No title

Archives of Sexual Behavior, Vol. 27, No. 6, 1998

G uest Essay

Boys an d Sexual Abuse: An En glish Opin ion D. J. West, M.D., Litt.D., FRC Psych .1

INTRODUCTION The upsurge of public anxiety about thre ats to childre n from se xual pre dators began with concern for the safety of girls. Influe nced by feminist protest against the perceived tendency of males to dominate and exploit females, research on child sexual abuse at first concentrate d on victimization of girls by fathers, stepfathers, or other males in the family circle. Research on the sexual abuse of boys developed later (Hunter, 1990). Awareness that women also may offend with minors, especially boys, came later still (Elliott, 1993). In view of the popular stereotype of priestly pederasts chasing after choir boys and the many newspaper reports of scoutmasters and the like “interfering ” with boys unde r their care, the initial lack of professional interest seems odd, but probably reflects a tradition that boys should be able to look after themselves. The experience of child care profe ssionals is ge nerally limite d to incide nts serious enough to lead to a complaint to police or social services. The impression gained from such cases is that any kind of sexual incident involving a child is likely to cause great and lasting harm (Wyre and Tate, 1995). The emotive terms adopte d in professional discourse — abuse, perpetrator, victim, survivor — have reinforced this idea and introduced a tone of moral revulsion alien to scientific inquiry. The highlighting by the media of horrendous cases of sexually motivated abductions and killings of children has spread the notion that all sexual interest in children is like ly to be violent and life-thre atening. The result has been an unprecedented public outcry against pedophile s and the introduction of extra penal measures. In Belgium, following the discovery in 1996 of the bodie s of girls who had been kidnapped, imprisone d and left to die by the murderous pedophile Dutroux, a quarte r of a million protesters marched through the capital demanding reform of police and judicial practice. 1

Institute of Criminology, University of Cambridge, 7 We st Road, Cambridge, CB3 9DT, United Kingdom. 539 0004-0002/98/1200-053 9$15.00/0

Ó

1998 Plenum Publishing Corporation

540

West

In the Unite d Kingdom, following similar le gislative move s in the Unite d State s, the Se x O ffe nde rs Act 1997 re quire s eve ryone , male or fe male , who is convicte d or cautione d for one of a sche dule of se xual offe nse s, including any involving a minor, to re giste r with the police and re port whe re the y are living. Police have authority to re ve al the whe re abouts of re giste re d offe nde rs to school he ads, pote ntial employe rs, and othe rs in the community, including, whe n conside red ne cessary, immediate ne ighbors. No othe r type s of criminal, not e ve n drug de ale rs or armed robbe rs, have bee n thought to require such measures. The provisions have to be enforced without discrimination and apply to offenders as young as 10 ye ars. The G uard ian (25 O ct. 1997) re porte d that a boy of 14 had be e n in clud e d following a convic tion for m isconduct with two e ve n younge r boys. The Se x O ffe nde rs Act 1993 lowe re d to 10 ye ars the age at which a boy can be charge d with rape . Had the four boys age s 10 and 11 trie d at the O ld Baile y for rape and the n for inde cent assault be en found guilty, the y too would have be e n place d on the re giste r (Guardian, 17 Fe b. 1998) . In January 1998, whe n se ve n men we re convicte d for group se x activitie s with e ach othe r (which is ille gal for male homose xuals) one of the m was found to have be e n 6 months unde r the age of 18 at the time . Although he prote ste d he was a willing participan t, not a victim, those who had had contact with him we re place d on the re giste r on grounds of pe dophilia (G uard ian , 23 Jan. 1998) . Young girls can also be offe nde rs. O n 26th O ct 1997 the Observer reporte d that two girls age s 14 and 11 had be e n caution e d for repe ate d sexual assaults on a boy of 6 for whom the y were baby sitting. The boy ’s pare nts, e nrage d that the girls had e scape d a trial, were thre atening a private prose cution. At le ast one acade mic has argue d that pursuing charge s against childre n for se xual assault s on othe r childre n is ofte n ine ffe ctual and dam aging to the childre n involve d and to the ir familie s ( Soothil l, 1997) . The Home O ffice has e stimate d that if the 1997 Act had applie d re trospe ctive ly some 125,000 me n in the community would have had to be registe re d as se x offe nde rs ( Marshall, 1997) . The y will not be entire ly immune , howe ve r, for the Crime and Disorde r Act, 1998, which is now in force , allows for anyone who has at any time be en cautione d or convicte d for a se xual offe nse whose be havior attracts suspicion or disapproval to be made subje ct to an orde r re stricting the ir move ments and activitie s for a minim um of 5 ye ars. The orde rs are to be instigate d by police or local authoriti e s “ for the purpose of prote cting the public ” and include the same re quire me nts as for those on the se x offe nde r re giste r.

Boys an d Sexu al Abu se

541

Given the climate of opinion and the e ase with which newspape rs le arn when a pedophile is released from prison, it is unsurprising to he ar of pedophile s and the ir familie s injure d by vigilante s, hounde d out of their homes by ne ighbors, or drive n from town to town by prote sters. The gove rnment has been oblige d to issue guide line s to discourage housing authoritie s from routine ly refusing to accommodate se x offende rs. Coinciding with this media-fe d moral panic, attention has exte nded to the sexual abuse of boys. In Belgium, the Dutroux scandal was swiftly followed by press alle gations of pedophile activity with boys made against the Deputy Prime Ministe r and othe r politicians who were known to be homosexuals (Reekie, 1997) . Be lief in the dire conseque nces of any kind of sexual involve ment with olde r persons now attache s to childre n of both sexes. Inde ed, male victims have fe ature d promine ntly in a spate of recent le gal case s in which adults have claime d compe nsation for lasting psychological damage alle ge dly caused by sexual abuse expe rience d many years previously, usually when at a residential school or childre n’s home or at the hands of now aging pare nts. Churche s have had to pay out large sums to settle claims against the ir priests by me n alleging the y were sexually abuse d as boys (Jenkins, 1996). It is putting re putation at risk to sugge st that the catch-all terms “ sex abuse ” and “ pedophile crime ” are being used for relative ly trivial as well as ve ry serious offenses. Neve rtheless, the re has always been tension between the findings of retrospe ctive surveys of adult populations, which sugge st a casual se xual encounte r with an olde r person during childhood is too common an occurre nce to be routine ly and seriously damaging and, in contrast, the experience of clinicians that incide nts of child sexual abuse provoke posttraumatic stress disorde r (PTSD), adult sexual maladjustment, and psychiatric illne ss. In the case of boys, the conflict of evide nce is particularly acute. The majority of pe rpetrators being male , clinicians obse rve that boys experience particular distress through the arousal of anxie ty about the ir sexual orie ntation, but men from nonclinical sample s, when questioned about their sexual past, often recall homose xual approache s whe n they were young which the y dismiss as inconse que ntial. Claim s that inte rge ne rational se xual contact, provide d it is conse nsual, ge ntle , and loving, is not harm ful, is supporte d by a certain amount of research (Constantine and Martinson, 1981; Geraci, 1997) . Protagonists of this view — often refe rred to as the pedophile lobby — have from time to time forme d organizatio ns, such as the North American Man/Boy Love Association ( NAMBLA), to prom ote tole rance of conse nsual sex be low the le gally pe rmitte d age . The se organizatio ns, from whate ve r country the y originate , se e m more conce rne d with men “ loving ” boys than with the commone r phe nome non of me n attracte d to small girls (O ’Carroll,

542

West

1980; Bronge rsma, 1986, p. 105) . The classical Greek tradition, in which young male s we re he ld to be ne fit from having an olde r male love r and mentor, is cite d re pe ate dly (Dove r, 1978) . Although he te rose xual pe dophile s, whe n in a situation to e xpre ss the ir ide as fre ely (Li et al., 1993) , put forward similar argume nts to justify the ir actions, the y have be en le ss incline d to propagandiz e . Acade mic re ports favoring the conclusion that some conse nsual pe dophile re lationshi ps are re lative ly innocuou s also te nd to con ce ntrate on inte rm ale affairs ( B aurm ann, 1983, cite d in Bronge rsma, 1990, p. 17; Be rnard, 1985) . The aim of this pape r is to re view the nature and significance of sexual incide nts be twe en boys and olde r pe rsons.

PREVALENCE ESTIMATES FOR MALES AND FEMALES Q ue stioning adults about their recolle ctions of se xual incide nts with olde r pe ople during the ir childhoo d has produce d wildly diffe ring statistical e stimate s ( Pe te rs et al., 1986) . Much de pe nds on what is meant by childhoo d and by the kind of incide nts counte d. Some authoriti es take childhoo d to me an be fore pube rty, or be fore 14; othe rs re gard as childre n anyone be low the le gal age for conse nsual se x, which can be anything up to 16 or 18 ye ars. The ve ry diffe re nt meaning of a sexual expe rience to a child of 7 and a lusty adole scent be come s obscure d. Some surve ys ask about actual ge nital contact or touching with cle arly sexual inte nt, othe rs e xte nd the que stions to include noncont act expe rie nce s such as e xposure to pornogra phy, adult se xual e xhibitionism , seductive posturin g, or ve rbal inde cency. E stimate s of the pre vale nce of child sexual abuse historie s are usually halve d if nonconta ct case s are exclude d (Nash and We st, 1985) . A minority of inquirie s limit the ir que stions to se xual be havior that was unwante d, assaultat ive , or pe rce ive d as ove rintrusive ; othe rs include be havior that was not fe lt abusive at the time. Most surve ys spe cify a minim um age gap be twe e n the child and the olde r pe rpe trator, but this usually allows for misbe havior of olde r juve nile s with childre n younge r than the mse lve s to be include d. Some surve ys are so ove rinclusive as to sugge st that it is a minority who are fre e from abuse ( Russe ll, 1984, p. 185) . There are othe r proble ms with re trospe ctive research. What pe ople are prepare d to re ve al is affected by how they are approache d. A confidential exchange with an unde rstanding inte rvie wer can yield more than a postal que stionnaire or an impe rsonal inte rvie w conducte d on market research line s. What pe ople can remember de pends on the passage of time

Boys an d Sexu al Abu se

543

and whe the r the y attach significance to an event. Findings from small samples that are conve nie nt for rese arch, such as psychology classe s, may be untypical and misle ading. In large -scale surve ys that aim to recruit a nationally representative sample , substantial numbe rs of subje cts may fail to respond to a que stionnaire or decline to be inte rviewed. Those who do respond may include a misleadingly high proportion of individuals who reply because they have something the y want to discuss. In a meta-analysis of North American surve ys, Gore y and Le slie (1997) estimated that about half the variability in prevale nce e stimates between surve ys was due to differe nces in ope rational definitions of abuse and diffe rences in response rate s. The y conclude d that an appare nt increase in child sexual abuse found in more re cent surve ys was like ly due to a decrease in response rate s rathe r than a real change in behavior. After excluding noncontact incide nts and adjusting for the effects of response rates and varying operational definitions the y estimate d that the aggre gate prevale nce of histories of childhood sexual abuse was 16.8% among female s and 7.9% among males. This is some what highe r than was found in a nationally re presentative British sample que stione d by the M.O .R.I. (Marke t O pinion Research International) organization, when 12% of females and 8% of males acknowledged having had a sexual e ncounte r with a mature individual when the y themselves were unde r 16 years (Bake r and Duncan, 1985) . Nearly all surve ys show girls more like ly to be affe cted than boys (Laum ann et al., 1994) . For e xample , among an English cohort of men born in 1953, by the time the y were 40, 0.7% had acquire d a conviction for a se xual offe nse involving some one unde r 16, usually a girl (0.6 vs. 0.1% ) ( Marshall, 1997) . Among 15-ye ar-old pupils in Finland, de pe nding upon the de finition of abuse , 6-8% of girls, but only 1-3% of boys, re porte d having had such e xpe rie nce (Sariola and Uute la, 1994) . In a surve y of an American stude nt population , We llman ( 1993) found the pre vale nce of e arly se xual abuse among male s and fe male s to be in a ratio of about 2:1. In an e xte nsive Canadian national surve y conducte d by the Gallup Poll organizatio n (Badgle y, 1984, pp. 175-193) , 23.5% of women reported having expe rienced unwante d sexual touching, nearly half (47.5% ) of whom recalle d the first such incide nt as having happe ned whe n the y we re unde r 16. In contrast, only 12.8% of male s re porte d having expe rie nce d unwante d se xual touching and only 18.6% of the se men said it had happe ne d to the m whe n the y we re unde r 16. In a British surve y of male s, using a sample from the e le ctoral re giste r, 20% of men re calle d some se xual e ncounte r with an olde r pe rson whe n the y were unde r 16 (West and Woodhouse , 1990, p. 99) . This compare s with 42% of English women re porting having had some such e xpe rie nce while unde r 16 ( Nash and We st, 1985) .

544

West

Men are said to be less confiding than women, so the prevalence figures for boys could be underestimates. Heterosexual men who think that to be targeted by a homosexual means being perceived as effeminate may not like to admit that it happe ned to them. Adolesce nt boys who have contacts with older females may not view this as abusive or mention it when questioned in later years, unless spe cifically asked about it. Among one sample of American college students (Fritz et al., 1981), the men were not very far behind the women in reporting childhood incidents, 7.7% of the women and 4.8% of the men saying they had been sexually molested by older persons when they were preadolescents. It is sugge stive of unusually comprehensive reporting that, in this survey, a majority of the molestations of males (60% ) were by females. Although incidents with older females are quite common, contacts between boys and older males are more frequent than might be expected given the relatively small minority of homosexual males in the community. Yet research suggests that the generality of homosexual men are, if anything, less likely to have pedophile interests than heterosexuals (Freund, 1981, p. 162; Howitt, 1995, pp. 44-49; Newton, 1978). Contrary to popular belief, mature male homosexuals oriented towards adult sex seem less likely to “regress” in later years to child molestation than do heterosexual males (Groth and Birnbaum, 1978). Some of the pedophiles who molest prepubertal children are essentially attracted by smooth, hairless undeveloped bodies and target boys and girls somewhat indiscriminately. Homosexually oriented seekers after adolescent boys find many who are amenable and often achieve innumerable contacts. A few such men can swell the numbers of male “victims.” Because the prevalence figure s yielded by retrospective surveys are so much a function of how abuse and childhood are defined, the precise statistics are of less interest than the fact that even the most conse rvative estimates show that sexual encounters with adults feature in the live s of a great many children and young persons (Li et al., 1993, p. 148). More females than males report experiences and describe more ongoing experiences at younger ages, but the number of boys involved is still considerable, especially when contacts with women are taken fully into account. FEMININE DISTRESS The retrospective accounts of abuse d men and women differ. Women more often describe distress at the time as well as long-te rm adverse effects. They tend to report even trivial-see ming incide nts with conside rable negative emotion. For example , Wellman (1993) found that male students took minor incide nts less seriously than women students who, even when they had had no such experiences themselves, were convinced that early sexual encounters must be harmful. Meta-analysis of publishe d surveys of nonclinical sample s (Bauserman and Rind, 1997; Rind and Tromovitch, 1997) have shown that, among

Boys an d Sexu al Abu se

545

nonclinical female samples, there is usually a substantial majority who report negative short-term reactions, whereas a majority in most male samples report positive or neutral short-term reactions. Complaints of long-te rm deleterious effects are also made much more often by females. Girls have different kinds of experiences from boys as well as perceiving sexual incide nts in a different light. Incidents with girls tend to start early and are more often incestuous (Baker and Duncan, 1985). If the severity of sexual abuse of girls is to be judge d by the degree of violence employed, or whether sexual penetration occurs, then the majority of incide nts are at the less severe end of the spectrum. Since its findings have proved broadly consistent with more modern research, it is worth looking back on the classic survey of American women by Kinsey et al. (1953, pp. 116-122). They were questione d about sexual encounters when they were unde r 14 with males who were at least 5 years older than themselves and not less than 15 years of age. Of the 4441 in the sample , 24% recalle d at least one encounte r. Of these women, 80% had experienced only one incide nt. Verbal or exhibitionistic approaches without physical contact accounte d for 62% of the reported incide nts. Only 3% involved coitus and only one of the 4441 women reported a serious injury incurre d as a result of sexual assault. Nevertheless, 80% had been emotionally upset or frightene d by their experiences. The authors concluded with what would nowadays be an extremely politically incorre ct comment: “It is difficult to understand why a child, except for its cultural conditioning, should be disturbed at having its genitalia touched ” (p. 121). In the British sample of women studied by Nash and West (1985) about a fifth reported sexual encounters involving physical contact with an adult at least 5 years older than themselves when they were under 16. Most consisted of bodily caressing and ge nital fondling; only 2% of respondents reported sexual intercourse. However, in spite of most intrusions being limite d, the great majority of women reported having reacted at the time with fe ar, confusion, ange r, or shame; neutral reactions of curiosity or amusement were each mentioned by only 14% of the abuse d women. This is a typical result. In the sexological literature there is comparatively little discussion of the possibility that girls ’ sexual encounte rs with adults can be other than harmful (Okami, 1991) or that many girls below the le gal age of consent are nevertheless sexually mature and active. Some years ago I participated in a television discussion on the legal age of consent led by Kilroy Silk, a former politician specializing in penal affairs. Included in the studio audience were some nubile girls, close to 16 years of age, who complaine d bitterly that their boyfrie nds had been arrested for having sex with them. It can be argued, of course, that the exploitation of young girls ’ readine ss for sex is an abuse because , whe n they are older, they may come to regret their earlie r behavior.

546

West

A small minority of adult women, however, do conside r as unproblematic childhood e xperiences which the majority would regard with horror. Nelson (1981), using a sample re cruited by adve rtisement, found a surprising numbe r of wome n with teenage ince stuous e xperiences which the y regarde d as positive . The publication of a biography that reveale d the artist Eric Gill ’s erotic inte rest in pube scent girls, including his own daughte rs, prompte d one of the m to comment to a newspape r re porte r: “ I don ’t think it harmed me at all. . . . We were all very fond of my fathe r. . . . We were old enough to say if we didn ’t want to go along with him ” (Bille n, 1992) .

MASCULINE INDIFFERENCE A majority of the childhood sexual e xpe rie nces re calle d by nonclinical sample s of adult men are, like those re calle d by women, of the le ss seve re varie ty, that is nonviole nt, nonpe ne trative , and ofte n limite d to propositioning without actual physical contact. Far fewer men than wome n asse rt that such expe rience s have had any significant e ffect. Childre n’s reactions are influe nced by adult attitude s to sex. Men are said to put gre ater value on physical pleasure , wome n to be more conce rned with relationships (Baldwin and Baldwin, 1997) . Male s tend to be more pe rmissive towards teenage se xual activity and to expre ss less guilt and anxie ty about sexual expe rie nce. Wome n are traditionally more conservative in sexual habits and sexual politics, more concerned about pornography and the prote ction of childre n from pre mature sexual relationships and he nce to be more like ly, as they grow up, to develop anxie ty about any premature expe riences they may have had themselves. Although gende r differences may be de creasing in modern Western culture s, it is still true that boys tend to be sexually less inhibite d and more adve nturous than girls. The promine nce of male genitalia and male e rections encourage s masturbation from an e arly age . Compe titive se xual displays, se xual horse play, and expe riments in mutual masturbation are more common between boys, but the se do not generally signify romantic crushes or presage a homose xual orientation. Sexual initiative s from adult male s are perceived as having more serious inte nt and are apt to provoke ange r at the time , but not usually the sustaine d anxie ty than occurs among girls. Vaginal penetration, the physical imprint of lost virginity and the risk of pregnancy are consequence s peculiar to the female. Pubertal boys can experience the pleasure of heterosexual intercourse without incurring these complications. Even the risk of acquiring HIV from intercourse is le ss for the heterosexual male. Far from feeling victimized by having been seduced into vaginal intercourse with an adult, boys may feel a sense of accomplishment.

Boys an d Sexu al Abu se

547

Girls who have been seduced, even if they were compliant at the time, when they later acquire conve ntional attitude s, may look back on the event with shame and guilt or with ange r against the perpetrator. It has been suggested that boys’ attitudes grow in the opposite direction as the y absorb a macho image and want to portray themselves as ever eager for and in control of sexual situations. Among the male student sample that was included in the West and Woodhouse (1990) survey, 45 recalled a boyhood encounter with a man, compared with 34 who had had an experience with a woman. The heterosexual encounters mostly occurred when the boys were pube rtal or postpubertal and the responde nts generally remembered them as enjoyable. One responde nt commented on an episode of mutual masturbation with a woman of 30 when he was 15: “It made me more confide nt about sex in general.” That boys are often aware of the interest in them displayed by homosexual pe dophile s, without be ing gre atly conce rned about it, was brought home to me by a legal case against a master at a boys ’ boarding school who was discovere d to have behave d inde cently with a pupil. Subse quent police inquiries revealed that many of the boys had known of and talke d among themselves about the teacher’s peculiaritie s. Some had actually been grope d by him on occasion, but until the affair became a public scandal they had looke d upon the behavior as a joke and had not thought to report it. Se lf-asse rtion, both ve rbal and physical, fe ature s in the culture of masculinity and in the upbringing of boys. Boys may the re fore be be tte r equippe d to re je ct unwante d se xual invitations. Wellman ( 1993) note d that roughly e qual num be rs of me n and wome n re porte d having expe rience d as childre n se xual approache s that did not le ad to actual contact, but wome n de scribe d more incide nts of physical inte rfe rence . This supports the notion that boys are le ss submissive and be tte r able to avoid escalation into unwante d intimacy. From a surprisingly e arly age boys se em able to recognize when there is a sexual meaning to tentative approache s from adult men and, in most cases, to make clear they do not welcome such attentions. In the British survey by West and Woodhouse (1993) most of the incide nts recalle d were approache s, propositioning or touching by men encounte red in public place s or during routine activitie s by such people as teachers and sports supervisors. The usual reaction was to rebuff the approach and avoid furthe r intimacy. Such incide nts were recalle d as having at the time arouse d curiosity, mild annoyance , or une asine ss and having bee n talke d about among friends of their own age , but rarely re vealed to pare nts or othe r adults. The reaction of a boy of 11, oblige d to share a bed with a frie nd ’s fathe r, who fondle d him during the night, was typical: “ I mean it was a nuisance that I was being kept awake . O ther than that it was slightly embarrassing ” (p. 53) .

548

West

Only a minority of this sample admitte d having been compliant. One responde nt, a heterosexual married man, re calle d a pe riod at age 13 when he was one of a group of boys who were take n on outings in a van by a man who e ngage d the m in se cretive masturbatory games, which they enjoye d in the spirit of share d sexual expe rimentation (p. 50) . Eventually, pare ntal suspicions brought the outings to an end. The nature of the situation and the characte ristics of the initiator can greatly influe nce a boy’s reaction. Another responde nt (p. 71) described how, when a boy of 13, he had enjoye d an occasion when a young man approache d him and a companion while they were bathing in a river and induce d him to participate in mutual masturbation. However, a year late r, when an old man seated close to him on a bus finge red his thigh he thought the be havior disgusting and the expe rience horrible . Psychoanalysts and others who have recorded the reminisce nces of men who identify themselves as exclusive ly homose xual find that most of them remember erotic interests and fantasie s, directed towards their own sex, having developed before pube rty and be fore any ove rt contacts had be en experienced. Some homosexually inclined boys strive to avoid early contacts through awareness of the taboo or from fear of appearing unmasculine (Friedman, 1988, p. 195), but others say they not only enjoyed boyhood encounters with olde r males but delibe rately encourage d or initiated them. This was certainly true among a sample of male homose xuals interviewed as part of the West¯Woodhouse survey. Potentially traumatic anal penetration, although more frequently experienced by boys, is not the most common pedophile sexual activity. Erickson et al. (1988) compare d what was done with boys and girls under 14 according to de scriptions supplie d by known child mole sters, men who might be expected to be at the more severe end of the scale of severity of offending. Fondling was much the most frequent activity with both boys and girls, followed by actual or attempted vaginal intercourse in the case of girls. With boys, fellatio was more common than invasive anal sex. Anal penetration can cause severe pain and result in tears and bleeding when forcefully performed on an inexperienced subje ct without preliminary lubrication and gradual dilatation. The physical sign of chronic anal dilatation may result from repeated, brutal, coercive penetration, but often it indicates that probably there has been a longstanding compliant relationship with the boy. The risk of potentially lethal HIV infection from unprote cted anal sex, impose d by taking advantage of a boy’s compliance and possible ignorance , is a particularly wicked act. Given the massive publicity about pedophilia, children today are alerted to the possibility of adults taking a sexual inte re st in the m. Inde ed, te achers’ unions have bee n concerned at the incre asing numbe r of unjustifie d and damaging alle gations of se xual improprie ty made against male te ache rs by

Boys an d Sexu al Abu se

549

disaffe cted pupils. In many se xual incide nts with olde r persons outside the home boys are effe ctive ly masters of the situation. The y can and do avoid contacts the y do not want. The ir pote ntial for reporting a pedophile assault is a powe rful means of control. O n occasion, boys collaborate in nonthre atening homose xual situations out of curiosity, wish to please , or genuine erotic inte rest. The risk-taking sexual aggre ssor, who use s thre at or force to se cure a boy’s compliance , is a re lative rarity. When the wide varie ty of sexual inte rchange s betwee n boys is appre ciated, it becomes easie r to acce pt the conclusion of Bause rman and Rind (1997) , from their surve y of findings from nonclinical sample s, that the majority of such incide nts are “ evaluate d by males as neutral or positive .”

THE AGE FACTOR Discussion so far has be en large ly base d on the re colle ctions of adults of incide nts whe n they were old e nough to have identifie d the behavior as sexual and to have remembered what happe ned. Adole scents and preadole scents have a de gree of unde rstanding and se lf-de te rmination in matters of sex. Younge r childre n ’s ide as about se x may be vague , but the y know when activitie s are ne cessarily secret and there fore forbidde n. Babie s and infants, howe ve r, can be manipulate d at a time whe n the y have no conce pt of what is happe ning to the m and are unable to speak to othe rs about it or to remembe r it in late r ye ars. From the recipie nt ’s standpoint there are great physical similaritie s be twe en adult atte ntion to their ano-ge nital hygie ne, innoce nt cuddle s and tickle s, and be haviors motivate d by adult erotic inte rest. It is not necessarily the case that se xual stimulation is always more traumatic the younge r the child. Traditionalists adhe re with moral fe rvor to the view that it is essential to shield childre n from pre mature se xual knowle dge , se xual exploration, or obse rvation of adult se xuality. It use d to be suppose d that childre n are ase xual until pube rty, or that in postinfancy the y pass through an asexual late ncy period. Empirical e vide nce shows these impre ssions to have be en derived from adults ’ reluctance to acce pt childre n ’s sexuality and childre n ’s secretive ne ss in the face of adult disapproval. Kinsey (Kinsey et al., 1948, p. 176) has been posthumously attacked for citing evide nce, alle gedly imprope rly obtained, that very young boys, some unde r a year old, can respond to masturbation with erections and appare nt orgasm. Such observations lend credibility to anecdotes about nursemaids pacifying infants by sexual stroking. Anthropological observers of permissive societies, before their cultures were affected by Western influences, report masturbation and copulatory practice s carried out fre ely by very young children and also, in some

550

West

cases, adults participating ope nly in the sexual stimulation of infants and young childre n (Ford and Beach, 1952, p. 188). Even in modern Western society, at least in relative ly uninhibite d culture s like Norway, infants left to their own devices, but observed unobtrusive ly by kinde rgarten teachers, are seen to engage in much spontaneous sexual be havior, including “bodily exploration, genital manipulation and coital training” (Gunde rsen et al., 1981). The relevance of these observations to child sexual abuse is twofold. First, claims made by many offenders that some young childre n seem to enjoy sexual stimulation by an adult gain plausibility. Second, with the possible exception of the promotion of precocious sexual interest and responsive ness, the adve rse effects of noninvasive sexual manipulation by an adult appe ar to be conne cted with the psychological meaning rathe r than the physical nature of the behavior. Howe ver, abse nce of physical damage , and the fact that some primitive socie ties have vie wed the se behaviors as unproble matic, is irrelevant to the potential psychological harm when a culture defines sexual contacts between adults and childre n as horrendous crime.

CRIMINOGENIC EFFECTS Men accused of sex offenses against childre n often report having been sexually abused themselves when they were young (Groth, 1979). Self-exculpatory recolle ctions are difficult to verify, but they are produced both by clerics, who are socially conformist except in their sexuality, as well as by the more socially deviant offe nders (Haywood et al., 1996). Clinicians and psy chodynamic theorists are seriously concerned that sexual molestation of boys may cause them to become molesters themselves. Molested girls are not thought to carry the same risk, though it has been sugge sted that when they become mothers they may fail to protect their childre n from sexual abuse or may display lack of parenting ability in other respects (Banyard, 1997). Wisdom and Ames (1994) have publishe d an important prospe ctive study of a cohort of childre n that include d a substantial numbe r with validate d child sexual or physical abuse historie s, toge ther with matched nonabuse d controls. The y were followe d up and their subse que nt arre st historie s analyze d. The majority of abuse d boys did not become offende rs, but there was a significant statistical association betwee n either early se xual or physical maltreatment and subseque nt offending, both sexual and nonse xual. Sexual abuse was specifically linke d with running away from home as a juve nile and with subse que nt prostitution, but it was not significantly more often associate d with the ge nerality of sex offending in adulthood than was a history of physical abuse or neglect. Physical abuse , but not sexual abuse , was significantly linke d with adult crimes of sexual viole nce.

