P1: FPN Archives of Sexual Behavior [asb]
PL110-327
April 19, 2000
11:56
Style file version Nov. 19th, 1999
Archives of Sexual Behavior, Vol. 29, No. 3, 2000
A Statewide Survey of Age at First Intercourse for Adolescent Females and Age of Their Male Partners: Relation to Other Risk Behaviors and Statutory Rape Implications Harold Leitenberg, Ph.D.,1,2 and Heidi Saltzman, Ph.D.1
In a statewide survey of a representative sample of adolescent girls in 8th–12th grades (N = 4201), information was obtained on age at first intercourse and age of their male partners. Excluding intercourse experiences where physical force was threatened or used, 31% had intercourse by age 15 and 45% by age 16. Contrary to the impression left by studies of teenage mothers, girls who first had sex between age 13 and age 15 or between age 16 and age 18 did not have a large percentage of much older partners (5 or more years older; 12 and 7%, respectively). The percentage of much older partners was higher, however, for girls who had sex in very early adolescence, ages 11–12 (34%). Much older male partners were associated with greater problem behaviors for girls who first had intercourse in very early adolescence (11–12), but less so for those who first had intercourse between age 13 and age 15 (truancy only) and not at all for those who first had intercourse at between 16 and 18. Regardless of partner’s age disparity, earlier age at first intercourse during adolescence was associated with a greater number of other problem behaviors. The implications of these findings for recent calls to enforce statutory rape laws more stringently to reduce teenage pregnancy were discussed. KEY WORDS: age at first intercourse; statutory rape; adolescent sexual behavior; teenage pregnancy.
INTRODUCTION Prompted by statistics showing that the majority of fathers of children born to teenage girls were adults (Landry and Forest, 1995; Males, 1992; Males and Chew, 1 Department
of Psychology, University of Vermont, Burlington, Vermont 05405. whom correspondence should be addressed. e-mail: H
[email protected]. Fax: 802-6563482.
2 To
203 C 2000 Plenum Publishing Corporation 0004-0002/00/0600-0203$18.00/0 °
P1: FPN Archives of Sexual Behavior [asb]
204
PL110-327
April 19, 2000
11:56
Style file version Nov. 19th, 1999
Leitenberg and Saltzman
1996), the United States Congress, as part of the 1996 Federal Welfare reform law, urged states to enforce statutory rape laws aggressively as a way to reduce teenage pregnancy. The findings from these studies received substantial news coverage and the impression left by the media was that most pregnant teenagers were the victims of predatory older men. As a result, there were many calls for stricter enforcement of statutory rape laws (Donovan, 1997; Elo et al., 1999). Recent research suggests, however, that these birth record studies of partner’s ages may have conveyed an inaccurate picture of the typical ages of male sexual partners of teenage girls. Perhaps it was not realized that many of the “teenage” mothers in these studies were 18–19 years of age. Thus, it is not surprising that the majority of their male partners were adults (20 or older). When Lindberg et al. (1997) reanalyzed some of these data, they discovered that once married and older teenagers were excluded, only 21% of unmarried 15- to 17-year-old females who gave birth had male partners at least 5 years older. Moreover, these cases represented only 8% of all births to the 15- to 19-year-old unmarried teenagers in the sample. Even this 21% figure is misleading because it is based only on teenage girls who gave birth, a sample that may be biased toward having older partners compared to most female adolescents who are having sex (Elo et al., 1999). In fact, a recent national study indicated that only 5.5% of unmarried girls between 15 and 17 years of age who were sexually active had partners who were 6 years older (Darroch, et al., Oslak, 1999). Similarly, Miller et al. (1997) found that only 11% of the partners of girls who first had sex at between 14 and 17 were 5 or more years older, and Elo et al. (1999) reported that 18% of women who first had intercourse at between 15 and 17 years of age had a partner who was 4 or more years older. Therefore, contrary to the impression left by the earlier studies of partners of teenage mothers, these more recent studies suggest that the vast majority of male sexual partners of teenage girls are not substantially older men but are instead teenage boys or young adult males who are about the same age or only several years older than the girls with whom they are having sex. If most male sexual partners of teenage girls are not substantially older than the girls with whom they are having sex, the social policy implications of more stringent enforcement of statutory rape laws are quite different than may have been originally contemplated or intended. Statutory rape refers to sexual activity with a minor who is legally not able to give consent because of his or her age (Oberman, 1994). In the United States, 28 states use age 16 as the age of consent, 15 states use age 18, 6 use age 17, and 1 uses age 14 (Donovan, 1997). These laws are clearly targeted toward partners of teenagers who have said they had consensually engaged in intercourse. If physical force is used, typically other charges pertaining to sexual assault rather than to statutory rape would be pressed. And if the child is younger then a teenager, typically various child sexual abuse rather than statutory rape statutes would be invoked.
P1: FPN Archives of Sexual Behavior [asb]
PL110-327
Partner’s Age Disparity
April 19, 2000
11:56
Style file version Nov. 19th, 1999
205
Parental objections and statutory rape laws notwithstanding, the reality is that a large percentage of adolescent girls in the United States is sexually active (Brooks-Gunn and Furstenberg, 1989); approximately 50% have had intercourse by age 16 (Mott et al., 1996; Seidman and Rieder, 1994), and about one-third by age 15 (Besharov and Gardner, 1997). Although statutory rape laws are designed primarily to protect teenagers from exploitation or abuse of power and authority by adults and to discourage adults from engaging in sexual activity with minors, many states (29) do not, in fact, require any age discrepancy whatsoever between the partners for statutory rape laws to be prosecuted. Although the research literature typically uses a 5-year age difference to define child sexual abuse (Finkelhor, 1984), only four states require a minimum of a 5-year age discrepancy for sex with an underage minor to be considered a crime (Donovan, 1997). Therefore, if the vast majority of male partners of adolescent girls in this country are similar in age to or only somewhat older than the girls with whom they are having sex, more strict and indiscriminate enforcement of statutory rape laws would mean that a large percentage of the teenage male population in the United States would be at risk for prosecution as sex offenders. Because of the serious policy implications involved, we thought it important to collect additional information on the age of male partners of underage minor girls who have had sexual intercourse. In a statewide representative sample of adolescent girls, we compared partner’s ages for girls who first had intercourse at three age periods in adolescence, 11–12 vs 13–15 vs 16–18. We wanted to determine if the age spread between partners varies as a function of how old the adolescent girl is when she has intercourse for the first time. We expected that the percentage of much older male partners would be greater the younger the adolescent girls are when they first have intercourse. Our reasoning was that younger girls would be most vulnerable to being exploited by older males, and since boys mature later than girls, they would also have fewer similar-age male sexual partners available to them. We also wanted to determine if girls who have sex with much older male partners exhibit more behavior problems (suicide attempts, substance abuse, truancy, and pregnancy) than girls who have sex with similar-aged or somewhat older partners. Although it has been shown repeatedly that girls who have intercourse at a younger age in adolescence tend to exhibit more behavior problems than similarage girls who delay initiation of intercourse (Irwin and Millstein, 1992; Jessor and Jessor, 1977), it is uncertain to what extent age of their partners matters in this regard. It is usually assumed that an older partner is more likely to have a harmful effect (e.g., Lamb et al., 1986) but this may vary depending on the age of the girl when she first has intercourse. For example, a 5-year age difference may reflect a more exploitive situation when a girl first has intercourse at age 12 compared to age 16.
P1: FPN Archives of Sexual Behavior [asb]
PL110-327
April 19, 2000
11:56
206
Style file version Nov. 19th, 1999
Leitenberg and Saltzman
METHOD Sample The Centers for Disease Control and Prevention (CDC) originally developed the Youth Risk Behavior Survey in 1990 as a nationwide tool to assess periodically the prevalence of various health risk behaviors among youth in the United States. The data in this study were derived from the Vermont Youth Risk Behavior Survey, administered in 1997 by the Vermont Department of Health, Office of Alcohol and Drug Abuse Programs, in cooperation with the Vermont Department of Education and the CDC. Nineteen high schools along with their 13 associated middle schools were randomly selected to obtain the statewide sample of 8th- through 12th-grade students. The overall response rate was 71% (school response rate of 87% times student response rate of 82%). A total of 8636 students was included in this sample. The survey firm that contracted with the CDC used a statistical weighting formula to compensate for any differences between the sample and the population of all 8thto 12th-grade students in Vermont to ensure that the sample was representative of the larger population of students in these grades in the state. In the present study we utilized data from female respondents only and only from those who completed the questions about age at first intercourse and the age of their partners. This included a total of 4201 girls, mean age 15.40 (SD = 1.45). At the time of the survey, 92% of these girls were between age 13 and age 17, with less than 1% age 12 and 7.7% age 18. No other demographic data regarding social class, race, ethnicity, parent’s education, or religion were available, but based on the composition of Vermont, it can be assumed that approximately 97% of the participants were Caucasian and that the vast majority of respondents were Christian. Vermont also is a largely rural state, containing 575,000 people with no large urban areas (the largest city in the state has only 40,000 residents). It ranks thirtieth among the states in per capita income. Survey Only those questionnaire items pertinent to this study will be described. Age at First Intercourse and Age of Partner A question regarding age at first intercourse that has been standardly asked in the Youth Risk Behavior Survey reads: “How old were you when you had sexual intercourse for the first time?” The response choices are (a) I have never had sexual intercourse, (b) 11 years old, (c) 12 years old, (d) 13 years old, (e) 14 years old, (f) 15 years old, (g) 16 years old, and (h) 17 years old or older. The Youth Risk Behavior Survey has been shown in prior research to have a good test–retest reliability, with a kappa of 71% for age at first intercourse (Brener et al., 1995).
P1: FPN Archives of Sexual Behavior [asb]
PL110-327
April 19, 2000
11:56
Style file version Nov. 19th, 1999
Partner’s Age Disparity
207
Two additional questions were added to the Vermont survey specifically for the purpose of this study. The first inquired as to whether “physical force was threatened or used against you when you had sexual intercourse for the first time.” If the respondent answered “yes,” she was excluded from the remainder of the data analysis. The second question asked, “How old was the person with whom you had sexual intercourse for the first time?” The response choices were (a) I have never had sexual intercourse, (b) 12 years old or younger, (c) 13 or 14 years old, (d) 15 or 16 years old, (e) 17 or 18 years old, (f) 19 to 21 years old, (g) 22 to 25 years old, and (h) 26 years old or older. Having respondents provide the exact ages for their partners as distinguished from having to choose between these alternatives was not an option because of CDC survey design and format. For the purposes of this analysis, therefore, we used the midpoint of the scale item, for example, if someone chose “(e) 17 or 18” we calculated it as 17.5, if someone chose “(f) 19 to 21,” we calculated it as 20, etc. Also, for “12 years old or younger” we entered only 12 and for “26 years or older” we entered only 26. We then categorized partners age differences into three groups: similar age (−1.5 to +1.5 years apart), somewhat older (2.0–4.5 years), and much older (5 or more years). No further information is available on the nature of these relationships, e.g., whether the partner was a “boyfriend,” how long the relationship lasted, or whether the girl felt good about it or felt manipulated or coerced even if no force was used or threatened. Other Risk (Problem) Behaviors The problem behaviors analyzed in this study were suicide attempts in the past year; alcohol use (number of days in which alcohol was drunk in the past 30 days); drug abuse as defined by a composite score for number of times marijuana was used in the past 30 days, number of times any form of cocaine, including powder, crack, or freebase was used in the past 30 days, and number of times in the past 30 days in which the respondent reported sniffing glue or gas or breathing the contents of aerosol spray cans or inhaling any paints or sprays to get high; truancy defined as number of days cut school in the past 30 days; and pregnancy defined as lifetime frequency. RESULTS Overall, 3.5% of the females in this sample had a first intercourse experience in which physical force was threatened or used, representing 9% of all first intercourse experiences. These incidents were omitted from all subsequent analyses. Table I shows the absolute and cumulative percentages of age of first intercourse for those girls in the sample who were currently 16 or older. As can be seen, 31% had experienced intercourse by age 15 and 45% by age 16.
P1: FPN Archives of Sexual Behavior [asb]
PL110-327
April 19, 2000
11:56
Style file version Nov. 19th, 1999
208
Leitenberg and Saltzman Table I. Age at First Intercourse, Absolute and Cumulative Percentagesa Age
Absolute %
Cumulative %
11 12 13 14 15 16 17+
1 1 3 10 16 14 6
1 2 5 15 31 45 51
a Omits
first intercourse experiences where physical force was threatened or used.
Table II. First Intercourse for Females: Percentage Distribution of Age Differences of Their Male Partnersa Partner’s age disparity (%) Age at first intercourse
Similar age (−1.5 to 1.5) %
Somewhat older (2.0 to 4.5) %
Much older (5 or more) %
11–12 13–15 16–18
37 45 72
29 43 22
34 12 7
a Omits
first intercourse experiences where physical force was threatened or used.
The percentages of male partners who were similar aged, somewhat older, or much older as a function of the females’ age at first intercourse are shown in Table II. The data pertaining to the age of the male partner were derived from the full female sample, not just from those who were currently age 16 or older. As can be seen in this table, the distribution is very different depending on the age at first intercourse. For those girls who had their first intercourse experience at ages 11–12, 34% of their partners were much older than them (5 or more years). For those who had their first intercourse experience between 13 and 15, however, only 12% of their partners were much older, and for those who had their first intercourse experience between 16 and 18 the percentage of much older partners was the smallest, only 7%. These percentages are all significantly different from each other according to chi-square analyses [34 vs 12%, χ 2 (1, N = 1001) = 16.92, p < .001]; 34 vs 7%, χ 2 (1, N = 511) = 28.83, p < .001; 12 vs 7%, χ 2 (1, N = 1206) = 5.72, p < .01]. The male partners were on average 3.62 years older (SD = 3.41 years) than girls whose first intercourse occurred between age 11 and age 12, compared to 2.41 years older (SD = 2.26 years) if the girl’s first intercourse occurred between age 13 and 15 and 1.54 years older (SD = 2.21) if first intercourse occurred between 16 and 18. A one-way ANOVA indicated that these differences were significant [F(2,1361) = 94.17, p < .001].
