is that most automated systems are not designed to behave like humans when it comes to teamwork. Take Norman’s (1990) case study of the Boeing 747 with the loss of engine power. In this example, an autopilot system attempted to compensate for the loss of power by balancing the control surfaces (almost an inverse of the Sioux City incident described above). However, as Norman rightly points out, the autopilot did not provide feedback to the flight crew on its actions – so when it could compensate no more, the human team members were faced with a drastically worse situation than if they had been informed earlier. From the teamworking perspective, it is the communication which is missing.

Complemation and common frames of reference Any good team is built up of members with complementary skills, such that the task demands can be met and the team’s goals achieved. If any of the team members happen to be machine, the situation should be no different. Schutte (1999) coined the term ‘complemation’ to sum up the principle of exploiting automation to enhance


203

human capabilities while compensating for their limitations. For instance, we could allow the human to provide creativity and adaptability, while letting the machine store information and make precise measurements – maximising the strengths of each team member. Such an approach is a good start in getting humans and automation to work together more effectively, and it has the added advantage of avoiding the potential ‘technophobic’ stance of previous views. We know enough about human performance to delineate the strengths of the human component in this team, and technological progress is improving computing power all the time. Nevertheless, this only really covers the taskwork element of the team – we still need to solve the teamworking aspect of the equation. A high-profile air crash involving a new generation of automated aircraft exemplifies the teamwork problem. On 26th June 1988, the new Airbus A320 was being demonstrated at an air show in France. The pilot was making a low pass with the gear extended, but on attempting to pull out at the end of the circuit, he found that power was not available, and the aircraft crashed into trees at the end of the runway. Explanations differ according to the cause of this accident, but the automation was implicated as the pilot fought the aircraft for control. It has been said of situations such as this that instead of fighting the computer, pilots should occasionally switch it off and look out of the window. Dekker (2004) argues that we should move on from this point of view, and instead of switching it off, we should be asking ‘how do we get along together?’ For automation to be an effective team player, it needs to support coordination and cooperation, which means the human and the machine understanding each other’s perspective as well as their abilities and limitations. Hoc et al. (in press) refer to a ‘common frame of reference’ in which the human and machine understand each other’s goals and work in tandem to achieve them, but this crucially depends on information constantly flowing between the agents (cf. Griffin and Young, 2008). The essential basis of coordination and cooperation, then, is communication. This echoes Norman’s point that it is all about feedback, but the difference is that information flows in both directions. Rather than considering more or less automation, then, coordination is all about the quality of the automation, not the quantity. In order to understand what high quality automation consists of in the teamworking context, we need to revisit some frameworks of automation.

Taxonomies and tasks Up until relatively recently, approaches to classifying automation in the ergonomics literature have focused on task-based divisions of labour. The original Fitts’ list, devised in 1951, specifies those activities for which human skills surpass those of machines, and vice-versa. Later developments put forward taxonomies based on levels of authority, to decide how the tasks are divided (e.g., Sheridan and Verplanck, 1978). More recently a similar notion has dichotomised the authority into ‘hard’ and ‘soft’ automation, depending on whether the machine or the automation has ultimate control over task decisions (see e.g., Young et al., 2007). Such models

204

M.S. Young

for determining allocation of function dominated for decades, until the likes of Bainbridge and Reason shared their realisations about the impact of such fixed demarcation of tasks on human performance. The landmark of Sheridan and Verplanck’s (1978) taxonomy was in the recognition that automation is not an all-or-none option, and that by exploiting different levels of automation it is possible to assist the operator rather than replace them (Kaber and Endsley, 1997). Thus the general consensus moved on to consider adaptive automation – intelligent systems which monitor the environment and regulate the level of automation to maintain an optimal state for the operator (Byrne and Parasuraman, 1996). Whilst adaptive systems may be a laudable objective, the practicalities of transferring control and the technological barriers have so far proved insurmountable. Thus adaptive automation has fallen out of favour somewhat, to be replaced by the more contemporary concepts of operator support (e.g., Young et al., 2007). Nevertheless, the work of Sheridan and Verplanck (1978) is still proving inspirational to the latest generation of automation frameworks, which do focus on operator support from more of a teamworking perspective. Parasuraman et al., (2000) redefined the levels of automation matrix to consider the action cycle across the task – from information acquisition, analysis, decision and action selection, to action implementation. It is noted that the continuum of manual-automatic levels of authority can be applied across each of these task stages to determine the optimal balance of automation across the whole task. Hoc (2001; Hoc et al., in press) took this approach a step further in considering more explicitly the aspect of human-machine cooperation to propose a more teamwork-based allocation of function model. Hoc (ibid.) talks in terms of managing interference between the human and the machine, working together to establish that ‘common frame of reference’. The model proposes three levels of cooperation and four modes of task automation in a matrix of human-automation interaction. At the meta level, human and machine establish models of each other’s operation and behaviour, in order to provide a platform for coordination. The plan level defines rules of engagement based on task context, and criteria for automatic support or task delegation. At the action level, the behaviours and rules are implemented according to the demands of the task. The modes of operation somewhat reflect a hybrid of Sheridan and Verplanck’s (1978) taxonomy and the information processing cycle in the model of Parasuraman et al. (2000), covering perception, mutual control, function delegation, and full automation. The crucial distinction in Hoc’s model is the establishment of this common frame of reference – essentially a mental model of system operation, but one which must be held by both human and machine about the other’s behaviour. One of the key aspects of this is the awareness of context and intent – vital in interpreting the actions implemented in a given situation. Hoc et al. (in press) provide the example of a lane-keeping system in cars – which warns the driver when straying out of his/her lane. Clearly, sometimes this activity is legitimate – when overtaking, for example. The system has access to the vehicle’s electronics, though, and so only provides a warning if the driver is moving across the lane markings without having used a turn signal. Whilst this is a crude rule, it illustrates the importance of intent


205

and context in maintaining that common frame of reference and hence the smooth dynamics of the team. Although the concept of operator support is taking hold, it could be argued that in many cases designers pay it mere lip-service, skirting round the ideals of teamwork and cooperation. Staying with the example of vehicle automation, many systems are ostensibly designed to support the driver (indeed, this exonerates manufacturers from any potential litigation, since drivers are always held to be responsible), but in most cases these fall into the old trap of simply automating what can be automated. Once again, the driver has to manage the automation as well as drive the car, rather than working with the system on a common goal. Young et al. (2007) capture the principle by suggesting that true driver support should act as a human co-driver – providing advice when needed, assistance when necessary, but largely remaining invisible in the background under normal conditions. This philosophy fosters human skills for the task, and in cases where automation does have to intervene, it will communicate efficiently with the driver and work on the basis of the task context and driver’s intentions. Needless to say, technology undoubtedly has some way to go before automation can be smart enough to know what the human wants to do. In the meantime, we can set out a new philosophy for design and implementation of automated systems based on the contemporary principles of cooperation, communication and coordination.

Re-training the ghost Thus far we have seen how traditional perspectives on automation – which view the human as a passive monitor and divide the task piecemeal according to fixed rules – have been superseded by a more flexible approach based on the human and machine working as a team. In designing optimal human-human teams, the aims are to have a balance of skills and good communication and understanding between team members. Likewise for human-machine teams, the ultimate objective would be to design an automated system with complementary taskwork skills (cf. Schutte, 1999) and good teamworking abilities (cf. Hoc, 2001). The former is not so far removed from a Fitts’ list type approach, and technology is increasing the range of skills available with automation (indeed, it is high time the Fitts’ list was updated). As far as teamwork is concerned, though, the technological barriers are much higher. Training and design are two sides of the same coin, something that Rigner and Dekker (2000) realise in considering the impacts of flight deck automation on crew resource management (CRM). CRM is essentially a teamwork programme which has become a key component of flight crew training (Wiener et al., 1993). As Rigner and Dekker (2000) point out, automation affects the dynamics on the flight deck and needs to be properly integrated into the training programme – CRM should focus on utilising automation as a resource, and coordinating its input to the task as a whole. This sounds promising from our teamwork perspective, but perhaps does not accord with the guiding philosophies of ergonomics since it relies on fitting the

206

M.S. Young

human to the task. Young et al. (2007) argue for taking the concept a step further, applying the principles of CRM training to the design of automated systems. This may be an interim measure until technology allows us to fully integrate automation into the team as described above, and it is envisaged here that CRM-designed automation can be achieved in two ways. Firstly, by erring on the side of ‘soft’ automation, thus leaving the human in active control and able to delegate tasks as appropriate – in line with the frameworks proposed by Parasuraman et al. (2000) and Hoc (2001). Secondly, as argued earlier in this paper, the teamworking aspect ultimately comes down to communication in both directions – which means a significant design effort on the control-display interface to optimise the flow of information (cf. Griffin and Young, 2008). As the ‘ghost in the machine’, automation can sometimes be seen as having a mind of its own – a problem which is all too evident in some of the incidents described earlier in this paper. A teamworking approach, using the principles of CRM to emphasise coordination, cooperation and communication, can serve to overcome these problems in human-automation interaction and lead us forward to a more ‘autopian’ ideal (cf. Hancock et al., 1996). Bringing the arguments full circle, this is not about a technophobic reluctance to automate – in fact, it is quite the opposite. Both Norman (1990) and Dekker (2004) would agree that automation has been ‘thintelligent’, and we could actually see a case for ergonomics driving technological progress, rather than the other way around. If we are to design automation to fit with a human team member, we need to reverse Fitts’ list and define those areas of human weakness where technological support is necessary – for example, hazard perception in driving. This will require a much greater effort than simply automating what we can, but it enables us to define how we should. Where Wiener and Curry (1980) argued that automation had already passed its optimal point, instead it seems the best is yet to come.

References Bainbridge, L. 1983, Ironies of automation, Automatica, 19(6), 775–779. Byrne, E. A., and Parasuraman, R. 1996, Psychophysiology and adaptive automation. Biological Psychology, 42, 249–268. Cannon-Bowers, J. A., and Salas, E. 1997, Teamwork competencies: the interaction of team member knowledge skills and attitudes. In O. F. O’Neil (ed.), Workforce readiness: Competencies and assessment (Erlbaum, Hillsdale, NJ), 151–174. Crichton, M. 1990, Jurassic Park (Knopf, New York). Dekker, S. 2004, On the other side of promise: what should we automate today? In D. Harris (ed.), Human factors for civil flight deck design (Ashgate, Aldershot), 183–198. Endsley, M. R. 1995, Toward a theory of situation awareness in dynamic systems. Human Factors, 37(1), 32–64. Faith, N. 1996, Black box: the air crash detectives – why air safety is no accident (Boxtree, London).


207

Griffin, T., andYoung, M. S. 2008, The flow of information. Clued Up (2008 Issue), 62–67. Hancock, P. A., Parasuraman, R., and Byrne, E. A. 1996, Driver-centred issues in advanced automation for motor vehicles. In R. Parasuraman and M. Mouloua (eds.), Automation and human performance: Theory and applications (Lawrence Erlbaum Associates, Mahwah, NJ), 337–364. Hoc, J. M. 2001, Towards a cognitive approach to human-machine cooperation in dynamic situations, International Journal of Human-Computer Studies, 54, 509–540. Hoc, J. M., Young, M. S., and Blosseville, J.-M. in press, Cooperation between drivers and automation: Implications for safety, Theoretical Issues in Ergonomics Science. Kaber, D. B., and Endsley, M. R. 1997, Out-of-the-loop performance problems and the use of intermediate levels of automation for improved control system functioning and safety, Process Safety Progress, 16(3), 126–131. Molloy, R., and Parasuraman, R. 1996, Monitoring an automated system for a single failure: vigilance and task complexity effects. Human Factors, 38(2), 311–322. Norman, D. A. 1990, The ‘problem’ with automation: inappropriate feedback and interaction, not ‘over-automation’, Phil. Trans. R. Soc. London B, 327, 585–593. Parasuraman, R. 1987, Human-computer monitoring, Human Factors, 29, 695–706. Parasuraman, R., and Riley, V. 1997, Humans and automation: use, misuse, disuse, abuse, Human Factors, 39(2), 230–253. Parasuraman, R., Sheridan, T. B., and Wickens, C. D. 2000, A model for types and levels of human interaction with automation, IEEE Transactions on Systems, Man and Cybernetics – Part A: Systems and Humans, 30(3), 286–297. Reason, J. T. 1990, Human Error (Cambridge University Press, Cambridge). Rigner, J., and Dekker, S. 2000, Sharing the burden of flight deck automation training, International Journal of Aviation Psychology, 10(4), 317–326. Schutte, P. 1999, Complemation: an alternative to automation, Journal of Information Technology Impact, 1(3), 113–118. Sheridan, T.B., and Verplanck, W.L. 1978, Human and computer control of undersea teleoperators (MIT Man-Machine Systems Laboratory Report, Cambridge, MA). Wiener, E. L., and Curry, R. E. 1980, Flight-deck automation: promises and problems. Ergonomics, 23(10), 995–1011. Wiener, E. L., Kanki, B. G., and Helmreich, R. L. 1993, Cockpit resource management (Academic Press, San Diego). Young, M. S., and Stanton, N. A. 2002, Attention and automation: new perspectives on mental underload and performance, Theoretical Issues in Ergonomics Science, 3(2), 178–194. Young, M. S., Stanton, N. A., and Harris, D. 2007, Driving automation: learning from aviation about design philosophies, International Journal of Vehicle Design, 45(3), 323–338.

AUTOMATION AND TECHNOLOGY IN 21ST CENTURY WORK AND LIFE S. Sharples Human Factors Research Group, Faculty of Engineering, University of Nottingham, Nottingham NG7 2RD, UK Technology is pervasive in 21st century life. Many types of automation technology change the ways in which we complete tasks in our work and home. Bainbridge (1983) identified a set of “ironies of automation”. These ironies were identified in the context of process control operations, but we now encounter automated systems in a wide variety of contexts, ranging through the home and work environments. Users of automation also often have a degree of choice as to the extent to which they use the automation available. This paper analyses the impact of automation in three different case examples – rail signalling automation, handheld computing for pedestrian navigation, and word processing interfaces, and considers the relevance of three of Bainbridge’s ironies to modern automation. A new irony, that addresses the need for the user to choose to seek to actively understand how the automation works in order to fully reap its benefits, despite the aim of that automation potentially being to actually conceal the complexity of its workings, is proposed for consideration.

Ironies of automation Bainbridge (1983) wrote a landmark paper that identified a set of ironies of automation – based around the consequence that “the more advanced a control system is, so the more crucial may be the contribution of the human operator”. These ironies covered the effect of automation on manual and cognitive skills, operator attitudes and considered both the impact of requiring an operator to monitor whether an automated system is working effectively, and the implications for the operator if they are required to intervene in control of the automated system. The nature of the proposed ironies were generally the unintended side-effects of automation that in many of Bainbridge’s examples actually potentially worsened, rather than improved, the situation for the operator and the effectiveness of the system as a whole. The systems primarily considered by Bainbridge were those that typically aimed to “replace human manual control, planning and problem solving by automatic devices and computers”. She acknowledges that even these highly automated systems still involve humans in a number of roles, including monitoring, maintenance and future development. The systems considered by Bainbridge primarily fall into higher level of automation, categorised for example by Parasuraman et al. (2000) as ranging from automatically executing actions and informing the human operator, to 208

Automation and technology in 21st century work and life

209

the computer deciding everything, acting autonomously and “ignoring the human”. More and more however we encounter automated technologies that rather than taking over task elements, advise or guide us on how to complete tasks, such as in navigation support systems or integrated automation within word processing applications. Critically, we also often have the opportunity to choose the manner and extent of use of the automation available. This paper specifically considers three of the ironies identified within Bainbridge’s landmark paper, and discusses their relevance to a set of example recent case applications. In addition, the paper discusses whether any new ironies are emerging with the use of current pervasive automated technologies. The three ironies of automation that will be specifically considered are (paraphrased from Bainbridge (1983): 1. A designer who tries to eliminate the operator still leaves the operator to do the tasks which the designer cannot think how to automate. 2. An automatic control system has been put in because it can do the job better than the operator, but the operator is being asked to monitor that it is working effectively 3. An operator will only be able to generate successful new strategies for unusual situations if he has an adequate knowledge of the process

Case 1: Railway signalling automation Rail signalling presents a particularly interesting context for examining automation due to the variety of technologies in use, ranging from manually controlled lever frames to VDU based automatic routing systems. Signalling systems that include automatic route setting technologies have been analysed in a number of different studies over the past ten years (e.g. Nichols et al., 2000; Balfe et al., 2007; Balfe et al., 2008). A typical VDU based signalling interface is shown in Figure 1. When Automatic Routing System (ARS) is used, the train routes are automatically set, based on timetabling information, and conflicts between two trains which are scheduled to travel on the same track resolved by conflict resolution algorithms. The route set by ARS is indicated by an area of the track in front of the train being highlighted, and signallers have the opportunity either to anticipate the route to be set by setting the route in advance of ARS, or by using functions within the system to force the automatic system to prioritise particular trains. When considering the ironies of automation, it is perhaps no longer fair to describe the elements of the task that are not automated as being those that the designer “cannot think” how to automate. Theoretically, if all train movements are timetabled and all information in the ARS system is up to date, the operator should be able to simply monitor the movements of the trains and not have to intervene. However, observations of and interviews with signallers (Balfe, in preparation) suggest that this is rarely or never the case. So why do operators still end up actively controlling the system? Reasons include that the system cannot automate all the train movements or the operator chooses to anticipate or override the automatic route setting. An example of this is that

210

S. Sharples

Figure 1. A VDU based rail signalling interface (black and white image of colour interface).

some train movements (e.g. movement of train stock into a depot) may not be timetabled, or trains may run so early or late that their timetabling information can no longer be used effectively. A signaller may also choose to anticipate ARS and set the route before the automatic system intervenes, or may manually block parts of track to force ARS to select an alternative, for reasons such as restrictions in track availability. An example of this is as follows: if a slow train is late running, the automatic system may hold up an express train in order to allow the slower train (scheduled to travel through that part of track at an earlier time) to pass. In fact, the signaller may prefer to allow the express train to travel onto the track first, as the slow train is already late, and this action presents the express train also experiencing a delay. This may only be achieved by a manual intervention with the system. This example also illustrates the relevance of the other two ironies – the automatic route setting is introduced to enable a single signaller to control a larger area of track, yet the signaller retains the responsibility to monitor that the route setting is working effectively (i.e. minimising delays on the area under control). Interviews conducted by Balfe (in preparation) particularly illustrate the relevance of the operator’s understanding of the automation technology and conflict resolution algorithms –signallers are not explicitly trained in ARS and instead learn “on the job”. Whilst some operators may report that they understand how the conflict resolution operates, reports of the rules themselves vary indicating a lack of transparency of the system and its underlying rules. One of the most striking things observed in visits to signal boxes however is the wide variation in attitudes towards and strategies used in completing the signalling task. All signallers have safety as an absolute priority and have confidence in the underlying interlocking that prevents conflicting routes being set. However, detailed observations of signaller activities (e.g. Nichols et al., 2000; Balfe et al., 2008) illustrate that different signallers at the same workstation will vary in the extent to which they manually set routes. Signallers appear to differ in the extent to which


211

they allow ARS to “take control” – some try to anticipate all routes and “beat” ARS to setting the routes – others are happy to monitor and only intervene when they wish or need to. Signallers do have the opportunity to “query” ARS to identify why certain route settings have been made. The use of this facility, along with discussion with colleagues and on-the-job training, contributes to the signaller’s mental model of how the signalling automation works. This first case suggests that all three of the ironies under consideration do apply to automation in rail signalling and that, in addition, even in an example task that is quite close to the “process” tasks originally considered by Bainbridge, the individual has control over the way in which they use the automation to support their decision making. Rather than simply monitoring the process, the signaller has an opportunity to interact with the system and the automated support to work with the automation.