Boys an d Sexu al Abu se

551

The se results are consiste nt with othe r evide nce that maltre ated children are at risk of becoming adult offe nde rs and that viole ntly mistre ated childre n are at particular risk of becoming viole nt offe nders, but they yie lded no evide nce for the suppose d unique link between early sexual molestation and becoming a mole ster. It seems like ly that any ge nuine link between early abuse and late r crime is mostly cause d by e xposure to the conglome ration of traumatic influe nces, including family pathology and criminal justice interve ntion, that so often accompany it.

WHY BOYS CAN BE HARMED BY PEDOPHILES The notion that sexual attention from women does not harm boys is widely held and has influe nced penal decisions. In 1997, a married woman who had become enamore d of a boy of 14 absconde d with him to Florida, where she was arrested and sent back to England to be prosecuted in a blaze of publicity. Had the couple ’s gende r been reversed, the olde r partner would undoubtedly have been imprisoned, but a High Court judge , noting that the boy appeare d to have suffered no long-te rm harm and did not conside r himse lf abuse d, put the woman on probation (G uardian, 5 Dec. 1997). Of course, the fact that many who have experie nced sexual contact with or approache s from adults when they were boys claim to have been unaffected in no way detracts from clinical evide nce that sometimes serious, lasting, and occasionally devastating effects may follow (King, 1997; Watkins and Bentovim, 1992). The case s se en in clinical or counse ling situations may be self-sele cted and untypical, but they highlight the psychological trauma some young victims expe rience (Mende l, 1995) . In an inve stigation of a sample of men who had contacte d the British counse ling organization “ Survivors ” about sexual assaults, a majority had be en assaulte d when the y were unde r 16 ye ars of age . Most of the men reporting e arly se xual abuse had be en subjected to anal pene tration, nearly all by an olde r male who was known to the m, most ofte n a family member. V ery fe w had had contact with police or he lping age ncies at the time . Unlike those assaulte d as adults, the gre at majority who had be en abuse d as boys had not be come inse cure about the ir he te rosexual orientation, yet, afte r an ave rage lapse of 17 ye ars, they still felt the nee d to unburde n the mselves. Social deprivation, family conflict, and poor parenting are responsible for a statistical, but not necessarily causal, link between early sexual abuse and later maladjustme nt. Mullen et al. (1993) studied a large sample of women in New Zealand, selected at random from the electoral rolls of Dunedin. Some form of unwante d sexual confrontation with an older person while they were unde r 16 was reported by 32% of the women, nearly all of whom recalle d

552

West

their experiences as unpleasant and distressing. The abused women and a control group with no memory of abuse were intervie wed about their personal history and give n tests to determine their current psychological adjustme nt. There was a very significant association between having been reared in a dysfunctional family with inade quate care and protection and e xposure to physical abuse or to sexual abuse . There was also a significant association between sex abuse history and adult problems of depression, anxiety, substance abuse , and eating disorders. However, these problems were heavily concentrated among women from adve rse backgrounds. Save for a minority who had bee n subje cted to sexual penetration, those from stable backgrounds showed no significant excess of adult proble ms despite an abuse history. The authors conclude d that it was when sexual irregularitie s were part of a matrix of adve rse factors that long-te rm proble ms were like ly to ensue. A male population might well have give n a similar result, particularly in view of evide nce that sexually abuse d boys more often come from poore r and physically abusive families (Finke lhor, 1984, p. 150). Childre n of neglectful pare nts, lacking prote ctive supe rvision, may find comfort in attention from outside rs. The psychiatrist Yate s (1979) described the precocious and abusive sexual live s of childre n in the Chicago slums, where erotic play in the hallways of tene ments was “ a substitute for toys.” Many of the young male stre et prostitute s in London studie d by West and de Villie rs (1993) described exploitative sexual contacts with adults when the y were childre n, but nearly all had come from disturbe d backgrounds or rejecting familie s which had rende red the m vulne rable . Pove rty, home lessness, unemployme nt, and a homose xual orientation were the reasons they gave for their e ntry into the sex trade ; none of them sugge sted that early sexual abuse was responsible . Sexual proble ms can be mistake nly attribute d to se xual abuse . In a sample of marrie d men attending a clinic for se xual disorde rs, a diagnosis of curre nt se xual dysfunction was unre late d to the pre sence or abse nce of a history of early sexual abuse , but was significantly associate d with curre nt une mployme nt (Sarwe r et al., 1997) . Okami (1991), in his surve y of positive reactions to sexual incide nts, found few responde nts commenting positive ly on sexual approache s from family membe rs. Incestuous situations may occur in familie s of low social standards, whe re there is much viole nce and drunke n misconduct and where the childre n are expose d to numerous damaging influe nces, but even where this is not the case the sexual involve ment of childre n within the family is indicative of unhappy situations. Se xual molestation by family members, espe cially pare nts, is unse ttling to childre n, not only be cause the inappropriate sexual behavior is unwante d, but be cause of the aura of guilt and secrecy surrounding the activity and the insecurity caused by becoming caught up

Boys an d Sexu al Abu se

553

in the e motional conflicts of adults which threate n the breakup of the home . Although boys are quantitative ly at lesser risk than girls from abuse in their own homes, the abuse s they may suffer include maternal seductions as well as molestation by fathe rs, stepfathe rs, and olde r siblings or family visitors of either sex. It see ms that, unlike girls, boys are rare ly disturbe d by sexual improprieties short of physical contact. In one surve y of psychological adjustme nt among men stude nts (after controlling for the inte rvening variable of dysfunctional pare ntal style , which so often confuse s the picture ) it was found that a history of noncontact abuse appe ared to have no effe ct, but a history of contact abuse did have some association with late r psychological malaise (Collings, 1995) . Terrifying thre ats or se rious viole nce of any kind are calculate d to induce PTSD in anyone, but particularly in childre n and particularly where genital pain and injury is involved. Small childre n can be so subdued by the overwhelming power of an adult aggre ssor that they can do no other than comply, although, like the adult victim of forcible rape, they may be in reality petrified with fear of what may happen to them. Domestic settings, in which the child feels trappe d, are the most like ly venues for repeated physical and sexual brutality. Violent assaults by outside rs are much rarer, if only because opportunities for pe rpetrating the se atrocitie s with impunity are limite d, since the child will report what happe ned and describe the offende r. Offenders who kill their child victims to avoid detection, or who engage in such viole ntly sadistic practices that the victim dies, often have severe, antisocial personality disorde rs which render their sexual interest in minors lethal. Though rare, they attract massive publicity. Four men were sentenced in 1989 to a total of 62 years of imprisonme nt for the manslaughte r of a 14-ye ar-old London boy prostitute, Jason Swift, who was smothered in the course of violent sexual activity. Still in the news 8 years later, one of them, following release, was so hounded that police had to take him into expensive protective custody (G uardian, 7 Nov. 1997). Even when he was later put unde r security in a mental hospital, local protesters demande d his removal. Many women victims of rape experience victimization a second time when they find themselves blame d for letting it happe n, or when they are subje cted through the criminal justice system to hostile cross-examinations in court (Holmstrom and Burgess, 1978, p. 236). The trauma can be even worse when the victim is a child. Measure s have been taken to protect childre n from the worst excesses of courtroom drama, but the English Crown Prosecution Se rvice Inspe ctorate has criticize d the frequent failure s to ide ntify child witnesses who need the protection of screens or vide o links when giving evidence. Boys are particularly liable to conde mnation from family and ridicule from peers if they are suspected of having collaborate d with a homose xual.

554

West

Repeated interrogations by police and social workers, the long wait for cases to be heard and the feeling of re sponsibility for the imprisonme nt of someone who may have been a friend, are inevitably stressful. Charge s may be bitterly contested by offende rs desperate to avoid conviction as a pedophile . In the absence of witnesses to what happe ned, proof of guilt is difficult and offenders may be acquitte d, leaving the child unde r suspicion of lying and expose d to retaliation if the accused was a member of the family.

CONTROL POLICIES The prote ction of childre n and young pe ople from serious sexual abuse is not helped by failure to control and/or treat known offe nders who are genuine ly dange rous. Howeve r, whe n the incide nts in que stion are more like breaches of moral rules than true assaults and have caused no obvious damage, there is que stionable justification for invoking le gal processes that may be detrimental to the suppose d victim. A balance has to be struck somewhere between unne cessary and harmful overreaction and dange rous unde rreaction. Scie ntific evide nce fails to support some of the assumptions unde rlying the pre sent severe penal policie s. Social workers in the U.K. are unde r a profe ssional obligation, and in the U.S. unde r a statutory require ment, to report suspe cts to the police . Informal methods of resolution are out of favor because of the assumed seriousne ss of any se xual incide nt involving a child and the presumption that, if the offe nde r is not appre hended and incarce rate d, offenses will continue inde finite ly. These views, which are prevale nt among profe ssionals, are held in extreme form by large se ctions of the public who, judging by recent television programs, want to see all pedophile s give n life imprisonme nt, if not e xecuted. Conseque ntly, the pursuit of offe nders through the courts tends to be give n priority over the inte rests or wishes of child victims. Systematic follow-up research has generally found relatively low reconviction rates among child molesters, 5% over 15 years, 19% after 24 years in one English survey (Gibbe ns et al., 1981) or 13% over 4 to 5 years in a meta-analysis of 61 publishe d surveys (Hanson and Bussie re, 1995, as cited in Grubin and Wingate , 1996). In an English cohort of men born in 1953, of those convicted of a sexual offense before the age of 40, 10% had a further sexual conviction within 5 years (Marshall, 1997). Reconviction rates have become lower in England in recent years. Among prisoners released in 1987, 6.5% of those with a past or current conviction for a sex offense were reconvicte d for a subseque nt sex offense during the ensuing 4 years, only a fraction of the percentage of reconvictions of property offenders (Marshall, 1994). Men who offend against children

Boys an d Sexu al Abu se

555

in their own family are reconvicted less often than offenders against unrelated children (Tracey et al., 1983). The most persistent recidivists are found among men who are fixated on hunting boys or young men for sex. The law is likely to change , but convictions and reconvictions in the U.K. have been augmented by the criminalization of sexual contacts with young men even if the y are willing participants and over the age when sex with girls is legal. In reality, risk of reoffending among child molesters is extremely variable . A he terose xual man who has fathe red childre n, shown no interest in minors until relations with his wife deteriorate d and a daughte r was coming to sexual maturity, who is de vastate d when an imprope r relationship is exposed, is in a different league from a single male who has neve r been sexually arouse d by adults and has organize d his life and his work around gaining access to young boys or girls. Child molesters are by no means all pe dophile s with a lifelong and exclusive sexual attraction to childre n; some are individuals who make use of childre n only when they are frustrate d, drunk, or their pote ncy is failing. Sociopaths indulge whate ver sexual impulse s please them without regard for moral rules or the welfare of others. Many children who have been the subject of nonviolent sexual incide nts could be spared the stress of a protracte d penal process if appropriate action, by way of tre atment and supervision, could be taken without re sort to the criminal justice process. This can be done in the Netherlands through the “confidential doctor,” a state official who receives reports from complainants or concerned observers, assesses the situation, directs those suspected of offenses to helping agencie s, reporting them to prose cution authoritie s only if they fail to cooperate . The system enables cases to be identifie d and dealt with that would otherwise never be reported, but it can operate only if alternative s to public exposure , criminal conviction, and imprisonme nt can be seen to be efficacious. Alternatives to a strictly punitive response appe ar more constructive and more humane, but conclusive scientific proof that “treatment works” is virtually unattainable . Due to the impossibility of matching treated and untreate d offenders on all re levant factors, allocation needs to be strictly random to produce clear evidence, but in practice this is extremely difficult to arrange , as well as being ethically dubious. Because of relative ly low reconviction rates, to obtain an adequate statistical comparison samples have to be of adequate size and followed up for long periods. Methodological purity is rarely if ever achie ved, but indications of treatment successes, derived from an appare nt cessation of offending and me asurable improvements in attitude, life style, and social circumstances, have been reported from a variety of projects (Marshall and Pithers, 1994). In the U.K., treatment usually consists of group therapy using cognitive behavioral methods and is more often done unde r the aegis of the prison or probation services (see Grubin and Thornton, 1994, concerning treatment in prison) than by psychiatrists and psychologists within the overstretched National Health Service .

556

West

A Home Office sponsore d survey of community-based treatment proje cts for child molesters concluded that only long-term treatment produced beneficial change in the more deviant (i.e., the fixated and recidivist) offenders, but that even short-term intervention helpe d the less deviant (Beech et al., 1996). Unfortunately, most community projects in England are small-scale , not well researched, lacking in appropriate expertise, and without adequate follow-up or satisfactory criteria for evaluation. Adequate provision of treatment could revolutionize the approach to serious sexual abuse of boys and girls. Treatment within the prison system has to battle against the distrustful and antitherapeutic culture of prison inmates, the rapid increase in the prison population, which reduces the resources directed to specialist programs, and the essentially punitive ethos of prison regimes. The need is for specialist treatment centers with a multidisciplined and well-trained staff capable of assessing offenders and devising care and treatment plans directed to their very varied needs. Some offenders would have to be treated under conditions of security, but this need not be within the prison system. In principle, the English legal system provides for these possibilities. Courts have wide discretion in sentencing and before coming to a decision can call for social and psychiatric assessments and reports. Treatment requireme nts can be attache d to probation orders to ensure compliance. Those sufficie ntly disturbed to be dealt with unde r the Mental Health Act can be compulsorily detained in hospital. Unfortunately, because facilities are scarce or inade quate , because many psychiatrists are reluctant to undertake responsibility for sex offenders and the community is so hostile to anyone with a pedophile labe l be ing out of prison, the legal provisions are underused and the potential benefits of a more discriminating and active approach remain untested. Meanwhile, a man whose only offense has been to engage in mutually desired sexual acts with another who happe ns to be a little below the legal age of consent has to be treated as a pedophile and risks being dealt with in the same way as predatory and aggre ssive offenders who ensnare and attack small boys.

CONCLUSION The proble ms caused by sexual incidents between men and boys could be handled more effectively and humanely if the moral outrage encouraged by the media were reduced. Genuine victims would be better protected if penal responses were more discriminating, recognizing gender differences and limiting draconian measures to manife stly harmful or dangerous behavior. In place of a blanke t require ment to involve the police and criminal justice procedure s, informal control for suitable cases, through social and therapeutic services, should be supported.

Boys an d Sexu al Abu se

557

The law setting a relative ly high “ age of consent” for males, enforced regardle ss of circumstances, is unne cessary. The fact that the specified age varie s so much between diffe rent jurisdictions highlights its dubious basis. Children are, in any case, protected from unwanted molestation by laws against inde cent assault. Most people feel that really young childre n are incapable of valid consent, but this could be recognize d by a presumption of absence of consent, unle ss proved otherwise, in the case of childre n unde r (say) 13 years old. Thirteen is already enshrine d in English law on “ unlawful sexual intercourse ” (the equivale nt of Ame rican “ statutory rape ” ), which provide s for long imprisonme nt if the girl is unde r that age . Further protection, such as already exists in some European countries, could be introduce d to forbid sexual contacts with olde r childre n by adults in positions of trust, such as their parents, teachers, doctors, or employers. More important, however, than any readjustment of criminal law, is the need for better informe d public opinion and recognition of the need to shie ld abused childre n from “ secondary victimization. ”

REFERENCES B adgle y, R. F. (1984) . Sexu al Offenses against Children , chap. 6, Canadian Gove rnment Publishing Ce ntre, Ottawa. B anyard, L. (1997) . The impact of child se xual abu se and family functioning on four dimensions of women ’s later pare nting. Child Abuse Neglect 21: 1095-1107. Bagley, C., Wood, M., & Young, L. (1994) . V ictim to abuse r: Mental health and behavioral se quels to child sex abuse in a community survey of young adult males. Child Abuse Neglect 18: 683-697. Baker, A. W., and Duncan, S. P. (1985) . Child sexual abuse : A study of prevale nce in Great Britain. Child Abuse Neglect 9: 457-467. Baldwin, J. D., and Baldwin, J. I. (1997). Gender differences in sexual interest. Arch. Sex. Behav. 26: 181-210. B au rmann , M. C. ( 1983) . Sexu alität, G ewalt u nd psychische Folgen , B undeskriminalamt, Wie sbaden, Germany. Bauserman, R., and Rind, B. (1997) . Psychological correlates of male child and adole scent sexual expe rience s with adults: A review of the nonclinical literature. Arch. Sex. Behav. 26: 105-141. Beech, A., Fishe r, D., Beckett, R., and Fordham, A. (1996) . Tre ating sex offenders in the community. Hom e Office Research and Statistics Directorate, Research Bulletin, 38: 21-25. Bernard, F. (1985) . Paedophilia: A Factual Report, Enclave , Rotterdam. Billen, A. (1992) . Servant of Christ and the nymph. The Observer: 8 Nov., p. 58. Brongersma, E. (1986, 1990) . Loving Boys (2 Vols.), Global Acade mic, Ne w York. Collings, S. J. (1995) . The long-term effects of contact and non-contact forms of child sexual abuse in a sample of university men. Child Abuse Neglect 19: 1-6. Constantine, L. L., and Martinson, F. M. (eds.). (1981) . Children and Sex, Little, Brown, Boston. Dover, K. J. (1978) . G reek Hom osexuality, Duckworth, London. Elliott, M. (1993). Fem ale Sexual Abuse of Children, Longman, Harlow. Erickson, W. D., Walbek, N. H., and Se ely, R. K. (1988) . Behavior patterns of child molesters. Arch. Sex. Behav. 17: 77-86. Finkelhor, D. (1984). Child Sex Abuse, Fre e Press, New York.

558

West

Ford, C. S., and Beach, F. A. (1952) . Patterns of Sexual Behavior, Eyre and Spotiswoode, London. Freund, K. (1981). Assessmen t of pae dophilia. In Cook, M., and Howells, K. (e ds.), Adult Sexual Interest in Children, Academic Press, London. Friedman, R. C. (1988) . Male Hom osexualities, Yale University Press, Ne w Haven. Fritz, G. S., Stoll, K., and Wagne r, N. N. (1981) . A comparison of males and fe males who were sexually molested as children. J. Sex Marital Ther. 7: 54-59. Geraci, J. (ed.) (1997) . Dares to Speak: Historical and Contem porary Perspectives on Boy Love, Gay Me n’s Press, Swaffam, Norfolk. Gibbens, T. C. N., Soothill, K. L., and Way, C. K. (1981). Sex offenses against young girls: A long-term record study. Psychol. Med. 11: 351-357. Gorey, K. M., and Le slie, D. R. (1997) . The pre valence of child se xual abuse: Integrative re view adjustment for potential response and measureme nt biases. Child Abuse Neglect 21: 391-398. Groth, A. N. (1979) . Sexual trauma in the life of rapists and child molesters. Victim ology 4: 10-16. Groth, A. N., and Birnbaum, H. J. (1978) . Adult sexual orientation and attraction to underage persons. Arch. Sex. Behav. 7: 175-181. Grubin, D., and Thornton, D. (1994) . A national programme for asse ssment and treatment of sex offenders in the English prison system. Crim. Justice Behav. 21: 55-71. G ru bi n, D ., an d W i ng at e , S. ( 1 99 6) . Se xu al offe n ce re cidivi sm : pre dict ion ve rsus understanding. Crim. Behav. Mental Health 6: 349-359. Gunde rsen, B. H., Mel ås, P. S., and Skå r, J. E. (1981) . Se xual be havior of preschool children. In Constantine , L. L., and Martinson, F. M. (eds.), Children and Sex, Little, Brown, Boston, pp. 45-61. Haywood, T. W., Kravitz, H. M., Wasylew, O. E., Goldberg, J., and Cavannaugh, J. L. (1996) . Cycle of abuse and psychopathology in cle ric and non-cleric mole sters of children and adolescents. Child Abuse Neglect 20: 1233-1243. Holmstrom, L. L., and Burgess, A. W. (1978) . The Victim of Rape: Institutional Reactions, Wiley, Chichester. Howitt, D. (1995). Paedophiles and Sexual Offenses Against Children, Wiley, Chichester. Hunte r, M. (1990) . Abused Boys: The Neglected Victims of Sexual Abuse, Fawcett Columbine, New York. Je nkins, P. (1996) . Pedophiles and Priests: Anatom y of A Contem porary Crisis, O xford University Pre ss, Oxford. King, M. (1997). Sexually assaulted males: 115 me n consulting a counseling service . Arch. Sex. Behav. 26: 579-588. Kinsey, A. C., Pomeroy, W. B., and Martin, C. E. (1948) . Sexual Behavior in the Hum an Male. W. B. Saunders, Philade lphia. Kinsey, A. C., Pomeroy, W. R., Martin, C. E., and Ge bhard, P. H. (1953) . Sexual Behavior in the Hum an Fem ale, W. B. Saunders, Philade lphia. Laum an n, E . O ., Gagnon , J. H., Mich ae l, R. T., and Michae ls, S. (1994) . Th e Social Organization of Sexuality: Sexual Practices in the United States, Chicago Unive rsity Pre ss, Chicago. Li, C. K., West, D. J., and Woodhouse, T. P. (1993) . Children ’s Sexual Encounters with Adults, Prometheus Books, Buffalo, NY. Marshall, W. L., and Pithers, W. D. (1994) . A reconsideration of treatment outcome with sex offenders. Crim. Justice Behav. 21: 10-27. Marshall, P. (1994) . Reconviction of imprisoned se x offende rs. Research Bulletin 36: pp. 23-29, Home Office, London. Marshall, P. (1997) . The pre valence of convictions for sexual offending. Research Findings 55, Home Office Rese arch and Statistics Directorate , London. Me ndel, M. P. (1995) . The Male Survivor, Sage , London. Mullen, P., Martin, J. L., Ande rson, J. C., Romans, S. E ., and He rbison, G. P. (1993) . Childhood se xual abuse and mental health in adult life. Br. J. Psychiat. 163: 721-732.

Boys an d Sexu al Abu se

559

Nash, C. L., and We st, D. J. (1985) . Sexual molestation of young girls: A re trospective survey. In We st, D. J. (ed.), Sexual Victimization, Gower, Aldershot, UK. Ne lson, J. A. (1981) . The impact of incest: Factors in self-evaluation. In Constantine, L. L., and Martinson, F. M. (eds.), Children and Sex, Little, Brown, Boston. Ne wton, D. E. (1978) . Homose xual be havior and child molestation: A revie w of the evide nce . Adolescence 13: 29-43. O ’Carroll, T. (1980) . Paedophilia: The Radical Case, Peter Owen, London. Okami, P. (1991) . Self-reports of “positive ” childhood and adult sexual contacts with older persons: An exploratory study. Arch. Sex. Behav. 20: 437-457. Peters, S. D., Wyatt, G. E., and Finkelor, D. (1986) . Prevalence. In Finkelhor, D. (ed.), A Sourcebook on Child Sexual Abuse, Sage, Be verly Hills. Re ekie, A. (1997) . Be lgium. In W est, D. J., and Gree n, R. (e ds.), Sociolegal Con trol of Hom osexuality: A Multi-Nation Com parison, Plenum Press, Ne w York. p. 296. Rind, B., and Tromovitch, P. (1997) . A meta-analytic review of findings from national samples on correlate s of child sexual abuse. J. Sex Res. 34: 237-255. Russell, D. E. H. (1984). Sexual Exploitation, Sage , Beverly Hills, CA. Sariola, H., and Uutela, A. (1994) . The prevale nce of child se xual abuse in Finland. Child Abuse Neglect 18: 827-835. Sarwe r, D. B., Crawford, R. I., and Durlock, J. A. (1997) . The relationship between childhood sexual abuse and adult se xual dysfunction. Child Abuse Neglect 21: 649-655. Soothill, K. (1997) . Rapists under 14 in the ne ws. Howard J. 36: 367-377. Tracey, F., Donnelly, H., Morgenbesse r, L., and Macdonald, D. (1993). Program evaluation: Recidivism research involving sex offenders. In Greer, J. G., and Stuart, I. R. (eds.), The Sexual Aggressor, Van Nostrand Reinhold, New York. Watkins, B., and Bentovim, A. (1992) . Male children and adolescents as victims. In Mezey, G. C., and King, M. B. (eds.), Male Victims of Sexual Assault, O xford University Pre ss, Oxford. Wellman, M. M. (1993) . Child sexual abuse and gender differences: Attitudes and prevale nce . Child Abuse Neglect 17: 539-547. West, D. J., and De V illiers, B. (1993). Male Prostitution, Haworth, New York. West, D. J., and Woodhouse, T. P. (1993) . Sexual e ncounters betwee n boys and adults. In Li, C. K., West, D. J., and Woodhouse, T. P. (eds.), Children ’s Sexual Encounters with Adults, Prometheus, Buffalo, NY. W isdom, C. S., and Am e s, M. A. (1994) . Criminal conse que nce s of childhood se xu al victimization. Child Abuse Neglect 18: 303-318. Wyre, R., and Tate, T. (1995) . Murder of Childhood , Penguin, Harmondsworth. Yate s, A. (1979) . Sex without Sham e, Temple Smith, London.

Archives of Sexual Behavior, Vol. 27, No. 6, 1998

Sexual Harassm ent: Identifyin g Risk Factors Elizabeth A. O’Hare, M.A., 1 an d William O’Don oh u e, Ph.D.2

A new model of the etiology of sexual harassment, the four-factor model, is presented and compared with several models of sexual harassm ent including the biological model, the organizational m odel, the sociocultural m odel, and the sex role spillover model. A num ber of risk factors associated with sexually harassing behavior are examined within the framework of the four-factor model of sexual harassm ent. These include characteristics of the work environm ent (e.g., sexist attitudes am ong co-workers, unprofessional work environm ent, skewed sex ratios in the workplace, knowledge of grievance procedures for sexual harassm ent incidents) as well as personal characteristics of the subject (e.g., physical attractiveness, job status, sexrole). Subjects were 266 university female faculty, staff, and students who completed the Sexual Experience Questionnaire to assess the experience of sexual harassment and a questionnaire designed to assess the risk factors stated above. Results indicated that the four-factor model is a better predictor of sexual harassm ent than the alternative m odels. The risk factors most strongly associated with sexual harassment were an unprofessional environm ent in the workplace, sexist atmosphere, and lack of knowledge about the organization ’s formal grievance procedures. KEY WORDS: sexual harassme nt; risk; sexism; attractivene ss; se x role.

INTRODUCTION In recent years, the problem of sexual harassme nt has gained increasing attention by researchers. Although both wome n and me n may experience sexual harassment, re search shows that women are more like ly to be sexually harassed (Gruber, 1997; Gutek, 1985). Moreover, women are more like ly to experience negative job-re lated conseque nces of sexual harassment such as quitting or losing a job due to unwanted sexual advance s (Dansky & Kilpatrick, 1997). For these reasons, this study focuses on the sexual harassme nt of women by men. 1 2

Departme nt of Psychology, Northern Illinois University, De Kalb, Illinois 60115. Departme nt of Psychology/298, Unive rsity of Nevada, Reno, Ne vada 89557. 561 0004-0002/98/1200-056 1$15.00/0

Ó

1998 Plenum Publishing Corporation

562

O’Hare an d O’Don oh ue

Three type s of se xual harassme nt have bee n ide ntifie d (Gelfand et al., 1993) . The se include ge nde r harassme nt, unwante d sexual attention and sexual coe rcion, each of which consists of a varie ty of verbal and nonve rbal behaviors. Gende r harassme nt involve s be haviors that ge nerally result in conve ying hostile and degrading attitude s about wome n such as by genderbase d hazing. Unwante d sexual attention consists of behaviors that are more widely re cognize d as harassing such as re pe ated attempts to establish a romantic relationship afte r re fusal; unwante d touching; and sexual imposition or assault. Se xual coe rcion, the le ast common, ye t most unive rsally recognized type of harassment, involves bribery or threats (either explicit or implie d) for sexual coope ration. There is large variation in reports of the numbe r of wome n who have experienced harassme nt. The lifetime incidence rates range from about 28% (Cammaert, 1985; Ellis et al., 1991) to 75% (Lafontaine and Tredeau, 1986) . Actual incide nce rates are difficult to de te rmine because of the differe nces in rese arch methodology among studie s (e.g., sample size and dive rsity, de finition/cate gorization of sexual harassme nt, and time frame). In a review of 18 studie s, Grube r (1990) compute d the median percentage of wome n who have expe rience d se xual harassme nt at some point in their working care er as 44% . These estimates support the belie f that the proble m of sexual harassment is widespre ad. Because of the large number of women who reporte dly expe rie nce se xual harassm e nt, an unde rstanding of its effe cts is also important and has be en the focus of se veral studie s. Research indicate s that the conseque nces of sexual harassme nt to both the organization and to the individual are substantial. Organizations are face d with the costs of lawsuits and damage awards as well as with other indirect costs such as job turnover, absenteeism, reduced productivity, and me dical insurance claims (United States Merit Systems Protection Board [USMSPB] , 1981). Pe rsonal costs to victims of se xual harassme nt include disruption s in careers and re lationships with co-worke rs (Gutek and Dunwoody, 1988; Gute k and Koss, 1993; Schne ide r and Swan, 1994) , as well as emotional and physical symptoms including anxie ty, de pre ssion, irritability, ange r, sleeplessness, weight loss, uncontrolle d crying and stomach proble ms (Dansky and Kilpatrick, 1997; Loy and Stewart, 1984; Gutek and Koss, 1993). Women who have be e n subje ct to se xual harassme nt also expe rie nce a gre ate r life time risk of posttraum atic stre ss disorde r ( PTSD) and major de pre ssion than wome n who have ne ve r be e n se xually harasse d (Dansky and Kilpatrick, 1997) .