P1: FPN Archives of Sexual Behavior [asb]
PL110-327
Partner’s Age Disparity
April 19, 2000
11:56
Style file version Nov. 19th, 1999
209
Table III examines whether there is any difference in the frequency of girls’ problem behaviors (suicide attempts, alcohol use, drug abuse, truancy, and pregnancy) as a function of how much older their first intercourse partner was and how old they were when they first had intercourse. Following an initial 3 × 3 MANOVA across all five problem behaviors, which showed a main effect for partner’s age difference [F(10,1252) = 3.99, p < .001], a significant interaction between partner’s age difference and age at first intercourse [F(20,1242) = 2.56, p < .001], and a significant main effect for age at first intercourse [F(10,1252) = 13.88, p < .001], one-way 3 × 3 ANOVAS and post hoc Newman–Keuls analyses were conducted for each problem behavior. As can be seen in Table III, the effect of partner’s age difference varied dramatically as a function of age at first intercourse. For each dependent variable except truancy, there was a significant interaction between partner’s age difference and age at first intercourse such that partner’s age difference seemed to matter a great deal if first intercourse occurred in very early adolescence (11–12) but much less so if first intercourse occurred at age 13–15 and not at all if first intercourse occurred in late adolescence (16–18) [for suicide, the interaction F(4,1301) was 3.81, p < .004; for alcohol use, F(4,1362) = 4.86, p < .001; for drug abuse, F(4,1388) = 3.96, p < .003; for pregnancy, F(4,1367) = 8.33, p < .001]. Post hoc Newman–Keuls analyses indicated that girls who had intercourse at age 11–12 with a much older partner had more suicide attempts, more substance abuse, and a greater incidence of pregnancy than girls who had intercourse with similar-aged partners ( p at least <.05 in each comparison). Although the means suggest greater truancy with much older partners, these differences were not statistically significant. Sex with somewhat older partners was also associated with more suicide attempts and more substance abuse but not with greater truancy or pregnancy than similar-aged partners ( p at least <.05). For girls who first had intercourse between age 13 and age 15, the Newman– Keuls comparisons revealed that those with much older partners engaged in only one problem behavior with greater frequency than those with similar aged partners, namely, truancy ( p < .05). Surprisingly, alcohol use was greater for those who had somewhat older partners than for those who had either similar-aged or much older partners. Otherwise, age of partners made little difference. For girls who first had intercourse between age 16 and age 18, the Newman– Keuls analyses revealed that partner’s age had no significant effect on any variable. Although not significant, the truancy data were similar to what was found for girls who first had intercourse at age 13–15; the mean truancy rate was higher in the group with much older partners than in the group with similar-aged partners (1.44 vs .83 days per month). There was a main effect for age at first intercourse for each problem behavior [suicide, F(2,1301) = 29.76, p < .001; alcohol use, F(2,1362) = 13.95, p < .001; drug abuse, F(2,1388) = 38.00, p < .001; truancy, F(2,1376) = 7.43, p < .001; pregnancy, F(2,1367) = 39.10, p < .001]. Subsequent Newman–Keuls
Same age Somewhat older Much older Same age Somewhat older Much older Same age Somewhat older Much older
11–12
16–18
1.42a 1.54a 1.69a
1.53a 2.26b 3.12c 1.53a 1.83b 1.57a .31a .36a .35a
.65a 1.11ab 1.56b .45a .48a .51a
.83a 1.04a 1.44a
1.49a 1.63a 2.15a .97a 1.09a 1.51b
Cut school in past month (1 = 1 day) (2 = 2 days)
.04a .03a .00a
.17a .31a .69b .11a .11a .14a
Ever pregnant (0 = 0) (1 = 1)
210
different at least at the p < .05 level.
11:56
a Omits first intercourse experiences where physical force was threatened or used. Means that do not share the same subscript were significantly
.20a .16a .00a
.49a .64ab 1.09b .27a .22a .27a
Illegal drug use in past month (1 = 1 or 2 times) (2 = 3 to 9 times)
April 19, 2000
13–15
Partner’s age disparity
Age at first intercourse
Days drank alcohol in past month (1 = 1 or 2 days) (2 = 3 to 5 days) (3 = 6 to 9 days)
PL110-327
Suicide attempts in past year (1 = 1) (2 = 2–3)
Archives of Sexual Behavior [asb]
Table III. Risk Behaviors for Adolescent Females Associated with the Age Disparity Between Their Age and Their Partner’s Age the First Time That They Had Sexual Intercourse
P1: FPN Style file version Nov. 19th, 1999
Leitenberg and Saltzman
P1: FPN Archives of Sexual Behavior [asb]
PL110-327
April 19, 2000
Partner’s Age Disparity
11:56
Style file version Nov. 19th, 1999
211
analyses revealed that girls who first had intercourse at age 11–12 fared significantly worse on every problem behavior measured than those who first had intercourse between 13 and 15 or 16 and 18 regardless of the age disparity from their partner at the time. Girls who first had intercourse at age 13–15 also exhibited more problem behaviors than those who first had intercourse at age 16–18, but not uniformly so. They had a significantly higher frequency of pregnancy and use of illegal drugs ( p < .05) but did not differ on suicide attempts, alcohol use, or truancy. DISCUSSION Consistent with recent findings in other samples (Darroch et al., 1999; Elo et al., 1999; Miller et al., 1997), and contrary to what people believed was the case based on studies of partners of teenagers who had just given birth (Landry and Forest, 1995; Males, 1992; Males and Chew, 1996), the vast majority of the male partners of the girls in our sample who have had intercourse were not substantially older than them. Instead, only 7% of girls who first had intercourse between age 16 and age 18 had a partner 5 or more years older than them and only 12% of girls who first had intercourse between age 13 and age 15 had male partners 5 or more years older. For the small number of girls who had sexual intercourse between age 11 and age 12, however, the percentages of much older partners was much higher (34%), suggesting that girls this age are much more vulnerable to exploitation by older males. This is cause for serious societal concern, and in fact, offenders against girls so young are usually prosecuted for child abuse not statutory rape. Elo et al. (1999) and Miller et al. (1997) also reported that older partners were more common for younger adolescent girls. We also found that consistent with national statistics (e.g., Besharov and Gardner, 1997), a large percentage of underage minor girls are having sex, 31% by age 15 and 45% by age 16. Combined, these findings have several implications for recent calls to enforce statutory rape laws more strictly. First, it is simply not true that most teenage girls who have sex are being seduced and exploited by substantially older men. The data instead clearly indicate that their partners are typically similar aged or just somewhat older than they are. This means that if statutory rape laws were indeed more stringently enforced, many teenage and young adult males will be prosecuted. They will become convicted felons and be characterized as deviant sex offenders when their behavior is in fact quite common. Second, the results from this study, as well as Lindberg et al.’s (1997) analysis, calls into question the major reason that was originally given in support of stricter enforcement and harsher penalties for violations of statutory rape laws. The ostensible purpose was to reduce the prevalence of teenage pregnancy by prosecuting predatory older men. But if one were to restrict statutory rape prosecution only to men who were much older than underage teenage girls, one will
P1: FPN Archives of Sexual Behavior [asb]
212
PL110-327
April 19, 2000
11:56
Style file version Nov. 19th, 1999
Leitenberg and Saltzman
not make much of a dent in teenage pregnancy rates since these are not the typical males having sex with teenagers, and as Lindberg et al. demonstrated, births from unmarried 15- to 17-year-old females who had partners 5 or more years older represented only 8% of all births to teenagers. Thus, in order for this policy to have any chance of being effective, one would have to prosecute a large number of male teenagers since they are the usual partners of underage females who are having sex. But if society indeed went ahead and indiscriminately prosecuted statutory rape laws independent of the age of the offender, the “cure” may be as bad as the “disease.” It has also been pointed out that such a policy may backfire by inhibiting female teenagers from seeking reproductive health services because they will be afraid that their boyfriends will go to prison (Donovan, 1997). Encouraging young teenagers to abstain from intercourse until they are older while at the same time promoting contraception for those who choose otherwise seems a much more reasonable policy. In summary, what the data from the present study imply are not that statutory rape laws should be abolished—there are, after all, some older adults who do abuse their power and authority and exploit vulnerable young teenagers—but that they should be enforced with great discretion rather than with great zeal lest we criminalize a large segment of common teenage sexual behavior. Another purpose of the present study was to determine if having much older sexual partners (5 years or more) was associated with a greater number of problem behaviors in adolescent girls than having similar aged or just somewhat older sexual partners. The data indicate that the answer depends on the girls’ age of first intercourse. For girls who first had intercourse in very early adolescence (11–12), much older partners were associated with more suicide attempts, more alcohol and drug abuse, and a higher incidence of pregnancy. However, for girls who first had intercourse between age 16 and age 18, older partners were not associated with any greater number of problem behaviors except, perhaps, for truancy, which was greater though not statistically significantly so. The results for girls who first had intercourse between age 13 and age 15 were very similar to those who first had intercourse at 16–18, except this time truancy was clearly greater for those who had had sex with much older partners. There were no significant differences in this group as a function of partner’s age discrepancy, however, for suicide attempts, drug use, or pregnancy. In addition, for girls whose first intercourse experience occurred between age 13 and age 15, alcohol use was greatest when first sexual partners were just somewhat older rather than much older. It should be noted that the pregnancy data in the present study differ somewhat from those of Darroch et al. (1999). They found that pregnancy rates were several times higher among girls 15–17 whose partners were 6 or more years older than them compared to girls whose partners were no more than 2 years older. Although we found the same thing for girls who first had intercourse between age 11 and age 12, we did not see this in the girls in our sample who first had intercourse
P1: FPN Archives of Sexual Behavior [asb]
PL110-327
Partner’s Age Disparity
April 19, 2000
11:56
Style file version Nov. 19th, 1999
213
between 13 and 15 or between 16 and 18. A sample difference might account for the discrepant findings. Miller et al. (1997) found that in a black and Hispanic sample, condom use on first intercourse was less likely if male partners were older rather than peers and Darroch et al. employed a national sample which would have had much greater racial diversity than is the case in Vermont. Darroch et al. also did not restrict their analysis to first intercourse experiences. In longer-term relationships, there may be a greater tendency to want to get pregnant if one has an older partner who is independent and seemingly capable of supporting a family. Although a range of problem behaviors was assessed in the present study, there may, of course, be other harmful effects associated with older partner’s age that were not measured. This is a limitation that hopefully will be addressed by additional research. Future research also needs to examine whether the nature of these relationships vary as a function of how much older the sexual partner is, e.g., is the partner considered a “boyfriend,” how long does the relationship last, did the girl feel manipulated or coerced even if no force was used or threatened. It should be highlighted that regardless of partner’s age, the younger the girls’ were when they first had intercourse the greater the number of other problem behaviors observed, a finding that is consistent with the prior literature (Irwin and Millstein, 1992). This was especially evident for girls who first had intercourse in very early adolescence, 11–12. They were more truant, had made more suicide attempts, had more pregnancies, used alcohol more, and abused other substances more than girls who first had intercourse in their teens. In addition, girls who first had intercourse as a young teenager (13–15) exhibited significantly more behavior problems (pregnancies and substance abuse) than girls who first had intercourse as an older teenager (16–18). Although dysfunctional families (Newcomer and Udry, 1987) and prior behavior problems such as drinking, substance abuse, and truancy may have contributed to early experience with sex rather than vice versa (cf. Billy et al., 1988; Bingham and Crockett, 1996; Costa et al., 1995), it seems reasonable to hypothesize that very early sex may also exacerbate existing problems as well cause new ones, especially if the girl becomes pregnant or develops an STD or if she feels exploited or used, which is more likely to be the case the younger she is when she first has intercourse. Even if cause and effect is unclear, at the least the evidence suggests that on average it is not beneficial for young female teenagers to be having sex. These results provide further justification for keeping statutory rape laws on the books so long as they are enforced selectively and judiciously, e.g., only when a much older partner has taken advantage of an adolescent under age 16. It could be argued that without the presence of statutory rape laws in the background, many more older males than is currently the case would sexually exploit young teenage girls. We want to emphasize that although the present study was concerned only with the age disparity of heterosexual partners of female adolescents, future research should also examine the age distribution of sexual partners of male adolescents as
P1: FPN Archives of Sexual Behavior [asb]
PL110-327
April 19, 2000
11:56
Style file version Nov. 19th, 1999
214
Leitenberg and Saltzman
well as same sex partners for both adolescent girls and boys. Because the present study focused just on male partners of adolescent girls should not be taken as any endorsement by us of a double standard with regard to statutory rape laws. ACKNOWLEDGMENTS We thank the Vermont Office of Alcohol and Drug Abuse Programs, Department of Health, Agency of Human Services, for making this study possible. The assistance of Kelly Hale is especially appreciated. This article and the opinions expressed therein, however, are solely the responsibility of the authors. We also thank Bruce Compas, Karen Fondacaro, and Kay Jankowski for their comments on an early draft of the manuscript. REFERENCES The Alan Guttmacher Institute (1994). Sex and America’s Teenagers, The Alan Guttmacher Institute, Washington, DC. Alexander, C. S., Ensminger, M. E., Young, J. K., Smith, B. J., Johnson, K. E., and Dolan, L. J. (1989). Early sexual activity among adolescents in small towns and rural areas: Race and gender patterns. Family Plan. Perspect. 21: 261–266. Besharov, D. J., and Gardiner, K. N. (1997). Trends in teen sexual behavior. Child. Youth Serv. Rev. 19: 341–367. Billy, J. O., Lindale, N. S., Grady, W. R., and Zimmerle, D. M. (1988). Effects of sexual activity on adolescent social and psychological development. Soc. Psychol. Q. 51: 190–212. Bingham, C. R., and Crockett, L. J. (1996). Longitudinal adjustment of boys and girls experiencing early, middle, and late sexual intercourse. Dev. Psychol. 32: 647–658. Brener, N. D., Collins, J. L., Kann, L., Warren, C. W., and Williams, B. I. (1995). Reliability of the Youth Risk Behavior Survey Questionnaire. Am. J. Epidemiol. 141: 575–580. Brooks-Gunn, J., and Furstenberg, F. F., Jr. (1989). Adolescent sexual behavior. Am. Psychol. 44: 249–257. Caron, S. L. (1998). Cross-Cultural Perspectives on Human Sexuality, Allyn and Bacon, Boston. Centers for Disease Control and Prevention (1996). Sexually transmitted diseases and adolescents. State Legislat. 22: 7. Coley, R. L., and Chase-Lansdale, P. L. (1998). Adolescent pregnancy and parenthood: Recent evidence and future directions. Am. Psychol. 53: 152–166. Costa, F. M., Jessor, R., Donovan, J. E., and Fortenberry, J. D. (1995). Early initiation to sexual intercourse: The influence of psychosocial unconventionality. J. Res. Adolesc. 5: 93–121. Darroch, J. E., Landry, D. J., and Oslak, S. (1999). Age differences between sexual partners in the United States. Family Plan. Perspect. 31: 160–167. Donovan, P. (1997). Can statutory rape laws be effective in preventing adolescent pregnancy? Family Plan. Perspect. 29: 30–34. Elo, I. T., King, R. B., and Furstenberg, F. F. (1999). Adolescent females: Their sexual partners and the fathers of their children. J. Marr. Family 61: 74–84. Finkelhor, D. (1984). Child Sexual Abuse: New Theory and Research, Free Press, New York. Forrest, J. D. (1990). Cultural influences on adolescents’ reproductive behavior. In Bancroft, J., and Renisch, J. M. (eds.), Adolescence and Puberty, Oxford University Press, New York, pp. 234–253. Francoer, R. T. (ed.) (1997). The International Encyclopedia of Sexuality, Vols. I and II, Continuum, New York. Henshaw, S. K. (1993). Teenage abortion, birth, and pregnancy statistics by state, 1988. Family Plan. Perspect. 25: 122–126.