Case 2: Pedestrian geographical information support The second case of automation to be considered is pedestrian geographical information support. Work conducted by Nixon et al. (2008) has examined the way in which the design of handheld navigation devices affects navigation performance, and particularly confronts the challenge of displaying data on small screen displays. As display screens, batteries, processors and control interfaces are becoming smaller, the capability to integrate real time geographically linked information sources is becoming a reality; yet we still do not have consistent methods of presenting integrated spatial and non-spatial information in order to effectively support navigation tasks. Work with the fire service has identified potential for providing valuable support in safety critical environments, such as indicating presence of gas mains on a CAD diagram of a building to be evacuated, accompanied by updated registers of those expected to be working in the building at the time of evacuation. We are consistently presented with the challenge of presenting increasingly complex data in a shortened time frame on a smaller screen – if this challenge can be overcome then vital automated support can be provided in a safety critical environment. A prototype interface to be used in a pedestrian navigation guidance system is shown in figure 2. If actively implemented this system would use positioning technology to locate both the position and orientation of the user, and could work on the same route setting basis as an in-car system, where the user inputs their required destination, and the system automatically calculates the route needed to get to that destination most efficiently. If used in firefighting, theoretically this could be integrated with information about fire exits, or even predict the likely spread of fire or smoke. Such technology is a good example of automation of a task that has previously been conducted using a variety of artefacts in the real world, including maps or plans, local knowledge or even expectations of building or road layout based on previous experience of similar environments. The task of pedestrian navigation combines the use of knowledge in the world and knowledge in the head (Norman, 1988), where the use of electronic handheld navigation support constitutes an element of the “knowledge in the world”. Rogers and Scaife (1999) describe the use of

212

S. Sharples

Figure 2. Three variations of a prototype handheld pedestrian navigation display (black and white image of colour interface) (Nixon et al., 2008).

such technology to support active decision making as “computational offloading” – signifying the use of technology to reduce the amount of cognitive effort needed to solve problems. Artman and Garbis (1998) also highlight the importance of use of real world artefacts in naturalistic decision making in emergency situations. In the example of pedestrian navigation support, all elements of the navigation task are automated, leaving the user to only complete the physical interaction with the environment and check on reaching the destination that they have in fact reached the correct location. Therefore the first irony being considered, that the tasks that are hard to automate are those that still remain, is not so relevant here. However, the operator is to some extent still required to monitor that the automation is working correctly. This particular task very much relies on the combination of information in the world, head and on the interface being used. Consider the example of using a device to locate an office in a large tower building. If the system directs you towards floor 12, but on arrival on floor 12 you are greeted by company insignia that you do not recognise, you may assume that you have been directed inappropriately. It is then up to the user to decide which information to trust and act upon – that in the world, or that on the interface. Finally, the irony related to generation of new strategies on the basis of adequate knowledge of the process can be considered to be relevant if we consider the “process” to refer to the broader context of development of geographical knowledge. Whilst Zimmer (2004) showed that users of fragmented maps developed as high a level of spatial knowledge as those presented with full maps, we know that if knowledge is in the world and we do not need or try to transfer it to our memory then we do not tend to do so. An illustration of this is in mobile phone use, where we do not tend to memorise phone numbers that are stored in the phone’s memory. Indeed, in the context of in-car navigation systems, Burnett & Lee (2005) found that “drivers who received typical vehicle navigation system instructions remembered fewer scenes, were less accurate in their ordering of images seen along routes and drew simpler maps which included a smaller number of landmarks, when compared


213

to those using traditional methods.” Again, studies have shown large individual differences in the use of such systems, and also in the use of the landmarks in the real world to develop knowledge of the geographical layout of the environment. This second case particularly emphasises the continued relevance of the second and third ironies being considered – that an automatic control system has been put in because it can do the job better than the operator, but the operator is being asked to monitor that it is working effectively and that an operator will only be able to generate successful new strategies for unusual situations if he has an adequate knowledge of the process. The need for a broader view of the “process” is highlighted, with the general development of geographical knowledge or routes to be navigated being more appropriate for consideration. Again, rather than considering the user as a monitor of the system, the user is working with the system and other artefacts in order to construct knowledge and actively make decisions.

Case 3: Word processing interface support tools The final examples of automation to be considered are standard tools embedded with computer interfaces such as word processing. Within computer interfaces we are frequently offered different types of automated support, from automatic form fill-in, memory for passwords, graph drawing wizards, display only of frequently used menu items, automatic text formatting and paragraph spacings, as well as the Microsoft “paperclip”. However, as Lane et al. (2005) have pointed out, “even experienced users are inefficient in their use of graphical interfaces”. We frequently observe that these automated features are disabled or overridden, suggesting that they may not in fact be achieving their aim. Can the ironies of automation explain why? As we have witnessed applications such as Microsoft Word develop over the years, the level of automation has increased. However, there are still aspects of the word processing task that do not have automatic support. For example, Word 2003 has the potential to automatically recognise, and, if wished, correct some spelling mistakes, but it is not able to automatically check that the spelling of an individual’s name is correct if this name is not stored in the dictionary, and may actually inappropriately correct a spelling of a name that is very close to the spelling of something that is in the dictionary. In other words, “easy” spellings are automatically corrected, but “difficult” ones are not. The second irony, that the user is required to monitor that the automation is working correctly, also applies to word processing, and the added complexity here is that the third irony is also relevant – an adequate knowledge of the system is needed in order to develop new strategies. This is best illustrated by a separate example. If a user wishes to indent some text, then a person who is applying a “manual” metaphor will use the tab key, but one who is taking advantage of and embracing the application’s automatic functionality will select the “indent” icon or manipulate the paragraph format via the paragraph menu. As long as the space and paragraph characters are not visible, the layout will look the same, but the text might behave differently when edited (e.g. the formatting of one paragraph may influence the formatting of the subsequent added paragraph). In order to understand why

214

S. Sharples

the document is behaving in a certain way, and to avoid the presence of a form of latent error within the document, the user is required to have an appropriate mental model of the automatic formatting that has been applied. However, to develop the mental model they may have to either explore the document, by viewing paragraph and spacing marks, or analysing menu settings. But an action such as displaying paragraph and spacing marks results in a more cluttered display that may be more visually confusing and unpleasant, so users may be reluctant to take this route. Again, we see, and indeed many interfaces encourage and support, large individual differences in the way in which such applications are used. Individuals have the opportunity to choose their own menu layouts, visibility of paragraph markings, use of automatic support tools such as the Microsoft paperclip, and automatic formatting tools such as automatic spelling correction, formatting and heading identification.

Review of Ironies of Automation in current systems Whilst the wording of the ironies of automation relate to monitoring systems in general, the cases presented in this paper illustrate that the three ironies considered in detail are still relevant in current automation systems and can be applied to a broader context than process control. It may be appropriate to change the wording (paraphrased from the Bainbridge paper) from monitor to user, and from process to system. Indeed, the distinction between the user and the automation becomes blurred when we consider the overall process of decision making, and it is more appropriate to adopt a joint cognitive systems perspective (Hollnagel 2001). A key difference between the type of automation technology now being used from that considered by Bainbridge is the element of user choice – in most systems, ranging from setting our home video to dialling a number on our mobile phone, we have the opportunity to choose whether we use the automated approach or a more manual approach.

A new “irony”? However, the cases also suggest the need for a potential new irony of automation. The ironies proposed by Bainbridge that considered operator attitudes covered the issues of when a job is perceived as being “deskilled” by being reduced to monitoring, negative effect on operator attitudes, and that deskilled workers may insist on high pay level as remaining symbol of status which is no longer justified by job content. However, in addition to these issues, the cases above prompt the proposal of a possible new irony: Whilst automation may aim to “protect” the user from the complexity of the system, it is often only when a user enthusiastically explores the complexity of a system that they can often understand how the automation works, and thus harness its full potential. Automation is often only at its most effective when it is enthusiastically embraced and actively attended to by the operator. Automation is usually intended to reduce


215

the workload of the operator, and possibly not require them to attend to all the information that they may have needed to use with a non-automated system. In practice a user can only extend their understanding of the system, general knowledge and skill, to understand how the automation works and to enable them to complete more complex or unusual tasks, by actively choosing to attend and work to understand how the automation works. This motivational element is critical to the success of automation, particularly in systems which are not essential to use, and may well enhance the user experience (e.g. by producing a more neatly formatted word document, avoiding an automatic route setting system rigidly applying rules, or taking a more efficient navigation path). Users may need to be willing to change the way they work, to go through a learning curve (at the very least by reading instructions or help files, exploring different menus or observing expert users) and may initially become worse at a task before they can take advantage of the automation. They may also need to choose to attend to information to which the system does not in fact require or expect them to attend or to actively work to understand the way in which the system works to allow them to form a useful mental model.

Summary of ironies applied to case examples The wording of the three original and new ironies can therefore be summarised as follows, to reflect the wider context of current automated systems: • The designer who tries to eliminate the user still leaves the user to do the tasks which the designer cannot think how to automate • An automatic control system has been put in because it can do the job better than the user, but yet the user is being asked to monitor that it is working effectively • A user will only be able to generate successful new strategies for unusual situations if he has an adequate knowledge of the system • Automation may only maintain or extend operator skills or knowledge when it is embraced and actively attended to by the operator If we are to design systems that support operations effectively, it may be appropriate to in fact consider only two types of system: fully automated (i.e. with no user at all) or one where the user is always in control, with automation confined to assistance with information capture and data handling in order to enhance the speed and accuracy of decision making. Any human involvement will always require the user to have a good model of the system, so if this is built into the normal operation of the system the user is likely to be more alert and satisfied with their system use.

Conclusion The cases examined in this paper suggest that the ironies of automation considered do still have relevance to current automation technology. However, the concept of a “monitor” is now more generally a user, who may be required to conduct some monitoring, but is also likely to be more actively in control in some elements

216

S. Sharples

of the task, and will work with the automation, along with other technologies and artefacts, as part of a joint cognitive system. Similarly, the original paper by Bainbridge focussed on the “process” whereas it is now more appropriate to phrase the statements in terms of the overall system. In all three cases, the concept of individual differences in preferences for a style of automation use is evident. This, coupled with the element of user choice over the use of automation, leads to the consideration of a new irony, that despite the aim of some automation systems to protect the user from its complexity, it is only by understanding this complexity that the user is able to form a mental model and take advantage of the full capability of the automation by working with it as part of a joint cognitive system. This requires a deliberate act on the part of the user to embrace the automation.

References Artman, H. & Garbis, C. (1998) Situation Awareness as Distributed Cognition. In Cognition and Co-operation. Proceedings of the 9th Conference of Cognitive Ergonomics. Limerick, Ireland. Bainbridge, L. (1983). Ironies of automation. In J. Rasmussen, K. Duncan & J. Leplat (Eds.), New Technology and Human Error (pp. 271–283). Chichester: Wiley. Balfe (in preparation) Automation in rail signalling. PhD thesis, University of Nottingham. Balfe, N., Wilson, J.R., Sharples, S. & Clarke, T. (2007). Analysis of current UK rail signalling systems. Paper presented at the Human Factors and Ergonomics Society European Chapter conference on Human Factors for Assistance and Automation, Braunschweig, Germany, October 2007. Balfe, N., Wilson, J.R., Sharples, S. & Clarke, T. (2008). Structured Observations of Automation Use. In P. Bust (Ed) Contemporary Ergonomics 2008. Burnett, G.E., and Lee, K. (2005) The effect of vehicle navigation systems on the formation of cognitive maps, in G. Underwood (Ed.) Traffic and Transport Psychology: Theory and Application. Elsevier. Hollnagel, E. (2001). Cognition as Control: A Pragmatic Approach to the Modeling of Joint Cognitive Systems. Special issue of IEEE Transactions on Systems, Man and Cybernetics – Part A: Systems and Humans. Lane, D.M., Napier, H.A., Peres, S.C. & Sandor, A. (2005) Hidden costs of graphical user interfaces: failure to make the transition from menus to icon toolbars to keyboard shortcuts. International Journal of Human-Computer Interaction. 18(2), 133–144. Nichols, S., Bristol, N. & Wilson, J.R. (2001) Workload assessment in Railway Control. In D. Harris (ed) Engineering Psychology & Cognitive Ergonomics. Artman & Garbis (1998). Nixon, J. Sharples, S. & Jackson, M. (2008) Less is more? Navigating with different types of information on a small-screen device. Paper accepted for presentation at the Human Factors and Ergonomics Society’s 52nd Annual Meeting. New York: September 22–26.


217

Norman D.E. (1988) The Psychology of Everyday Things. New Your: Basic Books. Parasuraman, R., Sheridan, T. B., & Wickens, C. D. (2000). A model for types and levels of human interaction with automation. IEEE Transactions on Systems, Man, and Cybernetics – Part A: Systems and Humans, 30(3), 286–297. Rogers, Y. & Scaife, M. (1999) How can interactive multimedia facilitate learning? In Lee, J. (Ed.) Intelligence and Multimodality in Multimedia interfaces: Research and Applications. Menlo Park, CA: AAAI Press. Zimmer (2004) The Construction of Mental Maps Based on a Fragmentary View of Physical Maps. Journal of Educational Psychology 96(3), 603–610.

VIGILANCE AND HUMAN SUPERVISORY CONTROL – A POTTED HISTORY J.M. Noyes University of Bristol, UK The interaction of humans and machines in terms of task allocation and control issues is an important consideration in the design and operation of systems. With advances in technology and greater possibilities concerning ‘intelligent’ systems, the role played by the human is crucial to the success of the system. Too much human control, and the benefits of automation may not be realised; too little, and the human supervisor may be at a loss to deal with failures and malfunctions when they occur. This paper is the first of three papers being presented here on this topic. Its aim is to review the development of supervisory control since the 1940s in order to set the scene for the following two papers.

Clock watching A major milestone in the study of human supervisory control occurred in 1948, when Norman Mackworth devised the simple but very elegant clock test in order to evaluate vigilance in UK military personnel (Mackworth, 1948). Participants had to observe a clock-like device and look for double-step movements (the target) among single-step movements (the non-target). Typically, Mackworth found that there was a decline in detection rate with time on task and an increase in detection time. By carrying out a series of experiments, he demonstrated that individuals could improve their vigilance performance by having shorter working periods, taking breaks, being given feedback, and in the short term, by ingesting amphetamines (Mackworth, 1950). Mackworth’s work is seminal for various reasons. It draws attention to the problem of vigilance when humans are monitoring machine-based systems. The Second World War, unlike the First World War, had been fought mainly in the air, and this had necessitated equipment being designed and implemented at great speed without sufficient opportunity for testing and revision. As an example, radar operators had been placed on 8-hour shifts, and Mackworth had observed that a large number of targets were being missed after a relatively short period of time. Surprising though it may seem now, knowledge about human attention span and monitoring skills, especially when checking for low frequency targets, and the application of this knowledge to the workplace, was limited. Hence, Mackworth began to lay the way by studying human interactions with technology, and in particular, the allocation of functions issues which still test us today. 218

Vigilance and human supervisory control – a potted history

219

The ‘big issue’ in a world which has seen increasing use of technology over the last 60 years concerns human supervisory control. Too much human control and the benefits of automation are not realised; too little and the human operator is not ‘in the loop’ when failures and malfunctions occur. This is particularly evident on the civil flight deck where computer-based systems can very safely control the aircraft (and often with greater accuracy than the human pilot). However, when problems arise, systems will usually shut down one by one, before handing (ultimate) control back to the human. This person (or persons) then has to rectify a situation but with various gaps in their knowledge about what exactly has gone wrong. Thus, automation-induced dependence needs to be avoided. Can history help us? Automation is not a new problem. How over the last 60 years have we attempted to address this issue of human supervisory control? This paper will consider the historical development of advanced technology with particular reference to automation, and human and machine control of functions to see what we have learnt from the past.

1. Automate as much as possible In the immediate post-war years of the 1940s and ‘50s, the trend was to automate as much as possible. This was summed up by Bainbridge (1987, p. 271) who suggested that “the classic aim of automation is to replace human manual control”. The drive to automate may have been in part due to the novelty and newness of having systems and machinery, which could replace human activities. The obvious example is the development of computers around this time; for example, the replacement of the vacuum tube with the transistor heralded the second generation of computers, and hence, the creation of the transistor in the late 1940s as been cited as one of the greatest inventions of the 20th century (see, Herrick, 2003). Another factor driving the ‘automate as much as possible’ approach is likely to have been human error. Humans make errors, and generally, errors have negative connotations. One way to prevent human error is to keep people away from the system; ensuring that they do not have direct contact should mean they cannot make errors. In theory, this will work up to a point, since it is difficult, if not impossible, to remove human involvement completely from a system. Take a highly automated nuclear power plant. Humans may not be carrying out minute-to-minute monitoring activities, but they will be responsible for overall control, and stopping production should the need arise. In terms of human supervisory control, as a general rule, tasks which did not lend themselves to automation were allocated to humans. This subsequently became known as the ‘left-over approach’; that is, humans ended up with the tasks which the systems could not do. Not surprisingly, this approach had a number of drawbacks; the most obvious one being that the human was then left to do the tasks which were impossible for the automated system. Pragmatically, this does not make sense. It is likely the tasks which are difficult for the system are equally problematic for the human operator. As an example, take a commercial aircraft landing in adverse weather conditions. This may prove a hazardous exercise regardless of whether a

220

J.M. Noyes

manual or auto-pilot landing is attempted, and indeed, there will be some situations where a manual landing has to take place in preference to an auto-pilot landing, and of course, vice versa. In summary, the trend in the immediate post-war years was to automate as much as possible but this inevitably resulted in tasks, which could not be carried out by the system, being allocated to the human. This inevitably creates problems for the human in terms of supervisory control as they are likely to end up with the ‘hard’ tasks which are not possible for execution by the system.

2. Automate according to human and machine characteristics In the 1950s, the so-called ‘left-over approach’ was modified into one which took into account the attributes of humans and machines, and allocated the various functions according to the strengths of each faction. This was based on the work of Paul Fitts; he considered a number of functions relevant to the operation of a system in terms of how humans and machines would fare, and this became known as a Fitts’ List (Fitts, 1951). An example of some of the functions, which might be considered, is given in Table 1.

Table 1.

Example of a Fitts’ List approach (from Noyes, 2001).

