No title

Archives of Sexual Behavior, Vol. 26, No. 5, 1997

Sexual Orientation Differences in Cerebral Asymmetry and in the Performance of Sexually Dimorphic Cognitive and Motor Tasks Geoff Sanders, Ph.D.12 and Marian Wright, B.Sc.1

With each of the tasks in the present studies we expected to find the reported sex difference between heterosexual women and heterosexual men and we predicted a sexual orientation effect with the performance of homosexual men being similar to that of heterosexual women and different from that of heterosexual men. Study I aimed to replicate earlier findings by recording the performance of a group of homosexual men on a visuospatial task, the Vincent Mechanical Diagrams Test (VMDT), a dot detection divided visual field measure of functional cerebral asymmetry, and on five subtexts of the Wechsler Adult Intelligence Scale (WAIS). For each task the profile of scores obtained for the homosexual men was similar to that of heterosexual women in that they scored lower than heterosexual men on the VMDT, they showed less asymmetry, and they recorded a higher Verbal than Performance IQ on the WAIS. In Study 2, a male-biased targeted throwing task favored heterosexual men while, in contrast, on the female-biased Purdue Pegboard single peg condition heterosexual men were outperformed by heterosexual women and homosexual men. On neither of these two tasks did the performances of homosexual men and heterosexual women differ. One task, manual speed, yielded neither sex nor sexual orientation differences. Another, the Purdue Pegboard assemblies condition, revealed a sex difference but no sexual orientation difference. Failure to obtain a sexual orientation difference in the Some data reported in this paper were presented in preliminary form as pan of the Symposium on Sex Differences in Cognition: Biological Influences, at the International Congress of Psychology, Brussels, July 1992 and at the 19th and 20th Annual Meetings of the International Academy of Sex Research, June 1993 and 1994. 1Department of Psychology, London Guildhall University, Old Castle Street, London, El TNT, England. 2To whom correspondence should be addressed.

463 0004-0002/97/10004463$12.50/0 c 1997 Plenum Publishing Corporation

464

Sanders and Wright

presence of a sex difference suggests that the sexual orientation effect may be restricted to a subset of sexually dimorphic tasks. KEY WORDS: sex differences; male homosexuality; sexual orientation; cognitive abilities; motor tasks; cerebral asymmetry.

INTRODUCTION The existence of sex differences in two related areas of psychological interest, cognitive abilities and functional cerebral asymmetry, has been widely investigated and much debated. Maccoby and Jacklin (1974) reviewed the evidence for sex differences in cognitive abilities and concluded that, on average, women scored higher than men on verbal tasks whereas men scored higher than women on mathematical reasoning and spatial tasks. In a review of sex differences in cerebral asymmetry, McGlone (1980) concluded that men were more lateralized than women. Levy (1969) used ideas concerning the presumed availability of processing capacity in men and women to support a causal relationship between sex differences in cerebral asymmetry and those reported for cognitive abilities. She argued that male superiority in spatial ability resulted from the marked lateralization in men of verbal abilities to the left and spatial abilities to the right hemisphere. The presence of some verbal ability in the right hemisphere of women was said to account for both their superior verbal ability (a result of the additional verbal capacity in the right hemisphere) and their inferior spatial abilities (a result of reduced spatial processing capacity in the right hemisphere). Other investigators have argued against the existence of sex differences in both cognitive abilities (verbal: Hyde and Linn, 1988; spatial: Caplan et al., 1985) and in functional cerebral asymmetry (Fairweather, 1982; Hahn, 1987). The arguments these investigators present depend on the high proportion of studies that have failed to find significant sex differences. Two points are important here. First, a study may fail to reject the null hypothesis for reasons other than the absence of a sex difference, e.g., because the tasks employed are insensitive or the sample size too small. Second, if reports of significant sex differences are chance occurrences among a wealth of nonsignificant findings then the significant effects should split 50:50 in favor of women and men. Halpern (1992) made both of these points in her criticism of the conclusions drawn by Hyde and Linn (1988) from their meta-analysis of 165 studies of sex differences in verbal abilities. Although it is true that 109 (66%) of the studies analyzed failed to find significant effects, of the 56 that reported significant sex differences in verbal abilities, 44 (79%) found that women scored higher than men. A

Sexual Orientation and Task Performance

465

similar picture emerges from studies of functional cerebral asymmetry as measured by auditory, visual, and tactual procedures. For example, Hahn (1987) reviewed 41 such studies of children (ages 1-15 years) that yielded a total of 178 same-age paired comparisons. Although 142 (80%) of these comparisons failed to find significant sex differences, of the 36 that did, 30 (83%) found males to be more lateralized than females. Effect sizes for sex differences in the performance of cognitive tasks range widely from large, through moderate, to small (Halpern, 1992). Clearly, some tasks better than others tap the elements of cognitive processing that differ between the sexes and of the many tasks that have been investigated, some are sexually dimorphic, favoring women or men, and others are neutral, favoring neither sex. Indeed, some tasks appear anomalous: solving anagrams (a verbal task) is performed faster by men than women (Hyde and Linn, 1988) while location memory for objects (a spatial task) is performed better by women than men (Eals and Silverman, 1994; Silverman and Eals, 1992). In the investigation of sex-related differences task selection is critical. Such considerations are equally important when we consider an associated area of interest that has emerged in the last decade, the possibility that differences in cognitive abilities and cerebral asymmetry exist between sexual orientation groups (Sanders and Ross-Field, 1987). Using three male-biased spatial tasks, two versions of the Piagetian water level task and the Vincent Mechanical Diagrams Test (VMDT), Sanders and Ross-Field (1986a) found that heterosexual men performed better than both heterosexual women and homosexual men whose scores did not differ. Levy's (1969) view, that sex differences in task performance may be related to differences in functional cerebral asymmetry, was supported by the outcome of a related study (Sanders and Ross-Field, 1986b) in which heterosexual men showed the male-typical left visual field advantage for a dot detection task, whereas homosexual men and heterosexual women showed no visual field advantages. Early support for the existence of sexual orientation differences in cognitive performance was provided by Willmott and Brierley (1984) who administered five subtests of the WAIS to homosexual men, heterosexual men, and an undifferentiated but predominantly heterosexual group of women. They reported no differences in prorated Full-scale IQ but a higher Verbal IQ for the homosexual men than for the heterosexual men and a higher Performance IQ for the heterosexual men than for the homosexual men and the women. Following a review of possible biological influences on cognition, cerebral lateralization, and sexual orientation, Sanders and Ross-Field (1987) suggested that the influence of prenatal hormonal events may give rise to concordant patterns of cognitive abilities, cerebral asymmetries, and sexual

466

Sanders and Wright

orientation. Evidence for an effect of prenatal hormones on adult sexual behavior comes primarily from work on rodents and there may be problems in applying the model to human sexuality (Meyer-Bahlburg, 1984). However, a recent study of women prenatally exposed to the synthetic estrogen, diethylstilbestrol (DBS), found that they were more likely than controls to be rated bisexual or homosexual (Meyer-Bahlburg et al, 1995). It is unlikely, as those investigators conclude, that a single biological factor could determine human sexual orientation but the evidence is persuasive that prenatal hormones contribute to this process. If so, sexual orientation differences in cognition and cerebral asymmetry may reflect the organizational activity of prenatal hormones. Evidence for sexual orientation differences in cognition has emerged from other studies that employed tasks reported to generate sex differences. With spatial tasks at which men excel, some, but not all, investigations have obtained results that are concordant with the earlier findings. Similar patterns of sexual orientation group differences were found (i) with the water level task by Gladue et al (1990) and McCormick and Witelson (1991), but not by Tkachuk and Zucker (1991) or Gladue and Bailey (1995) who found no group differences, and (ii) with the Vandenburg Mental Rotations Test by Gladue et al. (1990) and Tkachuk and Zucker (1991), who found group differences, but not by Gladue and Bailey (1995). In other spatial tests, homosexual men obtained intermediate scores on the Primary Mental Abilities Test but performed like heterosexual men rather than like heterosexual women on the Differential Aptitude Test (McCormick and Witelson, 1991). Investigators who have used tasks that women perform better than men have failed to find sexual orientation differences (Gladue et al, 1990; McCormick and Witelson, 1991), however, these investigators used tests of verbal fluency for which sex differences are often small. Here we present two studies, the first aimed to replicate earlier reports and the second to extend the investigation by including sexually dimorphic motor tasks that had not previously been used in this context. STUDY 1

Study 1 set out to replicate Sanders and Ross-Field (1986a, 1986b) and to provide prorated verbal and performance IQ scores from a group of homosexual men for comparison with the data presented by Willmott and Brierley (1984). A group of homosexual men completed the VMDT, a divided visual field dot detection task and five subtests of the WAIS. On the basis of the previous studies we predicted that, in all these tasks, the performance of homosexual men would be similar to that of heterosexual

Sexual Orientation and Task Performance

467

women and that both groups would obtain scores that differed significantly from those of heterosexual men. Methods Subjects Subjects comparable with those who participated in the studies of Sanders and Ross-Field (1986a, 1986b) were recruited from the local population by personal contact or referral by those already participating in the study. The 13 subjects were of comparable age (Table I) and all were classified right-handed having completed seven items selected from the Harris Test of Lateral Dominance (Harris, 1974) and scored a minimum of five right dominant responses, including writing with the right hand. The homosexual men were self-declared, lifelong homosexuals who were satisfied with their sexual orientation and had no history of seeking treatment. Scores obtained on the Sexual Orientation Method (SOM), as modified by Sambrooks and MacCulloch (1973), indicate that the new group of homosexual men were comparable to the original group in terms of their homohetero-erotic arousal. A lesbian group was not included because volunteers proved difficult to recruit. Measures Vincent Mechanical Diagrams Test. The VMDT forms part of the National Institute of Industrial Psychology Engineering Test Battery (NFER, 1980). It tests visuospatial ability and typically generates a marked male Table I. Age Range and Mean Sexual Orientation Method (SOM) Scores: New Group of Homosexual Men (nHmM) Compared with the Homosexual Men (HmM), Heterosexual Men (HtM), and Heterosexual Women (HtW) from the Original Studies by Sanders and Ross-Field (1986a, 1986b) New group

Original Sanders and Ross-Field groups

nHmM

HmM

HtM

HtW

13

13

13

13

Age range (years)

19-41

22-43

20-43

20-40

SOM score (min 6; max 48) Homosexual scale Heterosexual scale

47.85 18.35

46.92 13.50

10.61 47 .56

15.00 45.50

Variable No. of subjects

46S

Sanders and Wright

advantage. Subjects are required to deduce the mechanical relationships between elements in two-dimensional space. Each test item is a drawing composed of either levers, cogs, or pulleys. For each item subjects must select from four examples the one that illustrates the movement that would be produced by the mechanism. The test was administered according to the standard instructions and the proportion of correct responses recorded within the time allowed used as a measure of performance accuracy. Divided Visual Field Dot Detection Task. The divided visual field paradigm has been widely used to investigate functional cerebral asymmetry. A performance advantage for stimuli presented in one visual field is interpreted as superior performance by the contralateral cerebral hemisphere. The simple detection of a black dot on a white ground does not generate a visual field advantage but the use of a range of contrast levels revealed a left visual field (right hemisphere) advantage which was greater in men than women (Davidoff, 1977). The present study employed a similar procedure. A 1-cm2 dot of light was projected briefly, at one of four contrast levels, 2° of visual angle to the left or right of a central fixation spot. An opaque slide was used for blank trials when no stimulus was presented. Subjects were required to indicate whether they had seen a dot of light appear on the screen. During practice trials tandem Polaroid lenses were adjusted to obtain a light level for each subject which, with a constant exposure duration of 11 msec, produced correct dot detections on about 70% of the trials. Experimental trials were presented in eight blocks of 12 trials within which dots at each of the four contrast levels appeared randomly once in each visual field together with four blank trials. The number of correct detections was calculated separately for the left and right visual fields. Further details are given in Sanders and Ross-Field (1986b). Wechsler Adult Intelligence Scale. The five WAIS subtests used were those originally chosen by Willmott and Brierley (1984) because they loaded highest for pure factors (Maxwell, 1960). Verbal IQ, Performance IQ, and Full-scale IQ scores were obtained by prorating. Comprehension, similarities, and vocabulary were used for the verbal scale plus block design and object assembly for the performance scale. Each subtest was administered and scored according to the standard instructions.