P1: FPN Archives of Sexual Behavior [asb]
PL110-327
Partner’s Age Disparity
April 19, 2000
11:56
Style file version Nov. 19th, 1999
215
Irwin, C. E., and Millstein, S. G. (1992). Risk-taking behaviors and biopsychosocial development during adolescence. In Sussman, E. J., Feagans, T. V., and Ray, W. J. (eds.), Emotion, Cognition, Health, and Development in Children and Adolescents, Earlbaum Associates, HIllsdale, NJ, pp. 95–102. Jemmott, J. B., Jemmott, L. S., and Fong, G. T. (1998). Abstinence and safer sex HIV risk-reduction interventions for African-American adolescents: A randomized controlled trial. JAMA 279: 1529– 1536. Jessor, R., and Jessor, S. L. (1977). Problem Behavior and Psychological Development: A Longitudinal Study of Youth, Academic Press, New York. Lamb, M. E., Elster, H. B., and Tavare, J. (1986). Behavioral profiles of adolescent mothers and partners with varying intracouple age differences. J. Adolesc. Res. 1: 399–408. Landry, D. J., and Forrest, J. D. (1995). How old are U.S. fathers? Family. Plan. Perspect. 27: 159–165. Lindberg, L. D., Sonenstein, F. L., Ku, L., and Martinez, G. (1997). Age differences between minors who give birth and their adult partners. Family Plan. Perspect. 29: 61–66. Males, M. (1992). Adult liaison in the “epidemic” of “teenage” birth, pregnancy, and venereal disease. J. Sex. Res. 29: 525–545. Males, M., and Chew, K. S. Y. (1996). The ages of fathers in California adolescent births, 1993. Am. J. Public Health 86: 565–568. Miller, K. S., Clark, L. F., and Moore, J. S. (1997). Sexual initiation with older male partners and subsequent HIV risk behavior among female adolescents. Family Plan. Perspect. 29: 212–214. Mott, F. L., Fondell, M. M., Hu, P. N., Kowaleski-Jones, L., and Menaghan, E. G. (1996). The determinants of first sex by age 14 in a high-risk adolescent population. Family Plan. Perspect. 28: 13–18. Newcomer, S., and Udry, J. R. (1987). Parental marital status effects on adolescent sexual behavior. J. Marr. Family 49: 235–240. Oberman, M. (1994). Turning girls into women: Re-evaluating modern statutory rape law. J. Crim. Law Criminol. 85: 15–79. Seidman, S. N., and Rieder, R. O. (1994). A review of sexual behavior in the United States. Am. J. Psychiatry 151: 330–341. Warren, C. W., Santelli, J. S., Everett, S. A., Kann, L., Collins, J. L., Cassell, C., and Morris, L. (1998). Sexual behavior among U.S. high school students, 1990–1995. Family Plan. Perspect. 30: 170–172.
P1: FLF Archives of Sexual Behavior [asb]
PL110-328
April 19, 2000
12:3
Style file version Nov. 19th, 1999
Archives of Sexual Behavior, Vol. 29, No. 3, 2000
An Instrument to Measure Safer Sex Strategies Used by Male Sex Workers Rodrigo Marino, ˜ Ph.D.,1 Jan Browne, Ph.D.,1 and Victor Minichiello, Ph.D.1,2
Several studies have related safe sex in the commercial sex encounter to the ability of sex workers to apply specific safer sex strategies. However, no instrument has been previously available to measure these skills. The Safer-Sex Strategy Scale (SSS) was developed for such purposes. The psychometric properties (reliability and validity) were evaluated with a sample of sex workers recruited from Australian cities. A questionnaire was administered to 184 sex workers aged 18–58 years using a convenience sampling method. Results indicated that the SSS can be conceptualized as a four variation construct and can be reliably (all subscales yielded at least a Cronbach’s α = .60) and validly measured. Additionally, the relationship of the SSS to various sex work experiences was explored. The results show that the length of time working as a sex worker and contact with sex work organizations and sexual health clinics appear to influence the level of agreement of using the safer sex strategies evaluated. KEY WORDS: safer sex; scale; male sex worker; sex industry.
INTRODUCTION Exposure to sexually transmissible infections has always been an occupational threat for sex workers. Recent studies show that in countries where sex work has been decriminalized and a professional “work” perspective promoted in the industry, sex workers are more likely to report a higher commitment to practice safer sex (Donovan et al., 1998). Within a professional work context, a male sex worker (MSW), when confronted with the client’s desire to practice unsafe sex, must choose what action to take to ensure a safer outcome. Broadly, these strategies include: exclusion of anal intercourse from the commercial sex repertoire (Kippax 1 School
of Health, University of New England, Armidale, New South Wales 2351, Australia.
2 To whom correspondence should be addressed. e-mail:
[email protected]. Fax: +61 267 73 3666.
217 C 2000 Plenum Publishing Corporation 0004-0002/00/0600-0217$18.00/0 °
P1: FLF Archives of Sexual Behavior [asb]
PL110-328
April 19, 2000
12:3
218
Style file version Nov. 19th, 1999
Marino, ˜ Browne, and Minichiello
et al., 1993; Bloor et al., 1993), being selective in the recruitment of clients (Bloor et al., 1992), and the adoption of specific strategies in the safer-sex negotiation (Bloor et al., 1993). Research has linked safe commercial sex to the negotiating and communication skills used by the sex worker in his work context (Bloor et al., 1992, 1993), but only Browne and Minichiello (1995) have defined and conceptualised the usage of safer sex strategies as independent constructs. Following a series of in-depth interviews with MSWs in Melbourne, Browne and Minichiello identified six safer sex strategies which sex workers use to achieve a safer sex encounter. These include the natural mode, the teaching mode, the alternative sex mode, the challenge mode, the trick sex mode, and the walk-out mode. The results from this qualitative study revealed that the development of an instrument that measures the sex worker’s commitment to using safe sex strategies may be useful. This information could be utilized to develop programs which encourage MSWs to negotiate safer sex by identifying which strategies could be reinforced or strengthened. It would also help us to identify which particular group of MSWs require further education on safer sex skills. If differences exist, for example, with regard to the adoption of specific safer sex strategies by work experience in the sex industry, then an instrument such as a safer sex strategy scale may reveal such patterns. For example, Bloor et al. (1992) have noted that a safer-sex outcome is partly dependent on the commitment of the MSWs to have safer sex, and “trick of the trade” strategies, which are learned and developed over time and are more likely to be used among experienced MSWs. Furthermore, research on gay men and heterosexual populations reveal that the decision to practice safer sex takes place within a context of peer influence and values (Kelly et al., 1989; Cohen, 1993; Kippax et al., 1993). It is possible that this context may also exist for sex workers, and that community involvement, in terms of contact with sex workers’ organizations, might be a key mediating factor, as suggested by Minichiello and colleagues (1998), in reinforcing safer-sex norms. This paper, first, reports first on the construction of a Safer-Sex Strategy Scale based on Browne and Minichiello’s classification model presented in Table I and, second, presents the results of tests on the psychometric properties of the scale. The paper presents the development of the scale in a sample of MSW from Australian cities and examines the psychometric properties of the scale. It then examines whether the distribution of responses with respect to safer sex strategies are influenced by the length of work experience in the industry and level of contact with community and health organizations that provide a service to MSWs. METHOD Instrument The items for the Safer-Sex Strategy Scale (SSS) were generated following the safe sex strategy model delineated by Browne and Minichiello (1995). The
P1: FLF Archives of Sexual Behavior [asb]
PL110-328
April 19, 2000
12:3
Style file version Nov. 19th, 1999
Measuring Safer Sex Strategies
219
Table I. Safer Sex Strategies Included in the Browne and Minichiello Model Strategy mode Natural Teaching Alternative sex
Challenge Trick sex Walk-out
a Source:
Description The MSW treats safe sex as a natural, expected, aspect of sex. This form of interacting includes thinking about safe sex as pleasurable. The MSW takes responsibility for teaching clients about AIDS and other STDs and nonpenetrative sex techniques, and reminds clients of the risk of unprotected sex. This mode involves proposing other forms of sex in which there is no exchange of bodily fluids or penetration. The worker may encourage an experimentation of client’s sexuality and new safer sexual experiences. The worker challenges the client’s arguments for unsafe sex by providing evidence against the underlying assumptions regarding safe sex and unsafe sex. Sex workers may use some “tricks of the trade” to put condoms on, or make the client believe he is penetrating him, when he is not. The client is so resistant to safe sex that the worker must choose between either walking out or having unsafe sex, the worker leaves if the client persists in refusing safe sex.
Browne and Minichiello (1995).
scale contains a total of six strategy variations (see Table I) and each variation originally included a total of five items. Prior to the finalization of the instrument and data collection, a series of meetings with expert outreach workers from three sex work organizations [Sex Workers Outreach Project (SWOP) in Sydney, Self-Health Queensland Workers in the Sex Industry (SQWISI) in Brisbane, and the Prostitutes Collective Victoria (PCV) in Melbourne] were organized to assess the content design and the format of the instrument and to ensure that the items and the instructions were understandable to the target population. The expert consultation stage resulted in the original scale being revised to an instrument which contained 18 items, 3 for each of the six strategy variations (see Appendix). Each item consisted of a sentence which described a situation and then asked the sex worker to reveal his level of agreement with using that particular course of action to obtain client compliance with safe-sex practices. A 5-point Likert type rating scale with anchors 1 (strongly agree) to 5 (strongly disagree) was used as the response format. The items were arranged in a random fashion in the questionnaire. The final version of the SSS was incorporated into a survey which explored the sociodemographic characteristics, attitudes, and sex work experiences of male sex workers (Minichiello et al., 1999). Sample With approval from the University of New England Ethics and Research Committee, male sex workers from Sydney, Melbourne, and Brisbane were invited to participate in the study. One hundred eighty-four sex workers gave their written
P1: FLF Archives of Sexual Behavior [asb]
PL110-328
April 19, 2000
12:3
220
Style file version Nov. 19th, 1999
Marino, ˜ Browne, and Minichiello
consent to participate. The sample was predominantly recruited from Sydney (44.8%), Melbourne (29.1%), and Brisbane (20.1%), with the remaining 6% from Perth and Adelaide. The age of the respondents ranged from 18 to 58 years; 46.7% were younger than 25 years old (mean age, 27.1 years). The majority (39.1%) worked as independent sex workers, 22.3% worked as escorts, 15.2% worked as street workers, and another 23.4% indicated that they worked in more than one type of sex work. With regard to the length of time working as a sex worker, 26.6% indicated periods of less than 6 months, while 40.2% of the MSWs indicated working for more than 5 years. Seventy-six participants (41.3%) indicated that they worked part-time as sex workers, and another 38.6% indicated that they worked on a full-time basis. The remaining 20.1% indicated that they worked either opportunistically, on weekends only, or when “they feel like it.” The majority (70.9%) had completed secondary education, with 41.8% having at least some additional tertiary education. About 64% of the sample had contact with either sex workers organizations or sexual health clinics “often or very often,” while less than 19% “never or very rarely” contacted these organizations. To evaluate whether the same MSW contacted both sex work organizations and sexual health clinics, the level of agreement on those who contacted one or the other was evaluated using the kappa statistic. The results reveal (κ < .20) that the pattern of contact with one of these organizations is different from the participants’ contact with the other; that is, the MSWs who visited the sex work organizations were not the same MSWs who necessarily visited the sexual health clinics. Procedures The questionnaire was distributed to a sample of sex workers between April 1998 to June 1998 by outreach workers employed by sex worker organizations. Respondents completed the questionnaire anonymously in their own time and were asked to return it directly to the outreach workers. RESULTS Internal Consistency Internal consistency refers to the degree to which items within the same dimension are correlated to each other. Internal consistency was examined on all of the three-item subscales, using Cronbach’s alpha. Each of the subscales was treated as an independent subscale. Respondents with missing data were excluded from the reliability analysis. The results reveal that for the natural subscale the α was .64, the teaching subscale had an α of .59, the alternative sex subscale had an α of .71, the challenge subscale had an α of .60, the trick sex subscale had an α of .58 and the walk-out subscale had an α of .75.
P1: FLF Archives of Sexual Behavior [asb]
PL110-328
April 19, 2000
12:3
Measuring Safer Sex Strategies
Style file version Nov. 19th, 1999
221
Validity As no standard criterion to assess this inquiry was available, the construct validity approach relied on factor analysis to explore the underlying factor structure of the constructs under study. If construct validity exists, the number of factors should approximate the number of dimensions assessed by each of the measures (Waltz et al., 1991). To determine content validity, the list of the items selected for the instrument was given to a panel of experts to assess the degree to which items adequately represent the contents in the domains of interest (Waltz et al., 1991). Furthermore, material from Browne and Minichiello’s (1995) in-depth interviews elicited a range of statements which were used in the construction of the items. This provided a degree of authenticity, which in turn contributed to the validity of the scale (Dawis, 1987). Construct Validity Before the analysis of the construct validity was undertaken, two items were eliminated to avoid redundancies (see Appendix items 3 and 14). Selection of items to be deleted within each subscale was on the basis of their internal reliabilities. The item that negatively affected the result was deleted. One of these items belonged to the alternative sex strategy, and the other to the original challenge strategy. Scores on the remaining 16 items were factor analyzed using principalcomponent analysis (PCA) extraction procedures. The number of factors in the final solution was determined by eigenvalues greater than 1.0. This analysis identified five factors with eigenvalues greater than 1.0. These five factors accounted for 59.7% of the total variance. The five-factor solution was retained and rotated using oblique rotation (oblimin) procedures, as it was assumed that these factors would be correlated. Items which correlated at least .30 with the corresponding rotated component were used to interpret factors (Waltz et al., 1991). The first factor (alternative sex) was composed of five items: two items addressing the alternative sex strategy and one each from the teaching, the challenge, and the natural strategies. The alternative sex strategy factor accounted for 24.7% of the variance. The second factor (trick sex) consisted of the three items from that subscale and explained 10.8% of the variance. The third factor contained four items each addressing the walk-out strategy and one from the challenge mode, and it accounted for 10.5% of the explained variance. The fourth factor contained three items, two items addressing the natural strategy and one from the teaching mode. This factor accounted for 7.3% of the variance. The fifth factor accounted for 6.5% of the variance and was composed of one item from the teaching strategy. With four exceptions, two items on the teaching strategy (see Appendix items 2 and 13), one item from the natural strategy and one item belonging to the walk-out strategy, none of the items loaded greater than .30 on any other subscale. However,
P1: FLF Archives of Sexual Behavior [asb]
222
PL110-328
April 19, 2000
12:3
Style file version Nov. 19th, 1999
Marino, ˜ Browne, and Minichiello
three items (Nos. 2, 13, and 5), had similar low loadings in more than one factor and, therefore, were highly nonspecific and eliminated from further analysis of the scale. A second PCA was conducted with the 13 remaining items. This analysis revealed four factors with eigenvalues greater then 1.0. Thus, the teaching and the challenge strategies were eliminated from the final version of the scale, although some of the items belonging to these strategies were still present in other factors. Internal consistencies were computed for each of the four PCA derived subscales, namely, the walk-out, the trick, the alternative sex, and the natural subscales, yielding Cronbach’s α coefficients of .75, .58, .61, and .53, respectively. Moreover, inspection of the corrected item–total correlation coefficient revealed that there was not a strong relationship between one item (Item 16) and the other items in the subscale, and therefore this item was affecting the alpha value of the factor derived from the natural subscale. This result indicated that if that item was omitted from the subscale, the resulting subscale had an α of .65. Thus, the final version of the scale was composed of 12 items on four of the six original dimensions of the safe-sex strategies scale. A third factor analysis was performed with the 12 items. Table II presents the factor structure matrix and loadings for the final solution and includes the eigenvalue and percentage of variance explained for each factor. The participants’ responses to these 12 items were averaged to yield four new variables with the name of the subscale (walk-out, trick sex, alternative sex, and natural) which represented the individual’s scores for that subscale. Finally, to examine if any of the resulting scales were so highly interrelated that they could be considered a parallel measure of a single construct, correlation analyses were performed using Pearson correlation coefficient. Table III presents the correlation matrix for the four subscales. Low but significant correlations were observed between the walk-out and the natural modes (r = .27). Two of the other modes were significantly associated at the .05 level of confidence; however, the strength of the associations indicate that the correlations were weak, at best explaining only 7% and as low as 2.5% of the variance. The other three modes were not significantly associated. To examine further if any pattern of relationships between responses to different scales existed, each subject’s response for each of the scales was plotted against each other to see scattergram representation. Scatterplots indicated that despite some moderate associations, no discernible pattern of response between two scales was found. To put this scale into practice, the scores of the four subscales served as the dependent variable in a series of Kruskall–Wallis one-way analyses of variance. This examined differences in the level of agreement of the strategies by three independent variables: length of work, contact with sexual health clinics, and contact with sex work organizations. These tests show that the strategies used varied significantly by length of work, contact with sexual health clinics, and contact with sex work organizations (see Table IV). The analysis by length of work revealed a significant effect on the level of endorsement of the trick sex strategies.