Function

Humans

Machines

Speed

Can work fast for short bursts, limited reaction times, comparatively slow, for example, seconds

Accuracy

Can be extremely accurate, but again difficult to sustain over periods of time Soon tire, subject to learning, fatigue and boredom. Poor, especially when required to sustain attention over long periods

Can work fast for long periods of time, much faster reaction times, for example, up to the speed of light Can maintain high standards of accuracy over long periods of time Excellent at repeating actions in a consistent manner Not an issue – if programmed to monitor, will continue to do this ad infinitum Poor (unable, for example, to carry out simple tasks such as edge detection) Completely reliable in that all items stored in memory can be quickly accessed, limited (but extremely large) capacity, formal structure Rule-based approach, good at deductive reasoning, poor in new situations Limited

Repetitive actions Attention span

Perceptual skills

Excellent especially at pattern detection and interpretation

Memory

Selective, impressive in that unprompted instant retrieval of items is possible, not always reliable, versatile and innovative

Decision making

Tend to use heuristics (‘rules of thumb’), good at inductive reasoning when need to draw on experience Excellent

Intelligence


Table 2.

221

Some shortcomings of the ‘list’ approach to allocation of function (from Noyes, 2001).

1. Some functions have to be performed by humans because machines cannot do them, and vice versa. 2. There will be some functions where allocation is not obvious, that is, they might be carried out by humans or machines or both. 3. It implies that tasks which machines do well need to be assigned to them leaving the remainder for the humans. 4. Direct comparison of attributes suggests that machines and people have opposite capacities. Obviously, this is not the case. 5. Some functions must always be performed by the human operator, for example, abort control. 6. Some functions may be allocated to the humans in order to make their jobs more interesting. 7. Technology changes faster than the Lists. 8. It does not cover all the reasons why the human may be allocated a ‘machine’ function. 9. Being skilled at a task is only one facet/dimension. 10. Humans have a need to work and feel valued. 11. They are also unique in that they have an emotional component. 12. Humans may have a preference to interact with other humans rather than machines.

This approach is also often referred to as a ‘compensatory principle’ (see, Fitts, 1951); this presumably derives from the fact that human and machine characteristics will compensate for each other. For example, humans are good at deductive reasoning whereas machines (computers) are better at inductive reasoning, so in a well-designed system, one aspect should compensate for the other to ensure the optimum type of reasoning is taking place. Later, these types of comparisons becoming known as MABA-MABA (“Men [sic] Are Best At – Machines Are Best At”) lists (see, Berry & Houston, 1993). The relatively simple approach to allocation of function manifested in the use of these types of lists has persisted over many decades. And as noted by Sheridan (2000, p. 203), they have been referred to as “a kind of gospel” for allocating functions to humans and machines. However, it is unlikely that Fitts and others intended them to be viewed in this way or indeed even for the list to be used for the allocation of tasks to humans and machines. Further, it is implicit in Fitts’ and MABA-MABA lists that humans and machines need to be compared in some way and this may not be the case. These are a few of the many shortcomings associated with the list approach; others are given in Table 2. In summary, allocation of function according to human and machine characteristics provides a relatively simple approach albeit with a number of shortcomings; however, it was an advance on the previous approach of automating as much as possible.

3. Automate flexibly taking into account overall system operation The 1970s led to a more integrative approach. It was recognised that human and machines/systems have their respective strengths and weaknesses, but they are both

222

J.M. Noyes

part of a common goal in terms of total system operation (see, Price, 1985). Thus, the emphasis shifted towards co-operation and the development of a symbiotic relationship between the two main players. This has become known as a complementary approach (Hollnagel, 1999). Engendered in this approach is flexibility; functions need to be allocated according to the goals and characteristics of the system at that time. On the flight deck, there may be occasions when the same task can be safely carried out by the human, and times when the systems with their ability to sample every number of milliseconds with great accuracy and then make appropriate adjustments are safer. Alternatively in an emergency, humans with their ability to reason inductively taking into account, for example, previous knowledge and understanding, may fare better. As the adage states, flight deck crew have an incentive to get home safely, which the computers lack. Further strives were also made in the 1990s concerning human error. As already mentioned, earlier systems were developed with an error-resistance approach; that is, humans make errors, so if at all possible, they need to be kept away from the machines. This was replaced with a view which accepts that human operators will make errors and this needs to be taken into account when designing the system and subsequent human-machine interactions. This became known as error-tolerance (Billings, 1997). Two simple examples might include the ‘undo’ key in word processing packages, and the ‘are you sure’ request when executing an operation which may have serious consequences (e.g. deletion of files). In summary, a more flexible approach to supervisory control came into existence in the 1970s and this probably still exists today. It is certainly evident in the approach taken to accommodating human error in the design of systems, which represents a U-turn in terms of the role of the human operator.

4. And in the 21st Century . . . The notion of flexibility persists. The human is still at the centre of activities even with highly automated systems. Take signalling on the rail network. The operator can choose to anticipate or override the automatic route setting (which they might due given their greater knowledge to the signalling system concerning train movements and track availability). Flexibility is also evident in the work on automation. For example, Kaber and Endsley (2004) developed a taxonomy comprising 10 levels of automation ranging from manual control to full automation for four different human/computer tasks (monitoring, generating, selecting, implementing). Moreover, they also suggested that other factors such as situation awareness, workload, and trust/confidence in the system are relevant when considering human supervisory control issues (see, Parasuraman, Sheridan, & Wickens, 2008, for a recent paper on these very three factors). If we take the first factor, situation awareness, this is a relatively new term developed in the 1990s for what is an old concept which was recognised during the Second World War (Watts, 2004). Although there are numerous definitions for


223

situation awareness, if taken at face value, there are obvious implications in terms of its benefits in supervisory control. Like situation awareness, workload (both cognitive and physical) and its measurement have a long history; however, it was not until the 1980s that some of the self-report measures for subjective workload gained maturity. For example, the development of the Modified Cooper-Harper scale (MCH; Wierville & Casali, 1983), the Overall Workload (OW; Vidulich & Tsang, 1987), the NASA-TLX (Task Load Index; Hart & Staveland, 1988), the Subjective Workload Assessment Technique (SWAT; Reid & Nygren, 1988). Trust and confidence in the system, that is, believing in the accuracy of the information being displayed, is also an important factor. As a general rule, operators tend to make the assumption that the information is correct; that is, they trust the automation. At a pragmatic level, this makes sense. In an aircraft at 35,000 feet, it is not useful to begin questioning the automation. However, studies by Muir and Moray (Muir, 1994; Muir & Moray, 1996) have shown some interesting findings with regard to operators’ trust; for example, they showed that there was an inverse relationship between trust and monitoring of the automation. They went on to suggest that “operators’ subjective ratings of trust and the properties of the automation which determine their trust, can be used to predict and optimize the dynamic allocation of functions in automated systems” (Muir & Moray, p. 429). As highlighted by the Parasuraman et al. (2008) these three factors are more recent developments and for consideration in supervisory control. A further addition, which has not attracted much interest from designers in recent times, is user expectations (see, Noyes, 2006). Expectations (i.e. how operators expect a system to perform) are known to be an important aspect of human-machine interaction. Not taking into account, people’s expectations can lead to a mismatch with their perceptions; this can lead to frustration and irritations as well as errors when humans interact with the system. Noyes suggested that past experience and knowledge facilitate development of expectations and these are important determinants in our attitudes towards technology. She described a questionnaire survey of the expectations and perceptions of automated warning systems by British Airways flight deck crew. It was found that the automation was more favoured by crew who had experience of using warning systems with a greater degree of automation. Conversely, those without this experience had more negative expectations. Thus, lack of exposure to a system may lead to more negativity. As a final comment, it is suggested that teamworking needs also to be considered. Traditionally, experimental research has focused on performance of the individual, especially in the laboratory setting, and in fact, there is scant mention of teams as opposed to individuals when considering the history of supervisory control. It would appear that this needs to be addressed more fully in future studies.

What can we learn from history? Over the decades, different approaches to supervisory control have been taken with varying degrees of involving/excluding the human. However, there is no straightforward or obvious answer in terms of how best to allocate function between the

224

J.M. Noyes

human and the system. In terms of system operation and human involvement, the general move has been towards flexibility and a symbiotic approach whereby humans and machines work in conjunction with each other. Certainly, history seems to have taught us that you cannot take the human ‘out of the loop’. Ultimately, all machine-based activities are controlled by humans at some stage (from the initial ideas through conceptual design and development to operation, maintenance and decommissioning). It may be that this situation will change in the future, but for obvious reasons, it is probably unlikely. A starting point, therefore, seems to be to begin with the human and ensure the human operators remain at the centre of machine-based activities, and throughout the life cycle. Ignoring the human is not an option.

References Bainbridge, L. 1987, Ironies of automation. In J. Rasmussen, K. Duncan & J. Leplat (eds.) New Technology and Human Error, (John Wiley & Sons Ltd., Chichester, England) Berry, L.M. and Houston, J.P. 1993, Psychology at Work: An Introduction to Industrial and Organisational Psychology, (Brown & Benchmark, Madison, WI) Billings, C.E. 1997, Aviation Automation: The Search for a Human-centred Approach, (LEA, Mahwah, NJ) Fitts, P.M. 1951, Engineering psychology and equipment design. In S.S. Stevens (ed.) Handbook of Experimental Psychology, (Wiley, New York) Hart, S.G. and Staveland, L.E. 1988, Development of NASA-TLX (Task Load Index): Results of empirical and theoretical research. In P.A. Hancock and N. Meshkati (eds.) Human Mental Workload, (Elsevier, New York), 5–39 Herrick, D.F. 2003, Media Management in the Age of Giants: Business Dynamics of Journalism, (Iowa State Press/Blackwell, Ames, IA) Hollnagel, E. 1999, Control versus dependence: Striking the balance in function allocation, CSERIAC Gateway, IX(4), 12–13 Kaber, D.B. and Endsley, M.R. 2004, The effects of level of automation and adaptive automation on human performance, situation awareness and workload in a dynamic control task, Theoretical Issues in Ergonomics Science, 5, 113–153 Mackworth, N.H. 1948, The breakdown of vigilance during prolonged visual search, Quarterly Journal of Experimental Psychology, 1, 5–61 Mackworth, N.H. 1950, Researches of the measurement of human performance. (Medical Research Council Special Report No. 268), (HMSO, London) [Reprinted H.W. Sinako (ed.) Selected papers on Human Factors in the Design and Use of Control Systems, 1961, (Dover, New York), 174–331 Muir, B.M. 1994, Trust in automation: Part I. Theoretical issues in the study of trust and human intervention in automated systems, Ergonomics, 37(11), 1905–1922 Muir, B.M. and Moray, N. 1996, Trust in automation. Part II. Experimental studies of trust and human intervention in a process control simulation Ergonomics, 39(3), 429–460 Noyes, J.M. 2001, Designing for Humans, (Psychology Press, Chichester, England)


225

Noyes, J.M. 2006, Expectations and their forgotten role in HCI. In C. Ghaoui (ed.) Encyclopedia of Human Computer Interaction, (Idea Group Inc., Hershey, PA) Parasuraman, R., Sheridan, T.B. and Wickens, C.D. 2008, Situation awareness, mental workload, and trust in automation: Viable empirically supported cognitive engineering constructs, Journal of Cognitive Engineering and Decision Making, 2(2), 140–160 Price, H.E. 1985, The allocation of function in systems, Human Factors, 27(1), 33–45 Reid, G.B. and Nygren, T.E. 1988, The Subjective Workload Assessment Technique: A scaling procedure for measuring mental workload. In P.A. Hancock and N. Meshkati (eds.) Human Mental Workload, (Elsevier, New York), 185–218 Sheridan, T.B. 2000, Function allocation: algorithm, alchemy or apostasy? International Journal of Human-Computer Studies, 52, 203–216 Watts, B.D. 2004, “Situation awareness” in air-to-air combat and friction. Chapter 9 in Clausewitzian Friction and Future War, McNair Paper no. 68 (revised edition; originally published in 1996 as McNair Paper no. 52). (Institute of National Strategic Studies, National Defense University) Wierville, W.W. and Casali, J.G. 1983, A validated rating scale for global mental workload measurement application. In Proceedings of the Human Factors Society 27th Annual Meeting, Santa Monica, CA (129–133) Vidulich, M.A. and Tsang, P.S. 1987, Absolute magnitude estimation and relativejudgment approaches to subjective workload assessment. In Proceedings of the Human Factors Society 31 st Annual Meeting, Santa Monica, CA (1057–1061)

HUMAN ERROR/ACCIDENTS

UNPACKING THE BLUNT END – CHANGING OUR VIEW OF ORGANIZATIONAL ACCIDENTS B.M. Sherwood Jones Process Contracting Limited Learning from incidents is a fundamentally different form of enquiry to investigations for the purposes of blame or compensation. However, the models we use to describe our understanding of incidents have still to take this into account. For example, ‘accident trajectory’ is a concept that builds in hindsight bias from the outset. Treating ‘the organization’ as a single entity does not reflect the complex network of organizational agents involved in most incidents. The lessons from the study of safety-related organizations (e.g. they run broken most of the time) have still to be absorbed. The concepts of ‘cause’ and ‘root cause’ still persist, even though ‘emergence’ is well established as a property of socio-technical systems. We are now at a stage where a fresh approach using systems thinking is both possible and useful. An example of such an approach, termed the Swiss Chateau Model, is proposed.

Introduction The book that really started the study of organizational accidents was Turner (1978). The three models presented ten to fifteen years ago by Moray (1994), Rasmussen (1997) and Reason (1997) have shaped recent thinking (the existence of Reason (1990) is acknowledged, but Reason (1997) was the influential publication). The lack of attention given to Moray (1994) and Rasmussen (1997) represents a missed opportunity to take a systems approach, but that seems to be the situation. The dominant model has been the Reason (1997) Swiss Cheese Model, shown in Figure 1 below (situated outside Gruyere). It has been very widely adopted and

Figure 1.

Swiss Cheese Model. 229

230

B.M. Sherwood Jones

undoubtedly been a force for good. Recently the Swiss Cheese Model has been the subject of some discussion (e.g. Perneger, 2005), with a significant contribution from Reason himself (Reason et al., 2006). Much of the discussion has been about the balance between individual and organizational (active and passive) failures. The evidence on this (Johnson, 2006a) seems to be that the “80% human error” figure continues to be unhelpful and a greater emphasis on organizational errors is still warranted.

Models of organizational accidents The discussion of models of organizational accidents is hampered by two issues. Firstly, the discussion is still in terms of ‘cause’. Organizational accidents are emergent properties of complex systems (Weinberg, 1975). As such, discussions about ‘cause’ are conceptually inappropriate. Discussion of the ‘influence’ of ‘patterns’ would be philosophically correct, and practically helpful (Snowden and Boone, 2007). Secondly, the purpose of current accident models is unclear; possibly they are to provide some sort of understanding or reconstruction. However, their structure induces massive hindsight bias, particularly the Swiss Cheese Model concept of an ‘accident trajectory’. Hindsight bias is the biggest problem in accident investigation (Dekker, 2002), and so current models are limited even in preventing a repeat accident. It could be argued that Leveson (2004) is an exception to the neglect of systems approaches. Her approach claims roots in socio-technical systems, and acknowledges the need to recognise emergence and the limitations of event chain models. However, the STAMP method emerges as a ‘new model of accident causation’ (Laracy, 2007 emphasis added), based on hierarchical control loops, rather than examining the emergent properties of a network of identified purposeful activity systems. Despite recognition of the nature of complexity (by citing Weinberg, 1975), the method uses ordered-systems thinking (Kurtz and Snowden, 2003) and is inapplicable to the un-ordered space occupied by networks of interacting agents. A full practical understanding of complexity (Snowden and Boone, 2007) has not been used. Despite acknowledging that change is the only constant, the approach, like traditional economics, makes the inappropriate assumption of a state of equilibrium (Beinhocker, 2007).

The ‘blunt end’ The emphasis in the organizational accident literature has been on the ‘sharp end’ despite recognition that the ‘blunt end’ affects safety through the effects of ‘the’ wider system on resources and constraints available to the sharp end (Cooke and Woods, 1994). Treating all the stakeholders from middle management to government regulators as a single system labelled ‘the blunt end’ is not helpful to our understanding of organizational accidents. Incident analysis needs to unpack the blunt end into more specific systems. The approach to incident analysis proposed by Dekker (2002) has yet to be adopted by formal investigatory bodies. Its adoption would enable analysis practice

Unpacking the blunt end – Changing our view of organizational accidents

231

to take a large step forward in both conducting analysis and in making recommendations. The singular, surprising, omission from his book (op. cit.) is extending the analysis from its exclusive focus on the sharp end to gaining an understanding of the evolving mindset in design offices or regulatory drafting groups. There has been little investigation of the role of policy and regulation. Johnson (2006b) is a notable exception, Macrae (2007) is a rare example of understanding the evolving mindset, and Lassagne (2005) is extremely rare in examining regulatory capture (though without using the term). It has occasionally happened that the regulator has been the object of investigation and prosecution (Booker and North, 2007). A suitable structure for the systems in the blunt end needs to be adopted. At a high level, this could be a model such as the Business Excellence model (EFQM, 1998), or a system engineering life cycle (ISO/IEC, 2008), supported by more localised models of good practice. The starting place needs to be action, rather than the accident. Such a change also removes the negativity of Generic Failure Types (Reason, 1997).

Requirements for a useful model From a practical point of view, it would be logical to try to prevent the next accident rather than the last one. The difference between the two tasks has a major impact on the model of organizational accidents required: the model would need to support a wider range of analysis and improvement actions. Features of a model that might help people to learn from incidents and prevent the next one include: 1. Recognition that the operator is the last line of defence, rather than the first. It appears to be an unintentional affordance of the Swiss Cheese Model that gives support to the idea that the crew should be the first line of defence. Pilots are always the first to arrive at the scene of the accident, but it is the responsibility of the other stakeholders to ensure that pilots are placed in manageable circumstances. The role of crew as a barrier should be seen as equivalent and downstream to that of say, a regulator. Human errors need to be seen as equivalent, rather than fundamentally different (cf. ‘active’ and ‘passive’). 2. Recognition that ‘the organization’ is not a single homogeneous entity, or a hierarchy, but rather a network of interacting agents. 3. A clear link to a framework for action and improvement. 4. The ability to reconstruct the evolving mindset (Dekker, 2002), rather than plot an accident trajectory. 5. Recognition that complex systems are not inherently safe, with a few holes in generally complete layered defences, but need constant effort to keep making them safe (Dekker, 2002, Cooke and Woods, 1994).

Proposed model – Swiss Chateau Model The proposed model has been situated in Swiss cheese country to represent evolution rather than revolution, and is the Swiss Chateau Model, illustrated in Figure 2

232

B.M. Sherwood Jones

Figure 2.

Figure 3.

Swiss Chateau Model.

Swiss Chateau Model elements.

below. From Reason et al. (2006) it appears that visual affordances of a model are important, and that a considerable part of the use and abuse of the Swiss Cheese Model has been from its affordances rather than the supporting theory. The Swiss Chateau Model emphasises that the outer defences to safe and effective operation are provided by the ‘enabling systems’ (ISO/IEC, 2008) i.e. the various organizational stakeholders. The inner defence is provided by the sharp end operators – the crew. There are no longer layered defences, and the single outer defence ‘runs broken’. Six enabling systems are identified, shown in Figure 3. Three systems are near the crew in temporal terms and generally operating on short timescales. Local Management is the immediate management and supervision. Recovery includes both containment and recovery. An important feature of a High Reliability Organization is the ability to bring in extra resources rapidly under circumstances of high demand. Learning From Experience is the system for learning from near-misses, surprise successes and incidents. In the case of a ship running aground, Crew is the on-watch


Figure 4.