Sexual Harass men t: Risk Factors

563

Although re se arch on se xual harassme nt has inve stigate d its pre vale nce as well as its conse que nce s both to individuals and to organizati ons, le ss cle ar are the cause s of se xual harassm e nt. Four major mode ls have be en propose d to e xplain se xual harassm e nt. The se mode ls include the natural/biol ogical mode l (Tangri and Haye s, 1997; Tangri et al., 1982) , the organizatio nal mode l ( Tangri et al., 1982) , the sociocultural mode l (Farle y, 1978; MacKinnon, 1979), and the sex-role spillove r model (Gutek and Morasch, 1982; Tangri and Haye s, 1997) .

Natu ral/Biologi cal Model The natural/biological model holds that se xual harassme nt is not actually harassme nt, and conse quently, does not have deleterious conse quence s, is not sexist, and is not discriminatory (Tangri et al., 1982) . Rathe r, be havior labe le d “ se xual harassme nt ” is a manife station of the natural attraction between men and women. This model assume s that men have stronge r se x drive s than wome n, and the refore, the y behave in a se xually aggressive manne r both in the workplace and othe r se ttings. Howe ver, the y have no inte nt to harass the individuals they pursue and some of their be havior may be perceived as unwante d but this is a natural consequence of their sexual asse rtiveness/ aggre ssiveness.

Organ ization al Model The organizational mode l assume s that organizations facilitate sexual harassme nt through power diffe rentials created by hie rarchical structures (Tangri et al., 1982) . Individuals in le gitimate positions of authority have the opportunity to abuse their power for their own sexual gratification through the harassme nt of subordinate s. Harassment offers a way for superiors to intimidate and to control their subordinate s. In addition to power differentials, other organizational characteristics are viewed in this mode l as contributing to the incide nce of sexual harassme nt. These include contact with the opposite sex on the job, the ratio of male s to fe males in the work place , occupational norms, job functions, job alte rnative s, and availability of grievance procedure s.

O’Hare an d O’Don oh ue

564

Sociocu ltu ral Mod el The sociocultural model addresses the societal context in which sexual harassment occurs. This model posits that sexual harassme nt in the workplace is a manifestation of general male dominance (Farley, 1978; MacKinnon, 1979). According to this model, harassment is one mechanism for maintaining male dominance over wome n, both occupationally and economically, by limiting the ir growth or by intimidating them to leave the work arena. This model holds that men and women are socialize d in ways that maintain this structure of dominance and subordination. Males are rewarded for aggressive and assertive behavior, whereas women are socialized to be passive, to avoid conflict, to be sexually attractive, and to feel responsible for their own victimization (Tangri et al. 1982). Vaux (1993) suggests that sexual harassme nt is a form of “ moral exclusion ” in which males feel entitled to their position of greater power in society. As a result, they justify their behavior and disregard any harmful consequences of their actions toward less powerful groups (i.e., women). Tangri et al. (1982) te ste d these three models in an analysis of the data from the USMSPB study (1982). O ve rall, the data did not provide support for the natural/biological model, but did lend limite d support for the organizational and for the sociocultural models. These analyse s sugge st that although the latte r two mode ls are informative , they are not sufficie nt to fully explain the causes of sexual harassme nt which appe ar to be more multidime nsional than is sugge ste d by any of the se models.

Sex-Role Sp illover Model A fourth mode l, the sex-role spillover mode l (Gute k and Morasch, 1982), attributes sexual harassment to the carryove r into the workplace of genderbased expectations which are irrelevant to, and inappropriate for, work. According to this mode l, sexual harassme nt is most like ly to occur in work environme nts where the sex ratio is skewed in either direction. For women in male-dominate d or in fe male-dominate d work, sex role becomes a more salie nt feature than work role, thus facilitating sexual harassme nt. In the male-dominated workplace, a woman ’s gender is a salie nt feature because of her singularity and distinctive ness. Thus, women in the male-dominate d workplace stand out, and are perceived in their sex role over and above recognition in their work role . In the female-dominate d workplace , sex role and work role ove rlap. Traditional female jobs tend to emphasize aspects of the female sex role (e.g., the nurturing role of teachers and nurses, the sex-obje ct role of cocktail waitresses, the helper role of administrative assistants), thus resulting in the job itself acquiring aspects of the sex role.

Sexual Harass men t: Risk Factors

565

Gute k and Morasch (1982) found that the ir mode l he ld up well unde r empiric al analysis. The y found that wome n who we re e mploye d in nontraditiona l jobs expe rie nce d more se xual harassme nt behavior and more ne gative conse que nce s from se xual harassm e nt than the ave rage working woman. The y also found that nontraditiona lly employe d women expe rie nce d more harassm e nt than wome n in traditi onal fe male jobs. This mode l combine s aspe cts of the organiz ational and sociocultural mode ls, and is the re fore more compre he nsive than any of the first thre e mode ls. Howe ve r, it fails to include organiz ational variable s othe r than sex ratio and ignore s pe rsonolo gic al variable s re late d to both the harasser and the victim that may be important factors. Although e ach of the mode ls de scribe d above has its stre ngths, each is limite d by a narrow focus on only one , or a limite d numbe r, of the dime nsions that may contribute to se xual harassm ent. Se xual harassment appe ars to be a phe nom e non de te rmine d by variable s at multiple leve ls including individual, organizational, and sociocultural. Thus a multidim ensional mode l is re quire d to ade quate ly de scribe its e tiology. In the follo win g se ctio n, a m ultidi me nsional m ode l, the four-factor mode l, which combine s individual, organizationa l and sociocultural factors is pre se nte d. This mode l is base d on the four-precondition mode l of sexual abuse , which atte mpts to de scribe the e tiology of child sexual abuse ( Finke lhor, 1984) .

Fou r-Factor Mod el The four-factor mode l is base d on the assumption that the variable s re late d to se xual harassm e nt can be groupe d into four factors which must be me t for harassm e nt to occur: 1. Motivation ( e.g., sexual attraction and/or powe r ne e ds); 2. Overcom in g Internal Inh ibition s against harassm e nt ( e .g., vie wing se xual harassm e nt as ille gal or immoral, victim empathy, outcom e e xpe ctancie s) ; 3. Overcom in g External Inhibitions against harassme nt (e.g., explicit grievance procedures and conse que nce s to harasse rs) ; and 4. Overcom in g Victim Resistan ce (e .g., he r ability to re cognize and stop pre monitory be havior to harassm e nt) . Factors 1 and 2 addre ss individua l variable s re late d to the harasse r, while Factor 3 addre sses situational factors, both organizational and sociocultura l, re le vant in sexual harassm e nt incide nts, and Factor 4 addre sse s individua l factors re late d to the pote ntial victim of se xual harassm ent. For sexual harassm e nt to occur, the conditions of all four factors must be satisfied ( Fig. 1) .

566

O’Hare an d O’Don oh ue

Fig. 1. The four-factor mode l of se xual harassment.

O ne of the major pote ntial advantage s of this mode l is that it combine s and accounts for all of the re le vant factors name d in the existing mode ls into one compre he nsive mode l. It include s sociocultural and organizationa l factors as well as individual factors as the y re late to both the harasse r and to the pote ntial victim . This mode l can also account for the wide varie ty of be haviors that can be conside re d sexual harassm ent in its most and le ast se ve re forms. For e xample , be haviors which are fre que ntly not conside re d to be harassing be havior by men such as sugge stive looks, se xual re marks, or pre ssure for date s (USMSPB , 1981) , could be motivate d by se xual attraction with no re al inte nt to de mean. The more se ve re forms of harassm e nt such as physical touching and pre ssure or bribe ry for se xual favors can be e xplaine d by othe r motivational factors such as the ne e d for powe r and control. A third pote ntial advantage of this mode l is that it re cognize s the role of the pote ntial victim of se xual harassme nt without blam ing the victim . While certain individual characte ristics of the victim may contribute to or facilitate harassm ent, the bulk of the re sponsibility lie s on the harasse r who must first be motivate d and take ste ps to ove rcome the re le vant inhibitors be fore the victim eve n be come s involve d. The purpose of this study is to be gin to explore the utility of the propose d mode l and thereby identify some of the spe cific variable s that act to inhibit or to facilitate incide nts of sexual harassme nt. Although the mode l is e tiological, this study re lie s on corre lational analyse s that may identify risk factors whose causal re lationship with sexual harassme nt can

Sexual Harass men t: Risk Factors

567

be evaluate d in future studie s. Because of the difficulty in ide ntifying men who sexually harass, it is difficult to assess the first two preconditions: the motivations and inte rnal inhibitors of the harasse r. This study focuses on those variable s that are assessible from the report of female victims (Factor 2: Overcoming Internal Inhibitors is not asse ssed). The following variable s relate d to each factor are evaluate d.

Factor 1: Motivation Physical Attractiveness. It is hypothe sized that women who are more attractive will report more sexual harassme nt be cause incre ased physical attractiveness increases the motivation of certain males to sexually harass. More specifically, women who are more physically attractive will be more like ly than women who are less attractive to experience sexual harassme nt in the form of verbal requests because these behaviors are aimed at establishing sexual or social relationships which normally involve sexual attraction. There may be other motivations to harass (e.g., ange r towards women) but these are not measurable in a sample of women.

Factor 3: Overcom in g External Inhibitors Privacy of work-space. Women who report having work environme nts that provide more opportunity for privacy will report more sexual harassme nt than women who work in less private environme nts. This can be explaine d by the sheer opportunity presented for the potential harasse r because there is less opportunity for anothe r worker to witness any unwante d sexual advance s, propositions, comme nts, or looks. Wome n who work in such isolation are more like ly to experience harassme nt involving physical contact than are women who work in more ope n environme nts. Knowledge of G rievance Procedures. Wome n who report knowle dge of the ir organization ’s grievance proce dure s for se xual harassme nt will report le ss harassme nt. In an organization where se xual harassme nt is re cognize d as a proble m, victims have a cle ar means of filing complaints, and offenders are punishe d appropriate ly, pote ntial harasse rs will like ly be de te rred from harassing. Sexist Attitu des. Women who work in organizations that they pe rceive to be characte rized by se xist attitude s and behaviors will re port more sexual harassme nt. Environme nts in which women are viewed as sex obje cts and as infe rior to men create a climate more conducive to the domination and denigration of wome n through se xually harassing behavior.

O’Hare an d O’Don oh ue

568

Unprofessional Environ m ents. Wome n who work in “ unprofe ssional ” environme nts (i.e ., where the re is swearing and drinking on the job, whe re employe es are treate d disrespectfully, and where e mploye es are expe cted to do activitie s that are not formally a part of their job) will report more sexual harassme nt than wome n in profe ssional environme nts. Skewed Sex-Ratio. Wome n who work in environme nts in which there are greatly une qual numbe rs of men and women will report more sexual harassme nt than women in inte grate d environme nts. As described in the sex-role spillove r model, for wome n in male dominate d and in female-dominate d work, se x role becomes a more salie nt feature than work role , thus facilitating se xual harassme nt (Gutek and Morasch, 1982) .

Factor 4: Overcom in g Victim Resistance Sex-role. Wome n who are type d as fe minine on the Bem se x-role inve ntory (Be m, 1974) will re port more se xual harassme nt. Traditionally, women have be e n socializ e d to be passive , nonasse rtive , to avoid conflict, to ple ase othe rs and to be se xually attractive . Women who fit the se tradition al fe minine ste re otype s are more like ly to acce pt sexual harassment as e xpe ctable be havior and to fe e l re sponsible for the ir own victimization, and thus would be le ss e ffe ctive in re sisting such be havior, or effe ctive ly discouraging pre monitory be havior. Job Status. Women in lower status positions will report more sexual harassment. These women are more vulnerable to harassme nt than are women in manage ment or supervisory positions be cause of a lack of job-relate d powe r over a potential harasse r. In addition, they may be less able to resist harassment because of less job security, fewer job alternative s, and less material independe nce. Because these women typically have less job security, they are more vulnerable to threats and promise s regarding their positions, and therefore, they are more likely to experience harassment in the form of inappropriate verbal requests which involve pressure, threats or promise s for sexual or social relationships.

METHOD Su bjects an d Procedu re A total of 560 que stionnair e s we re distributed to the fe male faculty, staff, and stude nts at a large Midwe ste rn unive rsity. Four hundre d fe male me mbe rs of the faculty and staff ( e xcluding individuals in the

Sexual Harass men t: Risk Factors

569

Departm e nt of Psychology) we re chose n at random from a comple te list of faculty and staff and we re re cruite d through the campus mail. Individuals who did not re turn the conse nt form afte r a 3-we ek pe riod were re contacte d through campus mail and we re again aske d to participate in the study. Stude nt participants we re re cruite d through the introduc tory psychology course . In addition, stude nts from seve ral campus sorority house s we re re cruite d to incre ase the variability in stude nt age and amount of work e xpe rie nce . The se stude nts were contacted through a sorority represe ntative who distribute d and colle cted the que stionnaire s at the group ’s monthly me e ting. Stude nts comple te d the same que stionnaire as faculty and staff participants. Thus, the ir e xpe rie nce s with sexual harassme nt in the work force , rathe r than in an education al setting, were asse sse d. Be cause the re is curre ntly no e vide nce to sugge st that diffe rent variable s would affe ct working wome n of diffe re nt age groups, the data from faculty, staff, and stude nt populations were combine d for all analyse s. O ne hundre d thirty-five que stionnaire s we re re turne d by faculty and staff re sulting in a re turn rate of 34% . O ne hundre d and thirty one que stionnaire s were re turne d by stude nts re sulting in a return rate of 82% . (The nature of data colle ction like ly accounts for the diffe re nce in re turn rate s be twe e n faculty/staff and stude nts. Faculty/staff participan ts were re cruite d through the mail with no face -to-face contact with the inve stigator, whe re as stude nt participants comple te d the que stionnair es in a large group with one of the inve stigators or a re pre se ntative pre sent.) Se ve nte e n incom ple te que stionnaire s we re e xclude d re sulting in a total of 249 valid que stionnaire s use d in the analyse s. The mean age of the participants was 30.8 years (SD = 12.2) with a mean of 12.3 ye ars of work expe rience (SD = 9.4) . Eighty-e ight percent of the participants were White , 3.6% Black, 2.0% Asian, 2.4% Hispanic, 1.6% othe r. Thirty-se ven percent of the wome n we re marrie d, 58.6% single , 5.6% divorce d, and 1% widowe d.

Materials

Participants comple te d a que stionnaire that assessed the incide nce of sexually harassing be haviors, pe rsonal characte ristics, and characte ristics of the work e nvironme nt in which the y had this e xpe rie nce (not limite d to the ir current position) .

570

O’Hare an d O’Don oh ue

Inciden ce of Sexual Harassm ent — The Sexual Experien ce Question naire The Sexual Expe rience Q uestionnaire (SEQ ), form W (Fitzge rald et al., 1988) was use d to assess the e xpe rie nce of sexual harassme nt of women in the workplace . The ite ms in the SEQ are groupe d into the three types of sexual harassme nt de scribe d above : Gende r Harassme nt (GH); Unwanted Se xual Attention (USA); and Sexual Coercion (SC). Response s on the SEQ are made on a 5-point Like rt scale (1-ne ve r, 2-once or twice , 3-sometimes, 4-ofte n, 5-most of the time). Fitzgerald et al. (1988) reporte d ade quate re liability and validity for the revise d SEQ. Alpha reliabilitie s range from .75 to .89 and 2-wee k te st-re te st re liability yie lded a coefficie nt of .86.

Work Environm ent/Organization al Characteristics Subje cts who re ported having e xperienced sexual harassme nt, we re aske d to provide specific information about the work environment in which the harassment was experienced. For purpose s of comparison, subje cts who reported never experie ncing se xual harassme nt were aske d to provide information about their current or most recent work environme nt. The following work characteristics were assessed. Sex-Ratio (Percentage of Fem ale Co-Workers). Ske wness of the sex-ratio was compute d as the absolute value of the difference between the pe rcentage of women and 50% . This resulte d in a range of value s from 0-50 where value s closest to zero indicate an integrate d environme nt and value s closest to 50 indicate a pre dominance of one gende r. Professional Atm osphere. A six-ite m scale consisting of the following behaviors which are typically believed to be unacce ptable in a professional environme nt was created: (i) use of obscene language on the job; (ii) use of alcohol on the job; (iii) disrespect to other employe es; (iv) expectations of performing menial activitie s not formally a part of the job; (v) inappropriate displays of emotion (e.g., crying, losing one ’s temper); (vi) unauthorize d use of company time and resources for personal use. Ratings of frequency of each behavior were made on a 7-point scale ranging from 1 (not at all) to 7 (very often) and summed. Analysis of the internal consiste ncy of the scale yielded an alpha value = .74. Awareness of G rievan ce Procedures. Subje cts rate d the ir knowle dge of the organization ’s grie vance proce dure s at the time of the harassme nt (if applicable ) or in the ir curre nt place of e mployme nt on a 7-point scale (1not aware to 7-fully aware ).

Sexual Harass men t: Risk Factors

571

Sexist Atm osphere. Subje cts rated their perception of sexist attitude s and be lie fs among male co-worke rs on a 7-point scale (1-not at all sexist to 7-sexist) . Privacy of Work Space. Subje cts indicate d the opportunity for privacy during their work day on a 7-point scale (1-no opportunity for privacy/spend the majority of time in an area that can be seen from all side s to 7-high opportunity for privacy/spe nd the majority of time in a room with a door that can be shut).

Personal Characteristics Sex-Role. The Bem Sex-Role Inve ntory (BSRI, Bem, 1974) consists of two scales, a Masculinity Scale and a Femininity Scale , which assess personality characte ristics that are typically associate d with feminine or masculine sex roles. A sex-role score was obtaine d by subtracting the Masculinity Scale score from the Femininity Scale score. High positive scores indicate d a stereotypical feminine role, while high negative scores indicate d a stronge r stereotypical masculine role. Bem (1974) reported inde pendence of these scales (ave rage r = ¯.03), reliability over a 4-week interval (average r = .93) , and lack of corre lation with the tendency to de scribe one self in a socially de sirable way (ave rage r = ¯.06). Job Status. Job status was indicate d by pay level in dollars per hour. Physical Attractiven ess. Subje cts were aske d to indicate on a scale of 1 (unattractive ) to 6 (attractive ) how physically attractive they be lie ved others would judge them. Rand and Hall (1983) inve stigate d the validity of self-ratings of attractive ne ss, and found the highe st corre lation between self and inde pende nt judge ’s ratings of attractive ne ss (r = .68) whe n subje cts were aske d how attractive the y believed othe rs would judge them.

RESULTS Table I presents the frequency of each type of sexual harassme nt. Individual scores of sexual harassme nt were compute d by collapsing the SEQ response s to dichotomous scores (1-ne ver, 2-5-once or more) and summing the response s on the relevant ite ms for each of the three type s of harassme nt. Thus each subje ct was assigne d a separate score for Gender Harassment, Unwante d Sexual Attention, and for Sexual Coercion. Theoretically, scores could range from 0-6 on the Gender Harassment and Sexual Coe rcion scales, and from 0-7 on the Unwante d Sexual Attention scale. Table II presents the correlation coe fficients for each of the risk factors propose d in the four-factor model with scores on each of the three type s of harassme nt.

30.9 39.0 33.7 41.8 63.1 36.5

46.6 47.8 67.9 69.9 52.2 82.3 97.2

90.8 89.2 92.0 88.4 92.4 93.6

Type 1: Gender harassment Suggestive stories/offe nsive jokes Crudely sexual remarks Wome n treated “ diffe re ntly” Condesce nding toward women Suggestive mate rials Se xist re marks about care er options

Type 2: Unwanted Se xual Attention Discussion of personal/se x life Unwanted sexual attention Atte mpts to e stablish romantic/sexual relationship despite discouragement Continue d to ask for dates after refusal Touching that made you fee l uncomfortable Unwanted stroking or fondling Unwanted attempts for sex that resulted in struggling

Type 3: Sexual Coercion Subtle bribery for se xual coope ration Threate ned for not cooperating Implied faster promotion for sexual cooperation Made to fe el it was necessary to coope rate in order to be treated well Felt you would be treated poorly for not cooperating sexually Treated badly for refusing to cooperate sexually

Ne ver

4.8 6.8 3.6 6.4 4.0 3.2

27.3 27.7 21.7 17.7 30.5 12.4 2.4

22.9 26.9 23.7 30.5 18.5 28.1

O nce or twice

Table I. Percentage of Sample (N = 249) Endorsing Each O ption

1.6 1.6 1.6 2.4 1.2 0.8

18.1 17.7 5.2 6.5 9.2 2.8 0.0

26.9 20.9 23.7 16.9 12.4 20.5

Sometimes

1.2 0.4 1.2 0.4 0.8 0.4

6.9 5.2 3.6 4.4 6.8 1.6 0.0

16.5 12.0 15.3 8.8 4.8 12.9

O ften

1.6 0.4 1.6 0.0 0.0 0.0

0.8 1.2 0.8 1.2 0.4 0.4 0.0

2.4 0.8 3.2 1.6 0.4 1.2

Most of the time

Sexual Harass men t: Risk Factors

573

Table II. Correlation Coefficients for Individual/Work-Environment V ariables and Sexual Harassment (by Type) Variable Attractiveness Sex role/fem ¯ masc Fe mininity Masculinity Pay Job status Grievance procedures Sex-ratio Privacy Unprofessional atmosphere Sexist environment a b

Gender harassment .05 a ¯.15 .02 ¯ b .20 b ¯.17a .12 ¯ b .22 b ¯.18 ¯.05b .50 .65b

Unwante d sexual atte ntion .12a ¯.11 .03 .18b b ¯.26a .13 ¯ b .32 ¯.07a ¯.12b .48 .54b

Sexual coercion .16b .05 .18b .11 b ¯.18a .15 ¯ b .19 .05 ¯.02b .32 .38b

p < .05. p < .01.

The corre lationa l analyse s we re followe d by ste pwise regre ssion analyse s for e ach type of se xual harassm e nt on the pre dictor variable s significant ly corre late d with e ach type . The data were first examine d to insure that all the assum ptions of re gre ssion analyse s, including line arity, normality, and constant variance we re me t. The variable “ pay” was found to have a positive ly ske wed distribution , the re fore , a log transform ation was pe rforme d to re duce ske wne ss. Tole rance value s were examine d to ensure the abse nce of colline arity among pre dictors. The variable “ se xist environme nt ” was e xclude d from all subse que nt analyse s involving Gende r Harassme nt be cause the re was an ove rlap be tween the ite ms on the Gende r Harassm e nt scale and this ite m. Table III pre se nts the variable s that we re e nte re d in e ach equation (according to the crite rion p £ 0.05 for the F value of the variable examine d on that ste p), and the ir diffe re ntial contributions to se xual harassment score s. The main contribu tor to Unwante d Se xual Atte ntion and to Se xual Coe rcion score s was wome n ’s pe rception of se xist attitude s among male co-worke rs. This variable accounte d for 28 and 15% of the variance in the se type s of harassm e nt re spe ctive ly. An unprofe ssional work environm e nt was the main contribu tor for Gende r Harassm ent and accounte d for 29% of the variance . This variable was also a significant pre dictor for Unwante d Se xual Atte ntion accounting for an addition al 8% of the variance for this type of harassm e nt. Knowle dge of grie vance proce dure s also pre dicte d both Ge nde r Harassm e nt and Unwante d Se xual Atte ntion, and high Masculin ity and Fe mininity pre dicte d Gende r Harassme nt and Se xual Coe rcion re spe ctive ly.

O’Hare an d O’Don oh ue

574

Table III. V ariable s Entered in the Stepwise Regression Analysis for Each Type of Harassment

b Variable

2

(after all variables were e ntered)

R (on each ste p)

.50 .16 .15

.29 .32 .34

.35 .29 .20

.28 .36 .40

¯ .38 .15

.15 .17

Gender Harassment Unprofessional environment Knowledge of grie vance proce dures BSRI ¯masculinity score Unwanted sexual attention Sexist e nvironment Unprofessional environment Knowledge of grie vance proce dures Sexual coercion Sexist e nvironment BSRI ¯femininity score

The four-factor model was compare d to the other models discusse d by testing separate regression equations for each of the models and comparing R2 value s. (Note : The biological model was not include d in these comparisons because of a lack of information about the harasse r.) Table IV presents the R2 value s for each model by type of harassme nt as well as the standardize d regression coefficients for each variable entered in each equation. The fourfactor model accounts for more of the variance in sexual harassme nt scores for Gende r Harassment, Unwante d Sexual Attention, and for Sexual Coercion than does the Organization model, the Sociocultural model or the Sex-Role Spillover model.

DISCUSSION Prevalen ce The most prevale nt type s of sexual harassme nt reported were Gender Harassment and certain forms of Unwante d Sexual Attention. (e.g., discussion of personal/se x life). Many of the behaviors most widely experienced by women (e.g., sexual remarks, telling suggestive stories, unwanted or discouraged attempts to establish romantic relationships) are not unive rsally perceived as harassing be havior, and wome n more frequently labe l the se behaviors as se xually harassing than do men (Fitzgerald and Ormerod, 1991; Powell, 1986). The fact that many men do not conside r these behaviors to be harassing may account for their highe r prevale nce.

Sexual Harass men t: Risk Factors

575

Table IV. Re sults of Re gression Analyses for Each Model of Sexual Harassment (Beta Coefficients) Type of Organizational Sociocultural Sex-role Four-factor a harassme nt model model Spillover model model

GH USA SC

N/A

Be m score

Job status (pay)

GH USA SC

¯ .15c ¯ .28b ¯ .22

¯.21c ¯.21 ¯.09

Grie vance procedures

GH USA SC

.18b .26b .16b

N/A

Sex-ratio

GH USA SC

¯ .11 ¯ .03 ¯ .04

¯.09 ¯.04 .04

GH USA SC

.02 .08 .10

N/A

N/A

Privacy

.08 .09 .12

Unprof. e nvironment

GH USA SC

N/A

N/A

N/A

.52 c .25 c .12

GH USA SC

N/A

Sexist atmosphere

N/A c ¯.49c .36 ¯

Size of work group

GH USA SC

.18b .17b .15b

Trave l/overtime

GH USA SC

.09 .16b .10

Harasser has power to affect job status

GH USA SC

2

R a

GH USA SC

¯ .19 ¯ .11 ¯ .13

c

.14c .21c .11c

.08 .12 .11

.05 .11 .13

¯.12 ¯.10 .05

¯ .15 ¯ .09 .06

¯ .00 .09 .08

GH USA SC

N/A

Attractivene ss

c

b

N/A

N/A

¯ .17 .06 .03

b

N/A

¯ .07 ¯ .07 .08 ¯ .06 b ¯ .16 ¯ .10 .12 .20 b .10

¯ .10 .06 .05

N/A

¯ .51 c ¯ .33

¯ .34 c ¯ .31

N/A

N/A

N/A

N/A

N/A

N/A

N/A

N/A

¯.23 ¯.05 ¯.02

c

.11c .35c .17c

c

.06 c .29 c .13 c

c

.34 c .44 c .21 c

GH = Gende r harassme nt; USA = Unwanted sexual atte ntion; SC = Sexual coercion. p < .05. c p < .01. b

O’Hare an d O’Don oh ue

576

Se xual Coe rcion is the le ast common form of sexual harassm ent. Howe ve r, it is still e xpe rie nce d by a significant proportion of women (i.e ., 6-12% ). It should be note d that, of the women who re porte d expe rie ncing the se forms of harassme nt, ve ry fe w had e xpe rie nce d the se be haviors more than two time s. Thus, while the numbe rs sugge st that a substantial num be r of working wome n have e xpe rie nce d or will expe rience such be havior at some point in the ir care e rs, a much smalle r numbe r of wome n are re pe ate dly subje ct to such be havior. While this study sugge sts that sexual harassme nt is in fact a widespre ad proble m, the generalization of the se re sults to all working women may be limite d by the nature of the sample and the method of data colle ction. This sample was limite d to the faculty, staff, and stude nts in a large Midwe stern unive rsity. Howeve r, subje cts’ response s were not confine d to the ir expe rie nce in the unive rsity se tting, but cove re d the ir entire work history, thus incre asing the type s of work e xpe rie nces include d, and therefore , increasing the re pre sentative ness of the sample . Sam pling bias may also limit the ge ne ralizability of the se findings in that re sponse rate s for mail surve ys have be e n shown to be influe nce d both by the e ducational le ve l of subje cts and by the le ve l of inte re st in the subje ct matte r (Fowle r, 1988) . This bias may result in an inflation of pre vale nce rate s. Although the pre vale nce rate s re porte d he re are highe r than those re porte d by othe r re se arch in which rate s are typically close r to 50% (Fitzge rald et al., 1988; Gute k, 1985, USMSPB , 1981) , seve ral factors indicate that the se rate s are not ne cessarily unre asonably inflate d. Fitzge rald and Shullm an ( 1993) sugge st that while inflation of pre vale nce rate s may have be e n a proble m in e arly se xual harassme nt re se arch, re cent re se arch (which indicate s rate s ne ar 50% ) may in fact unde re stimate pre vale nce for se ve ral re asons including narrow or vague de finitions of se xual harassm e nt (e .g., Gute k, 1985; USMSPB , 1981) and inquiry limite d to harassm e nt in a curre nt job or narrow time pe riod (USMSPB , 1981) . In this study, se xual harassme nt was de fine d e xplicitly and compre he nsive ly, and wome n we re aske d to re port the ir e xpe rie nce of harassm ent over their entire work history. These differences may account for the highe r pre vale nce rate s.