P1: FLF Archives of Sexual Behavior [asb]
PL110-328
April 19, 2000
12:3
Style file version Nov. 19th, 1999
Measuring Safer Sex Strategies
223
Table II. Final Scale Rotated Factor Structure Matrix and Loadings, Eigenvalues, and Variance Explained by Each Factor Factor 1
Factor 2
Factor 3
Factor 4
Walk-out 1. If the client does not want to negotiate safe sex, I just tell him to get someone else and to fuck off 2. If the client says: I am not going to wear a condom and that is it, I reply: Forget it then 3. You get sick of trying to convince the client that you only have safe sex, You get so fed up that you walk out 4. If the client says: If you are clean there is no need to wear a condom, I reply: I know that I am clean, but how can you know for sure that I am clean?
.83629 .82400 .66938 .66520
Trick sex 1. I use my mouth to apply condoms without clients knowing I am using them 2. If the client will not wear a condom, I do not say anything, but I put the condom on without the client knowing 3. If the client will not wear a condom, I use my hand or tightly clenched thighs to make the client believe he is penetrating me
.74790 .73053 .71132
Alternative sex 1. When the client will not wear a condom, I suggest safe options 2. I show my clients alternatives like how nonpenetrative sex can be fun 3. If the client does not want to wear a condom, I tell the client about the risks for both of us if we do not use condoms Natural 1. Safe sex is done naturally and does not interfere with the flow of sex 2. I treat safe sex as natural Eigenvalue Percentage of variance explained
−.84047 −.80728 .34939
−.55904
−.84732
2.99 24.7
1.72 10.8
−.82747 1.12 6.5
1.52 10.5
Table III. Correlation Coefficient Between Pairs of SSS Subscales Subscale
Trick sex mode
Natural mode
Walk-out mode
Trick sex mode Natural mode Walk-out mode Alternative sex mode
— .18∗ .09 .06
— .27∗∗ .16∗
— .14
∗ p < .05. ∗∗ p < .01.
P1: FLF Archives of Sexual Behavior [asb]
PL110-328
April 19, 2000
12:3
224
Style file version Nov. 19th, 1999
Marino, ˜ Browne, and Minichiello Table IV. Mean Scores and Standard Deviation of the SSS in Relation to Length of Time as MSW and Contact with Sex Work and Sexual Health Organizations Safe sex strategy mode [mean (SD)]
Length of time as MSW Less than 6 mo Between 6 and 12 mo Between 1 and 2 yr Between 2 and 5 yr More than 5 yr Contacts with sex workers’ organizations Very often Often Rarely Very rarely/never Contacts with sexual health clinics Very often Often Rarely Very rarely/never
N
Natural
Alternative sex
Trick sex
Walk-out
49 33 28 39 35
2.10 (0.77) 1.82 (0.79) 1.77 (0.81) 1.83 (0.81) 1.83 (0.75) *
2.00 (0.58) 1.85 (0.67) 1.86 (0.78) 2.06 (0.91) 1.96 (0.54)
* 3.18 (0.82) 3.22 (1.08) 3.29 (0.77) 3.30 (0.87) 2.71 (0.69)
2.06 (0.78) 1.97 (0.99) 2.11 (0.87) 1.81 (0.77) 2.05 (0.71)
39 80 35 30
1.65 (0.76) 2.02 (0.77) 1.97 (0.87) 1.77 (0.72)
1.77 (0.62) 2.08 (0.76) 2.05 (0.63) 1.77 (0.64) *
2.98 (0.87) 3.11 (0.82) 3.35 (0.73) 3.18 (0.98)
1.82 (0.79) 2.06 (0.86) 2.10 (0.78) 1.95 (0.78)
51 71 24 37
1.78 (0.84) 2.04 (0.73) 1.92 (0.73) 1.76 (0.83)
1.75 (0.65) 2.09 (0.62) 2.04 (0.71) 1.95 (0.85)
3.07 (1.00) 3.17 (0.81) 2.96 (0.76) 3.24 (0.84)
1.99 (0.95) 1.92 (0.72) 2.03 (0.75) 2.16 (0.86)
∗ p < .05.
The results indicated that those MSWs with a longer work experience (more than 5 years) agreed more with using the trick sex strategy ( p < .05). The distribution of scores according to level of contact with sex workers’ organizations showed significant differences in the natural strategy ( p < .05), with those who very often contacted sex workers organizations more likely to indicate a higher endorsement of this strategy. Additionally, the alternative sex subscale was close to reaching significance ( p = .07) by level of contact with sex workers’ organizations. Groups at both extremes of level of contact tended to agree more with this strategy. Those who never had any contact with sexual health clinics showed a higher endorsement of using the alternative sex strategy ( p < .05). In addition, the results indicated that level of endorsement of the natural strategy was close to significant levels ( p < .07). Here again, participants at the extremes of level of contact, that is, those whose contact with sexual health clinics was either very often or never tended to agree more with using this strategy than those in the middle categories. Little or no distinction was found for the walk-out modes ( p > .10) by level of contact with sex work organizations or sexual health clinics or by length of time in sex work.
CONCLUSION The study shows that safer sex strategies is a complex construct, but quantifiable along multiple dimensions that are relatively independent of one another
P1: FLF Archives of Sexual Behavior [asb]
PL110-328
Measuring Safer Sex Strategies
April 19, 2000
12:3
Style file version Nov. 19th, 1999
225
in a psychometrically valid manner. The dimensional structure of the scale was examined through the use of principal-components analysis. This preliminary analysis found that the dimensional structure of the Safer-Sex Strategy Scale consists of four interpretable and internally consistent dimensions. Based upon these analyses, it was possible to determine which items best measure these dimensions and which items could be dropped in order to construct the final version of the scale which consisted of 12 items (see Table II). Regarding internal consistency, with the exception of the trick sex subscale, all of the final subscales reached or exceeded the minimum of .60, which is considered an acceptable result for group comparisons (Nunally, 1967). Furthermore, considering the low number of items in each dimension, this may be regarded as a good result. The findings also provide substantial evidence of the validity of the Safer-Sex Strategy Scale. The empirical factor structure (four factors) partially resembled the theorised dimensions of this construct (Browne and Minichiello, 1995). The results of the Kruskall–Wallis tests suggest that strategies differ by length of time in sex work and by level of contact with sex work organizations and sexual health clinics. However, in interpreting these results, it must be remembered that the Safer-Sex Strategy Scale focuses on an assessment of the level of agreement to apply specific knowledge and skills, and therefore, it is based on what the MSW agrees with rather than what he is actually doing in a particular commercial sexual encounter. The findings have two important implications. First, the data suggest that contact with sex workers’ organizations (in the case of the natural sex strategy) and with sexual health clinics (in the case of the alternative strategy) may play an important role in determining the level of agreement (and probably on their application) of the various safe-sex strategies, together with the exposure to the sex work culture (in the case of the trick sex strategy), as measured by length of time in sex work. However, these differences by level of contact or length of time in sex work were more a matter of degree of the agreement for the different strategies rather than on the direction of the agreement of these strategies (i.e., agreement vs disagreement), as most responses varied only between strongly agree and agree. It must be kept in mind that this study was conducted in large cities in Australia, where information about transmission and prevention of HIV/STDs are relatively available and accessible. Accordingly, the MSWs may be endorsing these strategies because they possess information about safe sex negotiations. Second, the linear association between the length of time in sex work and the level of endorsement of the different strategies was considered to some extent obvious and has been discussed in the literature (Bloor et al., 1992; Browne and Minichiello, 1995). However, the effects of the level of endorsement of safer sex strategies according to contact with sexual health clinics and sex workers’ organizations were unexpected. The direction of the associations was not as straightforward, and in most cases, the differences did not show a linear relationship. The level of
P1: FLF Archives of Sexual Behavior [asb]
PL110-328
April 19, 2000
12:3
226
Style file version Nov. 19th, 1999
Marino, ˜ Browne, and Minichiello
endorsement at the extreme end of the scale (very often or very rarely/never) tended to approach similar levels (see Table IV). Further research is needed to identify the source of information for safer sex strategies among the sex work population who have different levels of contact with various community and health organizations. In addition, research also needs to assess the ulitity of the SSS with larger groups of MSWs to determine differences between subgroups of MSWs, for example, between homeless and nonhomeless MSWs and people who use drugs compared to people who do not use drugs, in terms of endorsement of the various safer sex strategies. The safer sex campaigns to date have relied mostly on instructing sex workers to use condoms, and likewise, many studies rely on measures of the workers’ commitment to use condoms to predict whether safer sex was an outcome achieved in sexual encounters (Browne and Minichiello, 1995). What this assumption fails to understand is that the availability of a condom on either a sex worker’s or a client’s possession may not necessarily result in the usage of that condom during a particular sex encounter. The literature has shown that power relationship and the interactional context may influence the usage of condoms in commercial sex (Minichiello et al., 1998). The utility of the Safer-Sex Strategy Scale is that it may provide a better understanding of what particular skills the sex worker possesses or lacks in facilitating or preventing him from achieving a safer sex encounter. The knowledge gained from such a scale may be essential to support the message that it is a public health responsibility not only for sex workers to carry and use condoms, but also to show them how a repertoire of safer sex strategies can be used in various situations to move the encounter toward ensuring that condoms are more likely to be used with their clients. ACKNOWLEDGMENTS This study was funded by a grant received from the National Health and Medical Research Council. The authors acknowledge the support received from the Prostitutes Collective of Victoria, Sex Workers Outreach Project, and the SelfHealth Queensland Workers in the Sex Industry. In particular, we acknowledge Kenn Robinson, Kirk Peterson, Brad Reuter, and John Jones for assisting with the collection of the data. We thank Maria McMahon from the Sex Work Outreach Project for her assistance in the development of the questionnaire. REFERENCES Bloor, M. J., McKeganey, N. P., Finlay, A., and Barnard, M. A. (1992). The inappropriateness of psychosocial models of risk behaviour for understanding HIV-related risk practices among Glasgow male prostitutes. AIDS Care 4: 131–137. Bloor, M. J., Barnard, M. A., Finlay, A., and McKeganey, N. P. (1993). HIV-related risk practices among Glasgow male prostitutes: Reframing concepts of risk behavior. Med. Anth. Q. 7: 152–169.
P1: FLF Archives of Sexual Behavior [asb]
PL110-328
April 19, 2000
12:3
Style file version Nov. 19th, 1999
Measuring Safer Sex Strategies
227
Browne, J., and Minichiello, V. (1995). The social meanings behind male sex work: Implications for sexual interactions. Br. J. Soc. 46: 598–622. Browne, J., and Minichiello, V. (1996). The social and work context of commercial sex between men: A research note. Aust. N.Z. J. Soc. 32: 86–92. Cohen, M. (1991). Changing to safer sex: Personality, logic and habit. In Aggleton, P., Hart, G., and Davies, P. (eds.), AIDS: Responses, Interventions and Care, Falmer Press, London, pp. 19–42. Dawis, R. V. (1987). Scale construction. J. Couns. Psychol. 34: 481–489. Donovan, B., Hart, G., and Minichiello, V. (1998). Australia. In Brown, T., Chan, R., Mugrditchian, D., Mulhall, B., Plummer, D., Sarda, R., Sittitrai, W., and Minichiello, V. (eds.), Sexually Transmitted Diseases in Asia and the Pacific, Venereology, Armidale, pp. 26–61. Kelly, J. A., St Lawrence, J. S., Hood, H. V., and Brasfield, T. L. (1989). An objective test of AIDS risk behavior knowledge: Scale development, validation, and norms. J. Behav. Ther. Exp. Psychiat. 20: 227–234. Kippax, S., Crawford, J., Davis, M., Rodden, P., and Dowsett, G. (1993). Sustaining safe sex: A longitudinal study of a sample of homosexual men. AIDS 7: 259–263. Minichiello, V., Marino, R., Browne, J., and Jamieson, M. (1998). A review of male to male commercial sex encounters. Venereology 11: 32–42. Minichiello, V., Marino, R., Browne, J., and Jamieson, M. (2000). Male sex workers in Australian cities: Socio-demographic and sex work characteristics. J. Homo. (in press). Nunnally, J. C. (1967). Psychometric Theory, McGraw-Hill, New York. Waltz, C. F., Strickband, O. L., and Lenz, E. R. (1991). Measurement in Nursing Research, 2nd ed., F. A. Davies, Philadelphia.
APPENDIX: SAFER-SEX STRATEGY SCALE (SSS) USED IN THE SURVEY The following are ways which sex workers can use to get clients to practice safe sex. How would you respond if you were in the situations described below? Mark the extent to which you agree or disagree with the statements. Strongly agree Agree Neither agree nor disagree Disagree Strongly disagree (1) (2) (3) (4) (5) 1. I treat safe sex as natural. 2. I take responsibility for teaching the client about safe sex. 3. If the client does not want to wear a condom, I tell him about the risks involved in not using condoms. 4. I use my mouth to apply condoms without clients knowing I am using them. 5. I carry condoms and lube and make them part of the sex act. 6. If the client will not wear a condom, I do not say anything, but I put the condom on without the client knowing. 7. If the client does not want to wear a condom, I tell the client about the risks for both of us if we do not use condoms. 8. You get sick of trying to convince the client that you only have safe sex. You get so fed up that you walk out. 9. Safe sex is done naturally and does not interfere with the flow of sex.