233

Longer term enabling systems.

bridge team. Local Management is the Master and his direct reports. Recovery includes the company response team, the relevant Coast Guard and rescue services, support from Class and passengers with mobile phones. Learning From Experience includes debriefings, company and other near-miss reporting systems. Even under normal conditions, Local Management has direct dealings with all enabling systems. Working arrangements do not reflect any abstract control hierarchy, and it is not clear why models of accidents should. The longer term enabling systems shown in Figure 4 below are as follows. Operations comprises the management of the enterprise that runs the system of interest e.g. a shipping company for a system of interest that is a ship. Design and Build embraces the supply chain for the system of interest. For an aircraft it might be Boeing and all its suppliers. Regulation comprises the regulatory system (which will be more than just a single regulatory body). Analysis of enabling systems, including the longer term systems, examines the evolving mindset e.g. why it made sense in a design office to remove an alarm channel and replace it with a different one. The frame of reference is not accident causation, but oriented to action e.g. shortfalls from good practice. If learning from the incident is to take place, then the lessons need to be absorbed into the body of narrative fragments used by the enabling systems, rather than residing in a report on a shelf. The ‘cycle of error’ (Cooke and Woods, 1994) can be avoided by choosing corrective actions that are appropriate to the Cynefin domain (Snowden and Boone, 2007).

Implications of a new model The first implication is a change in approach; many new stakeholders are put in the front line. This may help to shift the perceived importance of organizational and individual errors. The use of enabling systems as a structure assists analysis. It enables the evolving mindset of each of the stakeholders to be determined. This can be very difficult

234

B.M. Sherwood Jones

in conventional incident reports. The structure also aids improvement action by addressing the full range of stakeholders directly. If incident analysis is to help to prevent the next accident rather than the last one, then the capability of each enabling system needs to be assessed (ISO/IEC, 2004) against good practice, rather than an evaluation of outcomes against some assignment of cause. In practical terms, there are likely to be a good many fairly immediate implications for enabling systems; a pre-emptive self-assessment might be a good start. Incident analysis is in a transitional stage; there is an awareness that the sharp end is just the tip of an iceberg, but the field has not really taken a remit to go below the surface to the extent that would deliver real improvements in safety. Safetycritical organizations appear to be taking continuous improvement more seriously, schemes for near-miss reporting are growing, and a connection between ‘learning from incidents’ and more general organizational learning may well be accepted. A new framework is needed quickly if an opportunity is not to be missed (again). The need for a more usable incident analysis paradigm is pressing. This will require a move away from bulky incident reports and databases of categorised incidents. Our understanding of how to manage in a complex context has advanced considerably (Snowden and Boone, 2007), and social networking tools to apply that understanding are becoming available.

Acknowledgements The author would like to thank Lloyd’s Register for the opportunity to explore and act on many of these ideas, and two anonymous reviewers for very helpful comments. All errors are the author’s own and the ideas expressed are personal. Some of the material has been published previously under a Creative Commons licence.

References Beinhocker, E.D. 2007, The origin of wealth. Evolution, complexity and the radical remaking of economics (Random House Business Books) Booker, C. and North, R. 2007, Scared to Death: From BSE to Global Warming: Why Scares are Costing Us the Earth (Continuum International) Cook, R. and Woods, D.D. 1994, Operating at the Sharp End: The Complexity of Human Error in Bogner, M.S. (ed.) Human Error in Medicine (Lawrence Erlbaum) Dekker, S. 2002, The Field Guide to Human Error Investigations (Ashgate) EFQM 1988, Excellence Model (European Foundation for Quality Management) ISO/IEC 15288:2008 System engineering - system lifecycle processes (International Standards Organization) ISO/IEC 15504-4:2004 Information technology – Process assessment – Part 4: Guidance on use for process improvement and process capability determination (International Standards Organization)


235

Johnson, C.W. 2006a, On the Over-Emphasis of Human ‘Error’As A Cause of AviationAccidents: ‘Systemic Failures’and‘Human Error’in US NTSB and Canadian TSBAviation Reports 1996–2003 (http://en.scientificcommons.org/c_w_johnson) Johnson, C.W. 2006b, Establishing Public Policy as a Primary Cause of Engineering Failure: Did Market Deregulation Lead to the North American ‘Blackout’, August 14th 2003? in Balducelli and Bologna (eds.) Proceedings of the ENEA International Workshop on Complex Networks and Infrastructure Protection (ENEA) Kurtz, C. F. and Snowden, D. J. 2003, The new dynamics of strategy: Sense-making in a complex and complicated world, IBM Systems Journal, 42, 3 Laracy, J.R. 2007, Addressing System Boundary Issues in Complex Socio-Technical Systems (CSER 2007 Proceedings) Lassagne, M. 2005, A Tale of Two Laws : Private and Public Regulatory Organizations in the Maritime Shipping Industry XXIe Colloque EGOS (European Group for Organization Studies) Leveson, N. 2004, A New Accident Model for Engineering Safer Systems, Safety Science, 42, 4, 237–270 Macrae, C. 2007, Interrogating the unknown; risk analysis and sensemaking in airline safety oversight. Discussion Paper 43 (LSE Centre for analysis of risk and regulation) Moray, N. 1994, Error reduction as a systems problem in Bogner, M.S. (ed.) Human Error in Medicine (Lawrence Erlbaum) 255–310 Perneger, T.V. 2005, The Swiss cheese model of safety incidents: are there holes in the metaphor? (http://de.scientificcommons.org/8444947) Rasmussen, J. 1997, Risk Management in a Dynamic Society: A Modelling Problem. Safety Science 27, 2/3, 183–213 Reason, J. 1990, The contribution of latent human failures to the breakdown of complex systems (Philosophical Transactions of the Royal Society) series B. 327: 475–484 Reason, J. 1997, Managing the Risks of Organizational Accidents (Ashgate) Reason, J., Hollnagel E, Paries, J. 2006, Revisiting the “Swiss Cheese” model of accidents EEC Note No. 13/06 (Eurocontrol Experimental Centre) Snowden, D.J. and Boone, M.E. 2007, A leader’s framework for decision making Harvard Business Review Turner, B. 1978, Man-made disasters (Butterworth-Heinemann) Weinberg, G.M. 1975, An Introduction to General Systems Thinking (Dorset House)

LESSONS FROM THE PAST TO GIVE TO THE FUTURE: LEARNING FROM INCIDENTS S. Evans Rail Safety and Standards Board, UK The Rail Safety and Standards Board (RSSB) Human Factors (HF) team have developed a systems approach to learning from incidents. This approach comprises three work streams which together aim to increase the UK rail industry’s capability to learn the HF lessons. Firstly, an overview of an ongoing HF analysis of formal inquiries and incident investigation reports will be given. This will describe a classification system and database developed by the RSSB HF team to facilitate the identification of the multi-causal nature of incidents. This work was initiated to highlight the benefits of identifying the type and frequency of different error types and performance shaping factors (PSFs) in order to inform industry of (i) the human contribution to risk and, (ii) identify future HF priorities for the industry. Secondly, an overview of a supporting HF awareness training course to aid the incident investigation process will be provided. Finally, HF integration with RSSB’s strategy to improve the validity and reliability of safety-related information recorded in the industry’s Safety Management Information System (SMIS) will be highlighted.

Introduction Understanding the human contribution to risk There is a requirement to understand the human contribution to risk in order to increase the UK rail industry’s capability to learn human factor (HF) lessons from incidents. This paper describes three inter-related HF work streams which together feed into RSSB’s overarching strategy, referred to as ‘Learning from Operational Experience’.

Background In 2005, RSSB carried out a review of the industry’s Safety Management Information System (SMIS) (RSSB, 2005). SMIS is an industry-wide database used to record safety information on past incidents. The information is used to analyse risks associated with incidents at a national level. In the railway context, incident types include signals passed at danger, train derailments, collisions, train maintenance, broken rail etc. (RSSB, 2006). Although SMIS is thought to be a good source of risk data, it does not currently provide causal incident information which is usable from a HF perspective. Specifically, SMIS does not provide sufficient information on the type and frequency 236

Lessons from the past to give to the future

237

of different error types, thereby increasing the difficulty of identifying the human contribution to risk (RSSB, 2007).

HF learning from incidents – a systems approach Feeding into RSSB’s ‘Learning from Operational Experience’ strategy are three HF work streams which aim to increase the UK rail industry’s capability to learn HF lessons from incidents. This includes: 1. HF review of railway incident investigation reports 2. HF Awareness Training Course to aid incident investigations 3. HF contribution to SMIS Vision to improve the validity and reliability of HF incident causation information

Work stream 1 HF review of railway incident investigation reports The RSSB HF team have developed an evidence-based approach to reviewing the multi-causal nature of railway incidents (RSSB, 2007).This has involved the HF analysis of investigation reports. The aim was to highlight the benefits of identifying the type and frequency of different error types in order to inform the rail industry of the human contribution to risk and identify HF priorities, an approach supported by Network Rail and the Office of Rail Regulation.

Methodology The HF review of railway incidents included the development of a classification system and database in order to analyse the causal and contributory factors of incidents across the UK railway industry. A six-stage process was implemented: 1. A review was carried out to evaluate a number of existing incident analysis approaches. The classification system used in this study was based on the error classification model used in TraceR (Shorrock, 2006) and review of other classification approaches (e.g. Human Engineering, 2004). Given that the original error classification model used in TraceR was developed in the aviation domain, some aspects of the model did not translate directly to rail incidents. Adaptations were undertaken to the approach by the University of Nottingham (RSSB, 2004). Based on the information in the railway incidents analysed and inputs from railway operational specialists, the TraceR classification system has been further developed to reflect the nature of railway incidents. Further details are provided in the final report (RSSB, 2007). The aim of the review was to ensure the classification of error method did not omit any key error types or performance shaping factor. 2. A Microsoft Access database was developed to facilitate data classification and entry. This database has been modified during the project to reflect feedback on usability and coverage of contributing incident factors.

238

S. Evans

3. HF specialists within RSSB were trained on the analysis technique, classification of incident reports and use of the database. This was a key step to ensure consistent classification across incidents. 4. Monthly review of incident reports (approximately 20 reports per month). 5. Investigation reports were analysed and entered into the database by the RSSB HF team. Inter-rater reliability studies have been carried out in the development of TraceR (Shorrock, 2006). To ensure internal consistency, each report was classified by a minimum of two HF specialists. In the future, data already classified in the database will be verified by a formal inter-rater reliability study. 6. Ongoing review and refinement of the entire database.

Critical elements of the approach Incident analyses to investigate human performance issues are based on key concepts contained in theories of accident causation. The first is that incidents can be complex and be contributed to by a number of distinct factors, termed ‘incident’ factors. Secondly, incident factors can contribute to an event at the individual, job and/or organisational levels (Reason, 1999). A separate TraceR classification was used to categorise performance shaping factors (PSFs). Key categories include personal issues (e.g. stress, fatigue, and motivation), team factors (e.g. relationship with supervisor), workplace design (e.g. workload) and training issues (e.g. skills fade). Understanding the context of the error is also important. Therefore two elements to capture the context are included in the classification: (1) job role and personal characteristics of the person undertaking the task (e.g. experience) and (2) the task in which the error arose (e.g. driver observation of a signal). Incident reports can also reveal incident factors that relate to (1) the event itself, (2) post event factors which occur after the event but which potentially increase risk (such as incorrect train movement following a SPAD), and (3) incident investigation factors, which identify how the process of incident investigation could be improved (RSSB, 2007).

Identifying the human contribution to incidents Over six months of classification effort, 116 incident reports were reviewed and the relevant data entered into the database. The majority of report types analysed were formal inquiries and investigations (95%), the remaining being Rail Accident Investigation Branch reports and local investigations. An incident report may contain more than one type of event (for example a SPAD followed by a derailment). The most frequently occurring type of event was ‘SPAD Driver and Signal’. However, the high frequency of SPAD investigations is likely to reflect an industry bias towards investigating SPADs, rather than their relative frequency when compared to all incidents (RSSB, 2007). Frequencies for events across the industry can be more accurately derived from the Safety Risk Model (SRM) and other safety intelligence data. The SRM is a comprehensive mathematical representation of hazardous events that could lead directly to injury


239

or fatality on the railway. It was developed using modelling techniques informed by a combination of incident data and expert judgement. The use of expert judgement reduces the problems that can arise with calculating risk estimates purely on the basis of data, which may be insufficiently representative of the underlying level of risk. Most of the data used to populate the SRM comes from SMIS. The method adopted in this study made a deliberate effort to look more widely than SPAD incidents by biasing the collection of events towards other event categories. 38% of the events analysed are from these other categories (e.g. derailments, collision, control of train movements, track worker protection from train movements). In the future, the event selection criteria could be further developed to match industry information requirements. Out of the 116 incident reports reviewed, 121 events were analysed (an incident report may contain more than one event). Events (e.g. derailments, SPADs, collisions) can be complex and can be contributed to by a number of factors, known as incident factors. Out of the 121 events analysed, 910 incident factors were identified. This represents an average of eight incident factors per event, illustrating the multi-causal nature of incidents. Of these 910 incident factors, 77% were errors and 23% were PSFs. This is interesting as it provides an indication of different areas that should be monitored and be placed under greater management control. The data can be analysed in a number of ways by cutting a ‘cross section’ according to the factors that are of interest. For example, it is possible to look at how incident factors relate to events. 82% of incident factors contributed to the event itself – these factors are key as they define why an event occurred. Interestingly, there was a relatively high frequency of post event factors (13%), which indicates that consideration must also be given to factors which arise after an event (e.g. correctly reporting a SPAD). 5% of incident factors related to the actual investigation process. To provide an overview of errors, the broader error types are summarised in figure 1. Person errors are related to a specific person in a defined job role and this category is used if an error can be attributed to a specific individual. Design errors cover both equipment design (e.g. train cab interface) and the wider system design. Management errors include safety management systems such as driver management, signaller management, Rule Book and Railway Group standards (RSBB, 2006). Diagram one illustrates that person errors dominate but there is a sufficient number of management and design errors to allow analysis of the different types of errors which contribute to incidents. The bias to human errors may reflect a bias in the investigations to consider the errors by front line staff rather than investigate the management systems which support them. Further review of the investigation process would be required to ascertain if this is the case.

In the spotlight – performance shaping factors The database allows errors to be examined in closer detail, which is of course important, as errors are the factors which can be identified and rectified to prevent recurrence. However, performance shaping factors are also important because

240

S. Evans

Person error Management error Design error

Figure 1.

Error Types.

they increase the likelihood of incidents occurring. Logically, if PSFs can be removed or mitigated, then the likelihood of errors occurring could also be reduced (RSSB, 2007) Table 1 provides an overview of the frequency of PSFs by category. These categories are defined as part of the TraceR approach. The dominant category is workplace design (50%) which covers the design of workplace elements which affect human performance. Examples of workplace design issues include conspicuity of alarms and warning devices, train cab layout, trust in equipment, workload and task complexity. Personal factors (25%) are also significant human performance shaping factors and include issues such as alertness, emotional or occupational stress, fatigue, job satisfaction and previous incident history. Environmental factors include issues such as lighting, noise and distraction, people within the train cab apart from the driver, and temperature. Training, knowledge and experience capture issues relating to the availability of training, familiarity with the task (such as route knowledge), and supervision. Social and team factors include team morale, and team relations. Communications issues include accent/dialect, company specific rail jargon and communication equipment and reliability. Finally, procedural issues that influence human performance include procedure complexity, procedure accuracy and compatibility with other processes or documentation. The database can narrow its focus to very low level detail, so that the category of interest can be studied in increasing detail. For example, if workplace design is of interest, the database can supply information on sub-categories so we can see which factors influence human performance in varying degrees. Time pressure is identified as being the largest contributing factor of workplace design, which takes into account 18% of all performance shaping factors, followed by distraction (10%), routine signal sequence (6%) and workload (2%). If we want to look at driver distraction in more detail to ascertain the lower level contributory factors, the data tells us that distraction as a PSF can be caused by train faults (33%), track workers (17%), distractions external to the cab (8%), low workload (8%) and others. PSFs can also be investigated by job role. For signallers, the two most frequently


Table 1.

241

Overview of PSFs by category.

Performance Shaping Factor

Frequency

Percentage1

Workplace design Personal factors Environment Training, knowledge or experience Social and team factors Communications Procedures / documentation Total

106 53 19 17 10 4 1 210

50% 25% 9% 8% 5% 2% 0% 100%

1

Percentages rounded to nearest whole number

occurring issues are high workload and lack of familiarity. High workload is a factor which is an inherent part of the signalling task at certain times. Lack of familiarity is a factor which should be monitored, particularly as management errors in signaller competence were also identified as an area of concern.

Summary The HF review of incident investigation reports provides a powerful tool that has the capability of providing a comprehensive means of summarising rail incident factors for HF assessment. Usable outputs are being produced enabling incident factors to be built up across the UK rail industry. It is proposed that the database be used as an active tool to explore industry-specific issues and be provided for wider use throughout the industry to maximise its potential.

Work stream 2 HF awareness training course for incident investigators At the request of industry stakeholders, RSSB HF has developed a training course to raise awareness and promote understanding of HF causation (RSSB, 2008b). This training course has been developed as a ‘resource kit’, containing a number of different modules and tools that industry members can select and use as appropriate. The training course was developed for incident and accident investigators but would also be of interest to driver managers, trainers, and other personnel who have responsibilities for developing and maintaining staff competence. The content of the course is designed to introduce incident and accident investigators to HF. The course will be launched in February 2008 via ‘Train the Trainer’ sessions. These sessions will be delivered by the RSSB HF team to nominated trainers from Train and Freight Operating Companies. Company trainers will then deliver the training course to their colleagues in-house. The first training day provides a general introduction to HF and includes modules on human error and violations, understanding human performance, techniques for

242

S. Evans

investigation, and performance shaping factors at the individual level. The ‘Human Error’ module examines why errors and violations occur and provides tools to assist the incident investigator in the identification and classification of errors and violations. The ‘Understanding Human Performance’ module provides an overview of the human information processing system and explains how things can go wrong at the levels of perception, memory, decision making and situation awareness. Individual performance shaping factors are also discussed, such as fatigue, experience and training. Throughout the course, real-world examples are provided to illustrate how human error has contributed to incidents in the rail environment and other high-hazard industries, such as aviation, nuclear, and oil and gas. The second training day focuses on job and organisational factors that can contribute to incidents. Job factors, such as workload, equipment design and communication are discussed in detail, with an emphasis on how to identify risks that could lead to an incident. Organisation factors such as culture, leadership, systems and change are also discussed. The emphasis throughout the course is on practical application of the theory, and this is achieved through a range of exercises and group activities. Practical tutorials are included to provide trainees with hints and tips about classifying errors and violations, effective interviewing skills and techniques that can aid the investigation process. A toolkit is also provided to amalgamate the practical tutorials and is intended to aid the incident investigator whilst on-the-job. The toolkit includes useful questions for the incident investigator to ask while conducting a real-time investigation. These questions provide a checklist of prompts and reflect the current HF proposal for the industry-wide Safety Management Information System. As the training course has been primarily developed for incident investigators, it offers an excellent opportunity to map the process and terminology to the HF module in SMIS to ensure that both are consistent, which will maximise the accuracy of the data captured. This will be discussed further in the next section.