Procedure The subjects completed the SOM in their own time and all the other tasks in a single laboratory session. On arrival, subjects provided biographical details and then completed the handedness inventory, VMDT, WAIS, and dot detection task in that order. At the end of data collection the

Sexual Orientation and Task Performance

469

subjects were debriefed and paid a fixed sum to cover local travel expenses. Statistical analyses were conducted as follows: a Kruskal-Wallis one-way ANOVA and Mann-Whitney U tests for the VMDT, parametric ANOVA with Newman Keuls for the dot detection task and a related groups t test for the prorated Verbal IQ and Performance IQ scores. Results

Table II shows the proportion of correct responses recorded on the VMDT spatial task. A significant main effect of sexual orientation was revealed when the new group of homosexual men was compared with the original data for the heterosexual men and women (H = 19.41 p < 0.001). The scores for the new group of homosexual men, like those for the original group, were significantly lower than those recorded by the heterosexual men, U(13,13) = 34.5, p < 0.02), but not significantly different from the scores of the heterosexual women, U(13,13) = 48.5, p > 0.05. Table II also shows the number of correct dot detections made by each of the sexual orientation groups in the left and right visual fields. Separate two-way ANOVAs were used to compare the original and new group of homosexual men with the original data for the heterosexual men Table II. Visuospatial Ability and Cerebral Asymmetry: Mean Scores Obtained by the New Group of Homosexual Men (nHmM) Compared with Data from Homosexual Men (HmM), Heterosexual Men (HtM), and Heterosexual Women (HtW) Obtained in the Original Studies by Sanders and Ross-Field (1986a, 1986b) Sexual orientation (SOR) New group

Vincent Mechanical Diagramsa x proportion correct responses SD Dot detection taskb Left visual field (max. score 32) x correct dot detections

SD Right visual field (max. score 32) x correct dot detections

SD aSOR:

Original Sanders and Ross-Field groups

nHmM

HmM

HtM

HtW

0.65 0.22

0.65 0.22

0.85 0.08

0.50 0.09

23.62 2.06

20.38 3.90

24.38 5.52

22.15 3.36

21.92 2.36

21.38 3.52

19.31 3.15

23.08 3.28

nHmM/HtM/HtW,p < 0.001; HmM/HtM/HtW,p < 0.001. nHmM vs. HtM,p < 0.02; nHmM vs. HtW, ns. bSOR x Visual Field: nHmM/HtM/HtW,p < 0.01; HmM/HtM/HtW,p < 0.01. Left vs. Right Visual Field: nHmM, ns; HmM, ns, HtM.p < 0.01; HtW, ns.

470

Sanders and Wright

and women. Both analyses revealed significant interactions between sexual orientation and visual field, original group F(2,36) = 8.22 p < 0.01; new group F(2,36) = 6.37, p < 0.01. For the new group of homosexual men the difference between the left and right visual field scores was not significant, W(1,36) = 1.70 p, > 0.05. This finding agrees with those in the original study where the visual field difference was significant for heterosexual men, W(1,36) = 5.07, p < 0.01, but not for the homosexual men or the heterosexual women, W(1,36) = 1.00 and 0.93, respectively, p > 0.05. Table III compares the prorated IQ scores for the new group of homosexual men with those reported by Willmott and Brierley (1984) for homosexual men, heterosexual men, and an undifferentiated but predominantly heterosexual group of women. The Full-scale IQ scores are similar for all four groups. The new group of homosexual men have higher Verbal than Performance IQ scores, t(12) = 23.31, p < 0.001, as did the homosexual men and the women in the original study and in contrast to the heterosexual men who had marginally higher Performance than Verbal IQ scores. In the Willmott and Brierley study there were no significant differences in Full-scale IQ scores, however, the homosexual men scored higher on Verbal IQ but lower on Performance IQ than the heterosexual men.

STUDY 2 In the light of the above findings, Study 2 was designed to investigate further the existence of sexual orientation differences using male- and feTable III. Prorated Full-Scale, Verbal, and Performance IQ Scores Obtained by the New Group of Homosexual Men (nHmM) Compared with Data for Homosexual Men (HmM), Heterosexual Men (HtM), and Women from Willmott and Brierley (1984) New groupa IQ scores

Original Willmott and Brierley groups

nHmM

HmM

HtM

Women

125.9 11.06

128.15 8.22

127.60 7.05

122.40 9.37

134.85 10.42

133.50 8.30

124.20 6.65

130.25 10.18

110.92 14.52

117.40 11.22

128.50 12.63

110.30 12.79

Prorated Full scale IQ X

SD Prorated Verbal IQ X

SD Prorated Performance IQ X

SD anHrnM:

VIQ vs. PIQ, p < 0.001.

Sexual Orientation and Task Performance

471

male-biased tasks known to generate substantial sex differences and which had not been employed previously in this context. Watson and Kimura (1991) argued that more ecologically valid tasks may yield larger sex differences. They suggested that sex differences in the performance of motor tasks may be related to the spatial sphere towards which the behavior is directed, with female advantages for behaviors performed close to the body and male advantages for behaviors directed at distant objects. Watson and Kimura found robust sex differences, favoring men, for targeted throwing and an interception task. Despite the similarity of these tasks to a number of sports, partialling out physique and sports experience had little effect on the sex differences indicating that they could not be explained by differences in physique and sports history. We selected the male-biased targeted throwing task for the present study. For the female-biased tasks we avoided verbal fluency in favor of motor tasks performed close to the body. Hampson and Kimura (1988) reported reciprocal changes in the performance of male-biased spatial and female-biased articulatory-motor tasks between different phases of the menstrual cycle with the female-biased tasks performed better during the midluteal phase than at menses. From their battery we selected the manual speed and Purdue Pegboard tasks. Our predictions were (i) that all the selected tasks would generate the reported sex differences between heterosexual women and heterosexual men; (ii) that the performances of the heterosexual women and the homosexual men would not differ on any of the tasks; (iii) that heterosexual women and the homosexual men would outperform heterosexual men on the female-biased tasks while the heterosexual men would perform better than the heterosexual women and the homosexual men on the male-biased tasks. Methods

Subjects The 45 subjects, ages 19-31 years, were recruited from a student population. All gave informed consent and, on the basis of their self-reported sexual preference, 15 subjects were allocated to each of three sexual orientation groups: heterosexual women, heterosexual men, and homosexual men. A lesbian group was not included because volunteers proved difficult to recruit. Handedness was defined by the preferred writing hand. All but four of the subjects were right-handed by this criterion and used this hand for all the unimanual tasks. The exceptions were one left-handed heterosexual man and three heterosexual women who wrote with their left

472

Sanders and Wright

hand although they used their right for throwing. All three women preferred to use their left hand for the manual speed task and Purdue Pegboard single peg condition and performed consistently better with that hand than with their right on these tasks. Measures Manual Speed. This finger tapping task formed part of a battery of female-biased tasks used by Hampson and Kimura (1988) to investigate changes in performance across the menstrual cycle. In common with other female-biased tasks, manual speed was faster during the midluteal phase than it was during menses with an effect size of about 0.4 standard deviation units. Subjects were required to operate a telegraph key with their index finger as fast as possible. Two 10-second trials were completed with each hand and the mean number of taps was recorded separately for the preferred and nonpreferred hands. Purdue Pegboard Single Peg and Assembly Conditions. These tasks, which are reported to favor women with an effect size of about 0.6, were administered according to the standard instructions (Tiffin, 1968). The unimanual single peg condition allows subjects 30 sec to insert as many individual pegs as possible into holes on a board. The test was performed twice with each hand and the score for the preferred and nonpreferred hands was calculated as the mean number of pegs correctly placed on two successive trials. The bimanual assembly condition requires subjects to complete four-component assemblies by inserting a peg followed by a washer, a collar, and a second washer in that order. The task is performed by the coordinated use of both hands with the preferred hand placing the peg and the collar while the nonpreferred hand places the washers. The score was calculated as the mean number of items (pegs, washers, and collars) correctly placed on two successive 60 sec trials. Targeted Throwing. Watson and Kimura (1991) described a marked male advantage with an effect size of 1.3 for this task. In the present study, subjects used an overhand throw with their preferred hand in an attempt to hit the center of a 45-cm diam board with a 21-g dart from a distance of 300 cm. The center of the board was positioned 150 cm above the floor. Radial error was measured to an accuracy of 0.5 cm. Any throw that failed to hit the board was recorded as an error of 22.5 cm. The score for each subject was the mean radial error for 20 throws. Since this task is very similar to the social and competitive game of darts that is more widely played in the United Kingdom than in Northern America, we should compare the requirements for the task and the game. The regulations for the

Sexual Orientation and Task Performance

473

game of darts stipulate that players throw from a distance of 93.25 inches (237 cm) and that the center of the target be 68 inches (173 cm) from the floor. Thus, for the targeted throwing task employed in the present study the throwing distance was 20% greater and the target height 15% less than that used in the game of darts.

Procedure The motor tasks were completed in the same order by all subjects: first the manual speed task; second the Purdue Pegboard single peg followed by the assembly condition; and finally the targeted throwing task. A detailed sports history questionnaire was distributed for return with postage charges prepaid. Data from each of the four tasks were analyzed separately by ANOVA with appropriate simple main effects. Results Manual Speed. Mean scores and standard deviations are shown in Table IV for the preferred and nonpreferred hands of the three sexual orientation groups. The rate of unimanual tapping showed no interaction between sexual orientation and the hand used, F(2, 42) = 1.79, p > 0.1, and no difference between the sexual orientation groups, F(2, 42) = 0.46, p > 0.5. As would be expected, higher tapping scores were achieved with the preferred hand, F(l, 42) = 63.51, p < 0.0001. Purdue Pegboard Single Peg Condition. Mean scores and standard deviations are shown in Table IV for the preferred and nonpreferred hands of the three sexual orientation groups. For this unimanual task there was a significant interaction between group and hand, F(2, 42) = 11.49, p < 0.0001. Significant sexual orientation group effects were present for the preferred hand, F(2, 84) = 13.28, p < 0.0001, but not for the nonpreferred hand,F(2,84) = 2.37, p > 0.05. With the preferred hand, both heterosexual women, F(l, 84) = 26.11, p < 0.0001, and homosexual men, F(l, 84) = 9.80, p < 0.005, placed more pegs correctly than heterosexual men. Homosexual men tended to place fewer pegs than heterosexual women, their mean scores were 17.17 and 18.20, respectively, but the difference failed to reach significance at the 5% level, F(l, 84) = 3.92, p > 0.05. Purdue Pegboard Assembly Condition. Mean scores and standard deviations for the three sexual orientation groups are shown in Table IV. There was no overall sexual orientation group effect for this bimanual task, F(2, 42) = 2.47, p > 0.05, but paired comparisons revealed the predicted sex difference with heterosexual women scoring higher than heterosexual

474

Sanders and Wright

Sexual Orientation and Task Performance

47S

men, F(l, 42) = 4.94, p < 0.05. The homosexual men obtained intermediate scores and were not significantly different from either the heterosexual men, F(l, 42) = 1.20, p > 0.1, or the heterosexual women, F(l, 42) = 1.27, p > 0.1). Targeted Throwing. The mean radial error scores and standard deviations for each of the sexual orientation groups are shown in Table IV. There was a significant effect of sexual orientation, F(2, 42) = 5.67, p < 0.01. The error scores of the heterosexual women and homosexual men did not differ, F(l, 42) = 1.05, p > 0.3, but both heterosexual women, F(l, 42) = 10.83, p < 0.005, and homosexual men, F(l, 42) = 5.13, p < 0.05, threw less accurately than heterosexual men.

DISCUSSION Performance differences between sexual orientation groups are of interest because they may reveal differential organizational effects of hormones acting during critical prenatal periods (Sanders and Ross-Field, 1987; Gladue and Bailey, 1995). With each of the tasks in the present studies we expected to find the reported sex difference between heterosexual women and heterosexual men and we predicted a sexual orientation effect with the performance of homosexual men being similar to that of heterosexual women and different from that of heterosexual men. These outcomes were obtained in five of the seven tasks used. The present data from the VMDT (Table II), the WAIS Performance IQ (Table III), and the targeted throwing task (Table IV), confirm our previous findings (Sanders and Ross-Field, 1986a) that, on tasks favoring men, homosexual men and heterosexual women obtain similar scores which are lower than the scores obtained by heterosexual men. Data from the WAIS Verbal IQ (Table III) and Purdue Pegboard single peg condition with preferred hand (Table IV)) extend the generality of this relationship to tasks favoring women where the scores of homosexual men and heterosexual women were similar and both performed better than heterosexual men. As far as we are aware the single peg task and prorated Verbal IQ are the only tasks favoring women that have been reported to show a sexual orientation effect. The dot detection data from Study I (Table II) confirm our earlier finding (Sanders and Ross-Field, 1986b) that homosexual men, like heterosexual women, are less lateralized than heterosexual men. Overall, a concordant pattern of results emerges from these tasks with homosexual men matching heterosexual women in their pattern of performance (verbal better than spatial) and in their lack of marked functional cerebral asymmetry. The performance of both groups contrasts with the pattern of

476

Sanders and Wright

task performance (better spatial than verbal) and marked functional cerebral asymmetry shown by heterosexual men. These sexual orientation differences, like the sex difference studies on which they are based, support Levy's (1969) view that patterns of cognitive ability are related to patterns of cerebral asymmetry such that better verbal ability is associated with reduced lateralization while better spatial ability is associated with marked lateralization. Although these findings are compatible with a biological influence on brain organization and task performance, an environmental influence cannot be ruled out. Our targeted throwing task provides a good example. If heterosexual men play darts more than heterosexual women and homosexual men, this practice in the required motor movements and spatial judgments may contribute to, or even account for, their superior skill. However, a higher participation rate in the game of darts could be the result of, rather than the reason for, their greater skill because individuals might be expected to play games at which they enjoy a reasonable level of success. Too few of our sports history questionnaires were returned for us to be able to establish that the groups in Study 2 had similar dart-throwing experience, however, we have some data that are relevant to the question of practice effects. Table V shows the performance of 13 women who threw 12 darts with each hand at a target three times per week over a period of 5 weeks. The size of the target and the height of its center from the ground were the same as in Study 2 but the throwing distance was reduced from 300 cm to 250 cm. Over the 5 week period there was no significant change in preferred hand performance but, compared with Week 1, throwing with the nonpreferred hand became increasingly more accurate from Week 3 to Week 5 (Table V). Given that there was no significant improvement in preferred hand performance these data suggest that it is the motor rather Table V. Radial Error Scores Obtained by Women in a Target-Directed Throwing Task Showing the Effects of Practice Over a Period of Five Weeks Preferred hand Radial error (cm)

.