Risk Factors The data from this study sugge st that the stronge st organizational risk factors are a lack of knowle dge about grievance proce dure s for sexual harassment, an unprofe ssional atmosphe re, and the e xistence of sexist attitude s in the workplace . The se characte ristics may facilitate sexual harassme nt by

Sexual Harass men t: Risk Factors

577

lowering the e xternal inhibitors in the environme nt. In work environme nts where males hold sexist be lie fs about wome n (i.e., the y believe that women are le ss capable or do not have the same rights as men), it may be the case that wome n are treate d as infe rior and their rights are disregarde d. Thus, be haviors that constitute sexual harassme nt are not inhibite d because the y are consiste nt with the belief systems he ld by male s in the workplace . Similarly, in an unprofe ssional atmosphe re (e.g., workers are expected to perform menial dutie s that are not a part of their job, the re is fre que nt swearing and drinking on the job) , a ge ne ral atmosphe re of disrespect is enge nde red thus facilitating sexual harassme nt. Caution should be use d in the inte rpre tation of the se correlations because of the re trospe ctive nature of the que stionnaire . Wome n were asked to recall having expe rience d a numbe r of be haviors that constitute sexual harassme nt and the n aske d to de scribe the workplace in which the behavior occurre d. It is possible that re calling these e xpe rie nces may have colore d the ir pe rceptions of the workplace in which the harassme nt occurred and the refore, cause d the m to de scribe the environme nt in a more negative light. Such a “ ne gative halo effe ct” may have re sulted in inflation of the se corre lations. Prospe ctive studie s are necessary to clarify the re lationship between the se variable s. Prospe ctive studie s and studie s of male s are also important to more fully e valuate all four factors in the model. The prese nt study did not fully test this mode l give n that Factor 2 (overcoming inte rnal inhibitors) was not examine d and only one possibility — se xual attraction — was e xamine d in Factor 1. A ske wed se x ratio in the workplace was not associate d with a highe r incide nce of se xual harassm e nt as e xpe cted. Rathe r, Gende r Harassment was moderate ly corre lated with more inte grate d e nvironme nts. The se re sults are contrary to the results of re se arch on the sex-role spillove r mode l which have shown highe r pre vale nce rate s in e nvironm e nts and in occupations whe re the se x ratio is ske we d. It may be that the gre ate r numbe rs of both me n and wome n in an inte grate d environm e nt create gre ate r opportu nity for the ir inte raction, and thus a highe r pote ntial for harassm e nt. Consiste nt with prior re se arch, incre ased opportunity for privacy during the workday did not predict any type of sexual harassme nt. Howe ve r, a lack of privacy was mode rate ly correlate d with Unwante d Sexual Attention. This finding is counte rintuitive and difficult to explain. Howeve r, it may support the hypothe sis that certain behaviors that constitute sexual harassme nt are not vie wed as harassing by many men. If men do not perceive their be havior as wrong, the y are not like ly to confine their actions to more private work areas.

578

O’Hare an d O’Don oh ue

The relationships found between the risk factors related to victim resistance and the incide nce of harassme nt are not consistent with the hypothe ses. Although low job status was moderate ly correlated with all three types of harassment in the bivariate analyses, it did not emerge as a significant predictor in the regression analyses. It was e xpected that wome n in low status jobs would be more vulne rable to harassme nt because they are less threatening to potential harasse rs than are women in higher status positions. In addition, because lower status positions typically require less skill, these women might have less job security making them more vulne rable to threats and promise s and less able to resist harassme nt. It may be that because most sexual harassme nt is carried out by co-worke rs rathe r than by supervisors (USMSPB, 1981), that job status is not an important determinant. Se x-role socialization was associate d with sexual harassme nt, howe ver, not in the manne r expe cted. It was hypothe size d that women who were sextyped as feminine on the BSRI (i.e., score d high on the Femininity Scale and low on the Masculinity Scale ) would expe rience more harassme nt of all types. However, the conve rse was true for Gende r Harassment such that women who were sex-type d as masculine , (i.e., scored high on the Masculinity Scale and low on the Femininity Scale ) reported more of this type of sexual harassme nt. In addition, high scores on the Femininity Scale (without regard to Masculinity Scale score) pre dicte d more Se xual Coe rcion. It may be that while ste reotypically feminine wome n make easier targe ts for harassment because of low assertiveness skills, women who fit the more masculine stereotype may create a threat to some men’s feelings of power in the workplace . The se women may also place the mselves in more male -dominate d jobs thus incre asing the risk of sexual harassme nt. Therefore, these women may be more like ly to be come the targe t of harassme nt that is motivate d by a nee d for power and control. Powell ( 1986) reported that women high in masculinity te nd to perceive certain behaviors (e .g. sexual comme nts meant to be insulting) as sexual harassme nt more than wome n low in masculinity. Because the se women are more like ly to conside r these behaviors inappropriate , they would be more like ly to remember and report having these experiences. At the same time, women who are not offended by the behavior and do not conside r it to constitute se xual harassme nt would be less like ly to recall and re port its occurrence. Finally, physical attractive ness was, contrary to expectation, not found to pre dict se xual harassme nt. This may be due to the measure of attractive ness used. Attractive ne ss was base d on women ’s se lf-ratings which may be syste matically affe cted by the e xpe rie nce of harassme nt. Sexual harassment often results in psychological proble ms (Loy and Ste wart, 1984; Gutek and Koss, 1993) which may be associate d with lowe red self-este em and excessive ly negative se lf-e valuations.

Sexual Harass men t: Risk Factors

579

This analysis of risk factors provide s valuable information for organizations and for individuals regarding the prevention of sexual harassme nt. Lack of knowle dge about the grievance procedures and formal policy on se xual harassment in an organization is a particularly important risk factor because it is one that can be easily addre ssed. By stating sexual harassme nt policie s and procedures explicitly and ensuring employe e awareness, an organization can increase the external inhibitors for harassme nt in the environment. A lack of professionalism in the work environme nt is anothe r risk factor that can be addresse d within an organization. Professional conduct should be modeled by those in supervisory positions, and expectations of what constitute s acceptable and unacce ptable behavior should be clearly stated and enforced. Sexist attitudes and beliefs among males in the workplace may not be easily changed, as it is difficult to change an individual ’s well-establishe d belief system. However, manage ment can make clear that acting on such belie fs is unacceptable, ille gal, and will be followe d by the appropriate consequence s.

REFERENCES Bem, S. L. (1974) . The measureme nt of psychological androgyny. J. Consult. Clin. Psychol. 42: 155-162. Cammae rt, L. (1985) . How widespre ad is se xual harassment on campus? Int. J. Wom en ’s Stud. 8: 399-397. Dansky, B. S., and Kilpatrick, D. G. (1997) . Effects of se xual harassment. In O ’Donohue, W. T. (ed.), Sexual Harassm ent: Theory, Research and Treatm ent, Allyn and Bacon, Boston, pp. 152-174. Ellis, S., Barak, A., and Pinto, A. (1991) . Mode rating e ffects of pe rsonal cognitions on expe rience d and perceive d sexual harassment of women at the workplace. J. Appl. Soc. Psychol. 21: 1320-1337. Far le y, L. ( 1978) . Sexu a l Sh aked o w n : Th e Sexu al Harassm en t o f Wo m en o n th e Jo b, McGraw-Hill, Ne w York. Finkelhor, D. (1984). Child Sexual Abuse: New Theory and Research, Fre e Press, New York. Fitzgerald, L. F., and Orme rod, A. J. (1991) . Pe rceptions of sexual harassment: The influence of gende r and acade mic context. Psychol. Wom en Quart. 15: 281-294. Fitzge rald, L. F. and Shullman, S. L. (1993). Sexual harassment: A research analysis and agenda for the 1990 ’s. J. Vocat. Behav. 42: 5-27. Fitzgerald, L. F., Shullman, S., Bailey, N., Richards, M., Swecker, J., Gold, Y., O rmerod, A. J., and Weitzman , L. (1988) . The incidence and dime nsions of sexual harassment in academia and the workplace. J. Voc. Behav. 32: 152-175. Fowler, F. J., Jr. (1988) . Survey Research Methods, Sage, Be verly Hills, CA. Gelfand, M. J., Fitzgerald, L. F., and Drasgow, F. (1993) . The structure of sexual harassment: A confirmatory analysis across cultures and settings. Under review. Grube r, J. E. (1990) . Methodological problems and policy implications in se xual harassment re search. Population Res. Polit. Rev. 9: 235-254. Grube r, J. E. (1997). An epidemiology of se xual harassment: Evidence from North American and Europe. In O ’Donohue, W. (e d.), Sexu al Harassm ent, Allyn and Bacon, Boston, pp. 152-174. Gutek B. A. (1985) Sex and the Workplace. Josse y-Bass, San Francisco.

580

O’Hare an d O’Don oh ue

Gutek, B. A., and Dunwoody, V . (1988) . Unde rstanding sex and the workplace. In Strombe rg, A. H., Larwood, L., and Gute k, B. A. (eds.), Wom en and Work: An Annual Review, V ol. 2, Sage, Ne wbury Park, CA. Gutek, B. A., and Koss, M. P. (1993). Changed women and changed conseque nce s of and coping with sexual harassme nt. J. Vocat. Behav. 42: 28-48. Gutek B. A., and Morasch, B. (1982) . Se x-ratios, se x-role spillover, and sexual harassment of wome n at work. J. Soc. Issues, 38: 55-74. Lafontiane, E ., and Tredeau, L. (1986) . The fre que ncy, source s, and corre lates of sexual harassme nt among women in traditional male occupations. Sex Roles, 15: 433-41. Loy, P. H., and Stewart, L. P. (1984) . The extent and effects of the se xual harassment of working women. Soc. Focus 17: 31-43. MacKinnon, C. (1979) Sexual Harassm ent of Working Wom en: A Case of Sex Discrimination, Yale Unive rsity Pre ss, New Haven, CT. Powe ll, G. N. (1986). Effects of se x role identity and sex on definitions of sexual harassme nt. Sex Roles, 14( 1,2) : 9-19. Rand, C. S., and Hall, J. A. (1983) . Se x differe nce s in the accuracy of se lf-perce ive d attractiveness. Soc. Psychol. Quart. 46: 359-363. Scheidner, K. T., and Swan, S. (1994). Job-related, psychological, and health-re lated outcomes of sexual harassme nt, Paper pre sente d at the Ninth Annual Confere nce of the Society of Industrial and Organizational Psychology, Nashville, TN. Tangri, S. S., Hayes, S. M. (1997). The ories of sexual harassment. In O ’Donohue, W. (e d.), Sexual Harassm ent, Allyn and Bacon, Boston, pp. 112-129. Tangri, S. S., Burt, M. R., and Johnson, L. B. (1982) . Sexual harassme nt at work: Thre e explanatory models. J. Soc. Issues 38: 33-54. Unite d States Me rit Syste ms Protection Board (1981) . Sexu al Harassm ent in the Federal Workplace: Is It a Problem ? U.S. Gove rnment Printing Office, Washington, DC. V aux, A. (1993) . Paradigmatic assumptions in se xual harassmen t re search: Being guide d without be ing misled. Special issue: Sexual harassment in the workplace. J. Vocat. Behav. 42: 116-135.

Archives of Sexual Behavior, Vol. 27, No. 6, 1998

Determ in an ts of Item Non respon se in a Large Nation al Sex Survey Em il K upek, Ph .D.1

The study aim was to describe pattern s of item nonrespon se in a sex survey and iden tify factors associated with high nonrespon se. A ran dom sam ple of 4548 respondents to the National Survey of Sexual Attitudes and Lifestyles in Britain, 1990-1991 was assessed. Both respondent-wise and variable-wise patterns of incom plete response are described in terms of distinctive characteristics of nonrespondents and questions with a high nonresponse rate. Logistic regression was used to m odel the determ in ants of extreme non respon se. For at least 90% of respon den ts, the item non response was ignorable for m ost practical purposes and recall difficu lties were indicated as its m ain cause, while the sensitivity of the qu estions had a large effect m ain ly am on g those in the top 5% of the overall non respon se distribu tion. The latter were also distinguish able in terms of refusals to answer som e face-to-face questions. Several overlapping indicators of poor educational and social background were associated with elevated overall nonrespon se, particularly am ong the top 5% of the distribu tion. Thus recall accuracy rather than sensitivity of some sex survey questions was the key difficulty for a m ajority of the respondents whose overall nonresponse was satisfactorily low. The sensitivity of the topic had a large effect on item nonrespon se only for a small grou p of participan ts. KEY WORDS: survey methodology; item nonresponse ; se x research; ge nder differences.

INTRODUCTION The pattern of item nonresponse is an important issue as statistical modelling usually assumes that incomplete answers are missing at random (Little and Rubin, 1987; Woolson et al., 1992). Identification of both questions and respondents who 1

Departamen to de Psicologia, Universidade Federal de Santa Catarina, Campus Unive rsitario, Trindade , 88.010-970 Florianopolis-SC, Brazil. 581 0004-0002/98/1200-058 1$15.00/0

Ó

1998 Plenum Publishing Corporation

582

Kup ek

do not conform to the designers ’ expectations of comple te response was the main aim of this work as it can help future questionnaire designs to maximize survey efficiency. An exploratory approach looking into main determinants of item nonresponse was applie d to the National Survey of Sexual Attitudes and Lifestyles data. The survey was unde rtake n in 1990-1991 with primary aims to provide reliable data for modelling the spread of HIV and for assisting in the development of he alth e ducation programs designe d to reach those whose life style put them most at risk of acquiring HIV and other sexually transmitte d infections (Wellings et al., 1990; Wadsworth et al., 1993; Johnson et al., 1994). A random sample of 18,876 men and women age d 16-59 living in Great Britain was interviewed and invite d to comple te a questionnaire . Face-to-face interviewing was used for less intrusive topics and a self-comple tion bookle t for more sensitive que stions. O ve rall re sponse rate was 63.3% which is highe r than in pre vious studie s of se xual be havior in Britain and some othe r countrie s and comparable to large social surve ys of le ss se nsitive topics. The acce ptance rate was 71.5% among e ligible re sponde nts who could be inte rvie wed. A compre he nsive re vie w of the surve y me thodology and all surve y que stions is pre sente d e lse whe re (Johnson et al., 1994) . Item nonre sponse was ge ne rally low, particularly be aring in mind the se nsitivity of the surve y topic for some participants . Slightly le ss than 4% of the re sponde nts did not comple te the se lf-comple tion bookle t. O f those who did, the ite m nonre sponse was le ss than 5% for most of the que stions. This is a ve ry good re sult in comparison with othe r se xual behavior surveys which use d a varie ty of methodology conducte d in B ritai n and abroad ( Johnson et al., 1994; Bre akwell and Fife -Shaw, 1992a,b; Mc Q ue e n and Campostrini, 1994).

SUBJECTS AND METHODS All 4,548 subje cts who answe re d the long que stionnair e with expande d se ction on se xual attitude s and life style were include d in this analysis. The sample for the long que stionnai re was chose n by a random se le ction proce dure ( Johnson et al., 1994) in the sampling de sign phase . The final sample size achie ve d was 18,876 re sponde nts, 4,548 of whom re sponde d to the long que stionnair e . The combine d weight ( house hold and re gional we ight product) that I use d in pre vious analysis was re -

Determ inan ts of Item Nonr es pon se

583

scale d to sum the sample size of the long que stionnair e responde nts. All re sults pre se nte d he re are base d on we ighte d analysis. Some variable s were derive d from a range of que stions: social class, marital status, e duc ation, sm oki ng, and alc oh ol consumption . The ir compone nt parts were not of interest themselves and therefore were not include d in the analysis. Filtering of the que stions affected the total numbe r of que stions pe r re sponde nt, e.g., those without partne rs could not have incom ple te re sponse on any ite m re garding se xual be havior. The numbe r of que stions include d in the analysis of incom ple te re sponse was 239. Five cate gorie s of incomple te response (IR) we re conside re d: ( i) missing i.e . left blank; (ii) “ don ’t know ” (DK) and/or “ no vie w” questionnaire option; (iii) “ can ’t remember” que stionnaire option; (iv) “ re fused to answer” questionnaire option; and (v) self-comple tion only: responde nt-de fined “ don ’t know ” or “ can ’t re membe r” base d on the ir writte n comme nts in the space provide d for the que stion. Total IR was calculate d as a sum of all IR categorie s de fine d above . O ne pe rson could have only one category of IR pe r que stion. Refuse d answe rs category was explicitly de fine d as such in response options for some que stions in face-to-face inte rvie w only. While it was possible that unwillingne ss to answe r othe r que stions was expre ssed through othe r type s of IR, it was the explicitne ss of the former that made it a spe cial cate gory. A ge ne ral prope nsity to nonre sponse may be more strongly indicate d for the responde nts who ope nly re fuse to answe r a que stion than for those who use indire ct ways of doing it. Logistic regre ssion was used to model the de te rminants of extreme nonre sponse (top 5% of IR distribution) . No automatic sele ction of the variable s in the mode l was use d but gradual e limination of the least significant one followe d by refitting the model. Maximum like lihood estimation and Wald ’s chi-square crite rion for removal of variable s from the mode l at fixe d alpha level 0.05 were used.

RESULTS Thre e aspe cts of IR are pre se nte d: ke y IR compone nts and the ir relations hip, distinc tive characte ristics of re sponde nts with re spe ct to ove rall IR, and a logistic mode l of factors associate d with ve ry high IR rate .

584

Kup ek Table I. Pe rcentile Values for Categories of Incomplete Response

a

Perce ntile Category Missing Design de fined Don ’t know Can ’t reme mbe r Re fused Respondent defined Don ’t know or can ’t re member All a

90th

95th

99th

Highest value

10

20

60

145

3 0 0

4 1 0

9 2 1

28 8 4

0 12

0 21

1 69

4 148

All values are rounded to the neare st integer. Table II. Refusals to Answe r and Missing/DK Re sponse

a

No. re fused

Base (n)

Miss. % 10+

DK % 4+

0 1 2 3¯ 4

4492 28 19 9

9 77 92 100

10 12 34 45

a

All value s are rounded to the ne arest integer.

K ey Com pon ents of In com plete Respon se Ninety percent of respondents had no more than 12 incomplete answers in all IR categories together. This is 5% of 239 which is maximum possible number of incomplete answers. Missing and to a lesser extent DK answers make up the bulk of the total IR (Table I). Ninety-one percent of all respondents had both fewer than ten missing values and fewer than four DK answers. On the other hand, only 2% of all respondents had both more than ten missing values and more than five DK answers. Between regions, overall IR rate varied from 13% in Northern and Northwest parts of the country, 12% in West Midlands, around 10% in Southe ast, Southwe st, Greater London, Wale s, and Scotland, to 6% in East Anglia and 5% in East Midlands. It is interesting to note that Greater London area which had highe st proportion of refused intervie ws among the regions (Johnson et al., 1994) achieve d a reasonably low item nonre sponse rate. Out of maximum four possible answers refused, only 1% exercised this option at all (see Table I). The latter also had more missing and DK answers in comparison with those who did not refuse to answer any of the questions. Although 9% of the respondents without a refusal had more than 10 missing value s, 77% of those who refused to answer 1 question and 92% of those with 2 refusals had more than 3 missing value s on other questions (Table II).

Determ inan ts of Item Nonr es pon se

585

In addition to refusals to answe r, responde nts ’ satisfaction with the inte rview may be measure d by their readine ss to be contacte d again about the inte rvie w. Among those who refuse d to give the te lephone numbe r to be contacte d again if necessary, 17% had more than 12 incomple te answers — twice as many as among those who gave the te le phone numbe r. Spearman rank-orde r correlation was calculated as a measure of association between the compone nts of incomplete response. The numbe r of missing value s was correlated with the number of “can’t remember” (.25), the number of refusals (.17), and the numbe r of DK answers (.15). The latter were also correlated with “can’t remember” answers (.17). All these correlations are low and thus seem to indicate different aspects of incomplete response. Distin ctive Character istics of Non respon den ts Lower educational leve l and social class were associate d with highe r proportion of incomple te answers (Table III). Non-White ethnic groups had highe r rate of ove rall nonre sponse than the White one . Situational factors during the interview such as the presence of pare nt(s) or other adult( s) were associate d with increased proportion of incomple te answe rs, particularly if they were pre sent all the time. Difficultie s with literacy and language also increased the numbe r of incomple te answe rs dramatically, as did serious embarrassme nt of the responde nts as judge d by their inte rviewers. O n the othe r hand, age , de clare d re ligion, and marital status showed little diffe rential in ove rall IR rate . V irtually no ge nde r difference s in overall IR rate were found. Question s with High Prop ortion of Incom plete An swers The variable s with highe st proportion of incomple te answe rs (Table IV ) can be classifie d unde r three main headings: lack of information; recall difficultie s; and sensitive issues. The social class of the responde nts is the only question with sizable IR rate which is clearly of factual nature and regarding present rather than past situation. It was derived from a series of questions about the type of work, qualifications necessary for it, duties and responsibilities at work, and housing conditions. Many responde nts were unable to provide detailed information of this type about themselves. As a result, 11% of the respondents could not be assigne d social class. Some of them might have been unwilling to provide these items as they might have resembled income status e nquirie s which many people feel sensitive about. However, for only 6 respondents was it impossible to assign a household social class which has been used in most analyse s to date.

586

Kup ek Table III. Distinctive Characteristics of Nonrespondents: Perce ntages of Cases Excee ding 12 IR (90th Perce ntile V alue) V ariable

%

Base (n)

Household social class (SCPR) I Professional II Intermediate III Skilled nonmanual III Skilled manual IV Semi-skilled V Unskilled Other

4 7 9 11 13 17 14

319 1266 940 886 446 134 530

Education A-level & above O-level/secondary None Other

6 9 13 11

1687 1568 1183 91

Ethnicity White Black Asian Others

9 17 34 23

4296 90 84 56

Young adult 16-21 No/not applicable All the time Some of the time No information

9 15 15 14

4197 49 64 121

Other adult 21+ No/not applicable All the time Some of the time No information

9 14 12 14

4078 112 114 121

Parent( s) No/not applicable All the time Some of the time No information

10 24 12 14

4179 46 64 121

9 91 27 15

4342 16 62 128

Demographic features

Presence of other pe rsons

Difficulties in understanding Language No problem Severe problems Some problems Not answered

(Continued)

Determ inan ts of Item Nonr es pon se

587 Table III. Continued

V ariable

%

Base (n)

Literacy No problem Severe problems Some problems Not answered

9 62 30 14

4280 28 103 137

Other No problem Severe problems Some problems Not answered

9 50 26 15

4184 27 213 124

8 73 24 11 15

3222 48 264 896 117

Adverse emotional re action Respondent embarrassed Not at all Ve ry Somewhat Only slightly Not answered

Many re sponde nts found it difficult to recall details about third- and second-to-last partner. The details included the dates of the beginning and end of each sexual partne rship, the age of the partne r and type of the relationship — all of which have be en previously re porte d as difficult to revive in responde nts ’ memory (Johnson et al., 1994) . The patte rn of not answe ring the whole block of re late d que stions was name d block non respon se. To summarize , thre e main patte rns of incomple te answers have be en found: 1. The large majority of the responde nts (90% ) had an ignorable proportion of ove rall IR which was be low 6% of the numbe r of questions asked. The que stions with highe st proportion of IR were of a factual nature and/or those which were like ly to cause re call difficultie s. The opinions on issue s which some people might have seen as controve rsial (e .g., abortion, divorce, AIDS risk for lesbians) had somewhat elevated IR rates but still below 10% . 2. The re sponde nts in the range of 91-95 percentile of the overall IR rate had a rate which was conside rably ele vate d mainly due to the re call difficultie s. The IR rate was highe r for the que stions on se cond- and thirdto-last partne r than for suppose dly more se nsitive personal que stions. 3. The re sponde nts in the top 5% of the IR distribution. Up to 10% of them had severe proble ms with unde rstanding the que stions prope rly due to the lack of language and/or literacy skills necessary for the comple tion of the que stionnaire . This small group generate d extremely high value s

a

O pinion on divorce Attitude towards abortion AIDS risk — female homosexuals No orgasm or not satisfying — men How do you feel about your life ?

3rd-to-last partne r — sex 2nd-to-last partner — sex

3rd-to-last partne r, last occ-ye ar 3rd-to-last partne r, last occ = 1st? 3rd-to-last partne r, 1st occ-yr 3rd-to-last partne r, age 1st occ 3rd-to-last partne r — relationship 2nd-to-last partner, last occ = 1st? 2nd-to-last partner, 1st occ-year 2nd-to-last partner — re lationship

4091

9 8 8 7 8

3 2

4 3 3 4 3 2 2 2

11 10 10

Pre marital sex — own view at 16 First inte rcourse for partner Pre marital sex — parents view when re sponde nt was 16

0-90 10

Q ue stion

Respondent’s social class

“Occasion” abbreviate d as “occ.”

Base (n)

Sensitivity of the issue

Skippe d

Recall difficulties

Recall difficulties or no inf.

No inf.

Suspecte d reason for IR

219

12 14 15 17 9

25 33

26 25 25 26 26 33 34 34

14 24 10

14

91-95

238

18 16 25 30 13

83 85

84 84 84 84 84 85 86 85

20 29 21

22

Top 5

Percentile value s

4548

10 9 9 8 8

8 8

9 9 9 9 9 8 8 8

11 11 10

11

All

Table IV. Overall Incomplete Response Rate (Respondents Grouped by Perce ntile V alues of Total Number of Incomple te Answe rs)

Determ inan ts of Item Nonr es pon se

589

of the IR distribution with the ir ove rall IR rate five to nine times that of the ave rage of the othe r re sponde nts. The que stions on se cond- and thirdmost-re cent partne r were particularly affe cted with IR rate of around 85% .

Mod ellin g Extrem e Overall Incom plete Respon se The top 5% of obse rve d ove rall IR rate was chosen as a suitable cut-off point for extreme nonre sponse with these data. This corresponds to having more than 20 IR ove rall. A particular type of block nonre sponse was fou nd whe n focusin g on a sm all group of top 5% re spon de nts (n = 239) regarding ove rall IR rate : (i) The que stions on most recent he terose xual partne r were not answe red by 84% of the above re sponde nts. (ii) About two thirds did not answe r the que stions about last contact with the partne r of the opposite se x (new or not, age , condom use, sexual practices) and about last 4 weeks (numbe r of partne rs and occasions, numbe r of new partne rs, condom use ). (iii) Around two thirds did not answer the que stions on pre gnancy atte mpts in last 6 months, profe ssional he lp with infe rtility, STD clinic atte ndance , drug abuse , re fraining from sex because of the fear of STD, having met anyone with HIV and being teste d for HIV. (iv) Almost 40% were wome n who did not answe r the questions on expe rience of miscarriage and abortion, and one quarte r were men who did not answe r the que stions on paying for se x with wome n. Because of the high corre lation be twe e n many of the above que stions, only two were se le cte d for the logistic mode l: the que stions on ye ar in which last sexual occasion took place and on the experie nce on miscarriage or stillbirth. The former actually marked the block nonre sponse to almost all que stions on se xual partne rs, while the latte r was spe cific for women. The variables initially entered in the logistic model with binary outcome — exceeding 20 incomple te answers or not — were all those presented in Table III, as we ll as the que stion on the ye ar of re sponde nt ’s most re cent se xual expe rie nce , the e xpe rie nce of miscarriage or stillbirth for women, whe the r any face -to-face que stion was re fuse d, whe the r phone numbe r was give n for furthe r contacts regarding the inte rview, and responde nts ’ age . Just for this analysis, the inde pe nde nt variable s in the mode l did not contribute to the total IR to asce rtain the re gre ssion assumption of inde pe nd e nt rand om var iable s. The in de pe nde nt var iable s ’ contribution would have be e n a ve ry small fraction (le ss than 2% ) of the total IR mass, so this simple solution of the above issue se emed justifiable . The final logistic models showed satisfactory fit by the te st of Hosmer and Lemeshow (1989) (probabilitie s of 0.11 and 0.77 for the men and the

590

Kup ek

wome n model, re spe ctive ly, thus disconfirming the hypothe sis of bad fit). The chi-square for log-like lihood test also showe d highly significant deviance re duc tion for both logisti c mode ls fitte d. The mode ls corre ctly identifie d the re sponde nts excee ding 20 incomple te answers at the rate of 93-99% . This performance was maintaine d across the probability range , i.e ., both for case s who were ve ry like ly and ve ry unlike ly to exceed the thre shold of 20 incomple te answe rs. Large standard e rrors of some parame ters were due to small numbe r of responde nts for some variable levels compare d to the othe r le ve ls. Despite large confide nce inte rvals, nonre sponse to the self-comple tion questions on the year of the most recent se xual occasion and refusal to answer at le ast one of the face-to-face inte rview que stions were both associate d with large increases in probability of being among top 5% regarding overall IR rate (Table V). Ethnicity, e ducation, and pre se nce of othe r adults during the interview were significant factors for e xtre me nonre sponse for men but not for wome n. The men who de clare d themselves as be longing to an ethnic group othe r than White were almost three times more like ly to have more than 20 incomple te answe rs ove rall than the White one s, but this was only marginally significant. Men with no educational qualifications were thre e times more like ly to excee d the twe nty incomple te answe rs than those with Ale ve l or highe r qualifications and almost five times more like ly to exce ed that thre shold in comparison with those with any othe r qualification. Not answering the que stion on expe rience of miscarriage and olde r age were associate d with e xtre me nonre sponse for wome n (Table V ). The latte r were on average 5 years olde r than othe r women re sponde nts.