Work stream 3 HF contribution to SMIS Vision to improve the validity and reliability of accident causation information Improving the validity and reliability of industry-wide safety information The industry-wide database currently used to record safety information on past incidents is the Safety Management Information System (known as SMIS). RSSB, on behalf of the industry, is initiating a major programme to upgrade and enhance SMIS. This programme is known as the ‘SMIS Vision’ and is supported by a wide stakeholder group (RSSB, 2008c). One of the main drivers for the ‘SMIS Vision’ is to provide the industry with validated safety data that is accessible from one source. To be able to achieve this a number of improvements are needed to the current SMIS application and its related business processes. These include improving the flexibility and performance of SMIS, specifically for data entry inputters, industry agreement on the core and optional (company-specific) data requirement for collecting and recording


243

safety-related data. It also includes making improvements to the quality of data being entered into SMIS. These objectives ensure that key data is validated and only available from one source. The HF team have been contributing to ‘SMIS Vision’ by proposing improvements to the underlying SMIS HF causation module. The objectives of the HF contribution to the SMIS Vision include: • Increased accuracy: to improve the reliability and validity of HF related incident information. • Increased usage: to increase the frequency of HF related incident information. • Easy identification: to make the database as easy as possible for incident investigators to identify HF causal factors. • Easy input and extraction: to make it as easy as possible for interested parties to identify trends and patterns related to HF from the incident information entered in SMIS. The links with RSSB’s HF Awareness training course for incident investigators encourages increased usage of SMIS and greater accuracy in the recording of error classification and performance shaping factors within it.

Conclusion RSSB are working with stakeholders to develop an industry-wide strategy known as the ‘Learning from Operational Experience’. This paper has discussed three work streams that the RSSB HF team have contributed to this strategy. This included the HF review of railway incident investigation reports, a bespoke training course to raise awareness of HF to aid the accident investigation process, and refinement of the HF module in SMIS. These three work streams are integrated to ensure a systems approach to learning from incidents. The process of incident report review and classification in the database have highlighted how improvements could be made to incident investigation processes from a HF perspective. The HF Awareness training course offers a number of techniques and methods to capitalise on the database results to train those responsible for investigations in the rail industry. This course has also provided an excellent opportunity to map the process and terminology of HF issues covered in the course with that used in SMIS to maximise the accuracy and validity of the data captured. This work will also enable changes in causal patterns to be monitored to shape the HF strategy across the industry. These combined approaches deliver a strategy by which the UK rail industry can better learn HF lessons from incidents.

References Rail Safety and Standards Board (2005). The classification of causes related to human factors within SMIS.

244

S. Evans

Rail Safety and Standards Board (2007). Human Factors Review of Railway Incidents: Phase 1 Report. RSSB research project T635. Human Engineering. (2004). SPAD Hazard Checklist. http://www.opsweb.com Rail Safety and Standards Board (2004). Rail-Specific HRA Tool for Driving Tasks. RSSB research project T270 http://wwww.rssb.co.uk/publications.asp Rail Safety and Standards Board. (2008a). Understanding human factors: a guide for the railway industry http://www.rssb.co. uk/expertise/human_factors/ human_factors_good_practices_guide.asp Rail Safety and Standards Board (2008b). Human Factors Awareness Training Course. RSSB research project T635. Rail Safety and Standards Board (May 2008c). Human Factors Strategy Document: Human Factors Contribution to the SMIS vision. RSSB research project T635. Reason, J. (1999). Human error. Cambridge University Press. Shorrock, S.T. (2006). Technique for the Retrospective and Predictive Analysis of Cognitive Error (TRACEr and TRACEr-lite) in International Encyclopaedia of Ergonomics and Human Factors Edited by W. Karwowski. CRC Press, Boca Raton, Florida, USA, 2nd Edition, Volume 3 pp. 3384–3389.

THE DEVELOPMENT OF HSL’S SAFETY CLIMATE TOOL – A REVISION OF THE HEALTH AND SAFETY CLIMATE SURVEY TOOL C. Sugden1 , M. Marshall2 , S. Binch1 & D. Bottomley3 1

Health and Safety Laboratory Health and Safety Executive 3 Merlin Analytical Solutions

2

The Health and Safety Laboratory (HSL) have revised and updated the Health and Safety Climate Survey Tool (CST) so that it is a more reliable and valid psychometric instrument, and as a result it is more useful to both public and private sector organisations. The revised Safety Climate Tool, SCT, contains 40 statements that map onto 8 factors that are more coherent and meaningful than the old factors. The revised tool now takes less time to complete, is written in clearer English and leading or ambiguous items have been removed. The revised tool has been subject to a period of consultation and a series of piloting exercises at various companies in the manufacturing, construction, and major hazards sectors. This paper will outline how the changes were made and give details of the tool’s successful application to date in industry. New software is under development for release in 2009.

Introduction Organisational climate is based on an aggregation of employees’ perceptions of their experiences within an organisation (Dawson et al., 2008). Safety climate, as a term was initially used by Zohar (1980) to describe attitudes towards safety, and was derived from earlier work on organisational climates. Typically safety climate is explored through questionnaires exploring attitudes and perceptions regarding safety; it is a statistical construction of perceptions held in an organisation regarding safety and is a way to summarise these (Rousseau, 1988). Cox and Flin (1998) in a review of academic papers on the issues stated that the terms safety culture and safety climate are used interchangeably to refer to similar concepts. Safety culture, as a phrase, was first used by the International Atomic Energy Authority to describe the issues at Chernobyl at the time of their major incident (IAEA, 1986). The interest in culture arose in response to a realisation that organisational structure (i.e. the roles and their relationships, rules and procedures) was limited in achieving an organisation’s health and safety goals. Hence the formal systems should prevent accidents occurring, for example by prioritising resource allocation, assessing training needs, adopting risk assessment methodologies and choosing tolerable risk criteria. HSE stated that the explicit and implicit goal of a 245

246

C. Sugden et al.

safety management system is the development of a positive safety culture (HSE, 1991). This reflects the understanding that safety management systems only work if individuals are motivated to comply and conform with the organisation’s systems, which is where the need to understand the impact of safety culture arose. HSC (1993) defined safety culture as: “the product of individual and group values, attitudes, perceptions, competencies, and patterns of behaviour that determine commitment to, and the style and proficiency of, an organisation’s health and safety management. Organisations with a positive safety culture are characterised by communications founded on mutual trust, by shared perceptions of the importance of safety and by confidence in the efficacy of preventive measures.”

HSE’s Health and Safety Climate Survey Tool The Health and Safety Climate Survey Tool (CST) was published by HSE in December 1997 (HSE, 1997); from then until 2007 over 800 copies of the tool were sold. The CST was designed to gauge perceptions of the management of health and safety, and associated issues recognised as important to prevent occupational accidents and incidences of ill-health. The supporting documentation stated that safety climate is a measurable component of safety culture, and identifies tangible issues that should help improve an organisation’s safety culture. Furthermore they highlight the importance of carrying out climate surveys to promote employee involvement and engagement. The CST was a computer-based product that enabled organisations to customise a safety climate questionnaire for use within their organisation. The software allowed organisations to analyse the results of their survey and produce graphs and reports on the basis of these findings. The software allows comparisons to be made across organisational hierarchies, and other groups of workers based on statement responses. Differences are often found between these groups and HSE (1997) stated that understanding these differences is central to making improvements to an organisation’s safety culture. Early in 2007, HSE withdrew the tool from the market, as the software was not compatible with later versions of Windows. However, the Baker Panel report into BP’s Texas City incident (The B.P. U.S. Refineries Independent Safety Review Panel, 2007) and their pronouncements concerning safety culture revived interest in the topic. This project was initiated to address some of the CST’s perceived shortcomings (e.g. long-winded, out of date, repetitive) and to address perceived market needs. The remainder of this paper discusses the updating process.

Objectives of the revision to the Health and Safety Climate Survey Tool The initial aims of this work were: • Reduce the size of the survey tool such that it could be completed within 10 minutes, but not reduce its power as an evaluative tool;

The development of HSL’s safety climate tool

247

• Derive a factor set that was more internally coherent and comprehensive that that contained within the original tool, building on the historic accumulation of data; • Use plainer English; • Produce upgraded software.

Methods The original CST (HSE, 1997) had 71 statements grouped into 10 factors: • • • • • • • • • •

Organisational commitment and communication; Line management commitment; Supervisor’s role; Personal role; Workmates influence; Competence; Risk-taking behaviour & some contributory influences; Some obstacles to safe behaviour; Permit-to-work systems; Reporting of accidents and near miss systems.

Each statement required a response on a 5-point scale, ranging from “strongly agree” to “strongly disagree”. When coded for use in analyses, some questions had their responses reflexed to ensure that they all worked in the “same direction” (that is, a shared concept between a pair of variables is reflected in a positive correlation). Some users of the CST participated in a benchmarking scheme, which generated a large database of industry responses. This dataset was made available to the HSL project team, and comprised of almost 40,000 questionnaire returns from a range of industries. This was used as the basis for the statistical analyses and reworking of the survey tool. One fundamental requirement of this revision is that the statements are meaningful, i.e. there must be a strong conceptual/academic basis for the inclusion of each statement. Furthermore, individual statements are only of use if they have reasonable representation across the distribution of possible responses. Therefore variables with heavily biased response distributions have – by definition – relatively little variance and are likely to be of limited use in terms of distinguishing between strong and weak safety climates. Finally, questions that are unrelated to other questions (i.e. are uncorrelated) will also be of limited use in the final model. The revision of the safety climate tool followed the methodology outlined in Figure 1. There have been three stages in the revision of the original health and safety climate tool (HSE, 1997), namely: removal of statements which fail face validity tests, removal of statements with biased distributions, and removal of statements during exploratory factor analysis. These were then subject to factor analysis to create a comprehensive survey tool.

248

C. Sugden et al. Original Health and Safety Climate Survey Tool 71 Questions, 10 factors

Stage 1 - Face validity removal of 12 questions

Stage 2- Biased distributions Removal of 13 questions

Stage 3 - Exploratory factor analysis Removal of 9 questions Revision of safety climate question set 37 questions remaining

Factor solution for Revised Safety Climate Tool Addition of 3 questions

Piloting of Revised Safety Climate Tool 40 Item Question set Piloting loop 40 question, 8 factor SCT

Revised Safety Climate Survey Tool 40 Questions

Figure 1.

Overview of method.

Stage 1 – Face validity The project team removed statements that were not universally answerable by all respondents, were considered too conceptually similar to others, or were poorly worded. Each of the 71 variables from the original Climate SurveyTool was assessed to determine how well it was able to satisfy these criteria. To achieve this, a statement must be unambiguous in its interpretation, be relevant to the issue of safety climate and be liable to generate meaningful responses. 12 statements were unable to satisfy these criteria. For example: “My job is boring and repetitive” was removed as it was felt that understanding the responses to this question would have nothing meaningful to say about safety climate, and furthermore there are no actions that can be taken to resolve this by an organisation. Whereas, “People here do not remember much of the health and safety training which applies to their job” was felt to be ambiguous. In particular, it was felt that the word “people” could have a number of interpretations


249

from the respondent, their team to other people within the organisation. Furthermore health and safety training could cover either health and safety aspects of their job or specific health and safety training (e.g. risk assessment training), and that there was sufficient uncertainty concerning the question and the meaning of the response. Finally, all question relating to “permits to work”, were removed on the grounds that they were not relevant to the goals of the analysis in that they are not universally answerable.

Stage 2: Removing statements due to biased distributions The second stage of question removal involved looking at the distribution of responses for each of the questions in turn and identifying those with heavily biased distributions. Choosing a critical value for this type of exclusion is not straightforward as there is no objective criterion to measure against. Consequently, a compromise has to be reached between excluding the most heavily skewed variables whilst not affecting the pool of questions too radically by choosing too radical a criterion. In this case, a cut-off point of 70% at either end of the distribution was selected. That is, a variable was excluded if the sum of the “strongly agree” and “agree” categories or the “strongly disagree” and “disagree” categories equaled or exceeded 70% of respondents. In total, 13 statements were removed from the statement pool on this basis. Of these, one had also failed to meet the face validity criteria.

Stage 3: Exploratory factor analysis The previous two stages left 47 statements for inclusion in the analysis. A prerequisite of variables for inclusion in Principal Components Analysis is that they contribute something to that analysis. As the procedure works by pooling shared variance, it requires that each variable has some variance in common with other variables in order that it may be shared. Consequently, any variable that is unable to demonstrate sufficient shared variance should be excluded on the grounds that it adds very little to the analysis. Again, the precise threshold that was chosen was somewhat arbitrary as there is no objective criterion. However, that threshold should not discard too many variables to render the analysis meaningless yet should discard those that contribute the least. A cut-off point of 0.5 was chosen as this represents a point below which less than half of a variable’s variance is included in the analysis. Nine statements displayed communalities less than 0.5 and were removed from the analysis. Interestingly, many of these were statements that had been considered to be marginal on either face validity or distribution stages (but which had made it through that stage). The removal of these 9 statements left a pool of 38 statements for inclusion in the revised safety climate factor structure.

Development of a factor solution for the Revised Safety Climate Tool In order to examine the structure of safety climate with the reduced statement set, the Kaiser-Meyer-Olkin measure of sampling adequacy was used to confirm the

250

C. Sugden et al.

Table 1.

Factor definitions and Cronbach’s alpha.

Factor

Title

Cronbach’s alpha

1 2 3 4 5 6 7 8

Organisational commitment to safety ‘Health and safety’ oriented behaviour Health and safety trust Usability of procedures Engagement in health and safety Peer group attitude to health and safety Resources for health and safety Accident and near miss reporting

0.808 0.815 0.803 0.716 0.715 0.724 0.642 0.599

appropriate approach. A KMO score of 0.962 was achieved, suggesting that the suitability of Principal Components Analysis was satisfied. Given the overall adequacy of the model, the next step was to establish the amount of variance accounted for by that model. Six, seven and eight factor models were explored on the basis of Eigenvalues and the scree plot, as well as the meaningful constructs underlying these factors. In total 6 models were developed using orthogonal and oblique rotation, these were considered by the researchers, and the groupings evaluated to identify which factors best represented the notion of ‘safety climate’. It was felt that the eight factor model was most meaningful and conceptually better reflected current opinions of the key facets of safety culture. The eight factors solution was made up from 37 questions.

Factor reliability and validating the factors The reliability of the factors was assessed to determine their likely future usefulness. This determines the overall “strength” of the scale (i.e. the extent to which the individual questions appear to be measuring the same concept), as well as the contribution of each variable. Table 1 shows the results of the reliability analyses conducted on each of the eight factors. From table 1 it can be seen that the three “best” factors all have high reliability scores and the next three are also acceptable. Factors 1 to 6, therefore, represent reliable scales that should be replicable over time (achieving Cronbach’s alpha exceeding 0.7). Factors 7 and 8 had questionable reliability, however, they have fewer questions contributing to them and the appropriateness of reliability analysis is questionable under these circumstances. A proposed solution was to derive some additional questions that may load onto these factors to “boost” their reliability and pilot them to see how well they perform. Three additional questions were written to boost the reliability of factors 7 and 8. This made a 40 statement question set, comprised of eight coherent factors.

Piloting of revised question set Using this 40 item statement survey tool, a programme of pilot surveys were carried out to ensure statement clarity, highlight ambiguities and raise other issues concerning application of the survey tool.


251

Round 1 – an internal pilot on 20 individuals within HSL was carried out. This highlighted a number of issues, including typos, confusion over use of words (e.g. line manager and supervisor), use of idiom and jargon. Round 2 – the issues raised at Round 1 were explored at a major motor manufacturer when the questionnaire was distributed to around 40 employees. Subsequently semi-structured interviews were held with respondents to gather their perceptions about: • Questionnaire length; • Use of language, including idiom and jargon; • Specific questions identified in the first pilot as ambiguous, confusing, and hard to answer. At this stage a number of modifications were made to the survey tool to clarify some of the areas identified within these 2 pilots. For example a statement was reworded from: My immediate boss is receptive to ideas on how to improve health and safety My supervisor takes on board ideas on how to improve health and safety This change was felt to improve the statement and address the key concept in this instance. Other changes included changing the word ‘seldom’ to ‘rarely’ in several questions, and revising this question: Productivity is usually seen as more important than health and safety Getting the job done is usually seen as more important than health and safety Further pilots were carried out, adopting the same method in the chemicals manufacturing and construction sector. The process again involved semi-structured interviews to confirm some of the modifications and iterations to the question set. Nothing significant emerged from these. Additionally the motor manufacturer involved at the pilot stage rolled out the question set across their large site, and achieved responses from in excess of 600 employees. The piloting process was iterative and confirmatory to ensure that changes did not weaken or significantly change the meanings of statements.

Next steps At the time of writing in the region of 7000 individuals from a range of companies and sectors have completed the revised safety climate tool (SCT). Feedback from these companies has been positive. Companies state that the questionnaire is of the right length (e.g. reduces business burden), written in plain English, provides factors that are meaningful and has prompted organisations to identify solutions to address emerging issues.

Discussion and conclusions From the piloting and use of the revised safety climate tool (SCT) it appears that the technique is valuable to organisations as part of engaging their workforce in the process of safety improvement. Good practice use of the question set, would

252

C. Sugden et al.

involve its application across the organisation, and then carrying out focus groups to identify the reasons underlying respondent’s stated attitudes. This can then be used to develop action plans and obtain staff ownership and engagement. This paper reports on the process of revising a health and safety climate survey tool. Face validity, statistical validity and construct validity were considered in the revision of the safety climate tool. The SCT is comprised of 8 factors and 40 statements, of these 37 statements were retained from the previous model. The goals of the revision were to: • • • •

Reduce duplication and leading questions; Write in plain English; Retain statistical power and feel of the original; Renew the software tool.

To date the project has met the first three aims and work is being carried out on the fourth.