Nonpreferred hand

Difference from Week 1

Radial error (cm)

Difference from Week 1

Week

X

SD

(p)

X

SD

(p)

1 2 3 4 5

10.81 10.31 9.69 9.40 9.39

4.77 3.96 3.55 2.63 3.54

ns ns ns ns

16.68 15.40 14.32 13.68 12.33

4.90 5.53 4.75 4.40 4.84

< 0.05 < 0.01 < 0.001

ns

Sexual Orientation and Task Performance

477

than the spatial judgment component of the task that improves with practice. Further support for the conclusion that heterosexual men are more accurate at targeted throwing than both heterosexual women and homosexual men has recently been provided by Hall and Kimura (1995). They employed a task that was dissimilar from the game of darts because it required an underarm throw with Velcro-covered balls at a carpet-covered target. As in the present study, they also found that heterosexual men were more accurate than heterosexual women and homosexual men. Relevant practice as revealed by sports history and the physical factors of hand strength and finger size did not account for these effects. However, even in the absence of practice and physical advantage, psychosocial factors, such as expectation based on gender stereotypes (Signorella and Jamison, 1986), may have influenced performance on the targeted throwing and it seems likely that adult patterns of abilities reflect an interaction between nature and nurture. The present findings indicate that sexual orientation differences may be more likely to emerge in male-biased tasks where the sex differences tend to be larger. This view is supported by other studies of sexual orientation differences as noted in the Introduction. A failure to find sexual orientation differences in the absence of sex differences may indicate a lack of sensitivity in the task. However, the failure to find a sexual orientation difference in the presence of a sex difference demands closer attention, especially if sexual orientation differences are found with the same subject groups on concurrent tasks. Such was the case here in Study 2 (see Table IV) where the assembly condition failed to generate a sexual orientation difference although the expected sex difference did appear. We cannot use a lack of sensitivity in the task to explain the absence of a sexual orientation effect here. Other studies have also reported the absence of a sexual orientation effect in the presence of a sex effect. Hall and Kimura (1995) failed to find a sexual orientation difference with a Purdue Pegboard task. However, they used a bimanual condition requiring subjects to insert pairs of pegs, one peg with each hand, a procedure that is a simpler version of the bimanual assembly condition for which no sexual orientation differences were found in Study 2 (Table IV). Again, contrary to the outcome of Study 1 and the data reported by Willmott and Brierley (1984), Tuttle and Pillard (1991) found no sexual orientation differences on subtests of the WAIS; however, there are two differences between the studies which could account for the different findings. First, Tuttle and Pillard used a different selection of subtests from those employed here and by Willmott and Brierley. Second, they compared their sexual orientation groups on the scores obtained

478

Sanders and Wright

for each individual subtest and not on the basis of prorated IQ scores for the performance and verbal scales. Two other studies have failed to find a sexual orientation effect in the presence of a sex difference. First, McCormick and Witelson (1991) found that homosexual men obtained intermediate scores on the Primary Mental Abilities Test but performed like heterosexual men rather than like heterosexual women on the Differential Aptitude Test. Second, Gladue and Bailey (1995) found sex but no sexual orientation differences with Mental Rotation and the Water Level Task. Outcomes such as these, where a sex difference between heterosexual men and heterosexual women occurs in the absence of a sexual orientation difference between heterosexual men and homosexual men, suggest that the sexual orientation effect may exist for a subset of tasks only. Studies of a wider variety of sexually dimorphic tasks are required before we can draw firm conclusions regarding the relationship between patterns of cognitive/motor abilities and sexual orientation. It would be valuable to have more data from a range of sexually dimorphic tasks in order to identify task demands that do and do not generate sexual orientation differences. As a guide, it has been suggested that ecologically valid, rather than pencil and paper, tasks may be more likely to show these differences (Kimura, 1996). Ultimately, clarification of the crucial task demands could point to brain regions that may differ between sex and sexual orientation groups. The influence of prenatal hormones on cognitive abilities is presumably exerted through organizational effects on the brain which may be revealed by different patterns of cerebral asymmetry in sex and sexual orientation groups (Sanders and Ross-Field, 1987). Given the recent support for a prenatal hormonal influence on human sexual orientation (Meyer-Bahlburg et al., 1995) it is notable that the divided visual field dot detection task in Study 1 of the present paper replicates the finding of reduced functional cerebral asymmetry in homosexual men originally reported by Sanders and Ross-Field (1986b). Further support for reduced asymmetry in homosexual men has appeared in a recent magnetoencephalographic (MEG) study (Reite et al., 1995). Sex differences in cerebral laterality of the MEG-based source location estimates for the 100-msec latency auditory evoked field component (M100) have been reported by Reite et al. (1989, 1995). In women the sources are symmetrically located in the superior temporal gyri whereas in men the hemispheric locations are asymmetrical with the right anterior to the left. Reite et al. (1995) found the M100 source locations to be symmetrically located in the left and right hemispheres of homosexual men. Thus the relationship between cerebral organization and sexual orientation demonstrated by this MEG study is similar to that revealed by the divided visual field data from Study 1 and from Sanders and Ross-Field (1986b): In both cases homosexual men and

Sexual Orientation and Task Performance

479

heterosexual women exhibit a similar pattern that differs from that shown by heterosexual men. A recent brain imaging study (Shaywitz et al., 1995), which found unilateral left hemisphere activity in men but bilateral activity in women during the performance of the phonological component of a verbal task, is compatible with a sex difference in cerebral asymmetry and suggests another avenue for future sexual orientation research.

REFERENCES Caplan, P. J., MacPherson, G. M., and Tobin, P. (1985). Do sex-related differences in spatial abilities exist? Am. Psychol. 40: 786-799. Davidoff, J. B. (1977). Hemispheric differences in dot detection. Cortex 13: 434-444. Eals, M., and Silverman, T. (1994). The hunter-gatherer theory of spatial sex differences: Proximate factors mediating the female advantage in the recall of object arrays. Ethol. Sociobiol. 15: 95-105. Fairweather, H. (1982). Sex differences. In Beaumont, J. G. (ed.), Divided Visual Field Studies of Cerebral Organization, Academic Press, London. Gladue, B. A., and Bailey, M J. (1995). Spatial ability, handedness, and human sexual orientation. Psychoneuroendocrinology 20: 487-497. Gladue, B. A., Beatty, W. W., Larson, J., and Staton, R. D. (1990). Sexual orientation and spatial ability in men and women. Psychobiology 18: 101-108. Hahn, W. K. (1987). Cerebral lateralization of function: From infancy through childhood. Psychol. Bull. 101: 376-392. Hall, J. A. Y., and Kimura, D. (1995). Sexual orientation and performance on sexually dimorphic motor tasks. Arch. Sex. Behav. 24: 395-407. Halpern, D. F. (1992). Sex Differences in Cognitive Abilities, 2nd ed., Erlbaum, Hillsdale, NJ. Hampson, E., and Kimura, D. (1988). Reciprocal effects of hormonal fluctuations on human motor and perceptual-spatial skills. Behav. Neurosci. 102: 456-459. Harris, A. J. (1974). Harris Tests of Lateral Dominance, 3rd ed., Psychological Corp., New York. Hyde, J. S. and Linn, M. C. (1988). Gender differences in verbal ability: A meta-analysis. Psychol. Bull. 104: 53-69. Kimura, D. (1996). Sex, sexual orientation and sex hormones influence human cognitive function. Cur. Opinion Neurobiol. 6: 259-263. Levy, J. (1969). Possible basis for the evolution of lateral specialization of the human brain. Nature 224: 612-615. Maccoby, E. E., and Jacklin, C. N. (1974). The Psychology of Sex Differences, Stanford University Press, Stanford, CA. McConnick, C. M., and Witelson, S. F. (1991). A cognitive profile of homosexual men compared to heterosexual men and women. Psychoneuroendocrinology 16: 459-473. McGlone, J. (1980). Sex differences in human brain asymmetry: A critical survey. Behav. Brain Sci. 3: 215-263. Maxwell, A. E. (1960). Obtaining factor scores on the WAIS. J. Ment Sci. 160: 1060-1062. Meyer-Bahlburg, H. F. L. (1984). Psychoendocrine research on sexual orientation. Current status and future options. Prog. Brain Res. 61: 375-398. Meyer-Bahlburg, H. F. L., Ehrhardt, A. A., Rosen, L. R., Gruen, R. S., Veridiano, N. P., Vann, F. H., and Neuwalder, H. F. (1995). Prenatal estrogens and the development of homosexual orientation. Dev. Psychol. 31: 12-21. NFER. (1980). National Institute of Industrial Psychology Engineering Test Battery, Nelson, Windsor, U.K.

480

Sanders and Wright

Reite, M., Sheeder, J., Richardson, D., and Teale, P. (1995) Cerebral laterality in homosexual males: Preliminary communication using magnetoencephalography. Arch. Sex. Behav. 24: 585-593. Reite, M., Sheeder, J., Teale, P., Richardson, D., Adams, M., and Simon, J. (1995). MEG-based laterality: Sex differences in normal adults. Neuropsycholagia 33: 1607-1616. Reite, M, Teale, P., Goldstein, L., Whalen, J., and Linnville, S. (1989). Late auditory sources may differ in the left hemisphere of schizophrenic patients: A preliminary report. Arch. Gen. Psychiat. 46: 565-572. Sambrooks, J. E., and MacCulloch, M. J. (1973). A modification of the sexual orientation method and an automated technique for presentation and scoring. Br. J. Soc. Clin. Psychol 12: 163-174. Sanders, G., and Ross-Field, L. (1986a). Sexual orientation and visuospatial ability. Brain Cognit. 5: 280-290. Sanders, G., and Ross-Field, L. (1986b). Sexual orientation, cognitive abilities and cerebral asymmetry: A review and a hypothesis tested. Ital J. Zool. 20: 459-470. Sanders, G., and Ross-Field, L. (1987). Neuropsychological development of cognitive abilities: A new research strategy and some preliminary evidence for a sexual orientation model. Int. J. Neurosci. 36: 1-16. Shaywitz, B. A., Shaywitz, S. E., Pugh, K. R., Constable, R. T., Skudlarski, P., Fulbright, R. K., Bronen, R. A., Fletcher, J. M., Shankweiller, D. P., Katz, L., and Gore, J. C. (1995). Sex differences in the functional organization of the brain for language. Nature 373: 607-609. Signorella, M., and Jamison, W. (1986). Masculinity, femininity, androgeny, and cognitive performance: A meta-analysis. Psychol Bull. 100: 207-228. Silverman, I., and Eals, M. (1992). Sex differences in spatial abilities: evolutionary theory and data. In Barlow, J. H., Cosmides, L., and Tooby, J. (eds.), The Adapted Mind, Oxford University Press, Oxford, pp. 533-549. Tiffin, T. (1968). Purdue Pegboard Examiner Manual, Science Research Associates, Chicago. Tkachuk, J., and Zucker, K. J. (1991, June). The relation among sexual orientation, spatial ability, handedness, and recalled childhood gender identity in women and men. Poster presented at the annual meeting of the International Academy for Sex Research. Tuttle, G. E., and Pillard, R. C. (1991). Sexual orientation and cognitive abilities. Arch. Sex. Behav. 20: 307-318. Watson, N. V., and Kimura, D. (1991). Nontrivial sex differences in throwing and intercepting: relation to psychometrically-defined spatial functions. Pen. Indiv. Diff. 12: 375-385. Wilmott, M., and Brierley, H. (1984). Cognitive characteristics and homosexuality. Arch, Sex. Behav. 13: 311-319.

Archives of Sexual Behavior, Vol. 26, No. 5, 1997

Understanding Sexual Coercion Among Young Adolescents: Communicative Clarity, Pressure, and Acceptance Doreen A. Rosenthal, Ph.D.1,2

Young people's understanding of sexual coercion was studied. Boys and girls (N = 191) were asked to rate scenarios depicting sexual situations according to their perceptions of communicative clarity, the extent of pressure being applied to one partner, and the acceptability of the behaviors. Judgments of communicative clarity were given more readily when there was consent rather than dissent to sex. Clear communication was readily inferred even when there were no cues that this was the case. Boundaries of behaviors that were defined as constituting "pressure" were influenced by the outcome, that is whether sex did or did not occur, as well as the behavior itself. Ratings of acceptability closely followed those of pressure, although the relationships between perceptions of pressure and acceptability were stronger for girls than for boys. In general, there were few gender differences in perceptions of pressure and communicative clarity. Of concern was the finding that, for some respondents, pressure and acceptability were unrelated to the use of either physical or emotional force. KEY WORDS: sexual coercion; adolescents; communication; sexual roles.