DISCUSSION The lack of information see ms to be a re asonable explanation for incomple te answe rs to the general que stions such as those used to attribute social class. On the othe r hand, nonfactual que stions such as views, opinions, or attitude s may fe el sensitive for some responde nts even if the y do not ask about themselve s but are perceived as que stions evaluating social norms, e.g., views on pre marital se x, risk of AIDS infe ction for le sbians, opinion on divorce , attitude s towards abortion, and the importance of orgasm for men. “ How do you fe el about your life ? ” could have been se en as pe rsonal evaluation. Answe ring some que stions could be affe cte d by both re call difficultie s and true lack of information about past e ve nts (Nade au and Nie mi, 1995) . For e xample , whe the r the first inte rcourse for the re sponde nt was

At le ast one None

Non-White Unknown White

A-le vel/higher/degree O -level/CSE/other None

All the time Some of the time O ther forms of pre sence Not known No/not applicable

Not answered Answered fully

At le ast one None

Not answered Answered fully

Not answered Answered fully/n a

Refused to answer

Ethnicity

Education

Pare nt or other adult pre sent

Last sex-ye ar

Refused to answer

Last sex-ye ar

Miscarriage/stillbirth

a

Baseline.

Age

Leve l

V ariable

2513

116 2397

137 2376

35 2478

132 1903

78 87 174 56 1640

920 664 451

108 8 1919

21 2014

n

1.05

57.26 a 1.00

356.28 a 1.00

Women 86.51 a 1.00

535.16 a 1.00

20.66 3.40 7.15 3.75 1.00a

0.33 0.47 a 1.00

2.85 7.01 a 1.00

Me n 80.04 a 1.00

Adjusted O R

1.01

20.17

131.10

18.58

205.74

5.18 0.59 2.07 0.75

0.13 0.21

0.97 0.02

11.55

Lowe r

95% CI

1.09

162.43

968.19

402.72

999.00

82.34 19.62 24.63 18.84

0.85 1.02

8.38 999.00

554.78

Upper

Table V. Adjusted Odds Ratio (O R) and 95% Confidence Interval for Exceeding Twe nty Incomplete Answe rs for Me n and Women

592

Kup ek

also first inte rcourse for his/he r partne r is a factual que stion about a past eve nt. The que stions on re sponde nts ’ and the ir pare nts ’ approval of pre marital se x whe n the re sponde nt was 16 ye ars are not se eking ve rifiable answe rs and the re fore re quire d both trave ling back in time and re constructing some thing that might have ne ve r be e n e xpre sse d explicitly. The re sponse option “ Don ’t know/Can ’t re me mbe r” could have be en use d to acknowle dge this unce rtainty. The questions on se cond- and third-most-re cent partne r in the last 5 years are certainly personal but difficultie s in recalling events from that period could have also played a part. It is not the time elapsed alone that is necessarily the main factor in memory error but also the importance of the event or person involved which could strongly influence the recall accuracy (Krosnick, 1991; Friedman, 1993; Gaskell et al., 1993, Pickles et al., 1996). Previous partners might have been seen as less important than a conte mporary relationship which was more directly relevant to respondents’ lives. This view fits well with the notion that nonpersonal details of contemporary partner (e.g., about his/her job) were also often left unanswe red. Block nonre sponse to all the questions about second- and third-mostrecent partne r does not seem plausible from the perspective of distant events simply fading away as these were very different items. For example , it is difficult to believe that 8% of all responde nts could not remember even the sex of their second and third most recent partne r in the last 5 years. This again may indicate that for some responde nts the relevance of past partne rships to their contemporary life appe ared pretty remote. The block nonre sponse might have been partly due to the fact that these questions appe ared on the same page of the self-comple tion bookle t, so some of the responde nts could have skippe d that page . The logistic mode ls pointe d to the unwillingne ss to answe r the questions on last se xual partne r in the se lf-comple tion bookle t and at least one que stion on se x in face -to-face inte rview as ke y predictors of top 5% IR rate for both men and wome n. This indicate s that for a ve ry small numbe r of the re sponde nts, se x as a topic might have fe lt like too sensitive an issue to answe r about, particularly when it involve d disclosing information on last and often prese nt partne r. The que stion on miscarriage might have been perceived as too intrusive among a small numbe r of wome n with highest IR rate . Poor social background and presence of othe r adults during the inte rview were associate d with extreme nonre sponse for all responde nts but its significance could only be confirme d among men. Se nsitivity of some pe rsonal que stions doe s not seem to be the major factor in determining overall IR rate as ve ry few que stions of this type toppe d the list of variable s with high IR rate. This was found for all thre e main patte rns of incomple te response . Though high IR rate for the que stions

Determ inan ts of Item Nonr es pon se

593

of second- and third-most-re cent partne r among the top 5% cases of overall IR distribution might have been affe cted by the pe rsonal nature of these questions, it is still well above the IR rate for the question on having had an abortion. This indicate s the role of othe r factors influe ncing extreme nonresponse for a very small group of pe ople with serious language /literacy difficultie s or those who appe ared highly embarrasse d, as well as possible role of “ active forge tting ” about recent partne rs before the last one. The latte r re mains spe culative as no inde pendent measure of such mechanism can be derived from the que stionnaire . The type of re lationship with second- and third-most-re cent partne r could not cast any light on this issue as it was part of the same block nonre sponse . Howeve r, 12% of the responde nts with at least two partne rs reporte d a live -in relationship with second-most-rece nt partne r and the corresponding figure for third-most-re cent partne r was only half of it. This may indicate relatively small re levance of past partne rs for the conte mporary life of a large majority of responde nts without a live -in relationship in the past which is like ly to bind them to the ir partne rs more formally, e ithe r from the legal or from the social point of vie w (e .g., common frie nds) . The de terminants of incomple te response may be difficult to tackle in a large survey that is nationally representative in terms of ke y demographic characte ristics of the population, particularly if the topics are sensitive . O ve rsampling the strata where high refusal rate is expe cted, e.g., young highly mobile people in big cities (Johnson et al., 1994) , cannot be directly translate d for ite m nonre sponse . Spe cial que stionnaire module s can be de velope d for special groups of inte rest such as ethnic minoritie s. Inte rviewing me thods can also be adjuste d to particular circumstance s (e.g., self-comple tion allows more privacy but re quire s highe r language and literacy skills) . However, there are time and cost limits for any survey, both of which would be adve rsely affected by the dive rsification of survey module s and methods. Ethnographic methods have greater flexibility to adjust to the circumstances of individual responde nts but it is difficult to see how a large -scale survey can go far along those line s. However, focused research of a particular group of interest can be postfactum combined with survey findings providing both methods were designed in a way that makes it possible. Nevertheless, the vast majority of the respondents to this survey complied with the que stionnaire. This method still seems the most practical one for large -scale sex research.

REFERENCES Boulton, M. (1994) . The methodological imagination. In Boulton, M. (ed.), Challenge and Innovation: Methodological Advances in Social Research on HIV/AIDS, Taylor & Francis, London, pp. 1-21.

594

Kup ek

Bre akwell, G. M., and Fife-Shaw, C. R. (1992a) . Sexual activities and preferences in a United Kingdom sample of 16- to 20-year-olds. Arch. Sex. Behav. 21: 271-293. Bre akwell, G. M., and Fife-Shaw, C. R. (1992b) . Estimating sexual behaviour parameters in the light of AIDS: A review of rece nt UK studie s of young people . AIDS Care 4: 187-201. Bre akwell, G. M., and Fife-Shaw, C. R. (1994) . Using longitudinal cohort-sequential designs to study changes in sexual behaviour. In Boulton, M. (ed.), Challenge and Innovation: Methodological Advances in Social Research on HIV/AIDS, Taylor & Francis, London, pp. 25-38. Catania, J. A, Gibson, D. R., Chitwood, D. D., and Coates, D. J. (1990) . Methodological proble m s in AID S be havi oural re se arc h: Influe nce s of m e asu re m e nt e rror an d participation bias in studies of se xual be haviour. Psychol. Bull. 108: 339-362. Friedman, W. J. (1993). Me mory for the time of past eve nts. Psychol. Bull. 113: 44-66. Gaskell, G., Wright, D., and O ’Muirchaertaigh, C. (1993) . Reliability of surveys. Psychologist. 6: 500-503. Hosmer, D. W. Jr., and Leme show, S. (1989). Applied logistic regression, Wiley, New York. Johnson, A. M., Wadsworth, J., Elliot, P., Prior, S., Wallace, P., Blower, S., Webb, N. L., Heald, G. I., Miller, P. L., Adler, M., and Anderson, R. (1989) . A pilot study of sexual lifestyle in a random sample of the population of Gre at Britain. AIDS 3: 135-141. Johnson, A. M., Wadsworth, J., Wellings, K., Bradshaw, S. A., and Fie ld J. (1992). Sexual lifestyles and HIV risk. Nature 360: 410-413. Johnson, A. M., W adsworth, J., We llings, K., and Fie ld, J. ( 1994) . Sexu al Attitudes and Lifestyles, Blackwell, Oxford. Krosnick, J. A. (1991) . Response strate gies for coping with the cognitive demands of attitude measures in surveys. Appl. Cognit. Psychol. 5: 213-236. Little, R. J. A., and Rubin, D. B. (1987). Statistical Analysis with Missing Data, Wiley, New York. McQ uee n, D., and Campostrini, S. (1994) . Monitoring behavioural change in the population: A continuous data collection approach. In Boulton, M. (ed.), Challenge and innovation : Methodologica l advances in social research on HIV/AIDS, Taylor & Francis, London, pp. 39-56. Nadeau, R., and Nie mi, R. G. (1995). Educated guesses: The process of answering factual knowledge questions in surve ys. Public Opinion Quart. 59: 323-346. Pickles, A., Picke ring, K., and Taylor, C. (1996) . Reconciling recalled dates of developme ntal milestones, events and transitions: a mixed ge neralized linear model with random mean and variance functions. J. Roy. Statist. Soc. A. 159( Part 2) : 225-234. Wadsworth, J., Field, J., Johnson, A. M., Bradshaw, S., and Wellings K. (1993). Methodology of the National Survey of Sexual Attitudes and Lifestyle s. J. Roy. Stat. Soc. A. 156: 407-421. Wellings, K., Field, J., Wadsworth, J., Johnson, A. M., Anderson, R. M., and Bradshaw, S. A. (1990) . Sexual lifestyles under scrutiny. Nature 348: 276-278. Woolson, R., Clarke , W., and Lee per, J. (1992) . Missing data in longitudinal studies. In Dwayer, J. H., Feinlieb, M., Lippert, P., and Hoffme ister, H. (e ds.), Statistical Models for Longitudinal Studies of Health, Oxford University Press, Oxford, pp. 277-300

Archives of Sexual Behavior, Vol. 27, No. 6, 1998

Birth Order in Hom osexual Versus Heterosexual Sex Offenders Again st Ch ildren, Pubescents, an d Adults Ray Blan chard, Ph .D.,1,3 an d An thon y F. Bogaer t, Ph .D.2

Hom osexual m en erotically attracted to ph ysically m atu re partn ers typically have m ore older brothers than com parable heterosexual m en. We investigated whether late fraternal birth position s also occur in hom osexuals attracted to children or pubescents. Probands were 710 sex offenders from G ebhard et al.’s (1965) study of sexual offen ding. Hom osexual offenders against adu lts an d pu bescents had later fraternal birth po sitio ns than heterosexual offenders again st adu lts and pubescents, respectively; there was no difference between the hom osexual and heterosexual offenders against children. Results provide som e evidence that hom osexuality in m en attracted to im m atu re m ales is etiologically related to hom osexuality in m en attracted to m ature m ales. KEY WORDS: birth order; homosexuality; H-Y antigen; pedophilia; se xual offenders; sexual orientation; sibling sex ratio.

INTRODUCTION Rese arch spanning four de cade s and four countrie s has shown that homose xual men are , on ave rage , born late r than comparable he te rosexual men (Blanchard and Bogae rt, 1996a, 1996b; Blanchard and She ridan, 1992; Blanchard and Zucker, 1994; Blanchard et al., 1995, 1996, 1998; Hare and This study was supported in part by Standard Re search Grant 410-95-0003 from the Social Sciences and Humanities Research Council of Canada to Ray Blanchard. 1 Clarke Institute of Psychiatry, Toronto, Ontario, Canada; Departmen t of Psychiatry, Faculty of Medicine, University of Toronto. 2 Departme nt of Psychology, Brock Unive rsity, St. Catharines, Ontario, Canada. 3 To whom correspondence should be addressed at Clarke Institute of Psychiatry, 250 College Street, Toronto, Ontario M5T 1R8, Canada. 595 0004-0002/98/1200-059 5$15.00/0

Ó

1998 Plenum Publishing Corporation

596

Blanch ard an d Bogaert

Moran, 1979; Slate r, 1962; Zucker and Blanchard, 1994). Recent studie s have establishe d that this diffe rence in birth orde r is cause d by homose xual males having a gre ater numbe r of olde r brothe rs; they do not have a gre ater numbe r of olde r sisters, once the ir numbe r of olde r brothe rs has been take n into account (Blanchard and Bogae rt, 1996a, 1996b; Blanchard et al., 1998) . This phe nome non might the refore be describe d as the fraternal birth order effe ct. There are two re late d re asons for e xte nding this re search over the widest possible range of homose xual groups. The first reason is that the ubiquity of the fraternal birth orde r e ffect is one inde x of its probable importance as an e tiological factor in homose xuality. The second reason is that the presence of the fraternal birth orde r effect in dissimilar homose xual groups sugge sts that the sexual orientations of these groups have some etiological factor in common, despite their phenome nological differences. This logic is, of course, inferential rather than deductive . It is perfectly possible that one etiological factor produces a late birth position and homosexuality in one group; a comple tely different etiological factor produce s a late birth position and homose xuality in a second group; and yet another, distinct etiological factor produces a late birth position and homose xuality in a third group. The re is nothing in deductive logic to prohibit such a possibility. The a priori probability, however, of several comple tely different etiological factors, each of which — through pure coincidence — has identical effects on sexual orientation and identical effects on birth order, is rather less than the a priori probability of a single etiological factor that produce s the same effects in all groups of males, regardle ss of their psychological or demographic differences. Birth-orde r data, therefore, bear on the taxonomic relations of the various discriminable types of homose xual males. The homosexual groups that differ the most strongly from the typical homosexual man are those who de sire sex reassignment (homosexual transse xuals) and those who are sexually attracted to physically immature partne rs. The latter may be divided roughly into those most interested in prepube scent boys (pedophile s) and those most interested in pube scent boys (hebephile s). Research has found and confirme d that homosexual male-to-fe male transsexuals, despite their conviction that “inside ” they are really heterosexual women who have nothing in common with homose xual men, nevertheless exhibit the same high birth order observed in other homose xual groups (Blanchard and Sheridan, 1992; Blanchard et al., 1996). Less information is available regarding the other atypical groups, the homosexual pedophile s and hebe philes. One previous study has found that homose xual pedophiles are born later, relative to their brothers, than are heterosexual pedophiles (Bogaert et al., 1997), but it has not been replicated. No research has ever compared homose xual with heterosexual hebephile s on this variable.

Birth Order in Sex Offen ders

597

To addre ss this lack of information, the prese nt study inve stigate d frate rnal birth orde r, along with sibling se x ratio and pare ntal age , in men whose se xual historie s strongly sugge st homose xual or heterose xual pe dophilia, homose xual or he te rose xual he be philia, androphi lia ( the e rotic pre fe rence for physically mature men), or gyne philia (the erotic preference for physically mature women). The data came from those interviews originally use d by Ge bhard et al. ( 19 65 ) as the basis for the ir book, Sex Offenders.

METHOD Proban ds The Sex Offenders study e xamine d 1356 White males, most of whom had sexual offenses in one or more of 14 common cate gorie s (Gebhard et al., 1965, pp. 37-38) . The re were 721 men with offenses in one and only one of the six categorie s of prese nt inte rest and no offenses in any othe r cate gory. Our probands were the 710 of the se with comple te sibship data: 42 homose xual offende rs against childre n (i.e ., against boys younge r than 12) , 69 homose xual offende rs against pube scents (boys 12-15) , 156 homosexual offe nde rs against adults (men ove r 16) , 143 he te rosexual offenders against childre n (girls younge r than 12) , 127 he te rosexual offenders against pube scents (girls 12-15) , and 173 he te rose xual offe nde rs against adults (wome n ove r 16) . The e xtant information on offe nse history indicate s only that a man had at least one offe nse in a give n cate gory; the actual numbe r of offenses could not be re cove red. Thus, a heterose xual offender against childre n, for example , might have had only one or many offe nse s against prepube scent girls (and no offe nse in any othe r category) . A special note of clarification is require d regarding the homose xual offe nde rs against adults. Virtually all the ir offe nse s involve d engaging in, or soliciting, sexual activity with a consenting partne r. The behavior was ille gal eithe r because it occurred in a public place or else in a jurisdiction whe re all homose xual be havior was prohibit e d ( Gebhard et al., 19 65 , pp. 352-355) . The mean age of the probands was 33.61 years, (SD = 13.01). Their mean educational level was 9.01 grade s comple ted (SD = 3.75). Because this study used archive d data over 30 years old, colle cted from subje cts whose identitie s are never disclose d to outside researchers by the Kinsey Institute , issues regarding informe d consent did not apply.

598

Blanch ard an d Bogaert

Materials an d Proced ure The raw data for the Sex Offenders study are on-line at The Kinsey Institute for Rese arch in Sex, Gende r, and Reproduction (Bloomington, IN), whe re they were analyze d according to our instructions. In the se data, the two variable s, the mother’s and fathe r’s age s whe n the proband was born, always re fe r to the proband ’s biological pare nts; howeve r, the variable s, proband ’s numbe r of olde r brothe rs, olde r sisters, younge r brothe rs, and younge r siste rs, include half, step, and full siblings.

RESULTS Table I shows the mean numbe r of siblings pe r proband (excluding the proband himself) for e ach of the groups. The se diffe red in all re levant comparisons, significantly so for the offende rs against adults. Note that the p value s for these comparisons and for all othe rs in this article are twotaile d. The foregoing findings re quire d us to compute birth-orde r measures that would not be confounde d with sibship size. The global birth-orde r inde x was the numbe r of siblings older than the proband minus the numbe r of siblings younge r than the proband. Thus, the second row of data in Table I shows that the ave rage homose xual offender against adults had 0.51 more olde r than younge r siblings. In contrast, the average heterosexual offender against adults had ¯0.21 more olde r than younger siblings — or, in other words, 0.21 more younge r than olde r siblings. Two similar measures separately quantified the proband ’s birth order relative to his brothe rs and relative to his sisters. The fraternal birth-order inde x was the numbe r of brothe rs olde r than the proband minus the number of brothers younge r than the proband, and the sororal birth order index was the numbe r of sisters olde r than the proband minus the number of sisters younger than the proband. The results of comparing homose xual ve rsus he terose xual offe nders against childre n, homose xual versus he terose xual offenders against pube scents, and homose xual ve rsus heterosexual offende rs against adults, on the three birth-orde r measure s are give n in Table I. This table also pre sents the findings for pare ntal age at the proband ’s birth; the degrees of freedom for comparisons of pare ntal age vary, because some probands lacke d this information. The homose xual offe nde rs against adults and pube scents had highe r global and frate rnal birth-orde r inde xes than the he te rosexual offe nde rs against adults and pube scents. No othe r comparison was statistically significant, exce pt for the previously mentione d diffe rence in sibship size between the homose xual and heterosexual offende rs against adults.

_

3.80 0.75 0.43 0.32 33.34 28.29

3.43 0.48 0.14 0.33 33.93 28.74

O ffenders against pubesce nts No. of siblings Global birth order Fraternal birth order Sororal birth order Fathe r’s age at birth Mother ’s age at birth

O ffenders against children No. of siblings Global birth order Fraternal birth order Sororal birth order Fathe r’s age at birth Mother ’s age at birth 3.12 1.99 1.86 1.48 8.55 7.13

3.50 3.71 1.89 2.28 9.81 7.97

2.38 2.59 1.60 1.52 8.27 7.54

SD

Homose xual

2.57 0.51 0.37 0.14 32.85 28.44

x

O ffenders against adults No. of siblings Global birth order Fraternal birth order Sororal birth order Fathe r’s age at birth Mother ’s age at birth

Comparison

4.28 0.06 0.05 0.01 32.28 27.85

4.68

¯0.47 ¯0.22 ¯0.25 33.51 27.61

3.82

¯0.21 ¯0.18 ¯0.03 32.17 27.01

2.98 3.61 2.21 2.01 8.77 7.84

3.02 3.62 2.13 2.15 10.10 7.69

2.87 3.07 1.89 1.68 9.03 7.81

SD

He te rosexual

x

_

¯1.61 0.72 0.25 0.98 0.90 0.60

¯1.84 2.24 2.14 1.74 ¯0.10 0.53

¯4.25 2.27 2.81 0.96 0.62 1.51

t

183 183 183 183 131 142

194 194 194 194 149 153

327 327 327 327 251 264

df

ns ns ns ns ns ns

ns 0.03 0.03 ns ns ns

< 0.0001 0.02 0.005 ns ns ns

p

Table I. Comparisons of Homosexual Versus Heterosexual Offenders on Number of Siblings, Birth Order, and Parental Age at the Proband’s Birth

600

Blanch ard an d Bogaert

The variable, Sibling Sex Ratio, is most commonly reported as the ratio of brothers to sisters colle ctively reported by a give n group of probands. In White populations, the ratio of male live births to female live births is close to 106:100 (Chahnazarian, 1988; James, 1987). In the computation of inferential statistics, this value is more conve niently expressed as the proportion of males rather than the ratio of males to females, that is, 0.5146 (106/206). In this study, the sibling sex ratios (proportions) for homosex ual offenders against adults, pubesce nts, and children, and heterosexual offenders against adults, pubesce nts, and children, respectively, were 101 (0.5012), 96 (0.4885), 118 (0.5417), 106 (0.5136), 120 (0.5455), and 97 (0.4918). The se data were compared with the general population value (0.5146) using the z approximation to the binomial test. None of them differed significantly from the expected value .

DISCUSSION The homose xual offende rs against adults were born later among the ir brothe rs than were the he te rosexual offe nde rs against adults. This re sult confirms findings obtaine d with large groups of homose xual and he te rosexual volunte e rs closely matche d on de mographic parame te rs (Blanchard and Bogae rt, 1996a, 1996b) and re inforce s the conclusion that androphilic men have a highe r than e xpe cted frate rnal birth orde r. The novel finding of this study is that the homosexual offenders against pubescents were born later among their brothers than were the heterosexual offenders against pubescents, indicating that homosexual hebephiles also have a high fraternal birth order. It is not likely that this result arose primarily from an admixture of androphiles among the homosexual offenders against pubescents, despite the classification of these men on the basis of minimal information. There is very little incentive for an androphile to approach a pubescent sexually, for two reasons: (i) Adult male sexual partners are much more readily obtained than pubescent male sexual partners — in bars, bath houses, highway rest stops, parks, washrooms, and so on. This was likely just as true in the 1940s, when interviewing of the present probands commenced, as it is today. Thus, it is unlikely that a man who actually prefers physically mature homosexual partners would approach a pubescent purely on grounds of availability. (ii) Even in jurisdictions where all homosexual activity is prohibited, the legal consequences of approaching a pubescent sexually are liable to be much more severe than those of approaching an adult male. Thus, there is little reason for a man to risk approaching a pubescent rather than an adult unless he has a definite preference for the immature physique . In summary, the great majority of our homose xual offenders against pubescents probably were hebephiles, and it seems reasonable, absent contrary findings, to generalize our conclusions to that population.

Birth Order in Sex Offen ders

601

In this study, there were no birth-orde r diffe rence s be tween the homose xual and he te rosexual offende rs against childre n. Thus, we did not confirm the finding that homose xual pedophile s have a highe r frate rnal birth orde r than heterose xual pe dophile s (Bogae rt et al., 1997) . It is possible that our ne gative findings for offende rs against childre n do relate to the pre viously mentione d, main methodological limitation of this study, name ly, the minimal amount of obje ctive information on the proband ’s sexual history. A man whose convicte d offe nse s involve d only males may also have committed an equal or e ve n greate r numbe r of unde te cted offenses against female s, and vice versa. In practice , such case s are far more like ly to arise among the offe nde rs against childre n, be cause such men are much more prone to approach victims of both sexe s than are offenders against pube scents or adults. The re fore, the study may have faile d to find a diffe rence in frate rnal birth orde r be twe e n homose xual and he te rose xual offe nde rs against childre n merely because the se xual preference of the se men was le ss re liably classifie d on the basis of the ir convicte d offe nse s. Thus, furthe r the ore tical research should focus on this group, ideally in studie s with more e xtensive obje ctive information on the proband ’s sexual history. It should be stressed that any final conclusion that homose xual hebe philia, homose xual pe dophili a, or both are e tiologically relate d to androphilia would not imply that ordinary homose xual men (androphile s) are like ly to molest boys, any more than the conclusion that he te rosexual hebe philia or pe dophilia are re late d to gyne philia would imply that ordinary heterose xual men (gyne phile s) are like ly to mole st girls. A s in pre vio us stud ie s ( B lanc hard and B ogae rt, 199 6a, 19 96b; Blanchard and She ridan, 1992; see also Blanchard and Bogae rt, 1997), the negative findings for pare ntal age de monstrate that the high birth orders obse rve d in homose xual men are not mere ly artifacts of greate r mate rnal or pate rnal age . The ne gative findings for sibling sex ratio in all three homose xual groups continue the patte rn of results indicating that markedly high sibling se x ratios are reliably found in e xtre mely fe minine homose xual groups (Blanchard and She ridan, 1992; Blanchard et al., 1995, 1996; Zucker et al., 1997) but not in homose xual groups unse le cted for that characte ristic ( Blancha rd and Bogae rt, 1996a, 1996b; Blanchar d and Z ucke r, 1994; Blanchard et al., 1998; Z ucke r and Blanchard, 1994) . O ur re sults for both the offende rs against adults and the offenders against pube scents confirm the finding that sexual orientation is unrelated to sororal birth orde r (Blanchard and Bogae rt, 1996a, 1996b; Blanchard et al., 1998; Bogae rt et al., 1997) . The se re sults re strict the range of possible the orie s of the birth-orde r phe nomenon to those that can explain not only why olde r brothe rs incre ase the probability of homose xuality in later-born males but also why olde r siste rs neithe r enhance this effect nor counte ract

602

Blanch ard an d Bogaert

it. At pre se nt, only one publishe d explanation of the birth-orde r phenomenon explicitly predicts no relation betwee n siste rs and homose xuality. That is the hypothe sis that this phe nomenon reflects the progre ssive immunization of some mothers to Y-linke d minor histocompatibility antige ns (H-Y antige n) by e ach succe eding male fetus, and the concomitantly increasing effe cts of H-Y antibodie s on the sexual diffe re ntiation of the brain in each succeeding male fe tus (Blanchard, 1997; Blanchard and Bogae rt, 1996b; Blanchard and Klassen, 1997). O f course, othe r theories, psychosocial or biological, that could account for the ne gative findings regarding sisters as well as the positive findings regarding brothe rs, might be formulate d.

ACKNOWLEDGMENTS The authors thank Thomas G. Albright, John Bancroft, and Paul H. Gebhard, of the Kinsey Institute for Research in Sex, Gender, and Reproduction, for their ge ne rous assistance .

REFERENCES Blanchard, R. (1997) . Birth order and sibling se x ratio in homosexual ve rsus heterosexual males and female s. Ann. Rev. Sex Res. 8: 27-67. Blanchard, R., and Bogaert, A. F. (1996a) . Biodemographic comparisons of homose xual and heterose xual men in the Kinsey intervie w data. Arch. Sex. Behav. 25: 551-579. Blanchard, R., and Bogaert, A. F. (1996b) . Homosexuality in men and numbe r of older brothers. Am. J. Psychiat. 153: 27-31. Blanchard, R., and Bogae rt, A. F. (1997) . Drs. Blanchard and Bogaert reply [Lette r to the editor]. Am. J. Psychiat. 154: 137. Blanchard, R., and Klassen, P. (1997). H-Y antigen and homosexuality in men. J. Theoret. Biol. 185: 373-378. Blanchard, R., and Sheridan, P. M. (1992) . Sibship size, sibling se x ratio, birth order, and parental age in homose xual and nonhomosexual gender dysphorics. J. Nerv. Ment. Dis. 180: 40-47. Blanchard, R., and Zucker, K. J. (1994) . Reanalysis of Bell, Weinberg, and Hamme rsmith’s data on birth order, sibling sex ratio, and pare ntal age in homosexual me n. Am. J. Psychiat. 151: 1375-1376. Blanchard, R., Zucker, K. J., Bradley, S. J., and Hume, C. S. (1995). Birth order and sibling sex ratio in homosexual male adolesce nts and probably prehomose xual feminine boys. Dev. Psychol. 31: 22-30. Blanchard, R., Zucker, K. J., Cohen-Kettenis, P. T., Gooren, L. J. G., and Baile y, J. M. (1996) . Birth order and sibling se x ratio in two samples of Dutch ge nder-dysphoric homosexual males. Arch. Sex. Behav. 25: 495-514. Blanchard, R., Zucker, K. J., Siegelman, M., Dickey, R., and Klassen, P. (1998) . The re lation of birth order to sexual orientation in me n and women. J. Biosoc. Sci. 30: 511-519. B ogae rt, A. F., B e ze au, S., Kuban, M., and B lanchard, R. ( 1997) . Pe dophilia, se xu al orientation, and birth orde r. J. Abn. Psychol. 106: 331-335. Chahnazarian, A. (1988) . De terminants of the se x ratio at birth: Review of recent literature . Soc. Biol. 35: 214-235.