References The B.P. U.S. Refineries Independent Safety Review Panel, (2007), The report of the B.P Refineries Independent Safety Review Panel, http://www.bp.com/liveassets/ bp_internet/globalbp/globalbp_uk_english/SP/STAGING/local_assets/assets/ pdfs/Baker_panel_report.pdf (accessed 19th November 2008). Collins, A and Gadd, S (2002) Safety Culture: A Review of the Literature. Health and Safety Laboratory. Dawson, JF, Gonzalez-Roma, V, Davis, A and West, MA (2008) Organisational climate and climate strength in UK hospitals. European Journal of Work and Organizational Psychology, 17 (1), 89–111. Flin, R., Mearns, K., O’Connor, P. and Bryden, R. (2000). Measuring safety climate: identifying the common features. Safety Science, 34, (1–3), 177–192. Guldenmund, F.W. (2000). The nature of safety culture: a review of theory and research. Safety Science, 34, (1–3), 215–257. HSC (1993) ACSNI study group on human factors. Third report Organising for Safety. HSE Books. HSE (1991) Successful health and safety management. HMSO. HSE (1997) Health and Safety Climate Survey Tool. HSE Books. IAEA (1986) Summary Report on the Post-Accident Review Meeting on the Chernobyl Accident. International Safety Advisory Group, Safety Series 75-INSAG-1 (Vienna: IAEA). Pidgeon, N.F., (1991) Safety Culture and Risk Management in Organisations. Journal of Cross Cultural Psychology, 22, 129–140. Rousseau, D.M. (1988) The construction of climate in organisational research. In International Review of Industrial and Organisational Psychology, Eds. C.L. Cooper and I. Robertson. John Wiley and Sons, Chichester. Sugden, C and Healey, N (2008) Systematic review of key health and safety management literature. Internal report. RSU/08/06.

HUMAN FACTORS INTEGRATION

REPORTING LOCATION AND ENVIRONMENTAL FEATURES WHEN USING METAL DETECTOR OR PROBE IN A SIMPLIFIED MINE DETECTION TASK R.J. Houghton, C. Baber & J.F. Knight EECE, The University of Birmingham, Birmingham, B15 2TT, UK In this paper, the activities involved in searching for buried items with technical aids are explored. The domain of exploration is topical because it can be applied to the search for explosive devices by military and security forces, the ergonomics of which have received relatively little attention. A study is presented which tests the hypothesis that search involves separate cognitive processes for recognising locations and features. It is shown that, while both forms of search tested (i.e., using a trowel or a metal detector) perform similarly when considering location, the recall of features is impaired when using the metal detector. This is interpreted as evidence of different search strategies and information processing in the two conditions.

Introduction Despite its significance to police and security operations, there is relatively little research concerning how people search for buried or hidden objects in the field. In part, this might be because the relevant material could be operationally sensitive and hence, classified. A broader issue is that the psychological literature has located the act of searching largely within the domain of well-defined laboratory paradigms concerning the detection of abstract stimuli amongst distractors or within noise. In the present paper we report an experiment based on memory for buried items that bridges this gap between the laboratory and the field by considering search as an ongoing activity in which participants interact with their environment to discover target objects rather than passively observing. A key research question in this area is how technology might mediate or interact with the tasks of searching and reporting. In particular, we are concerned with the search for buried land-mines (see figure 1). Previous Human Factors research concerning mine detection equipment has suggested that it would be valuable to provide the user with some form ‘map display’ showing the location of objects as they were detected (Herman and Igelias, 1999; Herman et al., 2001). Studies of the use of mine detectors have characterised the task of mine detection as scanning an environment to compose a mental representation of space and the location of mines within it based upon auditory output from the metal detector (Staszewski, 1999; Stazewski & Davison, 2000). This implies that the expert is building a 2D map of the ground, and then interpreting this map against his knowledge of landmine patterns. In an emergency situation in which there is an absence of mine 255

256

R.J. Houghton, C. Baber & J.F. Knight

Figure 1.

Mine Clearance Patrol (image copyright M.O.D).

detection equipment, and/or where it is suspected low-metal mines are present that cannot be detected by metal detectors, an alternative body of technique involves the close physical inspection and probing of the ground. In the present experiment we examine whether these two different techniques might lead to differing forms of mental representation of the searched space, and thus change the quality of a subsequent verbal report. An aspect of mental representation that may be impacted by the degree to which search is mediated by technology and tools is the relative retention of environmental features (that is, the details of the small patch in which a target is located) versus the retention of global target location. Both aspects of a mine or explosive device are necessary parts of a report as they inform how further action can be a taken. Clearly one must know the position of a potential explosive device but knowledge of what is surrounding or covering the object can also have a bearing on how it would be recovered or acted upon. A distinction between spatial and identity representations has been suggested in the memory literature (e.g., Baddeley and Logie, 1999; Smith & Jonides, 1997). From this perspective, it might be argued that some technologies are better able to support one subsystem than the other. In the context of mine-detection, one might argue that a search involving a metal detector might be good for location (because the movement involved in using such a device might involve broad sweeps over local areas), but less good for detecting environmental features (because the sweeping might not support detailed local inspection); conversely, the use of a tool to dig into the ground might be good at supporting detecting environmental features (due to longer time spent at each location), but less good at location (because the inspection of each object might not be remembered as part of a spatial sequence so much as a temporal sequence. In previous studies, Baber and Houghton (in press) found that manual search led to superior recall in object identification over search using a metal detector, but that recall of location was similar for specified targets (although the metal detector condition led to better performance on a task involving the marking of allpossible targets). This implies that the studies which had identified improvements relating

Reporting location and environmental features when using metal detector

Figure 2a.

257

Layout of Markers on the Trays (trays containing metal objects are shaded).

Figure 2b. Trays laid out for experiment (key to letters: B = blue, Y = yellow, R = red, Bk = black). to support for the development of spatial patterns might not be related to the design of the technology per se, but rather to harnessing the ability to process location information.

Method The experiment involved 14 undergraduate students (12 male, 2 female). None had prior experience of the task or the domain. Thus, the experiment is not concerned with the specific abilities involved in mine detection, but addresses the general ability to search with different forms of technology support. The task involved a set of 16 seed trays laid out in a 4 × 4 grid. Each tray was filled with sand in which metal objects were buried in 8 of the trays. Each tray had a marker (3 × 3 cm) with a coloured shape printed on it. The layout of the markers is shown in figure 2. The objective of all participants was to search the trays, as many times as they wished,

258


until they were satisfied that they had found all of the metal objects and would be able to report both the location of the objects they had found and the identity of the markers for each object’s location (in terms of colour and shape). Participants were informed that they would be asked to recall one set of this information per trial, i.e., either recall the location of the object or recall the identity of the markers. Participants only completed two searches (one for each form of recall). The reason for this was to minimise the chances that they would simply learn the layout of the trays and rely on recall rather than search. Participants were randomly allocated to one of two conditions: metal detector or trowel. Prior to the trials, participants received an initial explanation on the study. This emphasised that the study was concerned with how well people could remember where they had found buried objects, emphasising that recall could be in terms of location or in terms of some feature in the environment (e.g., grid reference A4 or ‘by the tree’). They were then shown two test trays, one containing only sand and one with a metal object buried in the sand. They were asked to check these test trays in order to get a feel for how the search would be conducted. Participants with the trowel probed and dug in the sand until they were satisfied that an object was, or was not, present. Participants with the metal detector had to position the head of the device on the sand and heard a beep when metal was present. These practise trays provided both groups an opportunity to familiarise themselves with the task prior to the study. When they were satisfied that they understood the aims of the study and the tasks, participants were led to the 4 × 4 grid of trays and asked to search until they would be able to identify where all the metal objects had been hidden. Following this, they were led away from the grid and asked to recall either location of objects or identify of markers. Recall tasks were counterbalanced across participants. Next the search task was repeated and the other recall task performed. The dependent variables for this study were: time to complete the search task; number of checks made; recall of location; recall of marker identity; subjective rating of workload (using NASA-TLX). Pair-wise comparisons were made between conditions using t-tests.

Figure 3.

Checking the trays using trowel (on left) or metal detector (on right).


259

Results The results can be divided into three sections: search activity, reporting performance and workload.

Search activity In terms of time spent conducting the search, there is a small (but non-significant) difference between the conditions. On average, participants in the Trowel condition spent 108 (±35)s on the task, and participants in the Metal Detector condition spent 84 (±29)s. However, there was a difference in the number of checks made. On average, participants in the Trowel condition made 16 (±0.5) checks, whereas participants in the Metal Detector condition made 28 (±6) checks. This difference is statistically significant: t (12) = 5.2, p = 0.0002. If we crudely approximate a search per location as Mean time/Checks made, then we can see that the Trowel condition results in around twice as time spent per check, i.e., 6.7 (±2.2)s per check versus a time of 3.0 (±0.7)s per check for the Metal Detector. This difference is statistically significant: t(12) = 3.729, p = 0.0029. A comparison of the average number of checks for each tray further supports the claim that the metal detector condition led to a higher number of checks than the trowel. However, as indicated by the figures in bold in figures 4a and 4b, there does not seem to be any relation between number of checks and likelihood of finding metal in a tray.

Reporting performance Recall performance was measured on two dimensions: location and shape. In terms of location, there was no difference between conditions, i.e., recall of location for the Trowel condition had a mean of 4.9 (±2.1) correct items per trial and the Metal Detector condition had a mean recall of 5 (±2.8) items per trial. The location data were also analysed in terms of the participants’ ability to correctly trade-off recall of the location against erroneous recall. This was analysed as a form of signal detection (although, of course, we are not analysing the detection of the targets but the ability to recall the found targets which is quite a different question). Calculating d as z(FA) − z(H), we get a d for the Trowel condition of 1.004, and for the Metal Detector condition of 0.912. This demonstrates that the performance in the two conditions was very similar in terms of recalling the location of the targets. There was a noticeable difference between conditions in terms of the recall 1.4 1.4 1.4 1.5

1.4 1.5 1.5 1.6

1.4 1.6 1.5 1.6

1.5 1.5 1.6 1.5

Figure 4a. Mean checks per tray (Metal Detector).

1 1 1.1 1.1

Figure 4b.

0.9 1 1 1.1 1.1 0.9 1.1 0.9 1 1.1 1 0.9

Mean checks per tray (Trowel).

260


Table 1.

Comparison of Recall Performance for the two Conditions.

TROWEL

Target

No Target

METAL

Target

No Target

Recall No Recall

0.64 0.36

0.29 0.71

Recall No Recall

0.68 0.32

0.29 0.7

20 18 Metal Detector 16

Trowel

14 12 10 8 6 4 2 0 Mental

Figure 5.

Physical

Temporal Performance

Effort

Frustration

Mean NASA-TLX ratings across the two conditions.

of the shape. The Trowel condition had a mean recall of 5.1(±1) items per trial, while the Metal Detector condition had a mean recall of 3.0 (±1.3) items per trial. This difference was statistically significant – t(12) = 3.5324, p = 0.0041.

Subjective rating of workload A mixed Analysis of Variance (workload dimension × condition) was performed and revealed a signficiant main effect of workload dimension [F (5,60) = 6.221, p < 0.001] but not of condition. Inspection of figure 5 shows that the individual differences between participants (as shown by the standard deviation) was sufficient to obscure any differences between condition. However, we note small differences on the Physcial and Effort scales (which could be attributed to the need to bend and crouch when using the trowel, rather than simply walk around as with the metal detector), and also on the Mental and Frustration scales (which could imply that the search task was perceived as a little easier with the metal detector).

Conclusions Previous work into the ergonomics of mine detection had tended to focus on supporting spatial ability. The relative over-sampling of locations by the metal detector users in the experiment supports Stazewski’s account that some form of sequence


261

construction (possibly physically rehearsed using the metal detector itself as much as held in memory) is occurring: “. . . he located mines by sequentially finding edge points defined by the detector’s auditory outputs, holding them in memory and creating spatial patterns in his mind’s eye by linking contiguous edge marker locations” (Stazewski, 2007). However, this did not eventually lead to better spatial report of target locations than in the trowel condition (a caveat to this being that the spatial array in the present experiment only consisted of 16 cells). In terms of report of environmental features, the trowel users whose search was characterised by a slower, more effortful progress through the search array, showed superior performance compared to metal detector users. One interpretation of this is that feature identification is not well supported by the metal detector precisely because it engenders ‘scanning’ behaviour, which might suggest a possible problem in terms of conducting search of the environment and the location at the same time. One could suggest that, to some extent, military practice recognises this potential difficulty because in ideal circumstances it tends to divide search for mines into separate phases of detector-based search, marking and finally inspection. The present results suggest that in the development of future technologies to support search, consideration should be given to how the output of various devices can be combined in such a way as to streamline this process by either supporting both location and featural representation (which may not be possible if trade-offs between the two can not be reconciled) or else to automate either the representation of space (through positioning systems) or the representation of environmental features (through digital imaging) to allow the operator to allocate their attention to a specific dimension of an ongoing search task.

References Baber, C. and Hougthon, R.J., in press, Transitional information in a simplified mine detection tasks, The European Chapter of the Human Factors and Ergonomics Society Annual Meeting, TNO Soesterberg, October 2008 Baddeley, A. and Logie, R.H., 1999, Working memory: the multiple component model, In P. Shah and A. Miyake (eds) Models of Working Memory, Cambridge: Cambridge University Press, 28–61 Herman, H. and Iglesias, D., 1999, Human-in-the-loop issues for demining, In A.C. Dubey, J.F. Harvey, J.T. Broach and R.E. Dugan (eds.) Detection and Remediation Technologies for Mines and Mine-like Targets IV, Proceedings of SPIE, 3710, Bellingham, WA: SPIE, 797–805 Herman, H., McMahill, J.D. and Kantor, G., 2001, Enhanced operator interface for hand-held landmine detector, In A.C. Dubey, J.F. Harvey, J.T. Broach and R.E. Dugan (eds.) Detection and Remediation Technologies for Mines and Minelike Targets VI, Proceedings of SPIE, 4394, Bellingham, WA: SPIE, 844–851 Smith, E. E. and Jonides, J., 1997, Working memory: A view from neuroimaging. Cognitive Psychology, 33(1), 5–42 Staszewski, J.J., 1999, Information processing analysis of human land mine detection skill, InA.C. Dubey, J.F. Harvey, J.T. Broach and R.E. Dugan (eds.) Detection

262


and Remediation Technologies for Mines and Mine-like Targets IV, Proceedings SPIE 3710, Bellingham, WA: SPIE, 766–777 Staszewski, J. J., 2007, Spatial thinking in the design of landmine detection training. In G.L. Allen (Ed.) Applied Spatial Cognition, 231–267. New Jersey: Lawrence Erlbaum Associates Staszewski, J.J. and Davison, A., 2000, Mine detection training based on expert skill, In A.C. Dubey, J.F. Harvey, J.T. Broach and R.E. Dugan (eds.) Detection and Remediation Technologies for Mines and Mine-like Targets V, Proceedings of SPIE, 4038, Bellingham, WA: SPIE, 90–101

PREDICTIVE OPERATIONAL PERFORMANCE (PrOPer) MODEL K. Tara Smith HFE Solutions, Dunfermline Fife, Scotland Operational Performance Assessment or Operational Analysis is a key tool that can provide vital insight to force composition, interaction of new equipments, strategy, etc. However to date the human component has to the most part been filled by assumptions and rough measurements from similar tasks. This in turn means that issues like training, skill level, human workload and human error are not accounted for in the results. In recent years there have been a number of models developed that start to address this gap between Human Factors and the Operational Analysis fields. Two key developments are: • the STORM (Socio-cultural Teamworking for Operational Research Models) Model developed by DSTL and • the 4CCWM (For Close Combat Workload Model) developed by HFE Solutions This paper presents a conceptual framework that integrates these two models and discusses the integrated model which provides an initial pathway to predict operational performance from task and team composition. This new PrOPer model provides a coherent framework that allows an analyst to consider the consequence of workload in teams and scenarios. It addresses the major components of the human/task attributes and predicts the human performance impactors, thereby deriving a resultant performance and accuracy prediction. It is hoped that this approach will eventually remove the need for the OA models to make assumptions regarding human elements.

Introduction In the past, there has been much dissatisfaction at the lack of Human Factors models to feed into Operational Analysis (OA) studies. There is a recognised lack of human performance models that can be directly applied to OA due to the level of fidelity of the models. This paper describes one approach to address this longstanding issue. The development of this Predictive Operational Performance (PrOPer) model was inspired by the realisation that the STORM model and the 4CC Workload Model, although they address different problem areas, take a fundamentally similar approach to describing the task drivers and personnel profiles, utilising similar data and techniques. 263

264

K. Tara Smith

This paper presents an overview of the resultant model that combines elements of these two fundamental pieces of work and proposes that this approach can be used to supplement an OA assessment. This project resulted in a demonstrator of the model: this demonstrator did not provide the full functionality but was constructed in a manner that would facilitate the easy implementation of additional elements and features. The combined approach embodied elements of a number of other models and approaches. These include; • Wickens’ multiple resource theory • McCracken and Aldrich’s visual, cognitive, auditory, and psychomotor (VCAP) theory • Hopkins’ training methodology • Smith’s finite time element approach • NASA TLX assessment approach • Tuckman’s team maturity process • Noble’s knowledge enablers • Mathieson and Dodd’s model of organisational and social factors. Put simply, the objective was to be able to describe the social and cultural impactors, training levels, team structure, tasks and task demands to produce a model which can then predict; • • • •

Likely delays due to high workload Risk of human error A profile of high workload “hotspots” Task sensitivity to workload.

This prediction allows the examination of variations in team make-up, training levels and task allocation.

Context In any safety-critical industry (which includes the military environment), tasks often have to be performed within strict timescales. This means that the workload on the individual can be critical. The nature of human factors is not just looking at equipments and environments; it is understanding the fundamental capabilities of the human in the system and relating those to the demands of their tasks in the context of the environment in which they have to work. The PrOPer model allows the analyst to represent the task demands caused by the equipment and the environment and their dynamic nature, which can be driven by external events and relates it directly to the capabilities of the user. Other workload approaches have tackled this area by modelling the workload associated with the equipment: the approach taken in this model follows Hilary Putnam’s paradigm which is that weather systems are mainly driven by the inherent properties of water molecules, however studying water molecules does not allow

Predictive operational performance (PrOPer) model

265

you to predict the weather. The model allows the representation of an abstract level of workload for a whole team as well as for an individual person. Wickens’ multiple resource theory suggests that if this workload can be shared between the different cognitive resources then the accumulative workload is less likely to reach a critical threshold: the development of this model is intended to predict likely workload “hotspots” for the different cognitive resources thus allowing the possibility of mitigation strategies such as moving a task from visual to auditory or adding additional personnel. The model embodies Wickens’ multiple resource model to allow these workload aspects to be investigated. Tuckman’s work regarding the dynamics of teams and the subsequent work carried out by Mistry and Waters have created predictive models that allow the modelling of different scenarios and the consequences of changes to the team make-up. PrOPer embodies the STORM model, which was created to model the consequences of the make-up of the team. One current theme within military research is to do “better things” as well as “doing things better”: i.e. as well as improving current practices and equipments (doing things better) to introduce new and innovative approaches to solving the same problems (doing better things). This inevitably leads to requiring a better understanding of how things function and the underlying relationships between individual tasks and the overall performance of the manned system(s). Additionally, with the increasing costs of development, it is beneficial to be able to predict the consequences of decisions regarding equipment, manpower, etc., without committing to expensive development programmes. Therefore the model needs to represent the full operational context: i.e. hours or days of activity and 10s or 100s of individuals. The creation of predictive OA and HF models, if utilised as part of the procurement and investigation strategies, can have a major effect on identifying and fine-tuning strategies that improve manpower profiles, training, equipment and other cost drivers. As the model is data-driven it allows the same assessment to be applied to different solutions, providing better confidence in the comparative results. Currently the MoD requires a programme to be assessed through a series of financial gateways: in effect a “through-life cost versus operational effectiveness” trade-off. Although the human component of that trade-off is significant to determining the success of the programme, currently little data is transferred up from the human factors work to that level of decision-making. This model provides easily understood comparative data to be fed into the Government assessment/trade-off process, whilst still providing a view of the detailed human-related causal factors. Initial validation of the results of the demonstrator was conducted as part of the NEC4CC programme: the user trial was conducted over five days, however only involved three subjects. The trial analysed a 12-hour scenario representing a full Company (60–200 soldiers) and focussing on the three Company Commanders. Data collected during the trial was analysed against the model of the same scenario and concluded that the model represents over a 90% approximation to the study data. Further validation trials are planned both within NEC4CC and on other MoD research projects. This paper does not cover the validation activities or the additional development plans.