INTRODUCTION

We examined the understanding that young people have of sexual situations in which the possibility exists for the behavior of one partner to This study was supported by a grant from the Australian Research Council and by funding from the Victorian Health Promotion Foundation. 1Centre

for the Study of Sexually Transmissible Diseases, La Trobe University, Locked Bag

12, Carlton South 3053, Victoria, Australia. 2To whom correspondence should be addressed.

481 0004-0002/97/1000-0481$12.50/0 c 1997 Plenum Publishing Corporation

482

Rosenthal

be interpreted as exerting pressure on the other partner to engage in unwanted sex. Coercive sexual behavior among university or college students, labeled variously as courtship violence, date (or acquaintance) rape, sexual harassment, and unwanted sex play, has been a focus of interest among researchers for decades. It is clear that many young people (the overwhelmingly majority of whom are women) experience unwanted sexual intercourse (Craig, 1990; Erikson and Rapkin, 1991; Gavey, 1991; Koss et al, 1987; Koss and Oris, 1982; Muehlenhard, 1988; Muehlenhard et al, 1991; Muehlenhard and Linton, 1987; Patton and Mannison, 1995; Richardson and Hammock, 1991), however it seems that men are reluctant to admit to these behaviors (Craig, 1990; Patton and Mannison, 1995). In the context of concerns about young people's sexual well-being, in part as a result of HIV/AIDS and other sexually transmitted diseases which can have long-term serious consequences, it has become increasingly important to understand more about the ways in which young people negotiate a sexual encounter. This includes understanding the nature and meaning of sexual coercion and identification of possible antecedents and correlates of unwanted sexual behaviors (Gavey, 1992; Holland et al, 1991a). There have been several theoretical explanations of coercive sexual behaviors among heterosexuals. These include Craig's (1990) situational model in which it is proposed that individual personal characteristics, the situational context, and cognitive processes interact to produce coercive behavior on the part of men. Thus the personal dispositions of young men, such as their arousal patterns or their attitudes towards women, interact with the situational context. For example, opportunities may be perceived to manipulate a social situation to their sexual advantage such as use of alcohol or emotional threats, or to interpret messages in accordance with their own intent. As Patton and Mannison (1995) noted, the model has limited explanatory power for women. There is little room for the complex sexual socialization experienced by young men and women and the proscriptions, based on gender, which apply to sexuality. An analysis of sexually coercive practices and experiences in terms of understandings of heterosexuality and male/female relationships is provided by writers such as Gavey (1992), Holland et al. (1991a, 1991b, 1992), Hollway (1984), and others. Gavey wrote of the "technologies of heterosexual coercion" in which young women themselves take part of the responsibility for enduring coercive practices because these are part of "normal" heterosexual relationships. Holland et al. (1991a) argued that the sexual pressure experienced by young women results from the patriarchal power enjoyed by young men and women's subordination by men. This is expressed along a continuum of sexual violence, encompassing gentle persuasion as well as

Understanding Sexual Coercion

483

aggression and forceful rape (Kelly, 1987; Koss et al, 1987). These latter approaches suggest that, in order to understand sexual coercion, we need to take account of how young people interpret and understand sexual encounters which range from those where overt and explicit pressure is applied, such as forced rape, to those in which pressure is subtle or implied. To date, the sexual encounters studied in most American literature fall into the category of "date rape" among college students. The date rape literature deals with situations in which some men perceive sexual coercion as justifiable, and resistance by the woman to a sexual encounter is merely viewed as a hurdle to overcome (see, e.g., Feltey et al., 1991; Muehlenhard and Linton, 1987; Shotland and Goodstein, 1992). In other research, dispositional features of coercive males such as "hypermasculinity" or attention to selective interpretation of women's cues have been identified (Beaver et al, 1992; Craig Shea, 1993). Several studies (e.g., Abbey 1982, 1987; Abbey and Melby, 1986; Craig Shea, 1993) have demonstrated that men interpret women's cues as more sexually motivated than do the women themselves. This "misinterpretation" of women's responses may be a contributing factor in men's coercive behavior and, taken together with the widespread belief of some men that women often say no when they mean yes (Muehlenhard and Linton, 1987), suggests that the role of clarity in communications about sexual events needs to be addressed. Most studies of sexual coercion have sampled college students. This population, while important, is only one subset of populations that need study. In particular, we need research on younger adolescents because of their sexual vulnerability. These younger boys and girls may have more difficulty in communicating about sex, may be less sensitive to partners' needs, and may be more likely to respond to their immediate desires. Furthermore, it is at the early stages of the development of sexual relationships that particular patterns of behavior are laid down and are likely to be continued in later interactions. In one study of high schoolers (Davis et al., 1993), a majority of boys (60%) reported that it was acceptable for a boy to force sex on a girl in one or more situations. Of interest is that some girls agreed, especially in situations where the girl had "led him on," or where the couple had had sex before. This raises the question of the extent to which understandings of pressure and the acceptability of behaviors are gender-related. Although it is clear that many expectations about sexual practices are strongly gender-determined (Moore and Rosenthal, 1993), there is some evidence from respondents of both sexes that the same rules (e.g., how to encourage or discourage a sexual encounter) guide the sexual behaviors of boys and girls (Rosenthal and Peart, 1996). Whether boys and girls share the same ideas about definitions of sexual practices as coercive and/or acceptable, and

484

Rosenthal

whether the biased reading of communications evident among college students extends to younger respondents are the focus of the research reported here. The study was designed to examine a range of sexual encounters, some of which we believed could be defined a priori as coercive. We used scenarios to manipulate the level of pressure being exerted by one partner over the other (ranging from overt physical force through indirect pressure to no pressure) and the clarity of the communicative response by the other partner (ranging from statements of desire for, or rejection of, sex through indirect communication to no communication). We were interested in the extent to which sexual coercion was influenced by communicative clarity and perceptions of consent as well as the boundaries of behaviors defined as constituting pressure and thus coercive and unacceptable.

METHOD Subjects The sample consisted of 191 young people (98 boys and 93 gals), ages 15-17 years (x = 16.3), 81% of whom were living with their parents. Respondents were attending recreational youth groups in metropolitan Melbourne. The majority (71%) were still at school and the sample was predominantly working class. Most of the respondents (87%) were born in Australia. Of the remainder, 10 (5%) were Asian-born and the others were European or of English-speaking descent. Measures As part of a study of young people's understandings of sexual coercion, respondents were administered a questionnaire that contained scenarios depicting 16 situations with two protagonists where sex may or not ensue. The scenarios were designed to vary in the extent to which the desire to have or not have sex was clearly articulated, whether sex did or did not occur, and whether pressure was exerted by one individual on the other. In all cases, items and their wording were chosen following discussions with young people of the same age and from similar backgrounds as the respondents. These young people were in agreement that neither heterosexuality nor male coercion of female partners should be assumed in the scenarios. Thus, all scenarios were worded so that responses could refer to same or different sex partners, and to males or females. Subsequent

Understanding Sexual Coercion

485

pilot testing in focus groups confirmed that the scenarios depicted situations that would be familiar and relevant to the respondents and that they would be able to understand both the instructions and the scenarios. Respondents were asked to rate, on a 4-point scale, the clarity of the communication ranging from 1 (not at all clear) to 4 (perfectly clear). They were then asked to rate, for the same scenarios, the amount of pressure applied from 1 (no pressure) to 4 (a lot of pressure). Finally, respondents rated how acceptable the situation in each of the scenarios would be to them from 1 (not at all acceptable) to 4 (perfectly acceptable). The following instructions and scenarios were presented to the adolescents: 1. Please rate how CLEAR it is to Person X what Person Y wants in each situation. In each case the encounter is new and begins with the couple X and Y kissing intimately. 2. Now we would like you to read the stories again and rate in your opinion, how much PRESSURE is being applied by X on Y. Remember, in each case the encounter is new and begins with the couple X and Y kissing intimately. 3. We are interested in what you consider to be ACCEPTABLE. Please read the stories again and rate how ACCEPTABLE each situation would be to YOU.

Scenarios 1. X and Y discuss whether sex is possible. Y says he/she does not want to have sex. X then threatens to leave. 2. X kisses Y, then suggests a walk. 3. X and Y try to excite each other. Both agree to sex and sex then occurs. 4. X and Y discuss whether sex is possible. Y says he/she does not want to have sex, but X holds Y down so he/she cannot get away and sex occurs. 5. X feels really excited and continues to kiss Y. Sex doesn't occur. 6. Before increasing excitement, X asks Y if he/she would like to have sex. Y agrees. 7. In order to encourage sex, X tells Y that he/she may be in love. Without further discussion, sex occurs. 8. X "pretends" to hold Y down, playfully removing his/her clothes and kissing Y until, without discussion, sex occurs. 9. X and Y discuss whether sex is possible. Y says he/she does not want to have sex. X then threatens to end the relationship.

486

Rosenthal

10. X asks repeatedly for sex. Y says he/she does not want to have sex. X keeps asking and sex occurs. 11. X tries to excite Y until, just before sex, X asks if sex would be okay. Y agrees. 12. X presumes that sex is going to occur. X removes Y's clothes and tries to excite him/her until, without discussion, they have sex. 13. X tries to excite Y. Sex doesn't occur. 14. Y passes out from drinking too much alcohol. X then proceeds to remove Y's clothing and to kiss and touch him/her, until Y wakes up, still drunk, and sex occurs. 15. Before increasing excitement, X tells Y that he/she would like to have sex. Y agrees. 16. X finds him/herself too embarrassed to suggest sex with Y, so sex doesn't occur. Procedure Following clearance from the University Ethics Committee, respondents were recruited through youth workers at various locations in metropolitan Melbourne and in the country. The questionnaire was administered to groups of volunteer respondents at each location. Informed consent was obtained from all respondents prior to administration of the questionnaire which was completed anonymously.

RESULTS Table I shows the number of respondents who believed that communication was quite or perfectly clear, that there was no or hardly any pressure, and that the situation described was quite or perfectly acceptable. Sex differences in perceptions of communicative clarity, pressure, and acceptability were examined by MANOVA (with some slight loss of respondents because of missing data). Communication Of interest is the finding that for all but two of the scenarios (14 and 16) the majority of respondents reported that Y clearly communicated his/her wishes, although only three of the scenarios (3, 6, and 15) attracted this response from more than three quarters of the boys and girls. It appears that agreeing to have sex (Scenarios 3,6,11, and 15) was more readily

Understanding Sexual Coercion

487

Table I. Number (Percentage) of Adolescents Responding Positivelya Communication

Pressure

Situation

Boys

Girls

Boys

Girls

1 2 3 4 5 6 7 8 9 10 11 12 13

59 (60.2) 59 (60.2) 79 (83.2) 47 (48.0) 55 (56.1) 75 (78.2) 68 (70.1) 68 (70.1) 50 (52.1) 43 (44.4) 75 (78.2) 55 (56.1) 59 (60.2) 34 (34.7) 77 (79.3) 41 (42.2)

63 (67.7) 49 (52.6) 77 (82.8) 61 (66.3) 53 (57.6) 74 (80.5) 47 (51.7) 48 (52.2) 53 (58.3) 47 (51.7) 61 (66.3) 46 (49.5) 51 (54.9) 36 (39.2) 76 (82.6) 37 (39.8)

19 (19.4) 62 (63.3) 69 (70.4) 21 (21.4) 52 (54.1) 66 (68.0) 48 (50.0) 42 (43.3) 16 (16.5) 20 (20.6) 54 (55.6) 53 (54.6) 59 (60.8) 34 (35.4) 65 (66.3) 51 (52.0)

18 (19.4) 63 (68.5) 75 (83.3) 15 (16.2) 67 (73.7) 73 (79.3) 39 (42.4) 42 (46.2) 12 (13.1) 12 (13.1) 52 (57.2) 50 (53.8) 61 (66.3) 29 (31.2) 77 (82.8) 69 (74.2)

14 15

16

Acceptability Boys 24 62 74 24 65 78 62 48 29 24 70 52 61 17 76 62

(24.5) (63.9) (76.31) (24.5) (66.3) (80.5) (63.9) (50.0) (29.9) (25.1) (73.7) (54.2) (63.6) (17.7) (78.4) (63.9)

Girls 11 (11.8) 68 (73.1) 73 (79.4) 19 (20.7) 61 (65.6) 71 (76.4) 42 (45.6) 44 (47.3) 21 (22.9) 19 (20.7) 64 (68.8) 34 (36.6) 68 (73.1) 15 (16.5) 72 (77.4) 66 (71.7)

aCommunication

response: quite/perfectly clear; pressure response: no/hardly any pressure; acceptability response: quite/perfectly acceptable.