Birth Order in Sex Offen ders

603

Gebhard, P. H., Gagnon, J. H., Pomeroy, W. B., and Christenson, C. V. (1965). Sex Offenders: An Analysis of Types, Harper & Row, New York. Hare, E. H., and Moran, P. A. P. (1979) . Pare ntal age and birth order in homosexual patients: A replication of Slater’s study. Br. J. Psychiat. 134: 178-182. James, W. H. (1987) . The human sex ratio. Part 1: A re view of the literature. Hum . Biol. 59: 721-752. Slater, E. (1962). Birth orde r and maternal age of homose xuals. Lancet 1: 69-71. Zucke r, K. J., and Blanchard, R. (1994) . Reanalysis of Bieber et al.’s 1962 data on sibling sex ratio and birth order in male homosexuals. J. Nerv. Ment. Dis. 182: 528-530. Zucke r, K. J., Gre en, R., Coate s, S., Z uge r, B., Cohen-Kettenis, P. T., Z ecca, G. M., Lertora, V., Money, J., Hahn-Burke, S., Bradley, S. J., and Blanchard, R. (1997) . Sibling sex ratio of boys with gender identity disorder. J. Child Psychol. Psychiat. 38: 543-551.

Archives of Sexual Behavior, Vol. 27, No. 6, 1998

Ch an ges in the Sexual Orien tation of Six Heterosexual Male-to-Female Tran ssexuals Christop her T. Daskalos , Ph.D.1,2

Researchers traditionally have assum ed that sex reassignm ent procedures do not ch an ge sexu al orien tatio n. Of 20 tran ssexuals of variou s types that were interviewed, 6 heterosexual male-to-fem ale transsexual respondents reported that their sexual orientation had changed since transitioning from male to female. These respondents stated that before transitioning they had been sexually orientated towards females. After transitioning, these same respondents reported that they were sexually orientated towards males. Five of the six respondents reported having variou s sexual encou nters with m ales sin ce tran sition in g. The respon den ts explained the changes in their sexual orientation as part of their emerging female gend er id entities. Three of the respon den ts claim ed that the use of female horm ones played a role in changing their sexual orientation. It did not appear that the respondents ’ post-transitional sexual attractions towards m ales were sim ilar to autogynephilic images and fantasies described by Blanchard (1991). KEY WORDS: transsexualism; se xual orientation; gender dysphoria; retrospective study.

INTRODUCTION Sexual orientation, a fundamental aspect of personality, is conside red to be the result of prenatal or early life events by many researchers (Bell et al., 1981; Whitam and Mathy, 1986; Mone y, 1988; Diamond and Sigmundson, 1997). Once determined, sexual orientation is thought to be difficult, if not impossible , to change. Sexual orientation is recognize d as an important demarcation among transsexuals, implying fundame ntal behavioral and de velopme ntal 1 2

Departme nt of Sociology, Arizona State University, Te mpe, Arizona. To whom correspondence should be addressed at 218 Woodland Parkway, # 245, San Marcos, California 92069. 605 0004-0002/98/1200-060 5$15.00/0

Ó

1998 Plenum Publishing Corporation

606

Daskalos

differences among them (Blanchard et al., 1987). Most researchers on transsexualism assume that sexual orientation remains consistent throughout the sex reassignment process. Although postoperative transsexuals may experience dramatic change s in anatomy, personal relationships, and social status, their sexual orientation is not presumed to change (Benjamin, 1966; Blanchard, 1985; Blanchard et al., 1987; Diamond and Sigmundson, 1997). Relying on retrospective accounts, this pape r describe s the change s in the sexual orientation of six male-to-fe male transsexuals during the sex reassignme nt process. Such changes hav e rarely been investigated (Tully, 1992) and present a challenge to current understandings of sexual orientation. The details of these changes in sexual orientation and the respondents ’ explanations for such change s are reported below.

METHODS In 1996, in-de pth inte rviews were obtaine d from a nonclinical, nonrandom sample of 20 transse xuals in a Southwe ste rn city. The participants inte rviewed live d full-time as membe rs of the opposite sex and were in the proce ss of obtaining or had alre ady obtaine d se x reassignme nt surge ry. Half of the responde nts were postope rative at the time of the study. The responde nts were assigne d to two groups, homose xual and nonhomose xual, base d upon the ir original sexual orie ntation as sugge ste d by Blanchard (1989) and the ir biological sex (e.g. homose xual male-to-fe male transse xual) . The nonhomose xual group containe d 13 subje cts, 12 hete rosexual male-to-fe male re sponde nts and 1 heterose xual fe male-to-male responde nt. The homose xual group containe d 7 subje cts, 4 homose xual male-to-fe male responde nts and 3 homose xual fe male -to-male responde nts. The respondents were aske d to discuss, among othe r things, the ir se xual orientation and romantic relationships. The inte rvie ws we re re corded, with the subje cts’ permission, and transcribe d. The subje cts’ remarks below, using pseudonyms, are the results of those retrospe ctive inte rvie ws. The re searche r also atte nde d numerous meetings of The Wild Flowe rs, a local gender support group, during which the obse rvations note d below were made . THE RESPONDENTS’ ROMANTIC RELATIONSHIPS While 14 of the 20 transse xual re sponde nts reported no change s in their se xual orie ntation during the ir transition from one sex to anothe r, 6 responde nts reported that their sexual orientation had change d during this time. These 6 responde nts were heterosexual male-to-female transsexuals; 4 were postope rative and 2 were preoperative at the time of the study. Before

Ch anges in Sexu al Orien tation of Tran ss exu als

607

their transitions from male to female, the se subje cts remembered having be en sexually oriented towards females. None of the 6 conside red themselves to have been gay or reporte d any episode s of homose xual behavior. Five of the 6 had been married to heterosexual women. One was the biological fathe r to three childre n. After transitioning, these same subje cts reported being sexually oriented towards males. Of these 6 responde nts, only 2 we re involve d in a sexual relationship with a male at the time of the study. Anothe r 3 responde nts reporte d that the y had been involve d in se xual re lationships with male s since transition ing. O nly 1 of the 6 he te rose xual male-to-female responde nts who had change d their sexual orientation reported never having any sexual contact since transitioning. Of the two heterosexual male -to-female responde nts who were sexually involve d with men at the time of the study, one was married to a heterosexual cross-dresser and the othe r had be en dating a hete rosexual man for 5 months. Jane , a postope rative transse xual, had been married to Richard for about 6 ye ars at the time she was interviewed. The y had me t at a meeting of a social support group for heterosexual cross-dre ssers. Richard, like Jane at the time , was a he terose xual cross-dre sser and Jane related that initially she and Richard were just friends. When Jane e ventually transitione d into a woman, Richard ’s friendship deve lope d into romantic interest. Becky had met her boyfrie nd Cliff during a church service. Becky, who was pre operative at the time of the study, said that Cliff had agre ed to wait until she unde rwent surge ry before they had any sexual contact other than kissing. Becky hope d to one day marry Cliff but doubte d such a marriage would have the blessings of their church. Three othe r he te rosexual male -to-female responde nts had been involve d in romantic re lationships with hete rosexual males since transitioning although the y were not so involve d at the time of the study. O ne postoperative re sponde nt named She rri had met Scott at a meeting of a social support group for heterosexual cross-dresse rs. She rri and Scott became romantically involve d afte r She rri began to live as a woman full time. Before Sherri had unde rgone se x reassignme nt surge ry, the ir se xual contact was limited to pe tting and fondling. After surge ry She rri found that he r fe elings for Scott had change d, due to his continue d une mployme nt and slove nly pe rsonal habits, and she terminate d their re lationship. Sherri emphasize d that he r sexual de sires for male s remaine d strong de spite this break-up. Sherri stated that she wante d to find a man to marry and take to her 20th high school class re union. Luann, who had been postope rative for less than a year at the time of the study, had date d seve ral heterosexual men from he r colle ge while she was still preope rative . She said that the men she had date d did not know of her transse xual status and because of this she did not allow the relationships to deve lop. Patty, preoperative at the time of

608

Daskalos

the study, had date d a heterosexual man she had met at he r place of employme nt. The man had been unaware of Patty’s transsexual status when they started dating. Patty said that because of her preoperative status, the sexual contact between the two had be e n limite d to kissing and petting. Patty said that the man broke off the romantic aspe ct of their relationship after Patty disclose d her transse xual status to him. One postope rative he terose xual male-to-female responde nt whose sexual orientation change d from woman to man reporte d that she had not been involve d in any se xual activitie s with men since transitioning. Penny related that, as a he terose xual man, he r inte rest in sexual activity with wome n had been rathe r low until she reached her mid-20s. Penny said that during this pe riod of he r life she “ shut down ” sexually and felt almost no sexual inte rests. She attribute d her low interest in se xuality to he r gender dysphoria. Penny remaine d shut down se xually until her 40s, when Penny began to transition into a woman. Pe nny state d that she began to notice that her sexual attractions, such as they were, were for men rather than for women. Penny stated she would like to someday be involve d in a heterosexual relationship with a man but had no immediate plans to become so involve d. THE RESPONDENTS’ EXPLANATIONS FOR THE CHANGES IN THEIR SEXUAL ORIENTATION

Pretran sition al Orien tation to Fem ales Five of the 6 responde nts, whose sexual orie ntation had change d, reporte d that the ir pretransitional sexual orientation towards fe males was an atte mpt to conform to tradition al masculine ge nde r role expe ctations. She rri ’s e xplanation for he r pre vious se xual orie ntation went as follows: No, I was always attracted to women [be fore transitioning]. This is where you can make the mind fit the situation. The mind is a wonderful thing, you can make it work or not work. Just be cause of the way society is, if you look like a man you have to act like a man or you are totally ostracized. If you are really a man in quote unquote normal society, you do the male thing; acting like a male includes be ing with a female.

Patty, who did not have much se xual e xpe rie nce as a man, married to appear “ normal ” : “ I married to be normal; you know I thought maybe this would help me be normal. She was a good frie nd and we like d each othe r.” Jane also marrie d to conform to traditional ge nde r expectations: “ I tried to do what society said I was suppose d to do. So I tried and I went through marriage and the whole nine yards and finally it just didn ’t work.”

Ch anges in Sexu al Orien tation of Tran ss exu als

609

Becky relate d that he r sexual orientation towards women was a conce ssion to traditional gende r role expe ctations: I got out of the military service at the age of 23 and I re alize d that I had only had two intimate relationships at that time; both had bee n with women due to the peer pressure from my family and socie ty and eve rything. Both of them were total failures. The y we re failures because certain parts of me didn ’t work so good. I got married, another one of my attempts to be quote normal.

Penny re membe re d he r he te rosexual relationships as an atte mpt to fit in:

I don’t know, I was just trying to be like eve ryone else, fit in, going through the same things that everybody e lse does and not doing too well with the wome n. I mean I had a satisfying sexual relationship but there was always some thing wrong in the back of my mind that se emed to up the re lationship.

O ne re sponde nt, Luann, explaine d he r orientation towards fe male s had little to do with sexuality as she state d: I think that I was genuinely attracte d to women in my earlie r ye ars for certain reasons. There we re certain aspects of women that I was attracted to. . .. I do understand why I was attracte d to women before; it wasn ’t for sex like in a typical male female relationship, it was for companionship and for the close ne ss and friendship. Those are the aspects of it that attracte d me to women. I wasn ’t ready to jump my wife ’s bones every time I had the chance , I wasn ’t like that.

Luann ’s orie ntation towards females was an e xpre ssion of he r emerging fe minine gende r ide ntity which is why she sought the m as companions and frie nds rathe r than as se xual obje cts. According to five of the 6 he te rosexual male-to-fe male responde nts whose se xual orie ntation had change d, the ir sexual orie ntation for fe male s were atte mpts to conform to traditional masculine ge nde r role expe ctations. O ne of the 6 such re sponde nts state d that her orie ntation for fe male s was an e xpre ssion of her emerging feminine ge nde r identity and had little to do with se xuality.

Posttran sition al Orien tation Towar ds Males The 6 responde nts e xplaine d that the ir posttransitional sexual orientation to male s repre se nted an aspe ct of the ir female gende r identitie s which the y could not e xpre ss or did not know existe d be fore transitioning. All of the se re sponde nts reporte d be ing sexually oriente d towards fe males be fore transitioning and five had be en married to heterose xual wome n. None of the se responde nts had ide ntifie d as be ing gay or reporte d having any homose xual expe rience s be fore transitioning. Becky comme nted on her sexual orientation afte r transitioning:

610

Daskalos Over a period of time I ’ve realized that I must be with a man, that is what is most comfortable for me. I re alize that the very few re lationships that I have had have be en with women. I love my ex-wife but I’m not in love with my ex-wife. Cliff is a different story. I’m in love with him, I have felt things that I have never felt be fore. I know that I can love a man and love him better than any woman he ’s ever been with.

Jane ’s se xual orie ntation be gan to change as he r ge nde r change d: I went from be ing completely heterosexual as a man to completely he te rosexual as a woman and I was very comfortable with both sides when I was in those roles. As I progressed on the gender side [of her personality] I expe rimented with the se xuality side to find out where I was comfortable. As I was acce pted at various points [in her gender transition] I became comfortable at that point sexually so my se xuality kind of rode with my ge nder. . .. Maybe I was bisexual to start with although I question whether I could be comfortable with a woman. At this point I don’t know.

As Jane transitione d from male to female , she be gan to acknowle dge her sexual orie ntation to males. When She rri be gan to transition into a woman she was able to let he r sexual orientation to males emerge, in part due to her use of female hormone s, discusse d in furthe r de tail below. As Sherri relate d: They [the hormone s] gave me the emotions to go with the feelings and let the true emotions come out because be fore [the transition] I wasn ’t attracte d to guys but then again how much of this was society? I couldn ’t be attracted to a guy when I was a gu y be cau se that would m ake m e gay. Ho w mu ch of that made m e heterosexual as a man? . . . The way that I look at myse lf now is that I am a normal hete rosexual female with slight lesbian tendencies. Right now if a very handsome man and a very attractive lady walked up, I would take the guy. Part of that choice is socie ty. I know, it’s “ normal.” That ’s what society dictate s as normal and I want to be a normal wife so I am attracted to me n.

Penny, whose sexual inte re st had always be en rathe r low, transitione d with assumption that she would re main se xually attracte d to female s. Pe nny found that he r sexual orientation had change d after transitioning and that she was now attracte d to male s. Two of the 6 subje cts re membe re d having had passing sexual attractions to male s before transitioning but had neve r acte d upon the se de sires and had trie d to suppre ss such attractions. After the se responde nts had transitione d into wome n, the y were able to acknowle dge their se xual orientation to male s. Luann recalle d having sexual thoughts involving males before transitioning: There were time s before I transitioned when I thought I was maybe gay because I was actually having some thoughts of other guys like in high school and that. Until I realized I wasn ’t having gay thoughts, I was having male to female thoughts but I was thinking of being with the guys as a female.

Ch anges in Sexu al Orien tation of Tran ss exu als

611

These se xual attractions, Luann eventually re alize d, were an aspe ct of her emerging feminine ge nde r identity. Luann fe lt unable to express these sexual attractions until she had transitione d: I was also attracte d to men but I he ld back, I kept the reins on that kind of e motion from coming out. You just make yourse lf not do that. I had a couple of e xperiences where I never acted on them [sexual orientation to males]. But they occupie d my thoughts quite a bit. You tend to kee p that in check and I think that I would consider myself bisexual in a way. Maybe I use d to be bisexual more and now that I have bee n able to straighten my life out [through se x reassignme nt] I am able to let those emotions [sexual feelings for males] come out.

Luann could not expre ss he r sexual orie ntation for males, which she considered an aspe ct of he r fe male ge nde r ide ntity, until she had transitione d into a woman. Patty said that her se xual orie ntation to male s did not be gin in earne st until she began to transition: I never thought about guys be fore I started this whole thing [transsexualism]. I may have had thoughts about it but I immediate ly stuffed them. But then again I never had a lot of lustful thoughts about women either. But when I transitioned, that change d. I met this guy last year and we fooled around a little bit.

Patty did not consciously acknowle dge he r sexual orie ntation for males until she had transitione d. As men, these 6 re sponde nts fe lt that the y had be en unable to expre ss or were unaware of the e xiste nce of the ir sexual orientation to males. As the ir female ge nde r identitie s emerged, the responde nts were able to acknowle dge and e xpre ss this se xual orientation.

THE ROLE OF HORMONES IN THE CHANGES IN SEXUAL ORIENTATION Three of the he te rosexual male-to-fe male responde nts thought that the use of female hormone s contribute d to their change in sexual orientation. Fe male horm one s are powe rful drugs and the ir effe cts upon the psychose xual functioning of male-to-fe male transse xuals are not well understood (Cohe n-Kette nis and Gooren, 1992). Prese ntly it is assumed by researche rs that the inge stion of fe male hormone s by male -to-fe male transse xuals doe s not affe ct the ir sexual orie ntation (Mone y and Tucke r, 1975; Cohe n-Kette nis and Goore n, 1992) . Three of the re sponde nts in this study, however, thought that the use of fe male hormone s playe d a role in the change in their se xual orientation. Luann re late d the following regarding the matte r of hormone s and sexual orie ntation: I don ’t think I would be as relaxed be ing with a man without the hormone s. Before I ever took hormones, like when I was just starting out and say I was in a make out se ssion with a guy, I just couldn’t ge t comfortable enough to do that. I think

612

Daskalos that hormone s just get you ever so much more in synch with yourself. They make you feel much more in line.

Sherri, as stated above, thought that the use of female hormone s provide d emotional depth to her feelings, allowing her “ true emotions ” (her sexual orientation) toward males to “ come out” . Patty stated that her use of female hormones had a “big impact” upon her, making it possible for her to be intimate with a male. These three responde nts thought that the use of female hormones playe d a role in changing their sexual orientation. Much remains to be discove red regarding the effects of hormone s upon the psychose xual functioning of transse xual subje cts, especially concerning their role in shaping sexual desires. Besides causing change s in physical appearance , the use of female hormone s by male-to-female transsexuals has been reported to reduce tension, increase feelings of relaxation, and induce a calming effect upon emotional disturbance s (Cohen-Kettenis and Gooren, 1992). Many of these effects were reported by the homose xual and heterosexual male-to-female responde nts in this study. Some heterosexual male-tofemale transse xuals reporte d a decrease in their sexual desire but no such decrease was reported by any of the homose xual male -to-female responde nts. There were no reports by the other 17 subje cts of any change s concerning sexual orientation due to, at least in part, the use of hormone s. The widespread use and extensive knowle dge of female hormone s on the part of the heterosexual male-to-female subje cts confounds attempts to unde rstand their role in shaping the sexual orientation of this group of respondents. The brand names and pharmacological effects of various female hormones were well-known and frequent topics of conve rsation at meetings of The Wild Flowe rs, a local ge nde r group. Also discusse d among the transsexuals in attendance at The Wild Flowers meetings were the illicit avenues available for obtaining specific hormone s as well as strategies for persuading doctors to prescribe certain brands of hormone s. At one meeting of The Wild Flowers, two preope rative transsexuals exchange d a number of hormone tablets as each was dissatisfied with the ir own prescription. O ne respondent, Esther, talke d of the networks for hormone information exchange and procure ment that she establishe d while a me mber of The Wild Flowe rs. Esthe r stated that the majority of her female hormone s were purchased from a source other than a pharmacy. She mentione d that Mexico continued to be a ready source of unpre scribed hormones despite the United States’ efforts to restrict their flow across the border. From the interviews and obse rvations, it appeare d that female hormones, their effects and ways to obtain them, were an important part of the lives of the heterosexual male-to-female responde nts. Any attempt to unde rstand the role that hormones play in changing the sexual orientation of he terosexual male -to-female transse xuals must consider their extensive and often illicit use by such persons.

Ch anges in Sexu al Orien tation of Tran ss exu als

613

AUTOGYNEPHILIA AND THE CHANGES IN SEXUAL ORIENTATION Blanchard (1989) has sugge ste d that autogyne philia (erotic arousal in men relate d to the image of the mselve s as women) is close ly associate d with nonhomose xual ge nde r dysphoria. While autogyne philia may have playe d an influe ntial role in the ir live s, the re sponde nts in this study provide d no re ason to believe that autogyne philia, as currently conceptualize d, was re sponsible for the change s in se xual orie ntation de scribed above . The change s in se xual orie ntation describe d by the re sponde nts went beyond autogyne philic fantasie s and image ry and involve d posttransitional sexual attractions towards males who had value as sexual and romantic partne rs in their own right. Becky relate d the following conce rning the first time she met he r boyfrie nd Cliff: “ I saw this ve ry good looking man, tall, handsome , great voice . I was kind of in a daze . I was transfixe d by this guy! He was good looking, he had a nice build, he smelled good. I didn ’t know the re were fe elings like that. I neve r had them before.” She rri ’s attraction to male s he ld both se xual and pe rsonal significance : [Making a se xual reference] I know something that would thrill me to death! I’m just be ing nasty now [laughs]. I consider myself a 51-year-old virgin and part of my burning desire is to be happily married and have a loving, caring husband who loves me for who I am and who cares about me. And I want to be able to return his love with ge ntleness and tenderness. I think it’s every girl’s dre am to have a happy, married life, to have a nice husband. I re ally want that.

The responde nts expre ssed de sires for men as se xual and romantic partne rs that went be yond the autogyne philic thoughts and image s de scribe d by Blanchard (1991). Furthe rmore, five of the 6 re sponde nts who re porte d a change in the ir se xual orientation had actually acte d upon their posttransitional se xual attractions towards males to various de grees. Two of the se five were involve d in long-te rm romantic re lationships with males at the time of the study. In summary, it did not appe ar that the change s in the responde nts ’ sexual orie ntation were similar to the autogyne philic image s and fantasie s describe d by Blanchard (1991) .

REFERENCES B e ll, A. P ., W e inbe rg, M. S., and H am me rsm ith, S. K. ( 1981) . Sexu al Preference: Its Developm ent in Men and Wom en, Indiana University Press, Bloomington, IN. Benjamin, H. (1966). The Transsexual Phenom enon, Julian, New York. Blanchard, R. (1985) . Typology of male-to-female transse xualism. Arch . Sex. Beh av. 14 : 247-261. Blanchard, R. (1989). The classification and labe ling of nonhomosexual gende r dysphorias. Arch. Sex. Behav. 18: 315-334.

614

Daskalos

Blanchard, R. (1991). Clinical observations and systematic studies of autogynephilia. J. Sex Marital Ther. 17: 235-251. Blanchard, R., Clemmensen, L. H., and Steiner, B. W. (1987) . He te rosexual and homosexual gende r dysphoria. Arch. Sex. Behav. 16: 139-152. Cohen-Kette nis, P. T., and Goore n, L. J. G. (1992) . The influence of hormone treatment on the psychological functioning of transsexuals. In Bockting, W. O ., and Coleman, E. (eds.), G ender Dysphoria: Interdisciplinary Approaches in Clinical Managem ent, Haworth, New York. Diamond, M., and Sigmundson, K. (1997). Sex reassignme nt at birth: Long te rm re view and clinical implications. Arch. Pediat. Adolescent Med. 151: 298-304. Money, J. (1988) . G ay, Straight, and In-Between: The Sexology of Erotic Orientation, O xford University Press: New York. Money, J., and Tucke r, P. (1975) . Sexual Signatures: On Being a Man or a Wom an, Little, Brown, Boston. Tully, B. (1992). Accounting For Transsexualism and Transhom osexuality, Whiting and Birch, London. W hitam, F. L., and Mathy, R. M. (1986) . Male Hom osexuality in Fou r Societies: Brazil, G uatem ala, the Philippines, and the United States, Praeger, New York

Archives of Sexual Behavior, Vol. 27, No. 6, 1998

BOOK REVIEWS An n ual Review of Sex Research (Vol. 6). Edited by Raym on d C. Rosen. Socie ty for the Scie ntific Study of Se x, Mt. V ernon, Iowa, 1995, 244 pp., $35.00. Reviewed by Scott L. Hershberger, Ph.D.

1

The sixth volume of the Annu al Review of Sex Research provide s re views of are as in se x re se arch which are of curre nt inte re st. What is most striking about this colle ction of se ven article s is what they have to say about the breadth of re search activity taking place in sexual scie nce today. As attested to by the se article s, se xual scie nce re ache s across psychology, sociology, medicine , biology, and a host of othe r discipline s, making it truly an interdisciplinary scie nce. In fact, the re is no discipline that exists in which se x research is not of some re levancy. In the ope ning e ditorial comme nt, Rose n write s that the major task for the e ditors was to produce a “ balance d overview of diffe re nt the ore tical and methodological approache s to sexuality research.” It is hard to conceive how this goal could have been bette r accomplishe d. Also heartening is the appe arance throughout the article s of several traditionally neglected themes. I mention just two: possible evolutionary e xplanations and a re ne wed inte re st in female sexuality. In the first article , Fagot re vie ws re se arch examining gende r role developme nt in childre n. Gender role developme nt involve s the process by which childre n acquire knowle dge concerning their own sex, the behaviors that are identifie d with one sex more than anothe r, and the differences in roles males and females are expected in play in society. Research is reviewed fr om th re e p re dom in ate the ore tic al pe rs pe c tive s: social le arning, cognitive -deve lopme ntal, and schema theory. The social learning point of vie w propose s that ge nde r role de ve lopme nt primarily occurs through environme ntal shaping; that is, childre n are “ socialize d” into spe cific sex role s. 1

Departme nt of Psychology, California State University, Long Be ach, California 90840. 615 0004-0002/98/1200-061 5$15.00/0

Ó

1998 Plenum Publishing Corporation

616

Book Reviews

On the other hand, cognitive -developme ntal theories view developme nt as occurring through a predetermined, invariant maturational sequence. Fagot contrasts the social learning and cognitive -developme ntal perspectives, which vie w childre n as passive participants in the ir deve lopme nt, with sche ma the ory, which characte rize s childr e n as active con structors of the ir developme nt. Gende r schema the orists argue that although childre n are socialize d into specific sex role s, how this information is organize d by childre n varie s. In many respects, Fagot ’s article is a traditional review of this often confusing area. For example , Fagot doe s the expected and distinguishe s from the very beginning “ gende r” from “ sex,” and explains her use of the terms “ sex typing, ” “ se x role , ” and “ ge nde r role .” E ve n though the author acknowle dge s he r own theore tical biase s, this doe s not appe ar to have compromise d her review. This is a balance d, thoughtful, thorough, and use ful re vie w. Given the space constraints, it is unde rstandable that biological approache s to unde rstanding gende r developme nt are not covered. Laan and E veraerd’s article concerns the importance of both physiological and psychological determinants of female sexual arousal. The basic argument advanced by the authors is that unde rstanding genital arousal alone is inade quate in unde rstanding sexual arousal in female s. Subje ctive sexual arousal, measured, for example , by asking subje cts to rate their genital sensations or to report the ir mood, is conside red as important an indicator of sexual arousal as “ obje ctive ” indicators of sexual arousal, obtaine d usually through vaginal photoplethysmography. The authors support this conclusion by de scribing a numbe r of studies, in many of which they have been directly involved. Several explanations are give n for why subjective experience appears to be more important in women than men. For example , perhaps the male erection acts as a kine sthetic cue that facilitate s recognition of sexual arousal; such kine sthetic cues are less available to women and thus subje ctive experie nces are more necessary to the recognition of arousal. Nonetheless, although Laan and Everaerd firmly state that “ It is clear that genital arousal is a poor predictor of subje ctive sexual experience ” (p. 69), the evide nce reviewed, and the studie s conducte d by the authors themselves, often contradict this. Although there is little doubt that subje ctive sexual arousal is important in women and less important in men, genital arousal, as physiologically measured, still appears to be the best predictor of sexual arousal in both sexes. Methodological proble ms and possible solutions in sex survey research are discusse d by Catania, Binson, van de r Strate n, and Stone . In encyclope dic fashion, the authors revie w validity issues re late d to self-reporte d sexual behavior, the pretesting of surve ys, indice s of measure ment error, re sponde nt influe nce s on m e asure me nt e rror ( which include difficultie s with the

Book Reviews

617

re sponde nt ’s me mory, se lf-pre se ntation bias, motivation, and race and cultural issues), instrume nt influe nces on measurement error (which include issues of question wording, order, and context), mode effects (i.e., how the survey is administe red), and inte rviewer effects. Throughout, the authors identify a numbe r of areas where much additional research is needed. It is inte re sting to note , howe ve r, that in are as whe re re se arch has be e n conducte d, the results sugge st that sex surveys are not as vulne rable to some of these effects as previously thought. For instance , much has been made of using terminology responde nts can unde rstand (e.g., “ vaginal intercourse ” is a term some may find ambiguous), but, as it turns out, most people appe ar to unde rstand what the most common terms appe aring on sex surveys mean. Researchers will find this article a very valuable resource. Baile y and Pillard present the evide nce that human sexual orie ntation is at least in part gene tically influe nced. Evide nce is re viewed primarily from behavioral genetic studie s, which include twin and family studie s, although the re cent mole cular ge netic work by Hamer and othe rs is also discusse d. As this article convincingly shows, sexual orie ntation is largely a biological, if not ge ne tic, phe nome non. This is not to say that many of the extant studie s are not without flaws. But the conflue nce of evide nce is overwhe lming. For this re ason, future ge ne tic studie s of se xual orie ntation should not be designe d solely to demonstrate heritability. Rather, future studie s should be designe d to elucidate the biological or ge netic processes involve d. The authors provide a fe w e xample s of how eve n simple twin studie s can be use d to do this. For e xample , gre ater ge nde r nonconformity in childhood could be re late d to a gre ater ge ne tic te nde ncy toward homosexuality (Meyer-Bahlburg, 1993). This hypothe sis could be confirme d by finding that homose xual probands with homose xual co-twins are more like ly to exhibit gre ate r ge nde r nonconformity in childhood than homose xual probands with nonhomose xual co-twins. Pain atte nuation in fe males through ge nital stimulation is the subje ct of Komisaruk and Whipple ’s article . Emphasis is place d on wome n with spinal cord injurie s. Through de scribing a numbe r of studie s the y and others have done , Komisaruk and Whipple convincin gly de monstrate the re liabilit y of the phe nome non in both nonhum ans and humans. The authors ’ programmatic rese arch in this are a has be en impressive, leading to the discove ry and de scription of the neural pathways responsible . In many respects, this article is the most technical in the volume , although the writing is still clear and compre hensible to the novice in this are a. In my opinion, this is se xual scie nce at its be st.