266

K. Tara Smith

The PrOPer model As there are still gaps in human factors knowledge, this model was constructed in such a way as to allow extensions to include new research without compromising the underlying data structures. Wherever practicable in the model described below, where research was not sufficiently mature to provide a capability of modelling that aspect, the aspect was represented in the model by a data structure that could in the future be acted upon by a model. An example of this is how fatigue affects the workload threshold.

Background The two basic pieces of work that fed into PrOPer were: • STORM: STORM models the impact of social and cultural factors upon teamwork. It is designed to improve the state of the art of modelling military command and control, in particular Headquarters (HQ) modelling, in support of Level 2 Operational Analysis (OA). Level 2 OA provides support to Capability Management and must operate at the system level of abstraction, working with models which explain the effects of interventions in equipment and other Lines of Development at this level. • 4CCWM: The Information Overload Study was part of the Network Enabled Capability for Close Combat (NEC4CC) research programme. Its objective was to create a individual model of workload combined with a team-level model which identifies the information needed to be passed around the team, as currently there are no formal models of the cognitive and physical workload demands on people within a close combat operational situation.

Conceptual framework The first step in creating the model was to create a top level conceptual framework to allow the description of the components of both STORM and 4CCWM within the context of operational performance. Conceptually, the framework is divided into four blocks: • Elements of Change: changes to equipment, training, etc will impact on the human/task attributes. • Human/Task Attributes: modifications to the individual aspects of the experience and capability of the humans will result in changes to the human performance impactors. • Human Performance Impactors: the performance of the humans who make up the teams and fulfil the tasks needs to be understood as their variation impacts directly upon the overall performance of the manned system. • Operational Analysis: focuses on overall performance, based on time to complete and/or accuracy of the predicted results. It needs to know about the people/teams and their logical deployment (OrBAT) and the external events that affect the operation.

Predictive operational performance (PrOPer) model Human / Task Attributes

Human Performance

267

Operational Analysis

Impactors

Equipment Design

Knowledge

Group Behaviour

Doctrine

Target Audience

Individual Behaviour

Training Design

Teamwork Performance

Attitudes Task Performance

Event Time Sequence

Manpower Tasks

Personnel Selection

Scenario Enviroment

Environmental Impactors on Performance

OrBat

Workload

Human Error Impactors on Performance

Success Prediction

Task Performance

Equipment Protective Equipment

Figure 1. Top level conceptual framework. By modifying the elements of change, the human/task attributes are altered, which affects the human performance impactors and thus impacts upon elements of the operational analysis. For example, will giving soldiers a better radio (changing the equipment design) improve the overall performance of the regiment? The conceptual framework provides an overview of the human factors domains that are utilised to generate the base-level data, how these impact on a programme and how they are represented in the operational analysis trade-off studies. The conceptual model includes a representation of the following elements to be able to answer and correspond to OA-related models in estimation of human workload and team capability; • Scenarios, goals and actions: to allow representation of different military vignettes • The team workload: to allow prediction of team performance • The individual’s workload: to allow prediction of individual’s performance • The components of the actions: to provide a generic set of components that can be mapped to any task • The demands of those action components: to be able to calculate the workload • The teams that are going to undertake the operation: to be able to represent different teams with different experience levels running through the same scenario • The characteristics and capabilities of the individuals in that team: to model their ability to undertake the activities • The overall team performance: to provide a standard metric of time and number of delays. These aspects are captured within the three basic components of the overall system model.

Top level architecture The following diagram shows the top level architecture of the model: i.e. the three basic components.

268

K. Tara Smith Individual Task List

Perceived Risk

Scenario, Goal & Task Level

Performance & Workload Level

Team Level Training Level of Individual

Resultant Task Workload Team

Team Maturity Level

Figure 2.

Resultant Task Delay Team

Components of the system model. S c e n a rio,Go a l & Ac tio n s Activity

Scenario Components Goal Triggering Events

Individual Task List

ORBAT Mission Command Mission–– Command Intent

Team

External Events

Individual

Operational Context Perceived Risk

Figure 3.

Scenarios, goals and actions.

The following sections describe each of these top-level components in turn.

Scenario, Goal and Task Level For modelling purposes, an assumption was made that a scenario can be fully described by a set of triggering events that happen at timed intervals throughout the scenario. The triggering events initiate goals which in turn are represented by a set of activities that have to be completed to satisfy the overall goal. Additionally, conditions can be set to ensure that an activity cannot be started prior to another activity being completed. As activities are assigned to individuals, this structure allows for a triggering event to initiate team-wide goals. As the model runs, it dynamically creates a task list for each individual. In the demonstrator, the perceived risk of the operation is being entered as a settable value and is not being derived from any aspects of the overall operational scenario.

Team Level The STORM model takes parameters that represent aspects of team dynamics such as skill levels, training levels, cultural coherence, etc., amalgamates these into intermediate values such as willingness to co-operate and from there derives what overall state the team is operating in, in terms of forming, storming, etc. This study

Predictive operational performance (PrOPer) model Forming

Storming

Norming

269

Performing

Activity Awareness

Goal

External Situation

Role, Task

Current Task

Others

Plan Assessment

Team Business Trust

Relationships

Decision Drivers Mutual Understanding

Task Skills Current Task Work

Team Initiation Training

Experience

Socialisation

Figure 4.

External Info Feed

Co-location

Underlying structure of STORM.

has taken the position that the overall team performance (team maturity) is provided by an instantiation of the STORM model within this overall conceptual model. The current instantiation of the STORM model only captures and models the characteristics of an entire team and does not consider the impact of the training level of an individual within the team. Thus there is an unmodelled relationship between the sum of the individuals’ training levels and the maturity of the team. This aspect is not modelled in the demonstrator or the conceptual model as no previous research work has been identified that relates these two elements, but is represented within the model by an additional criterion that indicates the training level of an individual. The conceptual model recognises that some activities can be dynamically allocated to different individuals within the team. It implements this by providing a simple look-up table that links an activity to “can be undertaken by”. This approach allows for the simple cases where an activity can be allocated to one or more individuals, however there are other factors that affect selection, for example, trust by the Commander, fatigue, etc. The demonstrator does not implement this aspect, however, the structure and data to allow it to be implemented in future versions is provided.

Performance and Workload Level The conceptual model adopted a slightly modified version of the structure utilised in the 4CCWM. The generic activity components are structured into four “banks” of activities that can be mapped to any individual task. These banks represent different aspects of the overall job, such as team-working and risk-monitoring. The conceptual model starts by representing the activity demands as a matrix of activity demands versus generic activity components, based on an expanded NASA TLX assessment method. Each bank of the generic activity components are then operated on to provide a resultant workload for each activity bank.

270

K. Tara Smith

Following the above integration, the model filters and scales the workload by what aspects of that activity are currently active within a specific time slot, then sums the workload from each activity assigned to that individual within that time slot. At this point, an assessment is carried out as to whether that individual has exceeded their workload threshold. The workload threshold is where the combination of task demands that are allocated to an individual are greater than the capability of that individual to perform the tasks in the time available. If the threshold is exceeded, then the lowest priority activity is pushed on to the next time slot. This approach is known as a finite time element approach. This mechanism only represents a simple “delay a task” coping strategy and does not allow for other coping strategies that an individual might adopt, such as passing off a task to a colleague, limiting the scope of a task in order to complete it quickly, etc. Additionally, it must be recognised that the workload threshold is dynamic and variable and that this should be taken into account when reviewing the final results of the model. The team workload is a composite of the workload of the individuals within the team and any composite delays to the completion of activities caused by an individual’s workload threshold. Note: as the workload threshold is decreased, the amount of time that the team is operating in and around their threshold will increase and the number of delays will also increase. It is therefore suggested that the model may be run at multiple thresholds to determine the sensitivity of the threshold to the resultant delays.

Results of the model (Outputs) The outputs of the modelling process can be examined at two levels: • The macro (operational analysis) level where the model can provide an overall assessment of how long an operation may take (including delays caused by high workload) and its subsequent measures of effectiveness. • The micro (human factors) level where the model can be used to examine the causes and sensitivity of workload levels. It is important to remember that the results from any modelling activity such as this must be interpreted and not just taken at face value. The following list provides examples of the interpretation that must be applied to the results for each of the desired outputs: • Likely delays due to high workload: the model reports tasks that have been delayed due to high workload for each timeslot calculated. The analyst needs to question whether the priorities at that point in time were appropriately modelled and whether there would be any other coping strategies that could be adopted by the individual, other than just delaying the task. • Risk of human error: although very high or very low workload are often considered the main drivers of human error, it is well established that when an individual considers the actual risk (safety or risk to operation) to be high, they take more care and adopt robust procedures to reduce the instances of uncorrected human error.

Predictive operational performance (PrOPer) model

271

• Task sensitivity to workload: this is in itself an interpretation of the two elements above, considered in the light of variations in workload threshold values. • A profile of high workload “hotspots”: collapsing workload into a single figure can lose critical information. For example, during a trial of the workload model it was found that although visual workload was often very high, it was always additional demands in linguistic/cognitive reasoning that pushed an individual over the workload threshold. For this reason, the model reports on five workload indices, which are tabulated against each other to establish the relationships between them for each individual and each timeslot. These indices are; – Perceptive Index – indication of spatial reasoning – Physical Index – indication of psychomotive processing – Cognitive Index – indication of linguistic reasoning – Demand Complexity Index – compilation of all task demand scores – Distractive Index – all task sensory demands.

Conclusion Current OA models tend to represent the human element in a simplified smanner: the PrOPer model, if used alongside an OA model, would allow a more realistic representation of that human component. This operational performance prediction model could potentially form one component of an overall modelling strategy. It forms the first practical integration of human factors data into an OA-style environment. This work could not have been carried out without the support of Dr Beejal Mistry of DSTL.

References Dr Beejal Mistry and Dr Mike Waters, March 2007, Social Networks & Culture in HQ Modelling. Calibration and Integration of STORM with extant C2 model DARNOS. Graham Mathieson, John Holt (HVR), Beejal Mistry, Neil Verrall, June 2004, Building Effective Representations of Social Networks and Culture in Models of HQ. Graham Mathieson, Beejal Mistry, Mike Waters, 2005, Coping with Social and Cultural Variables in C2 Modelling for Networked Enabled Forces. Smith, K Tara, November 2006, NEC for Close Combat WP 1.6.3 Information Overload Interim Report 72/06/R/300/I Issue 2. Wickens, C.D., 1984 Engineering Psychology and Human Performance, Harper Collins Pubs. McCracken, J.H. and T.B. Aldrich, 1984, Analysis of selected LHX mission functions: Implications for operator workload and system automation goals, Army Research Institute Aviation Research and Development Activity.

DISTRIBUTED SITUATION AWARENESS: APPLICATIONS AND IMPLICATIONS IN DEFENCE P.M. Salmon1 , N.A. Stanton2 , G.H. Walker2 & D.P. Jenkins2 1

Human Factors Integration Defence Technology Centre (HFI-DTC) Human Factors Group, Monash University Accident Research Centre, Monash University, Victoria, Australia 2 Transportation Research Group, University of Southampton, School of Civil Engineering and the Environment, Highfield, Southampton, SO17 1BJ, UK Situation awareness is a critical consideration in defence system design and evaluation. Despite almost two decades of research in the area, however, attempts to describe team situation awareness remain ambiguous. This article presents a model of distributed situation awareness, which uses the sub-concepts of compatible situation awareness and situation awareness transactions, to explain how teams develop and maintain situation awareness during collaborative activities. In addition, an accompanying modelling approach, the propositional network methodology, is outlined and demonstrated. Following this, a case study undertaken in the land warfare domain, involving a situation awareness-based analysis of a newly developed digital mission support system is summarised. In closing, the implications of both the model and case study findings are discussed with regard to the design of future defence systems.

Introduction The popularity of Situation Awareness (SA), the concept that focuses on how operators in complex systems acquire and maintain awareness of ‘what is going on’ (Endsley, 1995a), continues to rise. Notwithstanding this, and over two decades of scientific investigation, SA remains a contentious topic; the inescapable conclusion arrived upon by the authors following reviews of SA theory (Salmon et al., 2008) and measures (Salmon et al., 2006) is that there are still significant issues that need to be confronted. Of most relevance given the increasing presence of teams within complex sociotechnical systems (Fiore et al, 2003) are the limitations of existing models (e.g. Endsley, 1995a; Smith & Hancock, 1995 etc) and measures (e.g. Endsley, 1995b) when applied to collaborative systems. The aim of this article is to describe new approaches to both problems, including a recently developed distributed cognition based model of SA in collaborative environments and an accompanying modelling approach. Both are subsequently demonstrated using a case study example from the defence domain and, in closing, the implications of the model and case study findings for the design of defence systems are discussed. 272

Distributed situation awareness

273

Distributed situation awareness Concepts such as shared SA (e.g. Endsley & Robertson, 2000) and mutual awareness (Shu & Furuta, 2005) remain ambiguous. Certainly, shared SA seems to be a blunt characterisation that leaves key questions unanswered. For example, does shared SA mean that team members understand elements of a situation in exactly the same manner? Are team members with access to the same information really arriving at the same situational picture? How do different team member roles, tasks, experience and schema map onto the idea of shared SA? Can team SA be satisfactorily explained by looking at SA in the heads of individual operators? All are difficult, but pertinent questions and existing models do not seem to contain sufficient answers. As a corollary, many in the field have articulated the need for further explanation of team SA (e.g. Artman and Garbis, 1998; Gorman, Cooke & Winner, 2006; Salmon et al, 2008 etc). Following Artman & Garbis (1998) call for distributed cognition (e.g. Hutchins, 1995)-based models of SA for collaborative systems, Stanton et al. (2006) proposed the foundations for a distributed cognition-based theory of SA. These foundations were built upon by Stanton et al. (2009) who outlined a model of DSA, developed as a result of applied research in a range of military and civilian command and control environments. Briefly, Stanton et al’s model is underpinned by four theoretical concepts: schema theory, genotype and phenotype schema, Neisser’s (1976) perceptual cycle model of cognition and, of course, Hutchin’s (1995) distributed cognition approach. The model takes a systems perspective approach to SA and views SA as an emergent property of collaborative systems; SA therefore arises from the interactions between agents; it is associated with individual agents but it may not reside within them as it is borne out of the interactions between them. At a systemic level, awareness is distributed across the different human and technological agents involved in collaborative endeavour. Stanton et al. (2006; 2009) suggest that a system’s awareness comprises a network of information on which different components of the system have distinct views. Scaling the model down to individual team members, it is suggested that team member SA represents the state of their perceptual cycle (Neisser, 1976); individuals possess genotype schema that are triggered by the task relevant nature of task performance, and during task performance, the phenotype schema comes to the fore. It is this task and schema driven content of team member SA that brings the shared SA (e.g. Endsley & Robertson, 2000) notion into question. Rather than possess shared SA (which suggests that team members understand a situation or elements of a situation in the same manner), the model instead suggests that team members possess unique, but compatible, portions of awareness. Team members experience a situation in different ways as defined by their own personal experience, goals, roles, tasks, training, skills, schema and so on. Compatible awareness is therefore the phenomenon that holds distributed systems together (Stanton et al., 2006; 2009). Each team member has their own awareness, related to the goals that they are working towards. This is not the same as other team members, but is such that it enables them to work with adjacent team members. Awareness is not shared; instead the situation is viewed differently based on these personal and task-related factors. Each team members’

274

P.M. Salmon et al.

SA is however compatible since it is different in content but is compatible in that it is all collectively required for the system to perform collaborative tasks successfully. The question remains as to how DSA is built between team members? Of course, the compatible SA view does not discount the sharing of information, nor does it discount the notion that different team members have access to the same information; this is where the concept of SA ‘transactions’applies. Transactive SA describes the notion that agents within collaborative systems enhance the awareness of each other through SA ‘transactions’. A transaction in this case represents an exchange of SA information from one agent to another (where agent refers to humans and artefacts). Team members may exchange information with one another (though requests, orders and situation reports); the exchange of information between team members leads to transactions in the SA being passed around; for example, the request for information gives clues to what the other agent is working on. The act of reporting on the status of various elements tells the recipient what the sender is aware of. Both parties are using the information for their own ends, integrated into their own schemata, and reaching an individual interpretation. Thus the transaction is an exchange rather than a sharing of awareness. Each agents SA (and so the overall DSA) is therefore updated via SA transactions. Transactive SA elements from one model of a situation can form an interacting part of another without any necessary requirement for parity of meaning or purpose; it is the systemic transformation of situational elements as they cross the system boundary from one team member to another that bestows upon team SA an emergent behaviour.

Modelling distributed situation awareness: the propositional network approach Viewing SA as a systems level phenomenon has interesting connotations for its assessment. Individual-based SA measures such as SAGAT (Endsley, 1995b) are not applicable since they focus exclusively on the awareness ‘in-the-head’ of individual agents and overlook the interactions between them. Instead what is required is an approach that is able to model SA from a systems perspective, including the information that is distributed around the system, the usage of this information by different agents, how the information is combined together to form ‘awareness’ and the SA-related interactions between system elements. Accordingly, the propositional network methodology (Anderson, 1983) is proposed as a way of modelling a systems SA (networks have been used by cognitive psychologists to represent knowledge since the 1970s e.g. semantic networks). A propositional network is essentially a network depicting the information underlying a system’s awareness and the relationships between the different pieces of information. DSA is represented as information elements (or concepts) and the relationships between them, which relates to the assumption that knowledge comprises concepts and the relationships between them (Shadbolt & Burton, 1995). Propositional networks can be constructed from a variety of data sources, including observational and verbal transcript data, cognitive task analysis data, task analysis data or data derived from work-related artefacts such as Standard Operating Instructions (SOIs). Initially, concepts, or information elements, followed by


275

the relationships between them are defined. This involves undertaking a simple content analysis on the data and extracting keywords. These keywords represent the information elements, which are then linked based on their causal links during the activities in question (e.g. contact ‘has’ heading, enemy ‘knows’ plan etc). The output is a network of linked information elements; the network contains all of the information (and the relationships between the different pieces of information) that is used by the different agents and artefacts during task performance and thus represents the system’s awareness. Thus, the information elements related to DSA, the relationships between them and also the usage, ownership and sharing of these information elements as the scenario unfolds over time can be defined. To demonstrate, Figure 1 presents an extract of a verbal transcript collected during a study of DSA during land warfare activities; the keywords extracted from this transcript via content analysis are highlighted in bold and the ensuing propositional network is presented on the right hand side of the Figure.