recognized as clear communication than was saying no (1, 4, 9. and 10), while the absence of discussion (2, 5, 7, 8, 12, 13) was, nevertheless, interpreted by many respondents as clear communication. There were significant univariate effects of sex of respondent for Scenarios 4,F(1, 173) = 6.40, p < .05, and 7, F(l, 173) = 4.46, p < .05. In the former, girls were more likely than boys to report communication as clear (Boys' x = 2.49; Girls' x = 2.93) while the converse held for the latter scenario (Boys' x = 2.86; Girls' x = 2.58). Pressure The scenarios elicited more varied responses in ratings of the pressure applied to Y than in judgments of clarity. The majority of respondents reported that significant pressure was applied to Y in six scenarios (1, 4, 7, 8,9,14). Scenarios rated as using least pressure were those where Y agreed to sex which subsequently occurred (3, 6, 11, 15), or where there was no discussion but no sex followed (2,5,11,16). In those situations where there was no discussion and sex occurred, ratings of pressure were moderate when there was no overt threat by X but, rather, a gradual buildup of excitement in order to encourage Y's acquiescence. Those cases where Y

488

Rosenthal

explicitly stated that he/she did not want sex (and which were accompanied by physical or psychological threats on the part of X) were regarded as instances of pressure on Y by over 80% of respondents. There were significant univariate effects of sex of respondent for five scenarios: 3, F(l, 174) = 8.42, p < .01; 5, F(l, 174) = 10.11, p < .01; 10, F(l, 174) = 5.53, p < .05; 15, F(l, 174) = 10.25, p .01; and 16, F(l, 174) = 11.15, p < .001. There was also a nonsignificant trend for scenario 4, F(l, 174) = 3.43, p = .06. Girls reported that greater pressure was being applied by X in scenarios 4 (Boys' x = 3.32; Girls' x = 3.57) and 10 (Boys' x = 3.13; Girls' x = 3.44). Boys reported more pressure being applied for scenarios 3 (Boys' x = 2.02; Girls' x = 1.61), 5 (Boys' x = 2.38; Girls' x = 1.92), 15 (Boys' x = 2.13; Girls' x = 1.67), and 16 (Boys' x = 2.40; Girls' x = 1.87). Acceptability

Ratings of acceptability closely followed those of pressure. Scenarios in which high levels of pressure were reported (1, 4, 9, 10, 14) were rated as unacceptable by the large majority of respondents whereas those scenarios that described consensual sex (3, 6, 11, 15) or where sex did not occur (2, 5,13,16) were rated as acceptable by a majority of boys and girls. Falling between these were the scenarios where sex occurred without discussion but without perceived pressure. There were significant sex differences for Scenarios 1, F(l, 174) = 3.89, p < .05; 7, F(l, 174) = 10.43, p < .01; and 12, F(l, 174) = 7.81, p < .01. In all cases, boys reported the situation to be more acceptable than did girls (Boys' x scores: 1.75, 2.87, and 2.63; Girls' x scores: 1.47, 2.39, and 2.20 for Scenarios 1, 7, and 12, respectively). Relationships Between Communication, Pressure, and Acceptability

Rank-order correlations were calculated from the rankings of scenarios as shown in Table I. Although the rankings of scenarios according to absence of pressure and acceptability were remarkably similar (p = .87 for boys and .91 for girls) they were less so for communication and acceptability (p = .62 for boys and .43 for girls). It was not clear, however, whether these reflected significant relationships between the measures. Correlations between communication and acceptability and between pressure and acceptability were calculated for each of the 16 scenarios using, for each measure, respondents' scores on the rating scales (from 1 to 4). Correlations are reported separately for boys and girls in Table II.

Understanding Sexual Coercion

489

Table II. Correlations of Communication and Pressure with Acceptability Communication/acceptability Scenario

Boys

Girls

1 2 3 4 5

.06 .22

6

7 8 9 10 11 12 13 14 15 16 ap

bp

.33a

.19 .15 .36a .24 .16 .04 .22 .24 .38b

.12 .28a .41b

.00

Pressure/acceptability Boys

Girls

.11

-.11

.07 .37* -.14 .24 .34a .13

-.15 -.14

-.14 -.35" -.15 -.18 -.24 -.16 -.37b

.34a -.02

.05 .17

-.30° -.11 -.07 .07 -.23 -.27a -.03 -.16

.34°

.14

.00 .10 .13 .00

-.10 -.16 -.19 .03

-.34"

-.26" -.28a -.52b -.17 -.10 -.28a -.28a -.10

< 0.01. < 0.001.

In general, correlations were modest with similar patterns emerging for girls and boys in terms of the relationships between acceptability and communicative clarity but not pressure. For both boys and girls, correlations between communication and acceptability exceeded .30 for only 4 of the 16 scenarios. In three cases (Scenarios 3 and 6 for both sexes and Scenario 15 for boys only), communication was clear and positive. As might be expected, the relationship between pressure and acceptability appeared to be stronger for girls than for boys, with girls' perceptions of greater pressure correlating significantly with lower ratings of acceptability in 8 scenarios, compared with 2 in the case of boys. DISCUSSION This study sought to shed light on the understandings that young people have of communicative clarity and pressure in situations where sex may or may not result, and the acceptability of the protagonists' behavior. A surprising finding is the divergence of young people's views about what constitutes clear communication, pressure, and acceptability. There was wide variability in the responses of these 16-year-olds and few scenarios

490

Rosenthal

received positive or negative endorsement on any of the three measures by at least four fifths of the respondents. Even in the most extreme situations (e.g., Y explicitly saying yes to sex or X virtually raping Y as in Scenario 4), there was a substantial minority who held an unexpected view (unclear communication, little or no pressure). This suggests that we need to be careful in imposing normative views of communicative clarity and/or pressure on young people. It appears that communication was judged as clearer when that message was yes to sex rather than no, suggesting some support for the cultural norm that, in sexual situations, no can be interpreted as meaning yes. Of some concern was the finding that girls as well as boys held this view. There were few differences between girls and boys in reports of the clarity of the messages being conveyed. This calls into question the assumption that boys continue to pressure girls into having unwanted sex because they misunderstand girls' messages. If it is true that some boys failed to understand the message, it is also true that girls had difficulty in this regard. Given the brief and straightforward nature of our scenarios, it is unlikely that the failure of our respondents to report clear messages was due to actual inability to decode the communication. Rather, it appears that boys and girls were responding to social rules about sexual encounters in which communication, however explicit, is only part of the whole. These findings suggest that efforts to encourage young people to accept, at face value, the positive or negative messages that they receive requires disentangling communication from context. In part, this means that we need to ensure that young people of both sexes recognize that no means no and that relying on nonverbal messages of consent or otherwise is a risky practice. As might be expected, young people were more likely to report that pressure was being exerted in situations where explicit pressure tactics, either physical or emotional, were used (holding someone down, threatening to leave a relationship), especially if unwanted sex followed. Most (but not all) of these young people appeared to recognize that having sex after one of the partners has stated that he/she did not want to do so involved pressure. On the other hand, a situation in which there was a nonsexual endpoint was less likely to be regarded as involving pressure even when there was a persistent attempt at persuasion on the part of one protagonist. Again, there were more similarities than differences between boys and girls. It is difficult to detect a pattern in the differences that were observed, but there is a hint that girls were more likely to report pressure when hassled until sex occurs (Scenarios 4 and 10), and less likely than boys to report pressure when there was a sense of mutuality about proceedings (Scenarios 3, 15, and 16) whether or not sex occurred. These findings are consistent with those reported by Davis et al (1993) in which their female high school-

Understanding Sexual Coercion

491

ers were more likely to report that it was acceptable for a boy to force a girl to have sex with him in situations where it could be inferred that the girl had played a role in the buildup of sexual arousal. Acceptability of the sexual scenarios appeared to be determined primarily by the existence or otherwise of pressure. Both boys and girls agreed that scenarios depicting consensual sex were the most acceptable, followed by those in which no sex occurred. Sex without discussion but with goodhumored pressure was moderately acceptable but nonconsensual sex (or ambiguous situations where sex may or may not occur) was not. This rank ordering of acceptability levels is interesting since it indicates that the large majority of these young people agreed that sex between willing partners was a highly acceptable activity, a belief that appears to be borne out in practice (Moore and Rosenthal, 1993). Again, fitting with findings from other studies (Buzwell, 1996; Buzwell et al., 1992), boys had a somewhat higher threshold of acceptability for situations in which pressure was more overt. The patterns of correlations between acceptability and pressure or communication point to the complexity of the process that young people engage in when making decisions about sexual situations. A key finding is that for girls but not boys acceptability is related to perceived pressure. Clarity of communication, on the other hand, yields some puzzling and inconsistent correlations. For example, saying yes related to acceptability, but not in all cases and not for both sexes. Because in the scenarios communication and pressure were confounded, it may be a more useful strategy to examine the content of those scenarios deemed to be acceptable or unacceptable. Nevertheless, there are lessons to be learned about the impact of perceived pressure on the ways in which young people assess sexual situations as acceptable or otherwise. Finally, our finding that, on the whole, there are few sex differences agrees with other research which suggests that boys and girls are equally aware of the rules of sexual encounters (Rosenthal and Peart, 1996). However, when it comes to acting out these rules, we know that boys and girls behave differently (e.g., Holland et al, 1991a, 1991b, 1992; Moore and Rosenthal, 1993). The task for educators is to discover why there is a gap between knowledge and practice. The present study used written scenarios which, although pilot tested with young people, may not have been realistic enough to reflect real life experiences. Moreover, responses to these scenarios do not tell us what happens in real life as young people find themselves in a variety of sexual situations. To judge a scenario as unacceptable does not mean that it will not occur. While it is not possible to obtain real-life data, in situ, one strategy we are adopting is to conduct in-depth

492

Rosenthal

interviews with young people, to document what their own experiences and their reactions to these have been. From the present study, though, we can assert that sexual coercion has many forms, and that what one person may report as coercive behavior may be viewed as gentle persuasion by another. Of concern is that what appears to us as a clear example of coercion, even rape, can be regarded as acceptable behavior by a worrying number of these 16-year-old boys and girls. We need to know more about the characteristics of those whose judgments of scenarios such as these are at odds with those of the majority of young people and with our norms about acceptable sexual practices. Finally, the research suggests that if we are to reduce the number of unwanted sexual encounters between young people and increase their ability to negotiate so that the sexual needs and wishes of both boys and girls are met, then we must clarify the meaning of pressure and teach young people to be more effective communicators.

ACKNOWLEDGMENTS

The author thanks Rachel Peart and Anne Mitchell for their contributions to the study.

REFERENCES Abbey, A. (1982). Sex differences in attributions for friendly behavior: Do males misperceive females' friendliness? /. Pers. Soc. Psychol 42: 830-838. Abbey, A. (1987). Misperceptions of friendly behavior as sexual interest: A survey of naturally occurring incidents. Psychol. Women Quart. 11: 173-194. Abbey, A., and Melby, C. (1986). The effects of nonverbal cues on gender differences in perceptions of sexual intent. Sex Roles 15: 283-298. Beaver, E. D., Gold, S. R., and Prisko, A. G. (1992). Priming macho attitudes and emotions. J. Interpers. Violence. 7: 321-333. Buzwell, S. (1996). Constructing a sexual self. Doctoral dissertation, La Trobe University, Melbourne, Australia. Buzwell, S., Rosenthal, D. A, and Moore, S. M. (1992). Idealising the sexual experience. Youth Stud. Aust. 3: 10 Craig, M. E. (1990). Coercive sexuality in dating relationships: A situational model. Gin. Psychol. Rev. 10: 395-423. Craig Shea, M. E. (1993). The effects of selective evaluation on the perception of female cues in sexually coercive and noncoercive males. Arch. Sex. Behav. 22: 415-433. Davis, T. C, Peck, G. Q., and Storment, J. M. (1993). Acquaintance rape and the high school student. J. Adolescent Health Care 12: 220-224. Erikson, P., and Rapkin, A. (1991). Unwanted sexual experiences among middle and high school youth. J. Adolescent Health 12: 319-325. Feltey, K. M., Ainstie, J. J., and Gibb, A. (1991). Sexual coercion attitudes among high schools students: The influence of gender and rape education. Youth Soc. 23: 229-250.