P1: FLF Archives of Sexual Behavior [asb]
228
PL110-328
April 19, 2000
12:3
Style file version Nov. 19th, 1999
Marino, ˜ Browne, and Minichiello
10. I show my clients alternatives like how nonpenetrative sex can be fun. 11. When the client will not wear a condom, I suggest safe options. 12. If the client will not wear a condom, I use my hand or tightly clenched thighs to make the client believe he is penetrating me. 13. I take an active role in showing my clients signs and symptoms of sexually transmissible infections and tell them how to get treatment. 14. If the client does not like condoms, I suggest something safer. 15. If the client says, “I am not going to wear a condom and that is it,” I reply, “Forget it then.” 16. I carry safe sex information and give it to my clients. 17. If the client does not want to negotiate safe sex, I just tell him to get someone else and to fuck off. 18. If the client says, “If you are clean there is no need to wear a condom,” I reply, “I know that I am clean, but how can you know for sure that I am clean?”
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
Archives of Sexual Behavior, Vol. 29, No. 3, 2000
The Influence of Gender on Sex: A Study of Men’s and Women’s Self-Reported High-Risk Sex Behavior Lisa A. Cubbins, Ph.D.,1,3 and Koray Tanfer, Ph.D.2
An investigation is presented of the relationship between gender and five selfreported high-risk sex behaviors: ever having had casual sex, the lifetime number of vaginal sex partners, the lifetime number of anal sex partners, having had multiple vaginal sex partners over the short term, and having had multiple anal sex partners over the short term. The analysis was guided by a conceptual model that emphasized the constraints and opportunities for high-risk sex behavior that arise from an individual’s structural position and cultural context. Gender differences in high-risk sex behaviors were predicted to be due to differences in men’s and women’s family roles, work roles, religious behaviors, and past sex experience. In addition, the effects of certain sociocultural factors on the high-risk sex behaviors were expected to be dependent on an individual’s gender. The hypotheses were evaluated using national data from the United States on self-reported sex behaviors for men ages 20 to 39 years old and women ages 20 to 37 years old. Data analyses were conducted using ordinary least-squares regression and logistic regression. Findings provided mixed support for the predictions. Gender was not significantly related to short-term, self-reported high-risk sex behaviors once social and cultural factors were included in the statistical models. But it continued to predict lifetime behaviors. Several variables, including race, age, age at first sex, and marital status, had gender-specific effects on the self-reported high-risk sex behaviors. The study demonstrates how the effects of structural and cultural factors on sex behavior differ for men and women. KEY WORDS: vaginal sex; anal sex; casual sex; sex partners.
1 Department
of Sociology, University of Cincinnati, P.O. Box 210378, Cincinnati, Ohio 45221-0378. Memorial Institute. 3 To whom correspondence should be addressed, e-mail:
[email protected]. 2 Battelle
229 C 2000 Plenum Publishing Corporation 0004-0002/00/0600-0229$18.00/0 °
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
230
Style file version Nov. 19th, 1999
Cubbins and Tanfer
INTRODUCTION Understanding sex behavior has taken on a new imperative given the increased spread of sexually transmitted diseases (STDs). In the early part of the decade, the U.S. Department of Health and Human Services (1991) estimated that nearly 12 million cases of STDs occurred annually, with an estimated 50 million Americans contracting incurable forms of STDs. Recent statistics indicate that while the rates of some STDs have declined (e.g., syphilis and gonorrhea), others have increased (e.g., chlamydia) (U.S. Department of Health and Human Services 1997). To a large extent, change in the incidence of STDs is a function of the socio-behavioral aspects of STD acquisition. Particularly important are certain sex behaviors that may increase the probability of infection. These high risk sex behaviors include age at first sex, certain types of sex practices (e.g., vaginal versus anal sex), the frequency of sex contact, and the number of sex partners over the short term and lifetime. Scholars from a variety of disciplines have turned their attention to the determinants of sex behavior. Within studies of sex behavior and the spread of STDs, gender has been a central factor, as it presents both biological and social influences on sex behavior. In this article, we extended past research by examining gender differences in high-risk sex behaviors using national data from the United States for men ages 20 to 39 years and women ages 20 to 37 years. We present analyses of high-risk sex behaviors based on a conceptual model that emphasizes the constraints and opportunities for sex behavior arising from an individual’s structural roles in the family and economy, as well as the cultural context of his or her ethnicity, religious affiliation, and region of residence. Long-standing gender differences in sex behavior are attributable partly to the differing social roles occupied by men and women, along with gender-specific cultural proscriptions on sex behavior (Schwartz and Rutter, 1998). We build on this understanding of sexuality by considering how gender affects the relationships between these structural and cultural conditions and a set of high-risk sex behaviors. What role does gender play in sex behavior? One effect of gender on sex behavior is indirect, as a determinant of life experiences associated with sex behavior. Several theoretical models of sex behavior are based on the assumption that sexuality is strongly influenced by the broadly defined social context (e.g., Gagnon and Simon, 1973; Reiss, 1986a,b). For example, social learning theorists argue that all aspects of sexuality—attitudes, meaning, behaviors—are the result of a variety of social forces. Individuals develop their sexual identity through a learning process that is related to their cultural group, family, peers, and specific circumstance. In this process, gender has a special role in shaping an individual’s sexuality, as males and females experience separate influences in their sexual development. Part of the outcome of gender differences in sexual development may be seen in males’ and females’ early sex experiences. While recent studies have shown a
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
The Influence of Gender on Sex
April 19, 2000
12:8
Style file version Nov. 19th, 1999
231
decline in gender differences in early sex behavior (Clement, 1989), Laumann and colleagues (1994) found that males still have earlier average ages of sex initiation. The timing of first sex activity is an aspect of an individual’s sex history that may indicate the extent of his or her sex experience and openness. The earlier individuals become sexually active, the more likely they may be to engage in less common sex behaviors, such as anal intercourse. This difference in early sex experience, then, may contribute to gender differences in adult sex behaviors. Both men and women tend to prefer sex in committed relationships over causal sex (Schwartz and Rutter, 1998). However, several studies indicate that men and boys are more open to recreational or casual sex than are women (Glass and Wright, 1992; Laumann et al., 1994; McCormick et al., 1984; Simon and Gagnon, 1984). Some have shown that women view emotional involvement as a prerequisite to sex (Cohen and Shotland, 1996; Taris and Semin, 1997). Other studies provide evidence for this view indicating that women express less support than men for sexual permissiveness (Hendrick et al., 1985; Smith, 1994; Sprecher, 1989; Wilson and Medora, 1990). Further, men appear to be more assertive in initiating sex encounters than are women (Byers and Heinlein, 1989; Grauerholz and Serpe, 1985; O’Sullivan and Byers, 1993). Perhaps as a consequence, men tend to have more sex partners (Forrest and Singh, 1990; Keller et al., 1982) and are more likely to engage in casual sex (Clark, 1990; Herold and Mewhinney, 1993). Similarly, how men and women view sex generally may account for gender differences in the likelihood of engaging in sex relations outside of committed partnerships (e.g., marriage or cohabitation) (Laumann et al., 1994; see Darnton, 1994). In this case, distinctions in sex approach or orientation are translated into gender differences in sex behavior. A second area in which gender may have an indirect effect on sex behavior is through cultural characteristics and social roles. For example, religious behavior has an important role in sex behavior, but also is a characteristic that differs by gender. Women tend to be more religiously active, and have greater religiosity, than men (The Gallup Report, 1987; Miller and Hoffmann, 1995). As most religious doctrines promote certain rules on sex conduct, we expected sex activity to differ by religious affiliation and level of religious activity. Individuals affiliated with conservative religions (e.g., conservative Catholics or fundamentalist Protestants) or who have high participation in religious organizations have lower sexual permissiveness (Davidson et al., 1995; Smith, 1994), and are likely to experience stronger normative controls over their sex activity, than others. Thus, certain sex behaviors (e.g., premarital sex, casual sex, noncoital sex) may be less likely to be practiced by individuals affiliated with conservative religions or those who have a high level of religious activity (Kost and Forrest, 1992; Thorton and Camburn, 1987; Trussell and Westoff, 1980). Because women tend to have higher religious participation than men (Batson et al., 1993; Cornwall, 1988), as well as a stronger personal religious commitment (Benson et al., 1989), women may be less likely
P1: FMN Archives of Sexual Behavior [asb]
232
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
Cubbins and Tanfer
than men to engage in certain sex behaviors. In addition, formally or informally, religious institutions may have more stringent rules on women’s than on men’s sex behavior, consistent with traditional boundaries on women’s sex behavior. Men and women also differ in terms of certain social roles that may affect their opportunities to engage in sex. These include family roles and paid work roles that influence both time availability and access to potential partners. In terms of family roles, individuals who have ever been married will have, at the minimum, one vaginal sex partner. Given that marriage usually involves at least a minimal level of monogamous sex relations, men and women who have been married are likely to have a lower total number of sex partners than those who have never been married. However, marriage may present fewer constraints on men’s sex behavior than on women’s, as evidenced by the higher number of extramarital sex partners found for men (Laumann et al., 1994). Thus, while marriage may reduce the number of sex partners for both genders, the reduction may be smaller for men than women. The presence of children in the home also may restrict sex behavior, though this may pose a greater constraint on women’s than on men’s sex activity. This is because women tend to be the primary caretakers of children (Brines, 1994). At the same time, men are more likely to be employed, and to work more paid hours on average (U.S. Women’s Bureau, 1993). This may present countervailing effects on men’s high-risk sex behavior. As employment may expand the number of potential partners, it also may place constraints on the time available for sex activity. A second way in which gender may affect sex behavior is by conditioning (or interacting with) the effects of other social or cultural influences. A characteristic or status may significantly influence sex behavior for both genders, but the effect of that factor on sex behavior may be larger for one gender than for the other. For example, some studies have shown differences in sex behavior by race and ethnicity. Blacks tend to be more permissive than non-Blacks of premarital and extramarital sex, though the reverse is true for homosexual relations (Smith, 1994). Compared to non-Blacks, Blacks begin sex activity at an earlier age, though there is mixed evidence as to whether or not Blacks have a higher frequency of coitus than non-Blacks (Billy et al., 1993; Tanfer and Cubbins, 1992). However, these race differences may be due to uncontrolled for structural or cultural influences. Although Hispanics appear to have a higher average coital frequency than non-Hispanics (Laumann et al., 1994), there is some evidence that Hispanics tend to be less likely than non-Hispanics to engage in premarital sex (Darabi, 1987). In addition, Hispanic women seem to have fewer lifetime sex partners than do nonHispanic women (Kost and Forrest, 1992). Again, these ethnic differences may be explained by other social and cultural factors correlated with both ethnicity and sex behavior. Race and ethnicity also may be associated with gender-specific rules of conduct that have direct or indirect effects on sex behavior, either by affecting
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
The Influence of Gender on Sex
April 19, 2000
12:8
Style file version Nov. 19th, 1999
233
opportunities for sex or by imposing informal sanctions on sex behavior. This may translate into sex attitudes and behaviors that are both gender- and race-specific. This is illustrated in the study of sex in the United States by Laumann et al. (1994). Considering group differences in sex attitudes, the authors found that men were more likely than women to have a recreational attitude toward sex, as opposed to relational or traditional views of sex. Among men, Blacks were most likely to have a recreational attitude toward sex, followed by Whites and then Hispanics. But among women, the order differed: White women were most likely to have a recreational attitude, followed by Hispanic women, then Black women. In this case, the effects of race or ethnicity may be most apparent when gender is taken into account. Gender-specific expectations of sex behavior within race or ethnic groups may help explain sex patterns by race, ethnicity, and gender. Sex behaviors also may be subject to cultural proscriptions based on region of residence. For example, individuals living in the South may be subject to a communitywide conservative ideology regarding sexual permissiveness (Smith, 1994) that reduces the motivation (as well as the opportunities) for sex activity, particularly premarital and extramarital sex. We expected that the more conservative context of living in the South would have larger effects on women’s than on men’s sex behavior. Other factors may affect one’s attitude toward sex behavior. Education tends to have a liberalizing effect on sex ideology, thus promoting less restrictive sex activity (Reiss, 1986a,b; Smith, 1994). Further, more highly educated individuals may have more sex partners from which to choose given their higher status. In contrast, education may enhance knowledge of disease risks, thus reducing risktaking behaviors (Leigh, 1983; Ross and Van Willigen, 1997). Since women are more likely than men to adopt health promotion and disease prevention behaviors (Verbrugge, 1989), education may accentuate this tendency among women. As such, education may have gender-specific effects on high-risk sex behaviors. Thus, gender differences in sex behavior may be due to differences in men’s and women’s social roles and cultural experiences; that is, gender may influence sex behavior indirectly through structural and cultural factors. In addition, though men and women may occupy the same social context or have similar characteristics, the effects of these factors may differ by gender. In turn, certain cultural experiences may have gender-specific effects on sex behavior. These include double-standards in which high-risk sex behavior may be more accepted or less sanctioned for one gender versus the other. Likewise, some religions have guidelines or rules on sexual activities that are more restrictive of women’s than men’s behavior. Based on this reasoning, we present two hypotheses for gender differences in high-risk sex behavior. First, we expected that gender differences in social and cultural characteristics (past sex behavior, religious affiliation and behavior, paid work roles, family roles) will account for gender differences in high-risk sex behavior. This hypothesis posits an indirect effect of gender through social roles
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
234
Style file version Nov. 19th, 1999
Cubbins and Tanfer
and cultural factors. And second, we predicted that the effects of certain social and cultural conditions (religious affiliation and behavior, family roles, education, region of residence, race, ethnicity) on high-risk sex behavior will differ by gender. We tested these hypotheses by analyzing five high-risk sex behaviors: (1) ever having had casual sex, including one-time sex encounters; (2) the total, lifetime number of vaginal sex partners; (3) the total, lifetime number of anal sex partners; (4) having had multiple vaginal sex partners over the short term; and (5) having had multiple anal sex partners over the short term. In our tests, we also controlled for age.4 The high-risk sex behaviors we analyzed are theoretically important in demonstrating gender differences in sex behavior. They also are relevant to research on gender and STD acquisition, as past studies have shown that the number of sex partners is related to the risk, as well as the perception of risk, of STD infection (Baldwin and Baldwin, 1988; Cochran and Peplau, 1991; Tanfer et al., 1995). The extent to which men and women differ in these behaviors contributes to gender differences in the risk of STDs for individuals, as well as their partners. Our analysis of these sex behaviors, particularly the less studied anal intercourse, builds upon past sex research that tends to focus on vaginal intercourse. We addressed the two hypotheses through an analysis of national data on men’s and women’s sex activity in the United States. We investigated gender differences in the five high-risk sex behaviors through multivariate analyses controlling for a set of social and cultural characteristics. We then tested for gender-specific effects by including in the analysis interaction terms between gender and the social and cultural characteristics. METHODS Subjects The data used in this paper were collected from national probability household samples of men and women in the contiguous United States. The National Survey of Men (NSM) and the National Survey of Women (NSW) were conducted in 1991 using a multistage, stratified and clustered, disproportionate area probability sample. The Black population was oversampled to allow adequate representation in the database. Other population groups, such as homosexual and bisexual men and women, were not oversampled. The study population consisted of adult males ages 20 to 39 and adult females ages 20 to 37. The sample frame contained 17,650 housing units, of which 93% were successfully screened for eligibility. 4 We
also tested for the effects of four family of origin measures: family structure, mother’s education, father’s education, and whether the mother was employed for 6 months or more when the respondent was between 5 and 15 years of age. In analyzing recent sex behaviors, we also tested for total income in the year previous to the survey. None of these factors significantly affected the dependent variables, thus they were excluded from the final models.