618

Book Reviews

Sipski and Ale xande r provide a comple m e ntary article on the conse que nc e s of spinal c ord inju rie s on the se xu al functioning of wom e n. As the au thors point out, most of what is kn own about the e ffe cts of spinal cord injury on se xual func tionin g is spe cific to me n, with issue s spe cific to wome n havin g be e n re lative ly ignore d un til re ce ntly. W hat little that was known te nde d to portr ay a rath e r ble ak picture of se xuality following spinal cord in jur y, with de cre ase s in se xual de sire and satisfaction and incre ase s in physic al impairm e nt ofte n d e s cribe d . A lth ough S ips ki and A le xa nd e r agr e e tha t t he con se que nce s of spinal cord injury can be de vas tatin g, much can be done to ame lior ate the se e ffe cts throu gh e ducation an d cou nse ling. For that re ason alone , the re ce nt work de scribe d by the se author s is valuable . In ad dition , this work has also le d to in sigh ts into the se xual function of fe m ale s withou t injury. Habituati on of se xual arousal is the subje ct of O ve r and Koukounas ’s article . In the ir re vie w and discussion, the authors are committe d to making a distinction be twe e n “ product ” orie nte d research, which is concerne d mainly with showing that habituat ion to se xual stimuli follows the patte rns one would e xpe ct ( e .g., the more fre que ntly a stimulus is pre sente d, the more rapid the habituat ion) , and “ proce ss ” orie nte d research, which is conce rne d with why habituat ion occurs. The authors favor a proce ss-orie nte d approac h to re se arch in this are a, with particular e mphasis plac e d on the the ory that habituat ion occurs through change s in inform ation proce ssing ove r time . The re is much value in a proce ss-orie nte d approac h. For one , it he lps e xplain why sexual arousal doe s not always habitua te . Se cond, it le ads to the inve stigation of response s corre late d with, and possibly re sponsible for, the habitua tion. O ve r and Koukounas sugge st a num be r of inte re sting are as for future inve stigatio n, including individu al diffe re nce s in susce ptibility to sexual arousal habitua tion. In conclusion, I cannot re commend this volume highly enough to anyone wishing a fair and re pre sentative ove rview of se xual scie nce in the late 1990s. The article s provide e xce lle nt introductions to their re spe ctive are as, and should be re ad by all individuals with a serious intere st in se x research.

REFERENCE Me ye r-Bahlburg, H. F. L. (1993) . Psychobiologic research on homose xuality. Child Adolescent Psychiat. Clin. N. Am. 2: 489-500.

Book Reviews

619

Pr ofe s s in g Fem in is m : Cau tion ar y Tales fr om th e Str an ge Wor ld of Wom en ’s Stu dies. By Daphn e Patai an d Noretta Koertge. Basic Books, New York, 1994, 235 pp., $24.00. Reviewed by Lenore Manderson, Ph.D.

2

Women’s studie s programs date from the late 1960s/e arly 1970s, as interest in feminist theory and scholarship developed. There were clear advantages to the establishme nt of se parate programs to enable the consolidation of the field and to provide a mechanism to support its developme nt in re search and te aching. In much the same way, are a studies departments — centers of Asian studie s, Latin American studie s, and so on — se rve d this function also (and continue to do so); gay and le sbian studies programs — relatively few — similarly have provide d an intellectual focus for research and teaching in sexuality; and, in the past decade, cultural studies programs and departments have facilitate d the developme nt of this emerging field. These institutional developments have been supporte d by othe r acade mic activitie s, including confe rences and seminars, scholarly journals and other publishing, and the establishme nt of professional associations. While the establishme nt of separate programs was useful to legitimize feminist scholarship and women’s studie s, it is important to ask, What value now is there of maintaining this separation? How can training in discrete acade mic discipl ine s be incorporat e d within an inte rdisciplinary study progra m? Unde r what crite ria should course s be offe re d by cognate departments for inclusion in women’s studie s majors? What is the relation be twe en the ory and praxis? And, to what extent have wome n ’s studie s program s be e n able to accommodate the the ore tical dive rsity that has emerge d in fe minist scholarship, and what are the implications of this pedagogically and in research? The proble m now, as ide ntifie d by Patai and Koertge, is the extent to which women ’s studie s programs have become sites of ideological battle , with scholarship antipathe tic to politics. Patai and Koe rtge provide a numbe r of example s of this, referred to acronymically and regarde d entire ly as negative de ve lopme nts: IDPO L (identity politics), FEMPED (fe minist pedagogy) , TO TALREJ (the uncompromising re je ction of patriarchal cultural he ritage ), and BIO DENIAL (the repudiation of the biological basis of sexual difference). The issues are trivialize d by the use of this mnemonic/he uristic device, as they are by the flip subtitle of the book. 2

Tropical He alth Program, Australian Ce ntre for International and Tropical Health and Nutrition, The University of Queensland, Herston Road, Herston, Brisbane, Q uee nsland 4006, Australia.

620

Book Reviews

Patai and Koertge argue that some acade mic staff and some students have been militant in their insistence on the necessary integration of theory and practice , and in their lack of tolerance of different world views — for example , by assuming a homoge nous or “ correct” theoretical line (hence, “ political corre ctness” ). This is a position that can be only partially supporte d, since feminist theorists have — as their theory has become increasingly sophisticated — diversified considerably. But Patai and Koertge hold that there is no place within women’s studie s programs for students to weigh these philosophic shifts. Further, they argue that women’s studies programs are antagonistic to scholarship, in denying philosophic tradition and scholarly antecedents to contemporary feminist theory and empirical research. They argue that social constructionist theory has been so promote d as to preclude any biological basis to social life, to preclude any discussion of this as a possibility, and to assume canonical status rather than an analytic approach. Patai and Koertge go further to challe nge the validity of feminist models alternative to logical and positivist epistemologies and to treat somewhat sourly feminist critiques of “malestream ” thought. Hence, by implication, they dismiss French philosophical developments, articulated as “feminisms of difference ” (e.g., Irigary, Cixous, and Kristeva), which have had major impact on feminist theory in the past decade. Patai and Koertge are also critical of the pedagogic approaches within women’s studies programs, although there is no good reason to suppose that their teaching and learning alternatives to didacticism are unique. Indeed, the approach within some women’s studies programs appears to have much in common with problem-based learning and practicum approaches in medical schools. The authors are also critical of curricula, dismissing in this context the need for women ’s studies programs to draw boundaries and decide on the inclusion/exclusion of various courses as parts of a women’s studies program. The inclusion of any subject (Patai and Koertge ’s example is the sociology of the family), simply because it deals with or includes women, seems to me an inadequate rationale. Patai and Koertge are correct to point out the difficulties of identity politics in acade mia — a matter by no me ans unique to women ’s studies programs. It is, too, entirely legitimate to question the philosophical basis of any area of teaching and learning, and to question the degree to which the broad aims of scholarship are advanced through the institutionalization of particular approache s. I accept too the inhe rent proble ms when an acade mic program is built around ideological or philosophical commitment — this would be as true for Marxist theory as for fe minism, and hence for de partments of political economy as much as for women’s studie s. It is also true, of course, that ideological battles are part of the everyday intellectual sce nery in many humanitie s and social scie nces departments today, where their presence flags the intellectual liveline ss, rather than death, of their disciplines.

Book Reviews

621

De spite my own conce rn with ide ntity politics and ide ologic al adhe rence to particu lar line s of e nquiry and inte rpre tation, I am not convince d that this book is use ful in e lucidating the issue s. The proble m — apart from my difficulty with the some what lambastic style of the te xt — is one of e vide nce . Patai and Koe rtge are extre me ly critical of what the y pe rce ive to be de magogy in wome n ’s studie s. But it is bad scie nce to base the ir criticism on inte rvie ws with a se lf-se le cted population, to quote se le ctive ly from e -mail corre sponde nce to advance the ir argum e nts without acknowle dging alte rnative vie ws and prac tices, and to substan tiate the ir claim s with a fe w le ngthy quote s and case studie s. The book stands, the n, as the opinion of two disgrun tle d acade mics. The re is a ne e d for an obje ctive re vie w of wom en ’s studie s, as one hope s the re is, from time to time , in any fie ld of study. B ut this require s a more compre he nsive re vie w of the program s and the ir offe rings, a care ful and re liable surve y of acade mics and stude nts from those program s, and an analysis of the positio n of the fie ld and its politic al de ve lopm e nts in an historic and social conte xt. Since the scie nce ( and the art) of criticism is e ntire ly lacking in this volum e , it remains, inste ad, at the le ve l of opinion .

Em battled Eros: Sexu al Politics an d Eth ics in Con temporary America. By Steven Seid m an . R outle dge , Ne w Y ork, 1992 , 220 pp., $45.0 0 (hardback), $14.95 (pape rback) . Reviewed by David F. G reenberg, Ph.D.

3

In an earlie r book, Seidman (1990) chronicle d the de velopme nt of a distinctive sexual ethic in the Unite d State s. Ove r the course of the 19th century, fe ar of se xual activity (as manife sted, e .g., in fear of masturbation and in advice to limit the fre que ncy of marital coitus) diminishe d, and gre ater emphasis was place d on the positive role sex could play in expre ssing, preserving, and enhancing love betwee n spouse s. It was thought to make a positive contribution to mental he alth. Romantic love was sufficie nt to legitimate marital se x, inde pende ntly of procre ation.

3

Departme nt of Sociology, Ne w York University, New York, New York 10003.

622

Book Reviews

Enhanced by contrace ption, technique , and e xperimentation, sex was supposed to be satisfactory to both spouses. In contrast to Victorian sexual ideology, which tended to deny female sexual response altoge ther, and regarde d a woman ’s love as primarily spiritual, authors of sexual advice in the early 20th century gave a wife ’s sexual satisfaction as much importance as her husband ’s. Sex reformers hoped that if men found sex with their wives more satisfactory, they would be less like ly to seek a divorce , pursue extramarital affairs, patronize prostitute s, and acquire venereal diseases. Over the course of this century, the legitimation of sex through romantic love gradually extended to include premarital relationships between opposite -sex partners. Broad social and cultural change s in American socie ty, such as urbanization, ge ographical mobility, increased education, and employme nt for wome n helpe d make the se change s possible , but Seidman argue s that the efforts of se x re formers, including radicals like Emma Goldman and Max Eastman (and late r, Norman O. Brown and Herbe rt Marcuse ), and libe rals like Ben Lindse y, Margare t Sange r, Eustace Chesser, Alfre d Kinsey (and late r, Mary Calde rone , A le x Com fort, William Maste rs and V irginia Johnson, and David Reube n), along with e nte rtaine rs like Elvis Pre sley, and entreprene urs like Hugh Hefne r, were critical. Their e fforts made contrace ption available . Through the ir influe nce , pre marital se x and cohabitation, oral-ge nital contacts, anal inte rcourse , and group sex all be came more common and less stigmatize d. Erotically tinge d adve rtising be came pe rvasive . Film ce nsorshi p gradual ly broke down. Whereas the se x manuals of e arlie r generations were written explicitly for reade rs who were married or e ngage d, the ne w crop of se x advice manuals was written for and reache d a much wider audie nce . Sex aids be came commercially available . In this eroticize d culture , se x became value d for its own sake, even when not part of an ongoing intimate relationship. In some contexts, it could be a vehicle for se lf-fulfillme nt and ple asure , while serving as an expre ssion of e motional intimacy and love in othe rs. This ne w se xual culture contribute d much to the acceptance of homose xual relations. Propone nts and oppone nts of these change s saw them, not altoge ther accurate ly, as intimate ly linke d with other change s of the 1960s: generational conflict, fe minism, political radicalism, and the rise of a hippie counte rculture and a drug subculture . From the 1970s on, sexual libe ralization became increasingly controve rsial. The moral issues raise d in controve rsies ove r such issues as homose xuality, AIDS, sadomasochistic practice s, inte rgenerational se x, and pornography form the subje ct of Embattled Eros. Others have already chronicle d the “ sex wars” within feminism, but Seidman ’s treatment is broader.

Book Reviews

623

Writing in lucid, grace ful, and se nsible but highly repetitive prose , Seidman sugge sts that two broad configurations of sexual morality unde rlie the se de bate s. Libe rtarians typically se e se x as multivale nt, holding different, situationally base d meanings to actors. It may be sought and value d for a varie ty of reasons, and has a large capacity for contributing to psychological well-be ing and self-realization. So long as all partie s consent, no form of se xual activity should be banne d. Libe rtarians te nd to se e the oppone nts of libe rtarian se xuality as conse rvative s, whose re ligious convictions le ad the m to oppose se xual freedom. Se idman shows that this ove rsimplifie s a more comple x re ality. Criticism of libe rtarian positio ns has come from across the politic al spe ctrum , an d from a range of re ligi ou s com m itme nts and se xu al pre fe rence s. E ve n though conse rvative s have be en promine nt in the ir opposition to libe rtarian se xual philosophie s, libe rals and radicals have also faulte d them. What is common to much of the oppositio n, Se idman argue s, is romanticism. Se x, the romantics conte nd, is not just a ple asurable muscle spasm. It is ine vitably linke d to profound e motions. Whe re it is seve re d from love , whe n its conte xt is ple asure alone and not e motional intimacy as well, it le ads to re lationsh ips that are shallow and de pe rsonaliz ed, and to child abuse and abandonm e nt, rape , divorce , lone line ss, exploitation, and viral e pide mics. Promine nt fe minists have seen the libe rtarian posit ion as prom oting the kin d of se x that male s valu e , to the de triment of the te nde rne ss and affe ction that the y be lie ve wome n want in se x. Romantics do not re je ct se x; on the contrary, they value its pote ntial for contributing to human happine ss. Yet they are also aware that se x can le ad to unwante d pre gnancy, dise ase , alie nation, and viole nce . It can de stroy intim ate relationships as well as enhance the m. Conflicts between libe rt aria ns an d rom antics e m e rge d within le sbian-fe m in is m ove r sadom asoc hism ( S/M) , and among gay male s ove r “ promiscuity ” and man-boy love . Se idman ’s tre atme nt use fully de e pe ns our unde rstanding of some conte mporary se xual controve rsie s. He is e xe mplary in his atte mpt to be fair to all partie s. As a gay male , Se idman is sympathe tic to the libe rtarian position for he lping to make choice possible for se xual minoritie s. At the same time , he take s some of the romantics ’ criticisms se riously, recognizing that some of the ir conce rns may be le gitimate , eve n if exagge rate d or rhe torically linke d to large r social and political age ndas he doe s not share .

624

Book Reviews

I am not e ntirely satisfie d with Se idman ’s framing sexual controve rsies entire ly in te rms of the libe rtarian/roma ntic opposition. Se idman neve r mentions the Roman Catholic Church in his account of any of the sexual controve rsie s. The New Right is mentione d, but the role of Prote stant fundame ntalism in his account se e ms unde rstate d. Se idman give s “ public inte lle ctuals ” like George Gilde r, Midge Decter, and George Leonard a good de al of attention. Yet on some issues, such as gay rights, Decter’s (1980) unflatte ring depiction of gay male life on Fire Island seems less important politically than the organize d religious opposition. It is this opposition that has lobbie d legislators, mayors, gove rnors, and presidents, and campaigne d in favor of re fe re nda opposing gay rights. Seidman ’s emphasis, I think, ste ms from his being more inte reste d in exploring the moral argume nts raise d in the se controve rsies than in documenting political history. The bishop who oppose s gay rights and the distribution of condoms in the public schools on the grounds that the se contrave ne God ’s law doe s not raise argume nts that Seidman could take seriously enough to e ngage inte llectually. It is, of course, of some intere st that even the religious opposition draws quite ofte n on the argume nts of the romantics. How important have the debate s Seidman analyze s be en? In the case of AIDS policy, the y cle arly made a diffe rence . At least in the early ye ars of the epide mic, stereotype s and denunciations of the “ gay male life-style ” circulate d by the romantics promote d fear of gays, and enable d gove rnments to avoid funding research and treatme nt for those the romantics claime d had brought the ir illne ss on themselve s through their own sexual misconduct. At least within women ’s studie s programs and, pe rhaps, more wide ly, the critique of pornography advance d by cultural fe minists has won a following. O n the othe r hand, all reports indicate that wome n are responsible for a substantial and growing proportion of the market for commercially distribute d porn vide os. The lesbian-fe minist de bate s on S/M may not have matte red much to many. The influe nce of the se debate s on se xual conduct is less cle ar. Have books and e ssays e mphasizing the risks of casual sex reduced the numbe r of people going to single s bars, or the ir willingne ss to go home for the night with someone the y mee t there? We do not know. If the re has be en a be havioral shift, I sugge st that it has more to do with fear of he rpe s and AIDS (which Se idman mentions) than with the influe nce of Christophe r Lasch ’s books labe ling casual sex an immature flight from emotional intimacy.

Book Reviews

625

Seidman ’s final chapte r sets forth principle s for an ethic capable of pre serving the fre edom that libe rtarian move ments have helpe d to make possible , while avoiding the ir pote ntially harmful consequence s. O ne ought to start with a basic re spe ct for se xual patte rns that already exist, as they are probably accommodatio ns to exte rnal constraints and restrictions, and have positive value for the ir practitione rs. Se idman sees casual sex among gay American males in the se te rms. Judging sexual practice s without unde rstanding their meanings to those who engage them is a bad idea. It is thus local, situate d meanings that count, not the conformity of actions to external, absolute behavioral standards, such as religious proscriptions against specific sexual acts. Responsibility for the conse quences of sexual activity is important. Consent alone is not sufficie nt to make sex acceptable ; in addition, one must be concerned with its effects on one ’s partne r(s). Even if one believed that an eight-ye ar-old was informe d and mature enough to give meaningful consent to sex with an adult (something Seidman doubts), concern about possible consequences for the child would lead Seidman to disallow it. Though I find little to fault in Seidman ’s ethic, I am skeptical of the value of this chapte r. Because Seidman ’s principle s are broad and vague , the y offer limite d guidance in concrete situations. For example , should an attempt be made to suppre ss clitoride ctomy? It is ordinarily performed coercively on girls who are not old enough to give meaningful consent, and it seriously damage s them. Yet, like any custom, it could also be conside red an accommodation to particular circumstance s. As a traditional practice among some people s of Africa, Seidman tells us that we should approach it with respect. Seidman ’s goal, of course, is not to resolve such dile mmas, but to remind us of what we need to conside r when we think about them. As a practical matter, many of the debate s that Se idman chronicle s concern suppose d conse que nce s of se xual practice s that are difficult to de monstrate e mpirically. The e ffe cts of pornography on men ’s attitude s toward wome n, for example , are not e asy to measure . It may ne ver be possible to asse ss the m with sufficie nt confide nce to confirm or re fute feminist claim s about the harm pornography doe s to wome n. O ne suspe cts from the confide nce with which such claim s are advance d in the abse nce of re search that the facts are large ly immaterial to the claims-m ake rs. The claims themselve s appe ar to be e mbedde d in a world vie w that is held as tenaciously as any re ligious faith. That substantial fraction of the Ame rican population that be lie ve s sexual prohibitions have been give n to us by God will also have little use for Se idman ’s postm ode rn, antifoundat ionalist ethic.

626

Book Reviews

This ethic has, in fact, no way of justifying itse lf. It has no basis for persuading a supre me e goist to be conce rne d for the well-be ing of anyone but himse lf. Will the rapist re ad Seidman and then realize that he should ask his victim ’s consent be fore he rape s the m? Will child mole sters stop se ducing 8-ye ar-olds be cause Seidman alerts them to the importance of conside ring the long-te rm psychological e ffect of seduction? Those most in need of this advice are the le ast like ly to be influe nce d by it. Se idman hope s, I suspe ct, to join the ranks of the public inte lle ctuals who, through the ir writings, have he lpe d to re shape American se xual culture . I hope he succe e ds. I wish I could be optim istic about his chance s.

REFERENCES Decter, M. (1980, Septembe r). The boys on the beach. Com mentary 70: 35-48. Seidman, S. (1990). Rom antic Longings: Love in America, 1883-1980, Routledge , New York.

7 REVIEWS Professing Feminism - Cautionary Tales from the Strange World of Women's Studies

file:///D|/...ing%20%20Feminism%20-%20Cautionary%20Tales%20from%20the%20Strange%20World%20of%20Women's%20Studies.txt[02/11/2011 7:39:22 AM]

8 REVIEWS Embattled Eros - Sexual Politics and Ethics in Contemporary America

file:///D|/...%20REVIEWS%20Embattled%20%20Eros%20-%20Sexual%20Politics%20and%20Ethics%20in%20Contemporary%20Am.txt[02/11/2011 7:39:23 AM]

Archives of Sexual Behavior, Vol. 27, No. 6, 1998

Letters to the Editor

PHEROMONAL INFLUENCES To the Editor: Cutler, Frie dmann, and McCoy (1998) report that an unide ntifie d “ phe romone ” place d in aftershave lotion increase d human sociose xual behavior during a 6-wee k treatme nt pe riod (the phe romone appare ntly was “ Athe na Pheromone 10X ” which is marketed by Dr. Cutle r’s Athena Institute [see http://www.athe na-inst.com/10x.html] ) Of the six behaviors recorded, they conclude d that the group re ceiving the phe romone exhibite d gre ater increases ove r base line than the place bo group for two (sle eping next to a partne r and fre que ncy of inte rcourse). They also reporte d that after counting “ [t]he numbe r of individuals perceiving positive change s during any of the expe rime ntal wee ks ” (p. 7) a highe r proportion of “ pheromone ” users perceived positive results during the treatme nt. Re-analysis of the ir data, howe ve r, raise s se rious doubt about all of the se conclusions. Conside r first the subje ct’s perception of positive results. Cutle r et al. report 47% (8 of 17) of phe romone use rs and 24% (5 of 21) of the place bo group pe rceive d positive change s and that this was a significant diffe rence . Even in a one -tail, fair te st of a diffe rence betwee n two proportions (which Cutler et al. appare ntly use in the re mainde r of their comparisons) it is not (p > 0.10; Fisher Exact O ne -Tail Te st). While the test use d by Cutler et al. (Z te st of a proportion) may be appropriate to evaluate the significance of a single proportion against an anticipate d result, e.g., the obse rved proportion of heads in n tosse s of a coin versus the anticipate d 0.50, it appe ars to be used inappropriate ly by Cutle r et al. Next conside r sle eping ne xt to a partne r and pe tting. In Table III, the authors claim that 6 phe romone - ve rsus 1 place bo-tre ated men exhibite d an in cre ase ove r bas e lin e in sle e ping ne xt to a partne r and 7 phe romone - ve rsus 3 place bo-tre ate d me n had an increase in petting. The se numbe rs were determine d by requiring that the ave rage value over the 6 weeks of tre atment had to excee d the ave rage value during base line , and that a male ’s “ highe st weekly base line score was exceeded at le ast once 627 0004-0002/98/1200-062 7$15.00/0

Ó

1998 Plenum Publishing Corporation

628

Letters to the Editor

during the expe rimental pe riod, if his base line score was not already at maximum ” (p. 7). Including this final qualifie r effe ctive ly insure d that two men in the phe romone group would be include d in the group showing an increase ove r base line for slee ping next to a partne r and one would be include d for petting. If the base line scores are maximal, then it follows that it would be impossible for the mean treatme nt-value to exceed the phe romone . In all fairne ss, since Cutle r et al. include d individuals from the phe romone group because the subje ct’s treatme nt value s could not exce ed base line in one of the ir measure s, the y should apply the crite rion across the othe r measure . Had this be en done , two additional individuals from the place bo group would have bee n include d for sleeping ne xt to a partne r (S11 and S27) and one (S11) for pe tting. After making these change s, the p value s be come, or re main eve n more , nonsignificant (p = 0.13). This le ave s only se xual inte rcourse as pote ntially affected by the treatment. It is notable that the fre que ncy of inte rcourse for many men in both groups was zero during the 2 base line weeks. For the se individuals, a single inte rcourse during the 6 wee ks of treatme nt therefore counts as an increase ove r base line . In the place bo group the re were no individuals among the 10 whose base line scores were zero who had a single intercourse during the treatme nt phase . In the phe romone group the re were 4 of 11. Interestingly, 3 of the se 4 reporte d their only instance of sexual intercourse in the 6th and final wee k of the study. In a wee k-by-we ek analysis, Fisher’s Exact one -tail test yielde d p > 0.10 for all treatme nt We eks 1 through 5 and did not reach significance until Treatme nt wee k 6. With re spe ct to the unide ntifie d phe romone , in earlie r work, axillary extracts were prepare d (Pre ti et al., 1986; Cutler et al. 1986) and analyze d only for selected steroidal compone nts as de scribed in Pre ti et al. (1987) ; howe ve r the comple xity of the se extracts pre clude d any a priori assumption regarding the identity of the compone nts that may have cause d alte rations in the menstrual cycle s (a prime r phe romone effe ct). Neither Preti et al. (1987) nor any re fe rence s cited there in present any evide nce sugge sting influe nces on sociose xual variable s (release r phe romone effects) resulting from application of axillary e xtracts or the ir compone nts. How the n could Cutler et al. sugge st that “ a synthe tic ve rsion of a phe romone ” was de rived from the work of Preti et al. (1987) after “ re fining a proprie tary formula, characte ristic of heterose xual men ” ? To de rive this formula (particularly one that applie s to he te rosexual men in general, a feat in itse lf), active ingre die nts would have had to have be en isolate d in a bioassay-dire cted, analytical e ffort. Nowhe re in Cutler et al. are data or reference s provide d

Letters to th e Ed itor

629

that would de monstrate that such activitie s provide the foundation for the ir choice of the unname d, proprie tary, active ingre die nt(s). We the re fore sugge st that Cutler et al.’s (1998) “ Phe romonal Influe nce s on Sociose xual Be havior in Me n ” be com e “ Phe nom enologic al Influe nce s . . . in Men.” Charles J. Wysocki, Ph. D. Monell Center, an d Departm ent of Anim al Biology School of Veterinary Medicine University of Pennsylvan ia, Philadelphia G eorge Preti, Ph.D. Monell Center, an d Departm ent of Dermatology, School of Medicine University of Pennsylvan ia, Philadelphia

REFERENCES Cutler, W. B., Friedmann, E., and McCoy, N. L. (1998) . Pheromonal influences on sociosexual behavior in men. Arch. Sex. Behav. 27: 1-13. Cutler, W. B., Preti, G., Krieger, A. M. Huggins, G. R., Garcia, C. R., and Lawley, H. J. (1986) . Human axillary se cretions influence wome n’s menstrual cycles: The role of donor extract from men. Horm . Behav. 20: 463-476. Preti, G., Cutler, W. B., Christense n, C. M., Lawley, H. J., Huggins, G. R., and Garcia, C. R. (1987) . Human axillary extracts: Analysis of compounds from samples which influence menstrual timing. J. Chem . Ecol. 13: 717-731. Preti, G., Cutler, W. B., Krieger, A. M. Huggins, G. R., Garcia, C. R., and Lawley, H. J. (1986) . Human axillary se cretions influence wome n’s menstrual cycles: The role of donor extract from wome n. Horm . Behav. 20: 474-482.

RESPONSE TO WYSOCK I AND PRETI We find this lette r particularly disturbing be cause it contains significant misstate ments of the proce dure s used in both the re search and the data analysis and reve als that Preti and Wysocki misre ad Table II. The ir arguments then re ly on such e rrors and their conclusions are based on false assumptions, only to be compounde d by resorting to “ post hoc analysis ” to furthe r their argume nts.