Understanding Sexual Coercion

493

Gavey, N. J. (1991). Sexual victimization among New Zealand university students. /. Consult. Clin. Psychol. 59: 464-466. Gavey. N. J. (1992). Technologies and effects of heterosexual coercion. Fem. Psychol. 2: 325-351. Holland, J., Ramazanoglu, C, Sharpe, S., and Thomson, R. (1991a). Pressured Pleasure: Young Women and the Negotiation of Sexual Boundaries, The Women's Risk AIDS Project, Paper No. 7, Tufnell Press, London. Holland, J., Ramazanoglu, C, Scott, S., Sharpe, S., and Thomson, R. (1991b). Presure, Resistance and Empowerment: Young Women and the Negotiation of Safer Sex, The Women's Risk AIDS Project, Paper No. 6, Tufnell Press, London. Holland, J., Ramazanoglu, C, Scott, S., Sharpe, S., and Thomson, R. (1992). Risk, power and the possibility of pleasure: Young women and safer sex. AIDS Care 4: 273-283. Hollway W. (1984). Gender difference and the production of subjectivity. In Henriquwa, J., Hollway, W., Urwin, C. Venn, C, and Walkerdine, V. (eds.), Changing the Subject: Psychology, Social Regulation and Subjectivity, Methuen, London, pp. 227-263. Kelly, L. (1987). The continuum of sexual violence. In Hammer, J., and Maynard, M. (eds.), Women, Violence and Social Control, Macmillan, London, pp. 46-60. Koss, M. P., Gidycz, C. A., and Wisniewski, N. (1987). The scope of rape: Incidence and prevalence of sexual aggression and victimization. J. Consult. Clin. Psychol. 50: 455-457. Koss, M. P., and Oris, C. J. (1982). The Sexual Experience Survey: A research instrument investigating sexual aggression and victimization. J. Consult. Clin. Psychol. 57: 455-457. Moore, S. M., and Rosenthal, D. A. (1993). Sexuality in Adolescence, Routledge, London. Muehlenhard, C. L. (1988). Misinterpreted dating behaviors and the risk of date rape. J. Soc. Clin. Psychol. 6: 20-37. Muehlenhard, C. L., Goggins, M. F., Jones, J. M., and Satterfield, A. T. (1991). Sexual violence and coercion in close relationships. In McKinney, K., and Sprecher, S. (eds.), Sexuality in Close Relationships, Erlbaum, Hillsdale, NJ, pp. 155-175. Muehlenhard, C. L. and Linton, M. A. (1987). Date rape and sexual aggression in dating situations: Incidence and risk factors. J. Counsel. Psychol. 34: 186-196. Patton, W., and Mannison, M. (1995). Sexual coercion in dating situations among university students: Preliminary Australian data. Aust. J. Psychol. 47: 66-72. Richardson, D. R., and Hammock, G. S. (1991). Alcohol and acquaintance rape. In Parrot, A., and Bechhofer, L. (eds.), Acquaintance Rape: The Hidden Crime, Wiley, New York, pp. 83-95. Rosenthal, D. A., and Peart, R. (1996). The rules of the game: Teenagers communicating about sex. J. Adolescence 19: 321-332. Shotland, R. L., and Goodstein, L. (1992). Sexual precedence reduces the perceived legitimacy of sexual refusal: An examination of attributions concerning date rape and consensual sex. Pen. Soc. Psychol. Bull. 18: 755-764.

Archives of Sexual Behavior, Vol. 26, No. 5, 1997

Testosterone Treatment in Men with Erectile Disorder and Low Levels of Total Testosterone in Serum Zoran Rakic, M.D., Ph.D.,1,3 Vladan Starcevic, M.D., Ph.D.,2 Vesna P. Starcevic, M.D., Ph.D.,2 and Jelena Marinkovic, Ph.D.2

Since decreased serum levels of testosterone (T) do not necessarily predict good outcome of testosterone treatment for erectile disorder, the purpose,of this study was to determine which men with erectile disorder and decreased serum levels might benefit from treatment. From a sample of 31 men (\ age = 39 years), 15 (48%) with erectile disorder and decreased serum levels of T responded well after 8 weeks of testosterone treatment (100 mg of testosterone propionate in the sustained-release form given im once a week). Good treatment outcome was associated with several variables, but only high levels of luteinizing hormone (LH) and low values of the T/LH (testosterone/LH) ratio consistently emerged as significant correlates and/or predictors of effective treatment. Levels ofLH above 7.5 IU/L or the values of the T/LH ratio equal to or below 0.87 nmol/IU in patients with erectile disorder and decreased serum levels of T suggest that testosterone treatment may be effective. KEY WORDS: erectile disorder; impotence; testosterone; luteinizing hormone; testosterone/Iuteinizing hormone ratio.

NTRODUCTION Decreased levels of serum testosterone may cause a decrease in men's sexual desire, which may subsequently impair their sexual arousal and cause erectile disorder. Such sexual dysfunction may respond to treatment with 1Department

of Psychiatry, Clinical and Hospital Center "Dr. Dragisa Misovic," Bulear JNA 84,11000 Belgrade, Yugoslavia. 2Belgrade University School of Medicine, Belgrade, Yugoslavia. 3To whom correspondence should be addressed. 495 00044M02f97/1000-049S$12.50/0 C 1997 Plenum Publishing Corporation

496

Rakic, Starcevic, Starcevic, and Marinkovic

testosterone (Benson, 1994; Carani et al., 1990; Gooren and Rubens, 1987; Kwan et al., 1983; National Institutes of Health [NIH], 1992; O'Carroll et al., 1985). Men with a decreased sexual desire are usually not aware of it in the beginning and are more likely to be troubled by a gradually developing erectile disorder. In middle-aged men who develop such a dysfunction, it is particularly important to determine whether its cause is hormonal. Therefore, testosterone serum levels should be measured in such individuals, even though most men who seek help in sexual disorder clinics do not have a hormonal origin of their erectile disorder (Benson, 1994; NIH, 1992). In view of the clinical observation that some men with erectile disorder respond to testosterone treatment, while others do not, the purpose of this study was to determine which men with erectile disorder and decreased serum levels of total testosterone might benefit from this treatment. We sought to elucidate those factors that are associated with good outcome of testosterone treatment for erectile disorder.

METHODS Thirty-one men with erectile disorder and low serum levels of testosterone (T) participated in the study. They all sought help in the Sexual Disorders Clinic of the Clinical and Hospital Center "Dr. Dragisa Misovic" in Belgrade over a 3-year period (1990-1993) and were evaluated by one of us (Z.R.). The mean age of these patients was 39.25 years (SD = 7.57; range = 24-56). The majority (64.5%) were educated at the secondary school level. Erectile disorder was diagnosed on the basis of the DSM-III-R criteria (American Psychiatric Association, 1987). An intact capacity for penile erection was verified by the Prostin VR test (Upjohn Co.), so that it was not necessary to verify it additionally by measuring nocturnal erections. Moreover, to exclude other causes of erectile disorder, patients were selected on the basis of normal findings on the following: general physical examination, routine laboratory analyses, laboratory analyses of liver and renal function, assessment of hormonal status (thyroid, pituitary, and adrenal function tests), urological examination, examination of penile blood vessels, and analysis of chromosomes (karyotype). These procedures were performed by clinicians who did not participate in other stages of the study. Information was collected about any medications the patients might have been taking before the onset of erectile disorder and in the course of it. Those patients who were taking medications that are known to cause sexual dysfunction were excluded.

Testoterone Treatment for Erectile Disorder

497

To monitor and assess the patients' sexual activity, only those subjects were included in the study who maintained the same sexual partner throughout the duration of the erectile disorder, including the treatment period. A comparison (control) group consisted of 16 male volunteers, most of them hospital employees. They were subjected to the same evaluation procedures (physical examinations, laboratory analyses and assessments) as patients. Their mean age was 42.37 years (SD = 7.65; range = 31-59). The majority of these individuals (56.25%) also had a secondary education. There were no significant differences between patients and controls in terms of the basic demographic variables. Procedures The serum levels of T and luteinizing hormone (LH) were determined in the morning (at 8 AM) for 3 consecutive days in all patients and control group subjects. Results of these three measurements were expressed as mean levels. The T serum levels were measured by the means of the RIA method, according to a modification with extraction, provided by Biodata (Italy). Serum LH levels were also measured by the RIA method, in accordance with the modification provided by INEP-Zemun (Belgrade). The mean value for intra-assay variability for LH was 7.15 IU/L (SD = 0.36) and the coefficient of variability (CV) = 4.99%. The mean value for interassay variability for LH was 7.45 IU/L (SD = 1.24, CV = 9.86%). For intra-assay variability for testosterone, 3c = 1.160 ng/ml (3.944 nmol/L), SD = 0.031 ng/ml (0.103 nmol/L), and CV = 2.7%. Finally, for interassay variability for testosterone, x = 0.506 ng/ml (1.520 nmol/L), SD = 0.033 ng/ml (0.112 nmol/L), and CV = 6.5%. Normal T levels in the serum ranged from 12 to 35 nmol/L, while normal serum LH levels ranged from 1.05 to 10.5 IU/L. We have used the T/LH ratio to obtain a measure that would take into account variations in the levels of T and LH. Thirty-one patients with low testosterone levels in serum were treated with testosterone propionate in the sustained-release form. The dosage was 100 mg im once a week for 8 weeks. The main indicator of sexual activity was frequency of sexual intercourse, because information on that could be reliably obtained from both the patients and their sexual partners. Likewise, the key parameter of improvement in the treatment of erectile disorder was frequency of successful sexual intercourse (with normal erection), as reported by both partners. Information was also obtained on levels of sexual interest, sexual dreams

498

Rakic, Starcevic, Starcevic, and Marinkovic

and fantasies, and spontaneous and nocturnal erections, but these were considered less objective for the purpose of rating sexual activity. Sexual activity in the course of the treatment was continuously monitored by the second author (V.S.), who did not perform initial patient evaluations. He assessed sexual activity and outcome of an 8-week testosterone treatment using a 3-point scale based on the frequency of successful sexual intercourse, as reported by both partners. The following rating system was used: 0 = No improvement in sexual activity (no successful sexual intercourse, with persistence of erectile disorder); 1 = slight improvement in sexual activity (at least one successful sexual intercourse in the 15-day period preceding the assessment); 2 = obvious improvement in sexual activity (the frequency of successful sexual intercourse returns to that which was usual for the couple before the onset of erectile disorder). Statistical Analyses Student's t test for independent groups and the chi-square test were used for comparisons of independent variables between patients and controls. Pearson's correlation coefficient was used to examine the relationships between all variables within the group of patients and the control group. Stepwise discriminant analysis (Afifi and Clark, 1984) was performed with the aim of identifying variables that are significantly associated with good outcome of the testosterone treatment. The sensitivity analysis (Weinstein and Fineberg, 1990) was used for the purpose of determining the levels of hormones that might distinguish between responders and nonresponders to the testosterone treatment. RESULTS Good outcome of the testosterone treatment (defined as "obvious improvement in sexual activity") after 60 days of treatment was demonstrated in 15 (48.4%) patients. Poor outcome (defined as "no improvement in sexual activity, with persistence of erectile disorder") was found in 16 (51.6%) patients. There were no patients with slight improvement in sexual activity at the end of the treatment. The mean age of patients with good outcome of treatment was 41.00 years (SD = 6.46); the mean age of patients with poor outcome of treatment was 37.63 years (SD = 8.35). Table I compares several variables between the patients and the control group. In comparison with the control group, the mean level of T was significantly lower and the mean level of LH was significantly higher in the

Testoterone Treatment for Erectile Disorder

499

patients. As a result, the mean T/LH ratio was significantly lower in the patients. Patients were significantly younger at the age they married. In the control group, the Pearson correlation coefficient suggested a significant linear relationship between the levels of T and age at which the persons married (r = -.519). In the patients group, the following linear relationships (based on the Pearson correlation coefficients) were significant: Between higher levels of LH and good outcome of treatment (r = .796), between lower values of the T/LH ratio and good outcome of treatment (r = -.629), between gradual onset of erectile disorder and higher levels of LH (r = .455), between gradual onset of erectile disorder and onset of the disorder at a later age (r = .378), and between gradual onset of erectile disorder and good outcome of treatment (r = .571). Using the Pearson correlation coefficient, the same (repeated) relationships between T and LH were not found in both the patient group and the control group. Table II shows the results of the first discriminant analysis. It included all variables except for the T/LH ratio, which is not an independent variable. Variables associated with good outcome of treatment were higher levels of LH, greater number of children, older age (at the time when the hormone levels were determined), and gradual onset of erectile disorder. The correct classification index with such combination of variables was maximal: 100%. Table III shows the results of the second discriminant analysis, which excluded the levels of T and LH, but included the T/LH ratio along with other variables. In descending order of significance, variables associated with good outcome of treatment were lower values of the T/LH ratio, older age (at the time when the hormone levels were determined), lower educational level, younger age at marriage, and younger age at the onset of masturbation. Poor outcome of treatment was associated with shorter duration of erectile disorder and onset of sexual activity at a later age. The correct classification index with this combination of variables was somewhat less informative: 96.15%. Different results of the two discriminant analyses may be accounted for by the relationships between variables (as determined by the Pearson correlation coefficients), differences in the correct classification indexes, and greater prognostic value of the LH level than the T/LH ratio. The sensitivity analysis showed that good outcome of treatment was associated with levels of LH higher than 7.5 IU/L and T/LH ratio equal to or lower than 0.87 nmol/IU. Therefore, it is reasonable to expect good outcome of the testosterone treatment in patients with decreased levels of T if their levels of LH are above 7.5 IU/L and their T/LH ratios are below the corresponding value.