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
The Influence of Gender on Sex
April 19, 2000
12:8
Style file version Nov. 19th, 1999
235
In-person interviews were completed with 70% of the eligible men and 72% of the eligible women. Thus, the sample included 3321 men (1238 Black, 2083 nonBlack) and 1669 women (728 Black, 941 non-Black). The sample also included 241 Hispanic men and 123 Hispanic women, who may be of any race. The demographic distribution of the sample (e.g., by age, marital status, education) reflected that of the U.S. population at these ages. The weighted sample accounted for the effects of stratification, clustering, disproportionate area sampling, the oversampling of Black men and women, and the effects of differential nonresponse. As the final sample combined data from the NSM and the NSW, adjustments also were made so that the gender distributions within the age ranges of the sample reflected the gender distributions in the national population. Weighted data permitted generalizations from the survey data to the U.S. population represented by the sample. All data collection, coding, editing, and survey data processing were carried out by the Institute Survey Research at Temple University in Philadelphia. Advance notification was not given to potential respondents. After the screening interview, the nature of the survey was explained to the selected respondent both orally and in a letter that was given to the respondent. The latter served as the respondent’s informed consent. Respondents were told that they would be asked about their sex and health behaviors, including specific sex activities and disease prevention efforts. The interviews were conducted in person (i.e., face to face) by trained female interviewers using a standard survey instrument. Initial contacts with the respondents were in person, with interviews usually conducted in the respondents’ homes. The average interview length was 80 min. Although the survey design did not specify race matching of the respondent and the interviewer, approximately 75% of respondents were interviewed by a same-race interviewer. Data from social surveys are likely to be subject to errors. These are caused by selective nonparticipation, as well as by measurement error. The interview response rates in these two surveys were respectable for surveys of sex behavior. Surveys with similar response rates have been and continue to be used to obtain useful population parameters, as well as to examine complex causal relationships (Catania et al., 1992; Davis and Smith, 1991). When individuals choose not to participate in a survey, the statistical results may be biased if the reason for nonparticipation is directly or indirectly associated with the dependent variable. For example, if southerners are less likely to participate in sex-related surveys than others, our analysis of regional effects on high-risk sex behaviors might be incorrect. In the NSW and NSM surveys, poststratification weighting of the data was used to adjust for the error introduced by selective nonparticipation. Poststratification weighting schemes consist of comparing the survey sample characteristics to those of the population from which the sample was drawn, using external data sources on the U.S. population such as the U.S. Census or the Current Population Surveys (CPS). To calculate the appropriate weights, one creates
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
236
Style file version Nov. 19th, 1999
Cubbins and Tanfer
cross-tabulations of relevant social and demographic characteristics (e.g., age, race, gender, education) for both the sample and the population. One then takes the ratios of the sample to the population in each cell of the cross-tabulation, which are used as weights to adjust the sample (up or down) to resemble the population. Item nonresponse, another source of error in surveys, was trivial in the NSW and NSM surveys. It ranged from 0.1 to 1.2% for response-sensitive questions, such as those about one-night stands (0.1%), sex for money or drugs (0.1%), STD infection status (0.1%), anal intercourse (0.2%), number of vaginal sex partners in the last 4 weeks (0.4%), and coital frequency in the last 4 weeks (1.2%). Because of their very small number, we simply excluded these cases from the analysis. One source of measurement error that was of particular concern to us was misreporting. Obtaining direct measures of sex behavior is rarely, if at all, possible. Consequently, researchers rely on self-reports of such behavior, which tend to be subject to over- or underreporting. More importantly, such reporting appears to be gender-specific, with men tending to overreport and women tending to underreport (Smith, 1992). In addition, differences in reporting may be due in part to men and women in the sample having different numbers of potential partners (Phillis and Gromko, 1985), a factor we were unable to control in the analysis. Because of likely misreporting, we could not always determine, unequivocally, whether the behavioral differences were entirely a result of gender differences or if they were in part caused by gender differences in reporting. Based on past studies (Smith, 1992; Wiederman, 1997), we assumed, though, that men were relatively more likely to overreport sex behaviors than women were to underreport. Still, the NSW and NSM data on sex behavior are consistent with similar behaviors pattern observed in other surveys for both men and women (Catania et al., 1992; Laumann et al., 1994). Given this, we do not expect that the NSW and NSM respondents greatly misreported many behaviors of interest (e.g., coital frequency, anal intercourse, age at first intercourse). Concerns regarding errors in data on sensitive behaviors should not lead to unwarranted conclusions or outright rejection of survey findings on crucially important topics. In circumstances in which selective nonparticipation and misreporting are suspected, a reasonable and safe inference is that the estimates of sensitive behaviors from sample surveys represent the lower boundary estimates of the true rates in the population. In the case of these data, if the reporting bias were such that more people concealed behaviors in which they engaged than reported behaviors in which they did not participate, then we could, at least, make lower boundary estimates and inferences on causal relationships with relative confidence. Measures Self-reported data were used to construct the five dependent variables. Casual sex was defined as whether the respondent reported ever having had a onenight stand. This was based on the response to the question, “Have you ever had a
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
The Influence of Gender on Sex
April 19, 2000
12:8
Style file version Nov. 19th, 1999
237
one-night stand—that is, have you ever had oral, anal, or vaginal sex with a total stranger, with whom you have never had sexual contact again?” Although this question may provide a valid measure of one-time sex contacts, it likely underestimates the level to which individuals engage in sex with someone they do not know, such as with a stranger. For example, the first contact between sex partners may occur upon their first meeting when they are strangers, but subsequent contacts of the couple would be between nonstrangers. This situation would not be captured within our measure of casual sex, even though it does involve a situation of initial sex between strangers. This distinction should be kept in mind when interpreting the results of the analysis of ever having had casual sex. Total lifetime number of vaginal sex partners and the total lifetime number of anal sex partners were measured as the number of different partners with whom the respondent had ever had vaginal or anal intercourse, respectively. Two measures of high-risk sex behavior over the short term were constructed: whether or not the respondent reported having two or more vaginal sex partners in the period between January 1990 and the time of the interview and whether or not the respondent reported having two or more anal sex partners between January 1990 and the time of the interview. Among the predictor variables, dichotomous measures were constructed for gender, race (Black/not Black), ethnicity (Hispanic/not Hispanic), current paid work status (employed/not employed), and region of residence (South/not South). Marital status was measured in terms of both the total number of marriages and the current marital status (married or cohabiting versus not married nor cohabiting). Total number of marriages was used in the analysis of the lifetime high-risk sex behaviors, and the current marital status measure was used in the analysis of the short-term high-risk sex behaviors. Current paid work status was only used in the analysis of the short-term behaviors due to time order concerns in predicting the lifetime behaviors. Education was based on the highest number of years of completed schooling. As an indicator of religious activity, frequency of religious service attendance was categorized as once a month or more versus less frequent or no church attendance. Religious affiliation was categorized as Catholic, conservative Protestant, other religious affiliation, and no religious affiliation. The designations of conservative Protestant were based on the coding of the orthodoxy of beliefs for each Protestant denomination. This was accomplished with the aid of Professor Rodney Stark at the University of Washington, who classified each of the denominations reported by the respondent as conservative, moderate, or liberal. We collapsed the moderate and liberal categories into a single category to separate the mainline Protestants from conservative Protestants. Four measures of past sex experience were developed. Reported age at first vaginal, anal, or oral sex was used as an indicator of timing of sex initiation. Second, a dichotomous measure of past type of sex behavior was used to distinguish between any homosexual behavior and only heterosexual behavior in the past
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
238
Style file version Nov. 19th, 1999
Cubbins and Tanfer
10 years. This measure was based on a response to a question concerning sex behavior in the past 10 years, rather than sexual identity. Although the latter information may have helped explain patterns of certain sex behaviors, questions on sexual identity were not asked of these respondents. Two additional sex experience measures were constructed to indicate high-risk sex behavior over the short term. The first is based on the question, “With how many different (men/women) have you had vaginal intercourse since January 1990?” The second measure is the number of anal sex partners in the short term. Women were asked the question, “With how many different men have you had anal intercourse since January 1990?” For men, the sex of the partner was not specified: “Since January 1990, how many different partners have you had anal intercourse with?” Given the skewed nature of responses, we created two dichotomous measures for whether or not the respondent had (1) more than one vaginal sex partner or (2) more than one anal sex partner since January 1990. Unfortunately, with these data we are not able to distinguish serial versus simultaneous multiple partners. Although the age ranges of the NSW and NSM are somewhat narrow (approximately 20 to 39), we included a control for age. Studies on women ages 15 to 44 years have shown mixed results, with sex activity both increasing (Jasso, 1985) and decreasing (Udry et al., 1982) with age. Age was measured in years. A second control variable, months to interview since January 1990, was included to adjust for that variation. This variable was used in the analysis of the dependent variables for high-risk sex behavior over the short term. RESULTS The data analysis was done in three stages. First, we conducted descriptive analyses showing how men and women differed on the dependent, independent, and control variables. Second, we tested the first hypothesis on gender differences through a set of multivariate analyses of the dependent variables. For the intervallevel dependent variables (total number of vaginal sex partners and total number of anal sex partners), we used ordinary least-squares (OLS) regression. As all of the other dependent variables were dichotomous measures, we used logistic regression for their analysis (Hanushek and Jackson, 1977). The second-stage analyses serve as the baseline, linear models. In the third stage of the analyses, we tested the second hypothesis on whether there were significant gender differences in the effects of the social and cultural characteristics. This was done by adding to the baseline models a set of interaction terms between gender and all the other predictor variables. In reporting these findings, we present the baseline models, the gender-specific effects, and the results for the tests of significance for the interaction terms. Separate samples were analyzed for each dependent variable, with respondents selected based upon their past sex experience. In investigating the relationship between the predictor variables and casual sex, we limited the sample to those
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
The Influence of Gender on Sex
April 19, 2000
12:8
Style file version Nov. 19th, 1999
239
respondents who had ever had oral, vaginal, or anal sex. For the remaining dependent variables, the sample analyzed consisted only of those who had ever had the relevant experience. For example, only those who had ever had anal sex were included in the analysis of the total number of anal sex partners. Additional restrictions were used when analyzing whether the respondent had multiple vaginal or multiple anal sex partners since January 1990. If respondents reported having had only homosexual sex activity in the past 10 years, they were excluded in the analysis of multiple vaginal sex partners since January 1990 because it was not possible for them to have experienced the event under analysis. As with the problem of structural zeros in log-linear analysis (Bishop et al., 1975), the probability of these respondents having had multiple vaginal sex partners was zero. Similarly, women who reported having had only homosexual sex activity in the past 10 years were not included in the analysis multiple anal sex partners since January 1990. Based on the two hypotheses, the main questions the data analysis addressed were (1) What effect did gender have on the high-risk sex behaviors once the predictor and control variables were included in the analysis? and (2) Did the effects of the other variables in the model depend on gender, that is, did the predictor and control variables have gender-specific effects? As a preliminary step to answering these questions, we considered how men and women differed on the variables used in the analysis. Observed gender differences in sex behavior may be due to how men and women differ on relevant social and cultural factors. In Table I, we present prevalence and mean levels on the dependent variables across categories of the predictor variables. We also show these statistics separately by gender. Chi-square tests and analysis of variance were used to test for significant gender differences. Given the large number of statistical tests on the bivariate relationships between the predictor and the dependent variables, a Bonferroni correction was used in determining the significance of the statistics presented in Table I. Indicated α levels of p < .05 and p < .01 are associated with an α-level adjustment of p < .006 and p < .0001, respectively. Also in Table I, the continuous distributions of age, years of education, number of marriages, number of children living with the respondent, age at first sex, number of vaginal sex partners prior to January 1990, and number of anal sex partners prior to January 1990 were collapsed into categories. At the bivariate level, gender was significantly related to each of the five dependent variables, with men having higher percentage or mean levels on the five high-risk sex behaviors. Most of the other predictor variables were related to at least one of the high-risk sex behaviors. Only years of education and region of residence were not significantly related to any of the dependent variables. Minimal race and ethnic differences were found at the bivariate level. Blacks in the sample were significantly less likely than non-Blacks to have had casual sex. Nearly five times as many Hispanics as non-Hispanics had had multiple anal sex partners over the short term. These race and ethnicity differences, however, may be due to some of the other factors shown in Table I that are related to the dependent variables.
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
240
Cubbins and Tanfer
Table I. Prevalence and Mean Levels of the Dependent Variables Across Categories of the Predictor Variables Percentage Mean Mean Percentage >1 Percentage >1 ever had total vaginal total anal current vaginal current anal casual sex sex partners sex partners sex partners sex partners Gender Women Men Race Black Not Black Ethnicity Hispanic Not Hispanic Age 20–24 years 25–29 years 30–34 years 35 years or older Years of education Less than 12 12 More than 12 Current marital status Married /cohabiting Not married /cohabiting Number of marriages None One Two or more Number of children living with respondent None 1 or 2 3 or more Church attendance Once or more a month Less than once a month Religious affiliation Catholic Conservative Protestant Other religions None Current work status Employed Not employed Region of residence South Not South Age at first sex Before 17 years 17 years or older
** 26.9 46.7 ** 25.7 39.4
** 6.7 13.6
** 1.4 3.2
** 18.0 24.3
** 1.0 8.0
11.4 10.2
3.0 2.3
26.6 20.5
34.6 37.7 ** 29.1 36.6 40.5 45.1
9.6 10.4 ** 7.3 9.2 11.8 14.3
2.9 2.3 ** 2.3 1.9 1.9 3.7
24.3 21.1 ** 34.3 21.5 15.7 13.3
9.5 4.6 ** 16.6 3.7 ** 15.3 2.4 2.9 4.6
35.4 36.8 38.3 ** 34.6 41.4 ** 41.4 32.1 57.2 **
10.1 10.5 10.3 ** 9.6 11.6 ** 11.3 8.7 19.5 **
2.1 2.7 2.2 * 1.9 3.1 * 3.0 1.8 3.1
20.6 22.2 20.8 ** 8.1 41.0 ** 37.1 10.5 19.0 **
4.0 5.6 4.5 ** 0.7 11.3 ** 11.0 1.7 1.4 **
43.6 31.9 28.7 ** 26.6 44.7 ** 37.5 30.3 36.9 50.2 ** 39.0 30.8
12.0 8.9 8.6 ** 8.3 11.7 ** 9.6 8.9 11.0 12.8
2.9 1.8 2.0
9.3 0.3 1.1
2.2 2.4
33.5 10.8 6.3 ** 15.6 25.3
2.5 2.4 2.2 2.4
20.3 19.2 21.6 27.4
6.5 6.5 3.0 5.1
10.5 9.8
2.2 3.2
21.0 22.8
4.3 7.9
36.8 37.7 ** 50.6 26.0
10.7 10.3 ** 15.1 6.3
2.3 2.4
24.5 20.1 ** 30.0 13.0
5.3 4.7
2.6 2.1
1.6 6.3
6.8 2.0 (Continued )
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
The Influence of Gender on Sex
241 Table I. (Continued )
Percentage Mean Mean Percentage >1 Percentage >1 ever had total vaginal total anal current vaginal current anal casual sex sex partners sex partners sex partners sex partners Sex activity in past 10 years Any homosexual activity Heterosexual activity only Number of vaginal sex partners prior to January 1990 0 1–4 5–10 11 or more Number of anal sex partners prior to January 1990 0 1 2 or more Total N
**
**
**
*
**
77.2
17.0
5.3
39.2
27.3
36.2
10.2
2.1
21.1
3.1
**
**
**
**
8.0 19.3 52.4 78.4 **
1.4 3.5 8.6 30.3 **
3.3 1.6 1.6 3.1 **
9.5 13.6 26.0 39.3 **
12.0 4.5 2.7 4.7 *
32.6 58.7 80.2 37.4
8.8 15.9 27.4 10.4
1.3 1.2 5.1 2.4
19.4 30.4 37.9 21.4
7.4 1.2 8.9 4.9
4592
4546
768
4503
767
Note. Subsamples were specific to the dependent variable being analyzed; see texts for description of the subsamples related to each dependent variable. Unweighted N ’s are shown; all statistics are based on weighted data. Chi-square tests were used for percentage difference tests; analysis of variance was used to test mean differences. With a Bonferroni correction. ∗ p < .05; ∗∗ p < .01.