Defence domain case study We have recently been involved in various naturalistic studies within the defence domain, including the land warfare domain (see below), the military aviation domain, the naval warfare domain (Stanton et al., 2006) and the multinational warfare domain. For the purposes of this article we focus on one such study, which involved the analysis of land warfare missions using a recently developed digital mission support system (See Stanton et al., 2008). The land warfare study focused on a new digitised battle management system that provides support for battlefield planning and execution tasks. The system provides users with various tools to support the land warfare mission planning and execution process, such as mission analysis, decision support overlay matrix, and synchronisation matrix construction tools and a real-time Local Operational Picture (LOP) display presenting friendly and enemy location and movement information and also various collaborative tools such as instant messaging and email facilities. The aim of our analysis was to assess the impact on DSA that the new mission support system had during mission planning and execution activities. The study was undertaken during a recent operational field trial of the digitised system involving a fully functional Division, Brigade and BattleGroup (BG) undertaking missions using the new system. Analysts located at the Brigade and BG HQ observed mission planning and battle execution activities over the course of the three-week trial. Initially, an SA requirements analysis was undertaken using HTA and interviews with subject matter experts. The findings indicated that the SA requirements of the different Brigade and BG staff were distinct and compatible, rather than shared. The nature of the land warfare mission planning and execution ‘system’ is such that each ‘cell’ within the Brigade and BG has very different roles and thus different, but compatible, SA requirements. As a result, even when team members have access to the same information, they are typically using it very differently and in

276

P.M. Salmon et al.

Figure 1. Propositional network example; figure shows verbal transcript and the resultant propositional network that is constructed based on the identification (via content analysis) of keywords from the verbal transcript data. conjunction with different pieces of information. As a corollary of this, SA across team members is very different in nature but is compatible in that it fits together to enable planning and mission execution activities to be undertaken successfully. To demonstrate, we can look more closely at the mission-planning phase, which involves the Brigade and BG working through seven planning questions in order to develop, select and refine appropriate courses of action. During question 1 (the battlefield area evaluation phase, i.e. what is the enemy doing and why?), the engineer cell is primarily concerned with developing SA the ground, friendly and enemy forces use of the ground and the impact that the ground is likely to have on friendly and enemy operations. The intelligence cell, on the other hand, is primarily concerned with understanding the threat posed by the enemy, including the enemy’s capability, strengths and weaknesses, doctrine and subsequent modus operandi. Thus even when both team members have access to the same information regarding the battlefield area and the enemy (e.g. terrain, enemy location, movements and formation), they use and view the information in a very different manner; it is the relationship between concepts that makes up their distinct SA. There awareness is however compatible since it connects together in order to allow production of the question 1 products; each cell’s awareness of the battlefield, the enemy and the level of threat conjugates together to form the question 1 output. It is this unique combination of information elements and the relationships between them that makes


Engineer view

Has

Intelligence view

ts

effec

Impact

Ground

ws

Has

Kn o

ws Kn o

Friendly operations

Has

Doctrine

Ground

on

Has

277

affec

ts

Capability

ef fe c

Knows

Knows

Has

Enemy operations

Enemy Has

Use of ground

Has

Is

s Ha

Enemy

ts

Has

Threat level

Has Is Strengths

Is Weaknesses

Figure 2. Extract of propositional networks showing Engineer and Intelligence cells differing views on enemy and ground information elements; figure demonstrates how each component is using the information for their own ends and how its subsequent combination with other information makes each view unique. each team member’s SA compatible and not shared; the very fact that an actor has received information, acted on it, combined it with other information and then passed it onto other actors means that its interpretation changes per team member; this represents the transaction in awareness referred to earlier. The engineer and intelligence components views on the enemy and ground during question 1 are represented in Figure 2. For the DSA analysis analysts directly observed the planning and battle execution activities within the Brigade and BG HQ. Propositional networks were developed based on verbal transcripts, HTA and live observation of the activities undertaken. The DSA analysis of the planning and battle execution activities observed provided compelling evidence of the impact on DSA that the new digitised mission support system. Firstly, compatible SA requirements were not supported in any way by the digital system; the system provided the same tools, functionality, displays and information regardless of who was using it. Secondly, due to problems with the digitised system (poor usability, convoluted processes etc), the teams involved continued to use the traditional paper map processes in order to support and supplement the new digitally supported process. From this it was concluded that the mission support system did not adequately support the acquisition and maintenance of DSA during the activities observed; rather, a combination of the paper map process and the new electronic systems was used. Thirdly, there were many instances in which the SA-related information presented by the digital mission support system was in fact inaccurate and was not compatible with the real state of the world. This was particularly problematic during battle execution, where the information presented

278

P.M. Salmon et al.

on the Local Operational Picture was often out of date or spurious. For example, due to data bandwidth limitations voice transmission was given precedence over global positioning data regarding the locations and movements of entities on the battlefield and as a result of, contact reports and positional information was presented on the LOP often up to twenty minutes late. Fourthly, as a consequence of the problems discussed above, users reported low levels of trust in the SA-related information presented to them by the electronic system; as a result users often took measures to clarify the accuracy of the information (e.g. requests for clarification of location and status reports). This served to add to the planning and execution process and also adversely impacted the tempo of operations.

Discussion The concept of DSA has significant implications for the design of complex collaborative systems, such as digital mission support tools. For example, rather than encourage shared perspectives on a situation (which assumes that team members require identical awareness of shared situational elements), DSA suggests instead that the primary goals of digitised mission support systems should be to support the distinct compatible SA requirements of its different users and to support SA transactions between team members. Supporting compatible SA involves adopting a ‘right information, at the right time, to the right team member’ design philosophy. This involves the provision of role-based interfaces, displays and tools that are designed based solely on each user’s distinct SA requirements. Displays and interfaces should present only the SA information that is required by each user and users should not be inundated with information required by other agents but not themselves, nor should they have access to tools and functionality that they do not explicitly require. Whilst this means that more sophisticated systems may be required (i.e. customisable interfaces, role-based systems) our case study findings suggest that adopting this approach will support DSA more effectively. Moving onto designing to support SA transactions, the exchange of awareness between agents, our research suggests that designers need an in-depth understanding of exactly what it is that users need to know, when they need to know and what they need to know it for. This allows designers to create information presentation systems (e.g. displays, interfaces) that present SA-related information logically in conjunction with the information that it is likely to be used with. Mapping information together on displays and interfaces supports the SA transactions that need to be made. For example, a land warfare mission support system (analogous to the one analysed in the case study) could present new incoming information regarding a destroyed combat vehicle to combat service support staff (whose job it is to remove and deal with casualties, repair damage and replenish forces) in conjunction with information relating to routes to and from the vehicle, casualty evacuation routes, distances and projected times, combat effectiveness, medical support information (e.g. nearest hospitals etc), force replenishment requirements, repair requirements and also resource availability. In this way, the system is supporting the integration of information elements and thus the SA transactions regarding the broken down


279

vehicle. On the other hand, the same system could present the information regarding the destroyed combat vehicle very differently in light of different user needs. For example, when presenting the information to the Chief of Staff (who is ‘running’ the battle at the ops table), associated information presented could include the proximal units and their capabilities, the Commander’s effects schematic, the task ORG (resource allocation chart) and the combat service support staff ’s assessment. This information would then support the Chief of Staff in allocating the destroyed units tasks to another unit on the battlefield. Encouraging the design of systems that support the distinct SA requirements of different users and the exchange of awareness between users seems an obvious proposition, and others in the field have articulated the need to consider distinct user SA requirements (e.g. Bolstad, Riley, Jones & Endsley, 2002; Gorman et al., 2006); however, our experiences suggest that it is a principle that is not often followed. Rather, systems appear to be designed based on an understanding of what information they need to present to users, but without an understanding of who requires what information or how the information will ultimately be used by different elements of a collaborative system. As a corollary of this, compatible SA requirements are not supported and the grouping of information on interfaces and displays can be inappropriate. Supporting compatible SA and SA transactions therefore depends on designers having a detailed understanding of the DSA requirements of a system’s different users. Matthews, Strater & Endsley (2004) point out that knowing what the SA requirements are for a given domain provides designers with a basis to develop optimal system designs to maximize human performance rather than overloading workers and degrading their performance. It is therefore recommended that collaborative system design should be driven with a very clear specification of the compatible and transactive SA requirements of the different users. It is important to note identifying DSA requirements does not involve merely identifying the different pieces of information that need to be known, rather it involves going further and identifying what it is that needs to be known, how this information is used and what the relationships between the different pieces of information actually are i.e. how they are integrated and used by different users. Identifying the relationships between different pieces of DSA-related information allows designers to group information meaningfully in their end design. Only with this in-depth understanding of the process and the resultant SA requirements of different team members can compatible and transactive SA truly be supported through system design. With increasing technological capabilities, there is great potential for enhancing DSA in defence systems through the provision of digitised mission support systems; however this is accompanied also by a very real opportunity to create defence systems that can hinder, rather than support DSA development and maintenance. Key issues to pursue relating to the concept of DSA include what information should be presented, in what manner and to which elements of the warfare system, how information can be presented in a more timely fashion and how the accuracy of information presented by command and control systems can be enhanced and ensured. It is hoped that the DSA model and propositional network approach presented in this paper be used to further our knowledge of how defence systems should be designed to better support DSA.

280

P.M. Salmon et al.

References Anderson, J. (1983). The architecture of cognition. Cambridge, MA: Harvard University Press. Artman, H. & Garbis, C. (1998). Situation awareness as distributed cognition. In T. Green, L. Bannon, C. Warren, Buckley (Eds.) Cognition and cooperation. Proceedings of 9th Conference of Cognitive Ergonomics. (pp. 151–156). Limerick: Ireland. Bolstad, C.A., Riley, J.M., Jones, D.G., & Endsley, M.R. (2002). Using goal directed task analysis with Army brigade officer teams. In Proceedings of the 46th Annual Meeting of the Human Factors and Ergonomics Society, Baltimore, MD, pp. 472–476. Endsley, M.R. (1995a). Towards a theory of Situation Awareness in Dynamic Systems. Human Factors, Vol. 37, pp. 32–64. Endsley, M.R. (1995b). Measurement of Situation Awareness in Dynamic Systems. Human Factors, Vol. 37, pp. 65–84. Endsley, M.R., & Robertson, M.M. (2000). Situation awareness in aircraft maintenance teams. International Journal of Industrial Ergonomics, Vol. 26 pp. 301–25. Fiore, S.M., Salas, E., Cuevas, H.M., & Bowers, C.A. (2003) Distributed coordination space: toward a theory of distributed team process and performance. Theoretical Issues in Ergonomics Science, 4, pp. 340 –364. Gorman, J.C., Cooke, N., & Winner, J.L. (2006). Measuring team situation awareness in decentralized command and control environments. Ergonomics, Vol 49, pp. 1312–1326. Hutchins, E. (1995). Cognition in the wild. MIT Press, Cambridge Massachusetts. Matthews, M.D., Strater, L.D., & Endsley, M.R. (2004). Situation awareness requirements for infantry platoon leaders. Military Psychology, 16, pp. 149–161. Neisser, U. (1976). Cognition and reality: Principles and implications of cognitive psychology. Freeman, San Francisco. Salmon, P., Stanton, N., Walker, G., & Green, D. (2006). Situation awareness measurement: A review of applicability for C4i environments. Journal of Applied Ergonomics, 37, pp. 225–238. Salmon, P.M, Stanton, N.A., Walker, G.H., Baber, C., Jenkins, D.P. & McMaster, R. (2008). What really is going on? Review of situation awareness models for individuals and teams. Theoretical Issues in Ergonomics Science, Vol 9, pp. 297–323. Shadbolt, N.R., & Burton, M. (1995). Knowledge elicitation: A systemic approach. In J.R. Wilson and E.N. Corlett (Eds.) Evaluation of human work: A Practical Ergonomics Methodology, pp. 406–440. Shu, Y., & Furuta, K. (2005). An inference method of team situation awareness based on mutual awareness. Cognition Technology & Work, 7, pp. 272–287. Smith, K., & Hancock, P.A. (1995). Situation awareness is adaptive, externally directed consciousness. Human Factors, 37, pp. 137–148.


281

Stanton, N.A., Salmon, P.M., Walker, G.H., Jenkins, D.P. (2009). Genotype and phenotype schema and their role in distributed situation awareness in collaborative systems. Theoretical Issues in Ergonomics Science. 10(1), pp. 43–68. Stanton, N.A., Stewart, R., Harris, D., Houghton, R.J., Baber, C., McMaster, R., Salmon, P.M., Hoyle, G., Walker, G.H., Young, M.S., Linsell, M., Dymott, R., & Green, D., 2006, Distributed situation awareness in dynamic systems: theoretical development and application of an ergonomics methodology. Ergonomics, Vol 49, pp 1288–1311.

HUMAN FACTORS MODELS OF MINI UNMANNED AERIAL SYSTEMS IN NETWORK-ENABLED CAPABILITY C. Baber, M. Grandt† & R.J. Houghton EECE, The University of Birmingham, Birmingham, B15 2TT, UK † FGAN Research Institute for Communication, Information Processing & Ergonomics Human-System Integration, Wachtberg, Germany In this paper we explore the potential uses of human factors modelling techniques (descriptive, predictive and process models) in a range of activities involving mini-UAS (Unmanned Aerial vehicle Systems). The primary interest lies in defining the information and decision chains that relate to the use of intelligence from these assets, and the impact that configuring these chains in different ways might have on overall performance. Our interest lies in contrasting different configurations of command structures to see how workload, situation awareness and decision making can be affected, and how intelligence from UASs can be incorporated into Network-Enabled systems.

Introduction The work in this paper is concerned with models of mini unmanned aerial vehicle (UAV) systems. Mini-UAVs have been designed to be operated by two people (in comparison with teams of 8 to 80 in medium and large UAVs) and are being used for a wide range of intelligence gathering operations in the military and, increasingly, by Civil Uniformed Services. While we can accept that a two person team can launch, fly and monitor such a UAV, it is likely that this team will form part of a larger organisation that will receive the resulting intelligence and act accordingly. This organisation will provide support for the UAV team (in the form of Force Protection), or authorise the mission that is being undertaken (in the form of Tasking Orders). Thus, there is a need to consider the relationship between the operators of mini-UAVs and the broader organisational framework within which they will operate.

Unmanned Aerial Vehicles If one views military operations as ‘Find – Fixate – Strike’, then we can consider contemporary warfare as a shift in emphasis in these terms. For example, if the enemy is hidden then the goal of Find becomes far more challenging; if the enemy is operating in an environment with other people living and working nearby, then the goal of Fixate becomes not simply a matter of defining a location but also of providing confirmation of a legitimate target and the goal of Strike becomes one which 282

Human factors models of mini unmanned aerial systems

Table 1.

283

Comparison of UAVs.

UAV

ALT [ft] Crew Craft GCS

Intel

MQ-9 Predator Reaper RQ-2 Pioneer (US Navy) RQ-2 Pioneer (US Marine) RQ Luna RQ-7 Shadow RQ-11 Desert Hawk RQ Aladin

>20 k

80+

4

Pilot PO 2 × PO

Launch and Maintenance recovery crew crew

Contemporary Ergonomics 2009: Proceedings of the International Conference on Contemporary Ergonomics 2009

Contemporary Ergonomics 2000 (Contemporary Ergonomics)

Contemporary Ergonomics 2008: Proceedings of the International Conference on Contemporary Ergonomics (CE2008), 1-3 April 2008, Nottingham, UK (Contemporary Ergonomics)

Contemporary Ergonomics 2003

Contemporary Ergonomics 1998

Contemporary Sport, Leisure and Ergonomics

Proceedings of International Science Education Conference 2009

Applied Ergonomics

Contemporary Computing: Second International Conference, IC3 2009, Noida, India, August 17-19, 2009. Proceedings (Communications in Computer and Information Science)

Contemporary Pragmatism 6-2 December 2009 (Contemporary Pragmatism S.)

Ergonomics In Computerized Offices

Visual Ergonomics Handbook

Biomechanics in Ergonomics

Office Ergonomics: Practical Applications

Perspectives In Rehabilitation Ergonomics

Visual ergonomics in the workplace

Ergonomics for Rehabilitation Professionals

Person-Centred Ergonomics

Meeting Diversity in Ergonomics

Ergonomics and Health Aspects of Work with Computers: International Conference, EHAWC 2009, Held as Part of HCI International 2009, San Diego, CA, ... Applications, incl. Internet Web, and HCI)

introduction to Ergonomics

Introduction to Ergonomics, 2E

Algorithmic Learning Theory: 20th International Conference, ALT 2009, Porto, Portugal, October 3-5, 2009, Proceedings

Advances in Cognitive Neurodynamics (II): Proceedings of the Second International Conference on Cognitive Neurodynamics - 2009

The history of human factors and ergonomics

Canine Ergonomics: The Science of Working Dogs

Introduction to Ergonomics Instructor's Manual

Visual Ergonomics Handbook

Ergonomics for Rehabilitation Professionals

Information and Communication Technologies in Tourism 2009: Proceedings of the International Conference in Amsterdam, The Netherlands, 2009

2009 International Building Code

Contemporary Ergonomics 2009: Proceedings of the International Conference on Contemporary Ergonomics 2009

Contemporary Ergonomics 2000 (Contemporary Ergonomics)

Contemporary Ergonomics 2008: Proceedings of the International Conference on Contemporary Ergonomics (CE2008), 1-3 April 2008, Nottingham, UK (Contemporary Ergonomics)

Contemporary Ergonomics 2003

Contemporary Ergonomics 1998

Contemporary Sport, Leisure and Ergonomics

Proceedings of International Science Education Conference 2009

Applied Ergonomics

Contemporary Computing: Second International Conference, IC3 2009, Noida, India, August 17-19, 2009. Proceedings (Communications in Computer and Information Science)

Contemporary Pragmatism 6-2 December 2009 (Contemporary Pragmatism S.)

Ergonomics In Computerized Offices

Visual Ergonomics Handbook

Biomechanics in Ergonomics

Office Ergonomics: Practical Applications

Perspectives In Rehabilitation Ergonomics

Visual ergonomics in the workplace

Ergonomics for Rehabilitation Professionals

Person-Centred Ergonomics

Meeting Diversity in Ergonomics

Ergonomics and Health Aspects of Work with Computers: International Conference, EHAWC 2009, Held as Part of HCI International 2009, San Diego, CA, ... Applications, incl. Internet Web, and HCI)

introduction to Ergonomics

Introduction to Ergonomics, 2E

Algorithmic Learning Theory: 20th International Conference, ALT 2009, Porto, Portugal, October 3-5, 2009, Proceedings

Advances in Cognitive Neurodynamics (II): Proceedings of the Second International Conference on Cognitive Neurodynamics - 2009

The history of human factors and ergonomics

Canine Ergonomics: The Science of Working Dogs

Introduction to Ergonomics Instructor's Manual

Visual Ergonomics Handbook

Ergonomics for Rehabilitation Professionals

Information and Communication Technologies in Tourism 2009: Proceedings of the International Conference in Amsterdam, The Netherlands, 2009

2009 International Building Code

Recommend Documents