500

Rakic, Starcevic, Starcevic, and Marinkovic

Testoterone Treatment for Erectile Disorder

501

Table II. Significant Variables in Predicting Outcome of Testosterone Treatment— Results of the First Discriminant Analysis Variables (by order of entrance) LH levels No of children Age Mode of onset of erectile disorder (1 = abrupt, 2 = gradual) aLevel

Good outcome of treatment

Wilks's Xa

SCDF coefficientsb

.31 .26 .23

1.057 0.525 0.332

10.8 1.85 41.9

.218

0.266

2

X

Poor outcome of treatment

SD

X

SD

2.57

1.57

6.25

4.68 1.61 37.9

0

1.54

0.5

0.7

0.5 8.2

of significance for all lambdas is < .001. Canonical Discriminant Function.

bStandardized

Table III. Significant Variables in Predicting Outcome of Testosterone Treatment— Results of the Second Discriminant Analysis Variables (by order of entrance) T/LH ratio

Age Level of education Age at marriage Duration of erectile disorder Age at onset of partner sexual activity Age at onset of masturbation

SCDF Wilks's Ja coefficientsb

1.3

Good outcome of treatment

Poor outcome of treatment

x

SD

X

.585 .458 .419 .388 .367

0.69 41.9 2.08 23.54 4.31

0.16

0.669 0.656 0.611 -0.597

0.49 2.57 4.98

1.89 37.9 2.38 25.31 3.76

.33 .3

-0.45 0.437

18.9 14.0

2.39 2.00

19.2 15.0

6.2

SD 1.03

8.2 0.65 4.90 3.00 3.78 1.73

aLevel

of significance for all lambdas is < .001. ''Standardized Canonical Discriminant Function.

DISCUSSION Even though levels of LH were higher in the group of patients as a whole (which was expected because patients had low levels of T), LH values showed prominent variability, ranging from 1.6 IU/L to 18.7 IU/L. In contrast, LH values in the control group were far more consistent, ranging from 4 IU/L to 8.6 IU/L. This indicates that in terms of LH levels, the group of patients was quite heterogeneous. Levels of free testosterone, which is a biologically active form of the hormone (Carani et al., 1990; Davidson et al., 1983; Gooren and Rubens,

502

Rakic, Starcevic, Starcevic, and Marinkovic

1987; Vermeulen et al., 1972), were not determined in our laboratory. Some studies (Carani et al., 1990) suggest that a decrease in the level of total testosterone does not automatically imply a decrease in the level of free testosterone. This might help account for the observation that increased levels of LH were not found in some of our patients with low levels of T, and that these patients did not respond to the testosterone treatment. Discriminant analyses suggest that higher levels of LH and lower values of the T/LH ratio are most significantly associated with good outcome of testosterone treatment, and therefore they might predict good response. The same variables were identified as significant on the basis of the Pearson correlation coefficients. A significant correlation was found between gradual onset of erectile disorder on one hand, and higher LH levels, somewhat older age, and good outcome of testosterone treatment, on the other. This finding suggests that patients with gradual onset of erectile disorder tend to have idiopathic hypogonadism and, therefore, they may be more likely to respond to testosterone treatment. However, our results also show that such a relationship is not the rule. In men over 40 with gradual onset of erectile disorder and good outcome of testosterone treatment, which suggests a hormonal origin of erectile disorder (idiopathic hypogonadism), there may be an unusually early onset of andropause. Such a condition had been described more than 50 years ago in a few men in their early 40s (Werner, 1939). Although some authors disagree that it exists (Skolnick, 1992), it may reflect a genetically determined process of premature aging. Good outcome of testosterone treatment was also associated with greater number of children in the first discriminant analysis. This was probably an artifact, because most patients with good treatment outcome had one or two children; there was only one patient with seven children. Likewise, the second discriminant analysis identified lower educational level and earlier age at which patients married as variables associated with good treatment outcome. It appears that these two variables are related, since persons with less education tend to marry earlier. However, it is difficult to conceptualize a meaningful relationship between lower educational level and erectile disorder with decreased levels of testosterone; therefore, the association of this variable with good outcome of treatment in a relatively small number of patients does not appear convincing. The association of good treatment outcome with an earlier age of onset of masturbation does not appear to be clinically significant. Although an earlier age of onset of masturbation points to a stronger sexual drive (Kinsey et al., 1948), the latter seems unrelated to the development of idiopathic hypogonadism later in life.

Testoterone Treatment for Erectile Disorder

503

Poor outcome of testosterone treatment does not point with certainty to any particular cause of erectile disorder, and its etiology then requires further investigation. In conclusion, the diagnostic workup of all middle-aged men who develop erectile disorder should include measurement of the serum levels of T and LH. The decision to start testosterone treatment should be based on the levels of LH and values of the T/LH ratio, because decreased levels of serum testosterone per so may not be the cause of erectile disorder and do not necessarily predict good outcome of treatment. According to the results from our laboratory, levels of LH above 7.5 IU/L or the values of the T/LH ratio equal to or below 0.87 nmol/IU in patients with erectile disorder and decreased levels of serum testosterone, suggest that the testosterone treatment is more likely to be effective and should therefore be attempted.

REFERENCES Afifi, A. A., and Clark, V. (1984). Computer-Aided Muttivariate Analysis, Lifetime Learning Publications, Belmont, CA. American Psychiatric Association (1987). Diagnostic and Statistical Manual of Mental Disorders, 3rd ed., rev., American Psychiatric Association, Washington, DC. Benson, G. S. (1994). Endocrine factors related to impotence. In Bennett, A. H. (ed.), Impotence. Diagnosis and Management of Erectile Dysfunction, W. B. Saunders, Philadelphia, pp. 31-41. Carani, C., Zini, D., Baldini, A., Delia Casa, L., Ghizzani, A., and Manama, P. (1990). Effects of androgen treatment in impotent men with normal and low levels of free testosterone. Arch Sex. Behav. 19: 223-234. Davidson, J. M., Chen, J. J., Crapo, L., Gray, G. D., Greenleaf, W. J., and Catania, J. A. (1983). Hormonal changes and sexual function in aging men. J. Gin. Endocrinol. Metab. 57. 71-77. Gooren, L., and Rubens, R. (1987). Overview of the concept of andropause. In Zichella, L., Whitehead, M., and Van Keep, P. A. (eds.), The Climacteric and Beyond: The Proceedings of the Fifth International Congress on the Menopause, Parthenon Publishing Group, Sorrento, pp. 85-93. Kinsey, A. C, Pomeroy, W. B., and Martin, C. E. (1948). Sexual Behavior in the Human Male. W. B. Saunders, Philadelphia. Kwan, M., Greenleaf, W. J., Mann, J., Crapo, Z., and Davidson, J. M. (1983). The nature of androgen action on male sexuality: A combined laboratory-self-report study in hypogonadal men. J. Gin. Endocrinol. Metab. 57: 557-562. National Institutes of Health. (1992, Dec. 7-9). Impotence. NIH Consensus Statement. 10(4):l-31. O'Carroll, R., Shapiro, C, and Bancroft, J. (1985). Androgens, behavior and nocturnal erection in hypogonadal men: The effects of varying the replacement dose. Clin. Endocrinol. 23: 527-538. Skolnick, A. A. (1992). Is "male menopause" real or just an excuse? J. Am. Med. Assoc. 268: 2486. Vermeulen, A., Rubens, R., and Verdonck, L. (1972). Testosterone secretion and metabolism in male senescence. J. Clin. Endocrinol. Metab. 34: 730-735.

504

Rakic, Starcevic, Starcevic, and Marinkovic

Weinstein, M. C., and Fineberg, H. V. (1990). Structuring clinical decisions under uncertainty. In Weinstein, M. C., and Fineberg, H. V. (eds.), Clinical Decision Analysis, W. B. Saunders, Philadelphia, pp. 61-62. Werner, A. A. (1939). Male climacteric. J. Am. Med. Assoc. 112: 1441-1443.

Archives of Sexual Behavior, Vol. 26, No. 5, 1997

Hypersexual Desire in Males: An Operational Definition and Clinical Implications for Males with Paraphilias and Paraphilia-Related Disorders Martin P. Kafka, M.D.1,2

The longitudinal history and temporal stability of total sexual outlet (TSO) in a group of outpatient males with paraphilias (PA) and paraphilia-related disorders (PRD) was assessed. Based on extant normative data from contemporary population-based surveys of sexual behavior, it was hypothesized that a persistent TSO of 7 or more orgasms/week for a minimum duration of 6 months be considered as the lower boundary for hypersexual desire in males. In almost all statistical analyses, the PA (n = 65) and PRD (n = 35) groups were not statistically different. The mean current TSO (PA, 7.4 ± 5.7; PRD, 8.0 + 4.2) as well as the current average time consumed in all unconventional sexual behaviors (1-2 hr/day) were not statistically different. Unconventional sexual behaviors (i.e., related to PAs or PRDs) leading to orgasm constituted 77% of current TSO. In the combined group (n = 100), 72% (n = 72) reported a hypersexual TSO of 7 or greater. Age of onset of hypersexual TSO in the PAs (19.2 ± 6.8 years; range 10-43) and the PRDs (21.0 ± 8.6; range 10-46) and the duration of hypersexual TSO (PA, 11.1 ± 11.2 years; PRD, 10.5 ± 9.1) were not significantly different. Fifty-seven males (57%) reported a TSO of 7 or more for a minimum duration of 5 years. Clinical implications of reconceptualizing PAs and PRD as sexual desire disorders are discussed. KEY WORDS: paraphilia; paraphilia-related disorder, hypersexuality; total sexual outlet; psychosexual disorder.

Presented by poster at the International Academy of Sex Research, Provincetown, Massachusetts, September 20-24, 1995; and at a symposium of the Association for the Treatment of Sexual Abusers (ATSA), October 12-15, 1995, in New Orleans, Louisiana. 1Harvard Medical School, Boston, Massachusetts. 2To whom all correspondence should be sent at McLean Hospital, 115 Mill Street, Belmont, Massachusetts 02178.

505 0004-0002/97/1000-0505JlZ50/0 C 1997 Plenum Publishing Corporation

506

Kafka

INTRODUCTION Although the clinical constructs for some increased human appetitive behaviors and psychiatric disorders characterized by these appetitive dysregulations (e.g., eating, sleep, and psychoactive substance-use disorders) are well established in psychiatric nosology, there has been very little empirical data to describe or operationalize a clinical construct for increased appetitive sexual behavior. The term "hypersexuality" has been utilized to describe acute changes in sexual behavior usually induced by an organic agent, for example, in illness (Blumer, 1970; Huws et al., 1991; Jensen, 1989; Van Reeth et al., 1958), brain injury (Epstein, 1973; Miller et al., 1986; Monga et al., 1986; Zencius et al., 1990), or a medication effect (Boffum et al., 1988; Uitti et al., 1989; Vogel and Schiffter, 1983). Hypersexuality has also been a clinical descriptor for sexual impulsivity disorders (SIDs), the paraphilias (PAs) and paraphilia-related disorders (PRDs) (Brotherton, 1974; Cooper, 1981; Davies, 1974; Kaplan, 1995; Orford, 1978). In men with SIDs, however, there are almost no data to characterize the prevalence of an appetitive dimension of sexual impulsivity disorders and, if present, whether hypersexuality represents an acute, episodic, or chronic condition. Paraphilias are socially deviant, repetitive, highly arousing sexual fantasies, urges, and activities enduring at least 6 months and accompanied by clinically significant distress or social impairment (American Psychiatric Association [APA], 1994). The most common paraphilias described in the DSM-IV include exhibitionism, pedophilia, voyeurism, fetishism, transvestic-fetishism, sexual sadism, sexual masochism, and frotteurism. Paraphilias are predominantly male sexuality disorders with an estimated sex differences ratio of 20:1. Kafka (1994a; Kafka and Prentky, 1992a; 1992b) operationally defined PRDs as intensely sexually arousing fantasies, urges, and sexual activities that are culturally sanctioned aspects of normative sexual arousal and behavior. These activities, however, increase in frequency or intensity (for at least 6 months duration) so as to produce significant distress or social impairment, including interference with the capacity for reciprocal affectionate activity. This definition was intended to suggest that the major distinction between paraphilias and paraphilia-related disorders is the cultural boundary distinguishing "normal" from "deviant" social preference behavior (Kafka, 1994a; Marmor, 1971). Commonly reported PRDs include compulsive masturbation (Carnes, 1989; Earle and Crow, 1989; Krafft-Ebbing, 1886/1965; Marmor, 1971), protracted promiscuity (APA, 1987, p. 296; Hirshfeld, 1948; Krafft-Ebbing, 1886/1965; Money, 1986), dependence on pornography (Carnes, 1989; Earle and Crow, 1989; Reinisch, 1990), phone sex dependence (Carnes, 1989; Earle and Crow, 1989- Kafka, 1991; Kafka and Prentky,