Older respondents had higher scores than younger respondents on the lifetime sex behaviors, though this was reversed when considering the short-term behaviors of multiple vaginal or anal sex partners. A somewhat similar pattern of long-term versus short-term behaviors was observed for the number of marriages: a higher percentage of those never married had multiple vaginal or anal sex partners in the short term compared to other respondents. Respondents who have no children had the highest levels on each of the dependent variables, except total number of anal sex partners. Those with lower or no church attendance and those with no religious affiliation appeared to be most likely to have engaged in high-risk sex behavior. Notably, those who were members of a conservative Protestant religion were less likely to have had casual sex, and had fewer total vaginal sex partners, than those who were affiliated with other religions. In this sample, current paid work status was significantly related only to ever having had casual sex, with those not employed being less likely to have had casual sex. Finally, those with an earlier sex history (age at first sex) or a more varied sex history (sex activity in past 10 years and number of sex partners prior to January 1990) generally had higher percents or means on the high-risk sex behaviors.
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
242
Cubbins and Tanfer
We expected that gender differences in high-risk sex behaviors are partly accounted for by gender differences in structural roles and cultural influences. We focused on four types of influence: family roles, paid work roles, religious behaviors, and past sex behavior. Table I provided evidence of significant bivariate relationships between these predictor variables and the high-risk sex behaviors. In turn, Table II shows that each of these factors was significantly related to gender (χ 2 at p < .05). Compared to men, a significantly higher proportion of women had children living in the home, attended church frequently, and was affiliated with some religious group. However, proportionally more men than women were employed. In Table II. Percentage Distributions of Selected Predictor Variables for Women and Men Women Number of children living with respondent None 1 or 2 3 or more Church attendance Once or more a month Less than once a month Religious affiliation Catholic Conservative Protestant Other religions None Current work status Employed Not employed Age at first sex Before 17 years 17 years or older Sex activity in past 10 years Any homosexual activity Heterosexual activity only Number of vaginal sex partners prior to January 1990 0 1–4 5–10 11 or more Number of anal sex partners prior to January 1990 0 1 2 or more N
Men
Difference test
47.3 42.9 9.7
53.0 34.9 12.1
**
44.3 55.7
36.4 63.6
**
33.5 15.9 42.7 7.9
33.3 15.2 39.3 12.2
**
73.5 26.5
86.9 13.1
**
36.2 63.8
55.6 44.4
**
4.1 95.9
2.2 97.8
**
20.3 50.0 18.1 11.6
14.0 31.7 21.9 32.3
**
86.8 10.5 2.7 1531
85.0 7.2 7.8 3061
**
Note. Sample included only those respondents who had ever had oral, vaginal, or anal sex. Unweighted N ’s are shown; all statistics are based on weighted data. Chi-square tests were used to test for significant gender differences. ∗ p < .05; ∗∗ p < .01.
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
The Influence of Gender on Sex
243
terms of past sex behaviors, proportionally more men than women had oral, vaginal, or anal sex prior to age 17, and substantially more men than women had five or more partners before January 1990. Only a small percentage of the samples engaged in any homosexual activity in the 10 years prior to the survey, with a slightly higher proportion for women (4%) than for men (2%). Overall, the findings from Tables I and II showed that, in addition to gender, other social and cultural conditions were related to high-risk sex behaviors. In turn, these factors may partly explain observed gender differences in the high risk sex behaviors. We tested the two hypotheses through a series of multivariate analyses. In Table III, we present the results of the logistic regression analysis predicting if Table III. Logistic Regression Results Predicting Ever Having Had Casual Sex Baseline Model Gender Women Men Race Black Not Black Ethnicity Hispanic Not Hispanic Age Years of education Number of marriages Church attendance Once or more a month Less than once a month Religion Catholic Conservative Protestant Other religion None Age at first sex Sex activity in past 10 years Any homosexual activity Heterosexual activity only Region South Not South Constant Model χ 2 (df)
Effects for women
Effects for men
Gender difference in effect
−.58∗∗ (.08) — −.73∗∗ (.13) —
−1.51∗∗ (.24) —
−.03 (.15) — .06∗∗ (.01) ∗∗ .08 (.02) −.25∗∗ (.07) −.53∗∗
(.09)
−.26 (.17) —
**
.21 (.25) — .10∗∗ (.02) .10∗∗ (.03) −.60∗∗ (.12)
−.17 (.19) — .03∗∗ (.01) .05∗ (.02) −.04 (.08)
**
−.53∗∗
−.53∗∗
(.14)
(.12)
—
—
—
−.16 (.14) −.33∗ (.17) −.21 (.14) — −.19∗∗ (.01)
.13 (.25) −.01 (.29) .10 (.24) — −.23∗∗ (.03)
−.23 (.18) −.39 (.21) −.29 (.17) — −.18∗∗ (.02)
1.54∗∗ (.26) —
1.84∗∗ (.37) —
.13 (.09) — .78∗ (.35) 556.84∗∗ (13)
.25 (.14) — −.75 (.62)
**
.80∗ (.39) — −.04 (.12) — 1.50∗∗ (.44)
49.30∗∗ (12)
Note. Models included only those who had ever had vaginal, anal, or oral sex. Log odds are shown, with standard errors in parentheses. Gender-specific models were estimated through a single model that included interaction terms between gender and all remaining predictor variables. For all models, unweighted N = 4592; statistics are based on weighted data. ∗ p < .05; ∗∗ p < .01.
P1: FMN Archives of Sexual Behavior [asb]
244
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
Cubbins and Tanfer
the respondent reported ever having had casual sex. The second column in the table shows the Baseline Model. Gender-specific coefficients (shown in columns 3 and 4) were generated through a single logistic regression model that included interaction terms between gender and all the remaining predictor variables. The final column of the table shows whether or not the effect of the predictor variable significantly differed by gender. The Baseline Model in Table III indicates that even after controlling for the other factors in the model, gender was significantly related to ever having had casual sex. Women were only half as likely as men ever to have had casual sex. For this behavior, then, gender remains a salient factor, even after controlling for social and cultural characteristics that differ by gender and by sex behavior. In terms of casual sex, Hypothesis 1 is not supported. To address the second hypothesis, whether the effects on high-risk sex behavior differed by gender, we added a set of interaction terms between gender and the remaining predictor variables to the Baseline Model. A chi-square test between the Baseline Model and the gender interactions model showed that the two models were significantly different (χ 2 = 49.30, df = 12, p < .05). This indicated that our ability to predict ever having had casual sex was improved by taking into account gender-specific effects. Three of the interaction terms—gender with race, age, and number of marriages—were found to be significant, providing some support for Hypothesis 2. The gender-specific race effect shows that the diminished likelihood of ever having had casual sex for Blacks compared to non-Blacks was significant for women but not for men. Only for women, then, does there appear to be a significant race difference in the likelihood of ever having had casual sex. The significant gender–age interaction term suggests that age had a larger effect on ever having had casual sex for women than for men. After controlling for the other factors in the model, as the respondents aged women were more likely than men to have engaged in casual sex at some time. In turn, the number of marriages respondents had seems to be a gender-specific influence, with only women having a lower probability of ever having had casual sex, the more times they were married. Several other variables had similar significant effects for men and women. Education increased the chances of having had casual sex, as did having any homosexual activity in the past 10 years. The inhibiting effects of religious participation on casual sex was seen in the effects of church attendance and religious affiliation. Those with higher church attendance or who were affiliated with a conservative Protestant religion were significantly less likely to have engaged in casual sex than those with less frequent church attendance or no religious affiliation, respectively. But early sex experience increased the likelihood of having had casual sex. Contrary to expectations, neither ethnicity nor region significantly affected the likelihood of ever having had casual sex. Tables IV and V present the results of the OLS regression analyses of the lifetime number of vaginal and anal sex partners, respectively. As with Table III,
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
The Influence of Gender on Sex
245
Table IV. Ordinary Least-Squares Regression Results Predicting Lifetime Number of Vaginal Sex Partners Baseline Model Gender Women Men Race Black Not Black Ethnicity Hispanic Not Hispanic Age Years of education Number of marriages Church attendance Once or more a month Less than once a month Religion Catholic Conservative Protestant Other religion None Age at first sex Sex activity in past 10 years Any homosexual activity Heterosexual activity only Region South Not South Constant Adjusted
R2
Effects for women
Effects for men
Gender difference in effect
−3.96∗∗ (.52) — .52 (.77) —
−2.86∗∗ (1.10) —
4.06∗∗ (1.11) —
1.14 (.92) — .54∗∗ (.05) .43∗∗ (.11) −.65 (.42)
.39 (1.38) — .57∗∗ (.09) .26 (.17) −1.72∗ (.68)
1.40 (1.23) — .47∗∗ (.07) .47∗∗ (.15) .19 (.54)
−1.43∗∗ —
−.78 (.78) —
−.73 (.92) −1.77 (1.04) −.20 (.89) — −1.49∗∗ (.08)
−1.98 (1.48) −1.68 (1.64) −.40 (1.43) — −1.03∗∗ (.13)
3.86∗
(.54)
(1.58)
— .91 (.56) — 17.16∗∗ (2.15) .18∗∗
8.75∗∗
*
−1.76∗
(.75) —
(1.89)
.35 (1.18) −1.31 (1.35) .01 (1.13) — −1.72∗∗ (.10)
**
−8.63∗∗
**
(2.95)
—
—
.56 (.80) —
.24 (.79) —
7.67∗ (3.30)
**
21.05∗∗ (2.82) .19∗∗
Note. Models included only those who had ever had vaginal sex. Unstandardized regression coefficients are shown, with standard errors in parentheses. Gender-specific models were estimated through a single model that included interaction terms between gender and all remaining predictor variables. For all models, unweighted N = 4546; statistics are based on weighted data. ∗ p < .05; ∗∗ p < .01.
we first tested Hypothesis 1 with the baseline models and then tested Hypothesis 2 through analyses of gender-specific effects. First considering Table IV, we found that even after controlling for the other factors in the model, gender had a significant effect on the lifetime number of vaginal sex partners, not supporting Hypothesis 1. After taking the social, cultural, and control factors into account, men reported nearly four more vaginal sex partners than did women on average. However, we did find some support for Hypothesis 2 on gender-specific effects. When we added the gender interaction terms to the Baseline Model, we significantly improved our ability to predict the lifetime number of vaginal sex partners [model comparison test: F(12, 3146) = 6.15, p < .05]. As shown in the
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
246
Cubbins and Tanfer Table V. Ordinary Least-Squares Regression Results Predicting Lifetime Number of Anal Sex Partners Baseline Model
Gender Women Men Race Black Not Black Ethnicity Hispanic Not Hispanic Age Years of education Number of marriages Church attendance Once or more a month Less than once a month Religion Catholic Conservative Protestant Other religion None Age at first sex Sex activity in past 10 years Any homosexual activity Heterosexual activity only Region South Not South Constant Adjusted R 2
Effects for women
Effects for men
.72 (.69) —
−.02 (1.13) —
.90 (.86) —
−.15 (.58) — .13** (.04) −.07 (.08) −.43 (.23)
.18 (.95) — .02 (.07) −.02 (.11) −.10 (.45)
−.46 (.73) — .21∗∗ (.05) −.11 (.10) −.70* (.27)
.09 (.38) —
−.25 (.55) —
.43 (.52) —
Gender difference in effect
−1.43∗∗ (.35) —
.39 (.53) −.28 (.67) −.24 (.51) — −.07 (.05)
.24 (.79) .34 (1.25) .44 (.80) — .004 (.08)
.05 (.72) −.89 (.82) −1.02 (.67) — −.13∗ (.06)
3.08∗∗ (.63) —
−.03 (1.00) —
5.45∗∗ (.80) —
.19 (.37) — 1.09 (1.47)
−.03 (.57) — .95 (2.55)
.36 (.50) — .72 (1.83)
.09∗∗
*
**
.13∗∗
Note. Models included only those who had ever had anal sex. Unstandardized regression coefficients are shown, with standard errors in parentheses. Gender-specific models were estimated through a single model that included interaction terms between gender and all remaining predictor variables. For all models, unweighted N = 768; statistics are based on weighted data. ∗ p < .05; ∗∗ p < .01.