630

Letters to the Editor

Our Proced ures In accordance with acce pte d scie ntific practice , our hypothe se s were de ve lope d be fore the protoc ol was e stablish ed. The crite ria for classification of subje cts pre ce de d the unblindi ng and examination of the data. The study, data, and re sults were the n pre sente d to colle ague s at the scie ntific mee tings of the American Socie ty for Reproductive Me dicine (9/95) in Seattle , the Inte rnational Academy of Se x Research (6/96) in Rotte rdam, and the North American Menopause Socie ty (9/97) in Boston for exposure and colle gial fee dback.

Perception of Positive Results We re porte d “ A significantly highe r proportion of users perceive d positive re sults dur in g the e xpe rim e ntal 6-we e k pe riod than did the place bo use rs (47 vs. 24% ; Z = 5.05, p < 0.001) .” The Z test we use d was appropriate as it is commonly used “ for comparing proportions for dichotomous variable s” (V ogt, 1993) . As state d in our publication, the diffe rence be twe e n phe romone and place bo use rs in pe rception of positive re sults was not ne arly as substantial as the diffe re nce be tween the two groups in the ir actual be havioral re sponse . Wysocki an d Preti ’s Errors In their third paragraph, discussing Table s II and III of the Study, Wysocki and Pre ti state that “ since Cutler et al. include d (as incre ase s over base line ) individuals from the phe romone group be cause the subje cts’ treatment value s could not excee d phe romone in one of the ir measure s, they should apply the criterion across the othe r measure .” It is simply not the case that we applie d this criterion inconsiste ntly. In fact (i) the re was no subje ct “ in the phe romone group include d as an incre ase ove r base line (whose) . . . value s could not exce ed base line . . .” and (ii) contrary to the ir asse rtion both groups were tre ated ide ntically. No subje ct who starte d at maximum was treate d as an increase in either group.

Letters to th e Ed itor

631

Table II re ve als that two marrie d place bo subje cts (S11 and S27) recorde d “ sle eping ne xt to a partne r” eve ry day and, thus, were at maximum for both base line and expe rimental weeks; one of the se (S11) did the same for “ affe ction/pe tting and kissing. ” This stable behavior rendere d 3 cells of data incapable of demonstrating an increase ove r base line : “ Sle eping ” for S11 and S27, “ Affection ” for S11. O ve rall, 38 expe rimental subje cts e ach re corde d 6 be haviors generating 288 cells of data (38 × 6) which are available to search for an increase ove r base line due to e xpe rime ntal tre atme nt. Thre e of the se 288 cells score d no change from the maximum at which the y began. Wysocki and Pre ti sugge st that we score the se 3 cells as if the y were an incre ase ove r base line . Howe ve r, it is irratio nal to assign the se 3 ce lls as an incre ase be cause the re was no incre ase . In fact, inspe ction of the data of the se me n re ve al no incre ase in an y of the 10 cells of sociose xual be haviors re corde d ( 5 sociose xual be haviors for 2 subje cts) ; and those be haviors that did not re main stable actuall y de crease d (e .g., sexual inte rcourse de crease d for both me n in the e xpe rime ntal phase ) . As appropr iate to this double blind place bo controlle d study, the de finition of “ incre ase ove r base line ” was applie d equally to all ce lls of both groups. We did not score stable or de clining ave rage s as an increase . In their examination of the sexual inte rcourse data from the place bo group (4th paragraph) , the y again did not re ad the data in Table II accurate ly. Wysocki and Pre ti state : “ In the place bo group the re were no individuals among the 10 whose base line score s were zero who had a single inte rcourse during the treatme nt phase .” Subje ct 48, in the place bo group, was cle arly such as individual. His 8 wee ks of inte rcourse data shown in Table II were 0 0 0 1 0 1 1 0: meaning two base line weeks of “ 0,” and one intercourse each, in weeks 4, 6, and 7. We e mploye d strict, rathe r than lax, criteria for scoring a cell as an increase in behavior by se tting two crite ria: (i) highe r weekly ave rage and (ii) at le ast one score excee ding the highe st base line score when mathematically possible . W ysocki and Pre ti are in corre ct in statin g “ this re quire me nt e ffe ctive ly insure d that two men in the phe romone group would be include d as showing an incre ase over pheromone for sleeping next to a partne r.” We can see no subje ct for whom this second crite rion would have classifie d his cell as an increase in Table III (or II). We fail to unde rstand the source of the writers’ inaccurate re ading of the data since the y did not provide an example .

632

Letters to the Editor

Fig. 1.

Wysocki and Pre ti state : “ It is notable that the fre que ncy of inte rcourse for m an y m e n in bo th groups was ze ro during the 2 base line we e ks. ” W e ag re e and suspe ct that this was the re ason the me n chose “ to e nroll in a study de sign e d to in c re ase the romance in the ir live s. ”

On the Rob ustness of th e Fin din g in Th is Sam ple A de taile d inspe ction of Table IIA actually re ve als the robust nature of the se data. Table IIA highlights 4 colum ns of data de rive d from the informa tion alre ady provide d to illustrate the consiste ncy and stability of the findings we have pre se nte d. The se que ntial re cords within each group are re arrange d in orde r to re ve al the cohe sive nature of the phe romone ’s e ffe ct. Table IIA shows that twice as many phe romone as place bo subje cts re corde d an incre ase in at le ast one sociose xual be havior ( 76 vs. 38% ) . Thre e time s as many phe romone as place bo subje cts recorde d an incre ase in at le ast two be haviors. Four time s as many phe romone as place bo use rs re corde d an incre ase in at le ast thre e sociose xual be haviors. And, an “ infinit e ly” highe r proport ion of phe romone than place bo use rs showe d an incre ase in four or more be haviors be cause the plac ebo group had no ( 0) me n who me t this crite rion. Figure 1 displays the se data. We conclude that phe romone s not only produce d increase s in men ’s attractive ness to women but that the data are robust. As we stated in the publication, these initial findings nee d to be re plicate d.

Re: Th e Non ob viou sn ess of the For m ula While not ide ntifie d as such in the publication, Wysocki and Pre ti are corre ct that the formula te ste d was Athena Pherom on e 10Xtm which is

Letters to th e Ed itor

633

Table IIA. Consistency Patte rns in Behavioral Response Gain Subject Status

³ 1 ³ 2 ³ 3 ³ 4

Sexual intercourse

Sleeping

Affection

Informal date s

Formal dates

Phe romone S06 S10 S47 S16 S21 S45 S46 S34 S09 S39 S42 S02 S08 S07 S14 S36 S51

M ND ND ND ND ND ND ND M KSC M M KSC M D D M

Totals n %

x x x x x x x x x x x x x

x x x x x x x x x x

x x x x x x x

x x x

13 10 7 3 76 58 41 16.7

22223232 00000001 00000001 00000010 00000000 01111231 00000000 00000001 00011000 42445445 32233202 31111122 00000000 21120221 00000000 00000000 00000000

71767777 00012001 00000001 00000010 00001000 30121221 00000000 00000000 77576774 66445445 57655243 75774777 00000000 76777774 00000000 10000000 67263476

75777777 23453345 00001011 10100111 01021010 00122221 00000001 00000000 66045445 64475223 11111101 33213312 76777774 00000000 00000000 10000000 12000000

33000000 22210221 00010001 00100010 00010110 43001001 00100111 00000000 00000000 00001002 11021000 00000000 10112011 00000000 00000000 00000000 00000000

00002000 01003001 00001011 10000101 01011000 00120010 00001001 00001000 00000000 66544443 00010031 00000000 23301322 00000000 00000000 10000000 00000000

8 47

6 35.3

7 41.2

6 35.3

7 41.2

Placebo S33 S48 S18 S41 S13 S17 S29 S30 S03 S05 S11 S15 S20 S23 S24 S25 S27 S32 S35 S38 S40 Totals n %

ND ND D D D ND ND ND ND D M KSC D D ND ND M D ND D M

x x x x x x x x

x x x x

x x

8 4 2 0 38 19 9.5 0

00000000 00010110 22423434 00000000 11011000 01010000 00000000 01000000 00000000 00000000 24114410 00000000 22001102 04200000 00000000 00000000 12101011 10100100 00000000 20212210 01010101 2 9.5

00000000 00000111 00001101 01100112 04242522 11020110 25433535 06323534 12100001 00000000 00002300 00000000 00001000 23222241 20020121 00000000 01010010 01010000 00000000 00000000 00000000 00000000 01200000 00000100 00000000 00000000 00000000 00000000 20000000 01111110 77777777 77777777 00000000 00000000 41411140 00000000 34222343 54223303 22000000 04000000 04000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 77777777 13202202 00000000 10000000 21211300 00000000 00000000 00000000 00000000 20101110 414232221 20222221 77777575 11010101 00000000 1 4.8

3 14.3

2 9.5

00000010 12113323 10000503 11123301 33213120 00000010 00000001 01000010 00100000 32120222 00000000 00000000 32302030 00000000 11000110 00000000 00000000 32421310 00000000 21202110 00000000 7 33.3

634

Letters to the Editor

marke te d in support of Athe na Institute for Women ’s We llne ss, Inc. Wysocki and Pre ti opine (5th paragraph) that the proprie tary formula is not obvious base d on prior publishe d work. Dr. Cutle r agre es and note s that nonobviousne ss is a re quire ment for he r pe nding pate nt application on the formula. She cre ated this formula inde pe nde ntly but de rive d it from their previous collaborative work through 1987. Winn ifred Cutler, Ph.D. Athen a Institute for Wom en ’s Wellness Research Chester Springs, Pennsylvan ia 19425 Erika Friedm an n, Ph.D. Departm ent of Health an d Nutrition Sciences Brooklyn College Brooklyn, New York 11210 Norm a L. McCoy, Ph.D. Departm ent of Psychology San Fran cisco State University San Fran cisco, Californ ia 94132

REFERENCE V ogt, W. P. (1993) . Diction ary of Statistics and Methodology, Sage, Ne wbury Be ach, CA, p. 69.

INTERSEXUALITY: RECOMMENDATIONS FOR MANAGEMENT To the Editor: It is good to have long-te rm follow-up reports on the treatme nt and manage ment of inte rse x conditions. Inde ed, such pape rs as the recent study by Slijpe r et al. (1998) are nee de d to amass a colle ction of case s from which directions for future tre atment can be e xtracte d. Unfortunate ly the pape r is less he lpful than it might be . First it incorre ctly state s my thinking and recomme ndations for de aling with the tre atment of inte rse xe d childre n. Se cond, it incorre ctly reports some of my Support for this rese arch was provided by the Euge ne Garfield Foundation of Philadelphia, Pe nnsylvania, and The Quee n’s Me dical Cente r of Honolulu, Hawai ‘i.

Letters to th e Ed itor

635

findings. Third, it lacks clarity of prese ntation so one cannot be sure of the childre n ’s manage ment nor treatme nt outcome . And fourth, the methods and discussion se ction s om it im portant con side ratio ns for the inte rpre tation of their findings. Slijpe r et al. (1998) state “ Diamond (1996) is of the opinion that se x assignme nt and genital surge ry should be de laye d until the child can decide for itse lf. This means the child should ne ithe r be raised as a boy nor as a girl, but as an inte rsex person ” (p. 142) . Regre ttably that is not my opinion and I have cle arly writte n othe rwise. First, the reference to which they refer with their mistake n statement of my ideas does not exist. The pape r to which they probably refer should read Journal of Sex & Marital Therapy rather than Journal of Sex Research. In that particular 1996 pape r what I do say regarding intersexed infants is: 1. Manage ment should not be de cide d sole ly on the size and nature of the phallus (p. 164). 2. “ Postpone any cosme tic clitoral surge ry until the individuals can the mselves unde rstand the situation well e nough to participate in the decisions ” (p. 165) . 3. “ Concern is only re garding cosme tic surge ry. I have no he sitation about surge ry for medically thre atening reasons ” (p. 166) . 4. “ When possible , the childre n have a say in any cosmetic surge ry and absolut e ly [whe n possibl e ] be involve d in any sex reassignm ent ” (p. 166) . Inde ed, Keith Sigmundson and I (Diamond and Sigmundson, 1997a) explicitly say: “ In rearing, pare nts must be consiste nt in se eing their child as eithe r a boy or girl; not ne ute r. In our society intersex is a designation of me dical fact but not ye t a commonly accepte d social de signation ” (p. 1047) . We furthe r indicate to which sex we think assignme nt of persons with diffe re nt conditions would offe r the best pote ntial outcome . Such assignm ent doe s not ne cessarily follow the chromosomal se x or ge nital configuration. The findings they misrepresent come from our pape r docume nting the John/Joan case (Diamond and Sigmundson, 1997b) . Slijpe r et al. write : “ Although raise d as a girl, the patie nt . . . as an adult aske d for sex reassignme nt ” (p. 126). We had stated cle arly that: “ at age 14 ye ars, Joan [an XY male that had bee n reassigne d as a female] decided to switch to living as a male ” (p. 300) . This occurred on his own, against advice and despite inte nsive profe ssional counse ling by Mone y and othe rs and pare nting to have Joan acce pt life as a female . This Slijpe r pape r is confusing or misle ading in othe r ways. For instance in Table I the major cate gorie s are listed according to medical diagnosis but in Table III the major cate gorie s are give n in te rms of ge nital

636

Letters to the Editor

appe arance . This makes comparing outcome difficult. Furthe r many of the cate gorie s are compounde d in discussion. This obscure s a cle ar interpretation of findings. It would be bette r if e ach case were liste d unde r a major cate gory with the genital appe arance , se x of assignme nt, treatme nt, and outcome give n so the re ade r could be tter unde rstand what occurred and judge for him/he rself as to the significance of the findings. Some 20% of the cases offered by Slijpe r et al. involve those with the comple te androge n inse nsitivity syndrome (CAIS). With XY chromosome s and a female genital appe arance without ambiguity, such persons are certainly intersexe d. Since the work of the last dozen years or so, howe ver, there is almost neve r any doubt as to assignme nt as female and the CAIS pe rson se e ing he rse lf as fe male whe n adult. Slijpe r et al. say as much (p. 126). Thus, in a pape r conside ring sex assignme nt for intersexed individuals with ambiguous genitalia this population would be better discusse d separate ly. The subje ct population of 59 would there fore, for conside ration, be reduced by these 12. The same can be said of the two individuals with Leydig cell hypoplasia. The base line numbe r for XY individuals whose sex of assignme nt might be in doubt as females the n becomes, as a maximum, 40 (59 ¯ 5 [those assigne d as male] ¯ 12 ¯ 2 = 40) . It is also questionable if the two individuals spoke n of as “ still too young to have their ge nder role behavior evaluate d ” (p. 137) should be conside red since treatment outcome for them is far from knowable . The baseline would the n be 38 rathe r than 59. Such conside rations significantly change the outcome percentage s. It is also notable that the work of Slijpe r et al. doe s not conside r XXY individuals since they are among the more common intersexe d conditions. Were the y not se e n among the ir “ 10 ye ars ’ work with childre n exhibiting a physical inte rse x condition ” (p. 127) ? More than a few of the se individuals with Kline fe lter’s syndrome , although assigne d as males, later switch to living as females and/or exhibit androphilic orie ntation. Table III indicate s that 7 individuals deve lope d a ge nde r ide ntity disorder (GID) and 12 individuals de ve lope d a de viant ge nde r role (DGR). Elsewhe re we le arn that “ Deviant ge nde r role be havior was not only exhibite d by the girls with a ge nde r identity disorde r, it was note d in 25 (46% ) of the total group of girls ” (p. 137) . And late r read “ Although 87% of the girls with a physical inte rsex condition develope d in line with the assigne d sex, 13% de ve lope d a GID, but only one (2% ) faile d to accept the assigne d sex” (p. 142) . How are these figure s to be reconcile d? Unde rstanding of patie nt manage ment would also have be ne fitted from some additional information. Inte rse x cate gorie s are known for having degre es of manife station. For instance , CAH phe notype s can range from individuals having minor clitoral hype rtrophy alone to pe rsons having an exte nsive pe nile phallus with accompanying labial fusion to form a scrotal

Letters to th e Ed itor

637

sac. The accompanying display of male -like be haviors can be great or limited for e ithe r extreme. The designation of partial androge n insensitivity syndrome (PAIS) can range from a de signation of 1 (comple te male-appearing) to 5 or 6 on a scale of 1 to 7 (comple te fe male-appe aring) (Quigle y et al., 1995). Because Slijpe r et al. have cate gorize d the patie nts by etiology rathe r than phe notype , the re ade r has no way of knowing the degre e of ambiguity involve d in an individual case . This is crucial information needed to manage the child and pre dict outcome . A inte resting finding reporte d by Slijpe r et al. is that GID in the ir group become manife st at 3-5 ye ars of age . This make s sense since childre n during preschool years certainly become aware if they disagre e with the ir sex assignme nt (Diamond, 1997) . It is also reporte d that psychopathology also de ve lope d as late as 27 ye ars of age . Nowhere are we told, howe ve r, how old the individuals we re at e valuation. Some were , I assume , still te enage rs. Pe rhaps GID will be manife st late r. Many inte rse xe d (and transse xual) individuals don ’t change gende r until in their 30s or afte r. Much de pe nds on the options the individuals can conside r and how much help is give n by therapists, physicians or othe rs. As one example , I recently reporte d on a he rmaphroditic individual who, at the age of 28, transitione d from fe male to male only afte r a sympathe tic counse lor sugge ste d this as a viable possibility (Diamond, 1997). The switch was then imme diate and satisfying. O the r physicians or the rapists to whom the individual previously went for he lp re fused to discuss se x change as an option and he himself didn ’t realize that re assignme nt could occur. Se x change can and does occur at any age . And sex reassignme nt can at any age be successful if done at the individual ’s behe st rathe r than being impose d. It is not cle ar at what age e ach of the subje cts was que stione d re garding ge nital surge ry or se x assignme nt. It is re porte d “ From the age of 4, the childre n were able to e xpre ss the ir own opinions . . . about the le ng th of the clito ris or its e re ctile function ” ( p. 13 3) . W hile it is appropr iate to discuss such matte rs with childre n, I do not be lie ve it appropriat e to base surgical de cisions re lative to the future value of a child ’s ge nitals on the impressions of a 4- or eve n a 10-ye ar-old. Such childre n usually have little or no concept of e rotic masturbation , orgasm, or mutual ge nital e xpe rie nce s from which to judge what loss of the ir born-with-ge nitalia might e ntail (Diamond, 1996) . Also, at these young age s childre n are most susce ptible to pare ntal and profe ssional pre ssures. The y cannot at those young age s give informe d consent. The y are not aware of what they can lose (Chase , 1996) . The authors atte mpt to answe r why, despite early surge ry, and psychiatric counse ling to pare nts and patie nts, there was still such a high degre e of psychopathology in the ir sample . Pe rhaps the intersexed indi-

638

Letters to the Editor

viduals were manife sting disparitie s and conflicts the y saw in the ir live s which were not re cognize d by the ir family or therapists. The y might have fe lt they could not e asily and acceptably e xpre ss their true feelings either at home or at the clinic and psychopathological be havior was the result. Many inte rse xed persons have re porte d be ing de nied the opportunity to fully de clare their own de sires or have the m re spe cted. Many have be en denie d knowle dge of the ir own historie s (see e .g., Diamond, 1997; Diamond and Sigmundson, 1997b) . Thoughts of inte rsexed persons are often labe led as misguide d, de viant, or eve n psychotic since the y don ’t follow expe ctations of clear male ¯or-fe male thinking or be haviors. Too often it is only when the inte rse xed individual adhe res to the stere otype d ge nde r constructs of the ir clinicians that the y are conside red “ normal. ” I sugge st, instead of asking the inte rsexe d individuals to conform to these criteria we expand the clinicians ’ ideas and unde rstanding so that the inte rsexe d person is seen as normal within a wide r set of parame te rs. Can it be assume d that Slijpe r and colle ague s were re ady to allow the ir inte rse xe d patie nts to easily expre ss disappointme nt with their life lot or sex assignme nt? Would any such admission result in furthe r psychothe rapy to re inforce the original sex assignme nt re gardle ss of whe ther that is what the individual desire d? Conside r: “ inte nsive psychothe rapy” was applie d to those childre n who could not e asily accept the ir assignme nt as fe males (p. 136) . As Slijpe r and colle ague s state: “ The aim of the team was to pre ve nt the deve lopme nt of cross-ge nde r ide ntification in childre n born with a physical inte rse x condition, e spe cially in ne onate s born with ambiguous ge nitalia ” (p. 127). While that might seem an unde rstandable and commendable goal I offer inste ad that the patie nt be allowe d to redirect the goal of the therapy, if that is his or her wish, and the n re ceive help in the new dire ction even if it e ntails gross-ge nde r identification. O therwise the tre atment can be see n as intimidating and “ brow-be ating. ” With such tre atment the child might e ventually se em to “ go along ” and have been convince d, but actually be rese ntful and only be waiting for the opportunit y to pursue his or he r own dire ctions. Such was the case , for instance , with John/Joan (Diamond and Sigmundson, 1997b) . The high incide nce of psychopathology see n in Slijpe r et al.’s inte rsex population might, in part, be due to the lack of support for the individual desirous of se x re assignme nt or cross-ge nde r identification. It is my recommendation to “ Allow the child fre e e xpre ssion as to choice s . . . . Do not obfuscate ; knowle dge is power, e nabling the patie nts to structure the ir live s accordingly. ” (Diamond and Sigmundson, 1997a, p. 1048) . If the individual pre fe rs a cross-ge nde r identification, that possibility should be explore d and supporte d. This, too, is also the thinking of the Intersex Socie ty of North

Letters to th e Ed itor

639

America (ISNA, 1994) . Admitte dly, the re are not yet e nough case s reporte d in the lite rature where these sugge stions have be en followe d to know if, in the long run, the percentage of those be ing happie r adults will be any bette r than with the manage ment offered by Slijpe r and colle ague s. I do, howe ve r, think I se e gre ater contentme nt in those I counse l or for whom I am consulte d. Slijpe r et al. (1998) indicate d “ the te am policy was to corre ct the virilization of the external ge nitalia immediate ly afte r birth or as soon as possible afte r the di agn osis was m ad e so as to avo id cross-ge nd e r ide ntification ” (p. 132) . While this aids the individual who pre fe rs life as a fe male it mitigate s against those who might late r prefe r to be male . I re cently re porte d on one such hermaphroditic individual whose enlarge d clitoris was take n from he r at 18 months of age (Diamond, 1997) . This too might foste r psychopathology rathe r than reduce its like lihood. Many individuals be come aware of ge nital surge ry and se e it as a bodily insult rathe r than an aid. It is well to kee p in mind the expe rime ntal work of G o y, B e r c o vi t c h , a n d M c B r a i r ( 1 9 8 8 ) . T h e y d e m o n s t r a t e d th a t androg e nize d prim ate s cou ld show ge nita l m asculin ization withou t be havioral masculiniza tion and be havioral masculiniza tion without genital change s. Slijpe r and colle ague s (Slijpe r, 1984; Slijpe r et al., 1992) have found similarly for humans. This again is re ason to withhold surge ry until it is clearly de sire d by the individual. The surge on has no way of knowing to which ge nde r inte rse xe d individuals of many e tiologie s will aspire . Early surge ry reduce s the options available . Early and nonconse nsual surge ry also impose s anothe r set of risks. For many inte rse xed individuals it confirms for them, consciously or not, that the ir status at birth is monstrous and automatically in need of corre ction. Slijpe r et al. mention psychopathological dange rs in regard to vaginal dilation (p. 133) It should be re cognize d to pote ntially hold for all othe r surge ries as well. And certainly not of small conside ration, ge nital surge ry can damage future sexual functioning (Chase , 1996) . Furthe r, along these line s, it has also bee n shown that the appe arance of ge nitals, e ithe r the ir own or that of their pe ers, in the typical child is not crucial for classification of ge nde r until about the age of 9 (Goldman and Goldman, 1982) . At least for English-spe aking childre n, the y may be aware of ge nital diffe rence s but usually do not unde rstand they are significant for gende r assignme nt. It is not surprising that the groups showing the highe st incide nce of GID and DGR were those of XY karyotype that had been expose d to the highe st androge n tite rs ye t assigne d as female . Individuals with a transve rse penis or 17-KRD or exte rnal cloacae typically have difficultie s with prenatal

640

Letters to the Editor

dihydrote stoste rone production nee ded for male ge nital developme nt but sufficie nt prenatal testoste rone production for masculinization of the ne rvous syste m foste ring male identification (Imperato-McGinle y et al., 1979a, b; R ösler, 1992; R ösler and Kohn, 1983) . With developme nt and maturity the ir masculine be havioral biase s be come activate d and ge nital masculinization advance s. It is my recomme ndation to assign the se individuals as males (Diamond and Sigmundson, 1997a) . Lastly, the re asons individuals maintain a sex of assignme nt, despite fe eling to the contrary, are many and dive rse . It doe s not mean the y would not have it othe rwise (Diamond and Sigmundson, 1997b). I be lie ve that the most ethical and correct way to tre at inte rse xed individuals has two main rules. The first is that manage ment should be in light of the diagnosis, not in light of the individual ’s ge nitals. Whe n diagnosis and genitals coincide , all to the bette r. When they don ’t the prognosis should gove rn. The second rule is that the rights and thinking of the mature intersexed individual should have priority and no cosme tic surge ry be performed until that individual ’s voice is he ard. I hope that Slijpe r et al. and othe rs continue to study and revie w the are a of inte rsex and publish their work. My only cave at is the y do so with concern for accuracy, clarity, and conside ration of a wide r range of possibilitie s than demonstrate d in this particular pape r. Milton Diam ond, Ph.D. University of Hawai ‘i — Manoa John A. Burns School of Medicine Pacific Center for Sex & Society Honolu lu, HI 96822

REFERENCES Chase , C. (1996) . Re : Me asure me nt of evoke d potentials during feminizing genitoplasty: Te chniques and applications (letter). J. Urol. 156: 1139-1140. Diamond, M. (1996) . Response: Considerations for sex assignme nt. J. Sex Marital Ther. 22: 161-174. Diamond, M. (1997). Sexual identity and sexual orientation in children with traumatized or ambiguous ge nitalia. J. Sex Res. 34( 2): 199-222. Diamond, M., and Sigmundson, H. K. (1997a). Management of intersexuality: Guidelines for de aling with pe rsons with ambiguous ge nitalia. Arch. Pediat. Adolescent Med. 15 1: 1046-1050. Diamond, M., and Sigmundson, H. K. (1997b) . Sex reassignment at birth: Long-term review and clinical implications. Arch. Pediat. Adolescent Med. 151: 298-304. Goldman, R., and Goldman, J. (1982) . Children’s Sexual Thinking: A Comparative Study of Children Age d 5 to 15 Ye ars in Australia, North Ame rica, B ritain, and Swede n. Routledge & Kegan Paul, London.

Letters to th e Ed itor

641

Goy, R. W., Bercovitch , F. B., and McBrair, M. C. (1988). Behavioral masculinization is inde pe nde nt of ge nital mascu linization in pre natal ly androge nize d fe male rhe sus macaques. Horm . Behav. 22: 552-571. Im pe rat o-McG inle y, J., P e te rson, R. E ., G au tie r, T ., an d Sturla, E . ( 1979a) . Mal e pseudohermaphroditism se condary to 5 alpha-reductase de ficiency — a model for the role of androgens in both the developme nt of the male phenotype and the evolution of a male gende r identity. J. Steroid Biochem . 11(1B): 637-645. Impe rato-McGinley, J., Pe terson, R. E ., Stoller, R., and Goodwin, W. E. (1979b) . Male pse udohermaphroditism se condary to 17B-hydr oxyste roid de hydroge nase de ficiency: Gender role change with puberty. J. Clin. Endrocrinol. Metab. 49: 391. Quigley, C., De Bellis, A., Merschke, K. B., El-Awady, M. K., Wilson, E. M., and French, F. S. (1995) . Androge n receptor defe cts: Historical, clinical and mole cular pe rspectives. Endocrine Rev. 16( 3): 271-321. R ösler, A. (1992). Steroid 17B-hydroxysteroid de hydrogenase de ficiency in man: An inherited form of male pseudohermaphroditism. J. Steroid Biochem . Mol. Biol. 43: 989-1002. R ösler, A., and Kohn, G. (1983). Male pse udohermaphroditism due to 17B-hydroxysteroid de hydroge nase de ficiency: Studies on the natural history of the defe ct and effe ct of androgens on gender role. J. Steroid Biochem . 19( 1) : 663-674. Slijper, F. (1984) . Androge ns and ge nde r role be havior in girls with conge nital adrenal hyperplasia (CAH). In De V ries, G. J., De Bruin, J. P. C., Uylings, H. B. M., and Corner, M. A. (eds.), Progress in Brain Rese arch: Sex Differe nce s in the Brain (V ol. 61). Elsevier Pre ss, Amsterdam, pp. 417-422. Slijper, F. M. E., Drop, S. L. S., Mole naar, J. C., and De Muink Ke izer-Schrama, S. M. P. F. (1998) . Long-term psychological evaluation of intersex children. Arch. Sex. Behav. 27: 125-144. Slijper, F. M. E., van dr Kamp, H. J., Brandenburg, H., de Muinek Keizer-Schram a, S. M. P. F., Drop, S. L. S., and Molenaar, J. C. (1992). Evaluation of psychosexual developme nt of young women with congenital adre nal hype rplasia. J. Sex Educ. Ther. 18: 200-207.