Hypersexual Desire in Males

587

1992a), dependence on sexual accessories such as drugs (Boffum et al., 1988) (e.g., amyl nitrate, cocaine) or foreign objects (e.g., dildoes) (Agnew, 1986), and severe sexual desire incompatibility (Friedman, 1977; Hirshfeld, 1948; Kafka and Prentky, 1992a; Krafft-Ebbing, 1886/1965). An operational definition for male hypersexualiry should include dimensional criteria distinguishing the frequency of specific sexual behaviors as well as the time consumed by those behaviors in comparison to a "normal" male sample. Kinsey et al. (1948) reported that in a large sample of males (N = 5300), the total sexual outlet (TSO) could be represented by a continuous frequency distribution curve skewed to the right (the high frequency end of the curve). Kinsey et al. defined TSO as the total number of orgasms achieved by any combination of sexual outlets (e.g., masturbation, sexual intercourse, oral sex) during a designated week. In such a skewed distribution curve, the median best described the "average" individual. Kinsey et al. reported that the median TSO/week of males ages adolescence-30 was 2.14, whereas in the total population, ages adolescence-85, the median TSO was 1.99. Unfortunately, most recent population-weighted sample surveys of sexual behavior have neglected the TSO concept and have focused on the incidence, prevalence, and frequency of specific sexual behaviors primarily believed to be vectors for the transmission of sexually transmitted diseases, especially the human immunodeficiency virus. In these surveys, sexual intercourse, the most common human sexual behavior, often occurs one-three times per week and declines with age and duration of marital status (Bachrach and Horn, 1987; Billy et al., 1993; Hunt 1974; Laumann et al., 1994; Leigh et al., 1993; Seidman and Reider, 1995). It is noteworthy that these surveys have generally neglected to quantify the frequency/week of masturbation, the total sexual outlet, and the total tune spent in all sexual behaviors. Regarding the high frequency end of the distribution curve, Janus and Janus (1993, p. 25) reported that 15% of men between ages 18 and 50 (n = 1,860) reported daily sexual activity (i.e., TSO of 7 or more) and 7.5% of men ages 18-50 (n = 887; p. 31) reported daily masturbation. Other contemporary investigators, however, have reported that a substantially smaller proportion of the male population report a TSO of at least 7/week. Kinsey et al. (1948, p. 197) reported that 7.6% of American males (adolescence-age 30) had a mean TSO of 7 or more for at least 5 years. In that subsample of males, masturbation was the primary sexual outlet. Atwood and Gagnon (1987) reported that 5% of high school and 3% of college age white males (n = 1,077) masturbated on a daily basis, i.e., had a TSO of at least 7 per week. Laumann et al. (1994), in the most recent sexuality survey of American males and females, reported that only 33.6% of the male population (n = 1320; ages 18-59) reported masturbation

508

Kafka

once/week, 14.5% masturbated 2-6 times/week, 1.9% masturbated daily, and an additional 1.2% masturbated more than once/day during the past year (S. Michaels, personal communication, October 18, 1995). This study was designed to examine the temporal stability and longitudinal history of total sexual outlet in a group of outpatient males presenting to this clinician for evaluation/treatment of paraphilias and paraphilia-related disorders. SIDS refers to the combined total of PAs and PRDs. Hypersexual TSO refers to the maximum sustained average TSO for a defined minimum period of 26 weeks (6 months) after the age of 15 years. The duration of months was selected to conform to the duration criterion for paraphilias in DSM-III, DSM-III-R, and DSM-IV (APA, 1980, 1987, 1994, respectively) as well as the duration criterion suggested for paraphilia-related disorders (Kafka, 1994a; Kafka and Prentky, 1992a, 1992b). In this report, hypersexual desire refers to males who reported a hypersexual TSO of >7. The following questions are addressed in this report, (i) Do men with PAs and PRDs self-report extended periods of persistent TSO of at least seven or more sexual behaviors leading to orgasm per week? (ii) Do PA and PRD groups differ in the prevalence or frequency of high frequency sexual behavior? (iii) Do sex offender paraphiliacs differ in their lifetime pattern of TSO in comparison with nonoffender paraphiliacs or men with PRDs? (iv) Can an operational definition for hypersexual desire in men be described utilizing the dimensions of weekly TSO as well as mean duration of a maximum average TSO?

METHODS Demographic Data The subjects were 100 consecutively evaluated outpatient males seeking treatment with this investigator for paraphilias (n = 65) or paraphiliarelated disorders (n = 35). Data were obtained during three initial evaluation visits utilizing an Intake Questionnaire and Sexual Inventories previously reported (Kafka, 1991,1994a; Kafka and Prentky, 1992a, 1992b, 1994). All demographic and sexual data were individually reviewed with the study subjects to determine PA/PRD status. Study subjects were obtained through advertisement, or self-, therapist-, or forensic referral. Men included in this investigation had to report repetitive PA or PRD behaviors in the 6 months preceding the evaluation.

Hypersexual Desire in Males

S89

Sexual Diagnoses Paraphilias were diagnosed utilizing DSM-III-R criteria. Rape of an adult, however, was diagnosed as a parapbiiia based on the reported presence of that behavior regardless of its persistence. Paraphilia-related disorders were diagnosed using Kafka's criteria (Kafka, 1991,1994; Kafka and Prentky, 1992a). Although men with PAs could report lifetime PRDs, men with PRDs could not report any lifetime PAs to be included in the PRO group. This was done to insure that PA men were "socially deviant" with regard to sexual preference when compared to "nonparaphilic" PRD subjects. All males were rated on the Kinsey 7-point scale for sexual orientation (Kinsey et al., 1948) to determine heterosexual (Kinsey scores 0, 1), bisexual (Kinsey scores 2, 3, 4), or homosexual orientation (Kinsey scores 5,6). Exclusionary Criteria

Men receiving any psychotropic medications or reporting medical or neurological conditions that could affect current sexual behavior or self-report were excluded, as were those men whose sexual behavior occurred only during periods of general psychoactive substance abuse. None of the PA/PRD males were involved in concurrent behavioral therapies that would have artificially increased TSO, e.g., masturbatory satiation therapy (Marshall, 1979), during the baseline week or prior 6 months. Measurement of Current Sexual Behavior

The Sexual Outlet Inventory (SOI; Kafka, 1991, 1994a; Kafka and Prentky, 1992a, 1992b, 1994) is a clinician-administered rating scale that documents the incidence and frequency of sexual fantasies, urges, and activities of sexually impulsive males and females during a current or baseline week. In the SOI, sexual behaviors are divided into two broad subcategories, conventional and unconventional. Conventional sexual behaviors are intended to reflect the culturally normative concept of nonparaphilic "reciprocal affectionate sexual activity" (APA, 1980, 1987). Conventional sexual behavior does not require the presence of a stable affiliation, but merely that the intent of a sexual behavior contained a mutually consenting, relational context. Unconventional sexual behaviors generally corresponded to PA and PRD activities, including the primary use of PA/PRD fantasy dur-

510

Kafka

ing sexual activities, either alone or with a partner, All study participants defined their own sexual behaviors according to the dimensional guidelines proposed by the aforementioned definitions. In this study data derived from the SOI were used to code the following variables during the course of a specific baseline week: (i) the ISO of both conventional and unconventional sexual behaviors, (ii) a measure of the average period of time per day/week spent in all unconventional sexual activities, e.g., fantasies, urges, and genital activities. Sexual fantasy was defined as thoughts or visual imagery with sexual content or intention. For example, time spent fantasizing about an ongoing or wished for sexual/romantic relationship was rated as conventional sexual fantasy while time spent in fantasizing about problematic exhibitionism or promiscuity was rated as unconventional sexual fantasy. Sexual urge corresponded to an action with sexual intention but without genital activity because of selfcontrol or situational constraints. For example, unsuccessful pursuit of a partner for sexual activity could be either a conventional urge (e.g., the objective involved a romantic alliance) or an unconventional urge (e.g., a repetitive pattern of seeking a prostitute for a promiscuous liaison, cruising for an opportunity for exhibitionism). Sexual activity was defined as direct genital stimulation including but not limited to orgasm. Sexual activity was subdivided into two categories: masturbation utilizing internally generated fantasy, and "other" sexual activities. The distinction between internally and externally generated sexual imagery derives from my clinical experience with the treatment of these disorders as well as research that supports this distinction (Everitt and Bancroft, 1991; Jones and Barlow, 1990). Masturbation with conventional internally generated fantasy could include sexual fantasies of past, current, or anticipated partners, whereas unconventional masturbation involved the primary use of internally generated PA/PRD-related fantasies during masturbation. The "other" conventional sexual activities category included sexual intercourse, oral sex, anal sex, or mutual masturbation in a reciprocal sexual relationship as representative outlets. Masturbation with an externally assisted fantasy or activity (e.g., phone sex, pornography), mutual masturbation, oral sex, anal sex, or sexual intercourse as part of a PA/PRD ritual were exemplary "other" unconventional sexual behaviors. For example, sexual masochism, a PA accompanied by bondage, spanking, or humiliation, was self-rated as an unconventional sexual behavior, whereas masturbation accompanied by these masochistic fantasies but without the explicit behavioral enactment of the fantasy was rated as unconventional masturbation. The average time per day spent in current unconventional sexual fantasies, urges, and activities was measured in intervals designated in minutes

Hypersexual Desire in Males

511

as 0-1, 1-5, 5-15, 15-30, 30-60, 60-120, 120-240, 240-480, and 480-960

min. Determination of Temporal Patterns of Total Sexual Outlet Several questions in the Sexual Inventories specifically inquired about temporal patterns of TSO. In the 6 months immediately preceding the diagnostic evaluation, the maximum TSO/day and the average TSO/week and per month were determined. The maximum TSO/day in the past 6 months was assessed on a scale from 0 to >4, which was rated as 5. The lifetime maximum TSO/week and TSO/month since the age of 15 was determined using a scale of overlapping ranges: 0-5, 5-10, 10-15, etc. The lifetime highest TSO/week scale was scored up to 25-30 and >30. For statistical purposes, men were scored by the midpoint in their range scale. Those men reporting a week with more than 30 orgasms were rated as 32.5, considered the midpoint of the last defined range. The lifetime maximum TSO/month scale also utilized overlapping ranges but these went from 0-5 to 40-45. Any subject rating >45 was rated as 47.5. Determination of a "hypersexual TSO," i.e., the highest TSO that lasted at least 26 consecutive weeks after the age of 15 years, was made on a scale ranging from 0 to >7. The uppermost scores of the scales maximum TSO/day in the past 6 months, lifetime (since age 15) maximum TSO/week, lifetime maximum TSO/month, and hypersexual TSO lasting 26 weeks were truncated to improve self-report accuracy since it seemed reasonable to infer that subjects would be less accurate at recalling the specific number of orgasms at their highest extremes. A TSO of >7 was not necessarily the equivalent of daily sex leading to orgasm.

Statistics Means are accompanied by standard deviations. Categorical variables were examined by the chi-square statistic. Continuous variables were compared utilizing unpaired, two-tailed Student t tests and analysis of variance (ANOVA). The Scheffe F test for multiple comparisons was utilized to distinguish significant differences between subgroups in ANOVA. Pearson correlation coefficients are included when indicated. Alpha values of 10/week). In the combined sample, 75% (75/100) of the men reported 1 month (i.e., 4 weeks), with an average TSO of 7 or more/week and 38% (38/100) reported a TSO of at least 10/month after the age of 15. Hypersexual TSO: Onset and Duration The frequency distribution of hypersexual TSO is reported in Table III. The modal hypersexual TSO for both groups was 7 or more. In the combined group, 86% of the men reported a cumulative prevalence of hypersexual TSO of 5 or more for at least 26 weeks duration, 78% of men reported a hypersexual TSO of 6 or greater and 72% reported a hypersexual TSO of 7 or greater. The age of onset of hypersexual TSOs in the PAs (19.2 ± 6.8 years; Mdn = 17; range = 10-43) and the PRDs (21.6 ± 9.3; Mdn = 19; range = 10-46) and the duration of hypersexual TSO (PA, 11.2 ± 11.1 years, Mdn = 9; PRO, 10.6 ± 9.,; Mdn = 10 years) were not statistically different. Fifty-seven males (57%) reported a TSO of 7 or more for a minimum duration of 5 years.

516

Kafka Table III. Hypersexual Lifetime TSO of At Least Six Months Duration in Males with Paraphilias (PA) and Paraphilia-Rdated Disorders (PRD) No. hypersexual TSO

2 3 4 5 6 7 or more

PA group (« = 65)

PRD grouo (n = 35)

n

%

n

%

1 3 7 6 3

1.5 4.6 10.7

2.8 5.7 0.0 5.7 8.5

45

69.2

1 2 0 2 3 27

9.2 4.6

77.1

ANOVA was utilized to examine whether there was a statistical relationship between the lifetime total SIDs and parameters of hypersexuality, including maximum TSO in a single week and month since age 15, hypersexual TSO, and age of onset and duration of hypersexual TSO. To further delineate any differences between the groups for these variables, the lifetime total SIDS category was divided into three subcategories: low, 1-2 lifetime SIDs, n = 33, PA = 17 (26.1%), PRD = 16 (45%); medium, 3-4 lifetime SIDs, n = 43, PA = 26 (40%), PRD = 17 (48%); and high, 5-8 lifetime SIDs, n = 24, PA = 22 (33.8%), PRD = 2 (5.7%). The high group was predominantly paraphiliacs and the predominant subgroup were forensically referred paraphiliacs (10/24; 41.6%). These subcategories were not statistically different in mean current age, F(2, 97) = 0.98, p = 0.37, or referral source, %2(6) = 5.67, p = 0.46. ANOVA comparing lifetime SIDs with maximum TSO in a single week revealed no statistical difference between the groups, F(2,97) = 1.08, p = 0.34. ANOVA comparing lifetime total SIDs and maximum TSO/month revealed a statistical difference between the low vs. the high group, F(2, 97) = 3.28, p = 0.04; Scheffe F = 3.12, p < 0.05. In addition, there was a statistically significant difference between the groups in mean hypersexual TSO, F(2,97) = 7.55, p = 0.0009, with the low group reporting a lower mean hypersexual TSO (5.6 + 1.7) in comparison with both the medium group, 6.3 ± 1.0 (Scheffe F = 3.39, p =