last column in Table IV, the effects of several variables operated differently by gender. Controlling for other factors, Black women were found to have had nearly three fewer vaginal sex partners than non-Black women, even though Black men had over four more partners than non-Black men. As with ever having had casual sex, the number of marriages women had reduced their total number of vaginal sex partners, but this factor had no effect for men. This finding provides support for Hypothesis 2, showing that the race and marital history effects on total vaginal sex partners depend on gender. Two measures of past sex behaviors also had gender-specific effects. Age at first sex had a significant influence on the lifetime number of vaginal sex partners,
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
The Influence of Gender on Sex
April 19, 2000
12:8
Style file version Nov. 19th, 1999
247
and the effect was larger for men than for women. An earlier age at first sex seems to lead to a greater total number of vaginal sex partners for men than for women. We should note that in this analysis age at first sex does not measure duration of exposure as respondent’s age was also included in the model. Perhaps not surprisingly, having any homosexual activity in the past 10 years reduced the total average number of vaginal sex partners for men by 8.6 partners compared to men with only heterosexual activity over the same period. But for women, this past sex experience had the opposite effect: women with any homosexual experience had substantially more vaginal sex partners than women with only heterosexual activity in the past 10 years. While women’s homosexual experience should present little to no risk for STDs, it may have an indirect effect on STD risks as it relates to vaginal sex behaviors. Age, years of education, and church attendance each had significant effects on lifetime number of vaginal sex partners, though the influences were not genderspecific (see Baseline Model, Table IV). As expected, age and years of education increased the lifetime number of vaginal sex partners. Higher church attendance was associated with fewer vaginal sex partners. Hispanic status, religious affiliation, and region of residence had no effect on the lifetime number of vaginal sex partners. In Table V, we presented the models predicting the lifetime number of anal sex partners. Even after controlling for the other predictor and control variables, gender had a significant effect on the lifetime number of anal sex partners, with men having 1.4 more partners than women. As with lifetime number of vaginal sex partners, Hypothesis 1 is not supported by this analysis. Nor did we find support for Hypothesis 2, since the measures for religious behavior, family roles, region, race, and ethnicity did not have gender-specific effects on the lifetime number of anal sex partners. However, two of the control variables, age and sex activity in the past 10 years, had gender-specific effects. There was a slight increase in the number of anal sex partners for men as they aged; for women, age had no effect. Men who had any homosexual activity in the past 10 years had over five more anal sex partners than men with only heterosexual experiences. Sex activity in the past 10 years, though, had no effect for women. Overall, the model including the gender interaction terms significantly improved our ability to predict the lifetime number of anal sex partners [F(12, 567) = 2.74, p < .05]. Still, the low values for the adjusted R 2 shown in both Table IV and Table V indicate that there remained substantial variation in the dependent variables yet to be explained. The final two tables show the results of logistic regression analyses predicting whether or not the respondent had multiple vaginal sex partners (Table VI) or multiple anal sex partners (Table VII) during the short-term period between January 1990 and the date of the interview. In the bivariate analysis (Table I), gender was found to be significantly related to having had multiple vaginal sex
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
248
Cubbins and Tanfer
Table VI. Logistic Regression Results Predicting Whether the Respondent Has Had Multiple Vaginal Sex Partners Since January 1990 Baseline Model Gender Women Men Race Black Not Black Ethnicity Hispanic Not Hispanic Age Years of education Married or cohabiting Yes No Number of children living with respondent Church attendance Once or more a month Less than once a month Religion Catholic Conservative Protestant Other religion None Current work status Employed Not employed Age at first sex Sex activity in past 10 years Any homosexual activity Heterosexual activity only Number of vaginal sex partners prior to January 1990 Region South Not South Months to interview since January 1990 Constant Model (df)
χ2
Effects for women
Effects for men
−.32 (.24) —
.49∗ (.21) —
Gender difference in effect
.18 (.12) — .04 (.16) — .32 (.19) — −.07∗∗ (.01) .04 (.03)
.21 (.30) — −.08∗∗ (.02) .05 (.04)
.42 (.25) — −.06∗∗ (.01) .02 (.04)
−1.84∗∗ (.12) — −.39∗∗ (.07)
−1.60∗∗ (.17) — −.42∗∗ (.10)
−2.18∗∗ (.19) — −.26∗ (.11)
−.51∗∗ (.12) —
−.49∗∗ (.17) —
−.53∗∗ (.17) —
−.18 (.19) −.04 (.21) −.09 (.18) — −.09 (.04) — −.16∗∗ (.02)
−.42 (.24) −.29 (.28) −.32 (.23) —
.43 (.34) .66 (.37) .49 (.32) — −.02 (.19) — −.19∗∗ (.03) .10 (.53) — .05∗∗ (.01)
.37 (.70) — .03∗∗ (.004)
.32∗∗ (.12) — .06∗∗ (.02)
.44∗∗ (.17) — .07∗∗ (.03)
.20 (.16) — .05∗ (.02)
2.42∗ (.93)
*
* * *
−.07 (.20) — −.14∗∗ (.02)
.45 (.39) — .03∗∗ (.004)
2.38∗∗ (.56)
*
**
2.49∗∗ (.75)
879.81∗∗
32.85∗∗
(17)
(16)
Note. Models included only those who had ever had vaginal sex and excluded those who had only homosexual activity in the past 10 years. Log odds are shown, with standard errors in parentheses. Gender-specific models were estimated through a single model that included interaction terms between gender and all remaining predictor variables. For all models, unweighted N = 4503; statistics are based on weighted data. ∗ p < .05; ∗∗ p < .01.
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
The Influence of Gender on Sex
249
Table VII. Logistic Regression Results Predicting Whether the Respondent Has Had Multiple Anal Sex Partners Since January 1990
Gender Women Men Race Black Not Black Ethnicity Hispanic Not Hispanic Age Years of education Married or cohabiting Yes No Number of children living with respondent Church attendance Once or more a month Less than once a month Religion Catholic Conservative Protestant Other religion None Current work status Employed Not employed Age at first sex Sex activity in past 10 years Any homosexual activity Heterosexual activity only Number of anal sex partners prior to January 1990 Region South Not South Months to interview since January 1990 Constant Model χ 2 (df)
b
(SE)
−1.31 —
(.79)
.89 —
(.79)
1.05 — −.07 −.05
(.69) (.05) (.12)
−2.32∗∗ —
(.75)
−.67
(.55)
−1.53 —
(.81)
.13 .17 −.53 —
(.80) (.95) (.80)
.29 — −.09
(.70) (.06)
1.84∗∗
(.64)
— .10∗
(.04)
.09 — .02
(.56)
−.09
(2.77)
(.09)
101.98∗∗ (17)
Note. Model included only those who had ever had anal sex and excluded women who had only homosexual activity in the past 10 years. Log odds are shown, with standard errors in parentheses. Unweighted N = 767; statistics are based on weighted data. ∗ p < .05; ∗∗ p < .01.
P1: FMN Archives of Sexual Behavior [asb]
250
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
Cubbins and Tanfer
partners since January 1990. Once the other predictor and control variables were included in the model, gender no longer was directly related to the dependent variable (Table VI), thus providing support for Hypothesis 1. The influence of gender on the dependent variable operates through the independent and control variables included in the model. We also found support for Hypothesis 2, as the effects of race and marital status are dependent on gender. One of the past sex behavior measures, the number of vaginal sex partners prior to January 1990, also has gender-specific effects. For religious affiliation, the gender-specific coefficients for Catholic, conservative Protestant, and other religion were significantly different from each other but are not significantly different from zero. The model including the gender interaction terms was significantly better at predicting the dependent variable than the Baseline Model (χ 2 = 32.85, df = 16, p < .05). The significant race effect was confined to men, with Black men being 1.6 times more likely to have had multiple vaginal sex partners than non-Black men. Currently, being married or cohabiting reduced the likelihood of having multiple vaginal sex partners over the short term more for men than women. So opposite to the marital history effect on lifetime number of vaginal sex partners, being married or cohabiting had a larger influence in reducing this high-risk sex behavior among men than among women. The remaining significant gender-specific effect was found for the past sex behavior measure, the number of vaginal sex partners prior to January 1990. For women, each additional vaginal sex partner they had before January 1990 increased the predicted odds by .05; for men, the comparable increase in the predicted odds was .03. In this case, the number of past vaginal sex partners had a slightly larger effect on women having had high-risk sex behavior than on men. Several of the remaining predictor and control variables had significant effects (see Baseline Model). These were age, number of children living with the respondent, church attendance, age at first sex, region of residence, and number of months between January 1990 and the interview. Age, number of children living with the respondent, and level of church attendance each reduced the probability of having had multiple vaginal sex partners. Age at first sex followed the pattern observed in earlier tables; in this case, multiple vaginal sex partners are linked to earlier ages of first oral, vaginal, or anal sex. Factors that increased the probability of having had multiple vaginal sex partners over the short term were the control variable for time between January 1990 and the survey interview and the unexpected influence of living in the South. The results of the analysis of the final dependent variable, having had multiple anal sex partners since January 1990, are shown in Table VII. As was observed earlier with multiple vaginal sex partners, gender did not have a significant effect on the probability of having multiple anal sex partners over the short term once the other predictor and control variables were controlled. This provides support for Hypothesis 1, as gender was significantly related to the dependent variable in
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
The Influence of Gender on Sex
12:8
Style file version Nov. 19th, 1999
251
the bivariate analyses (Table I). However, none of the interaction tests for gender differences in the effects of the predictor and control variables were significant. As for having had multiple anal sex partners in the short term, then, Hypothesis 2 is not supported. Only five variables in the model were found to have any influence on the probability of having multiple anal sex partners in the short term. Being married or cohabiting significantly reduced the likelihood of having had multiple anal sex partners, as did attending church at least once a month. The three past sex behavior measures were all significant in predicting having multiple anal sex partners over the short term. This suggests that certain individuals engage in a collection of high-risk sex behaviors.
DISCUSSION In this study, we have investigated the relationship between gender and five self-reported, high-risk sex behaviors. These behaviors were ever having had casual sex, the lifetime number of vaginal sex partners, the lifetime number of anal sex partners, having had multiple vaginal sex partners over the short term, and having had multiple anal sex partners over the short term. Our analysis was guided by a conceptual model that emphasized the constraints and opportunities for high-risk sex behavior that arise from an individual’s structural position and cultural context. We expected that as men’s and women’s lives differ in their social roles and are subject to differing cultural influences, their sex behaviors also will differ. In our first hypothesis, we predicted that gender differences in high-risk sex behaviors were due to differences in men’s and women’s family roles, paid work roles, religious behaviors, and past sex behaviors. Once these factors were taken into account, we argued, gender differences in high-risk sex behavior would diminish. Second, we predicted that the effects of certain social roles and cultural factors (religious behavior, family roles, region, race, and ethnicity) on the high-risk sex behaviors would be dependent on an individual’s gender. We evaluated these hypotheses using national data on self-reported sex behavior in the United States for men ages 20 to 39 and women ages 20 to 37. We found mixed support for the hypotheses. Hypothesis 1 was supported by the analysis of the short-term sex behaviors, having had multiple vaginal or anal sex partners since January 1990. However, even after controlling for the social roles and cultural influences, gender remained a significant predictor of the lifetime experience measures of sex behavior. It is curious that gender was significantly related to the lifetime sex behaviors, though not the short-term sex behaviors, once the social and cultural factors were controlled. Perhaps since our measures of social and cultural factors are primarily based on the respondents’ current characteristics, we are not able to model adequately the ongoing, indirect influence gender has on these high-risk sex behaviors through past social roles and cultural experiences.
P1: FMN Archives of Sexual Behavior [asb]
252
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
Cubbins and Tanfer
Longitudinal data on these variables, if they were available, might provide a better test of Hypothesis 1 for the lifetime sex behaviors. It also is possible that the significant effect of gender on the lifetime sex behaviors is related to gender differences in reporting sex behaviors. This misreporting behavior may be more apparent with lifetime behaviors than short-term behaviors, given the longer period of recall (Smith, 1992; Wiederman, 1997). Thus, at least some of the significant difference in men’s and women’s lifetime number of vaginal sex partners that remains after controlling for the social and cultural factors may be due to gender differences in reporting behavior. For Hypothesis 2 on gender-specific effects, we did find that for three of the high-risk behaviors—ever having had casual sex, lifetime number of vaginal sex partners, lifetime number of anal sex partners—the effect of race was dependent on gender. In addition, marital behavior had gender-specific effects, though the pattern of influence was somewhat contrary to expectations. The effects of the other social roles and cultural factors did not vary by gender. For the combined effect of race and gender, we found that Blacks were less likely than non-Blacks ever to have engaged in casual sex, with the effect being larger for women than for men. But though Black women had significantly fewer lifetime vaginal sex partners than non-Black women, the reverse occurred for men. Similarly, Black men were significantly more likely than non-Black men to have had multiple vaginal sex partners over the short term, though no race difference was found for women. The overall pattern found in these three findings is that Black men have higher, and Black women lower, high-risk sex behavior than other groups. This effect is consistent with studies (Laumann et al., 1994; Weinberg and Williams, 1988) that Black men have more positive views toward recreational sex compared to other men, but the reverse is true for Black women compared to other women. Considered along with economic influences, these attitudinal differences may contribute to Black men and women having different numbers of acceptable sex partners. Since the late 1970s (Wilson, 1978), social science researchers have noted the high unemployment and low earnings among Black men. These economic conditions, along with high rates of incarceration, contribute to Black men’s lower likelihood of marrying compared to other men (Tucker and Mitchell, 1995). Further, this likely reduces Black women’s chances of forming the kinds of committed relationships in which they prefer to have sex relations. This may be the case among Black women with few economic resources of their own because of their likely dependence on men’s earnings (Schwartz and Rutter, 1998). But Black women with a high education or income may be even more selective in their choice of partners. However, given Black men’s willingness to have recreational sex, economic conditions and the accompanying difficulty of forming committed relationships will present fewer barriers to their having casual sex and multiple partners. These findings support the argument that to understand high-risk sex behavior, the intersection of race and gender influences must be
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
The Influence of Gender on Sex
April 19, 2000
12:8
Style file version Nov. 19th, 1999
253
taken into account. The combined statuses of race and gender create distinct social and cultural experiences for individuals that produce different opportunities and constraints for high-risk sex behavior. As such, policymakers should be attuned to how race and gender interact to affect risk-taking and health promotion behaviors. It is unlikely that a single program will be uniformly effective across all race– gender groups. Marital behavior also had gender-specific effects. Being currently married or cohabiting was associated with a lower probability of having multiple vaginal sex partners over the short term for both genders, though this relationship was stronger for men than for women. However, this finding is in contrast to the relationship found between the number of marriages and the lifetime number of vaginal sex partners. Over the long term, multiple marriages reduce the number of lifetime vaginal sex partners for women but are unrelated to men’s lifetime number of vaginal sex partners. This suggests that men’s cumulative vaginal sex partners are not diminished or increased by their marital behavior over the life course. But when men are married or cohabiting, they are less likely to have multiple vaginal sex partners over the short term than other men. Thus, marriage reduces the number of vaginal sex partners over the short term for both genders but over the lifetime only for women. While the overall effect of marriage is consistent with that in other studies (Laumann et al., 1994; see also Darnton, 1994), the lack of effect for men seems unusual. Perhaps before and after marriage, men’s sex activity increases to such an extent that having multiple marriages does not reduce the overall average number of vaginal partners for men. Essentially, men may make up for missed sex opportunities during marriage. The Laumann et al. (1994) finding on gender differences in the number of new sex partners before and after marriage is consistent with this interpretation. We found support for the role of several social and cultural factors in partially explaining differences in men’s and women’s high-risk sex behavior. Past sex behaviors were important factors in predicting lifetime and short-term high-risk sex behavior. In particular, early sex experiences appear to lead to distinct paths of sex activity, which partly account for gender differences in high-risk sex behavior. The earlier the age at first sex, the more likely was the respondent to have engaged in high-risk sex behaviors, in terms of having had casual sex and multiple partners over both the short and the long term. Although we expected that conservative religions would limit high-risk sex behavior, little support was found for this view. There were few differences in highrisk sex behavior between respondents affiliated with conservative religious groups (e.g., fundamentalist Protestants) and those with other or no religious affiliation. The only exception was in terms of the likelihood of ever engaging in casual sex. The more evident influence of religious behavior was in terms of church attendance. Compared to those with lower levels of participation, respondents who attended church at least once a month were less likely to have engaged in casual sex, had fewer total vaginal sex partners, and were less likely to have had multiple vaginal
P1: FMN Archives of Sexual Behavior [asb]
254
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
Cubbins and Tanfer
and anal sex partners over the short term. As found in other studies (Davidson et al., 1995), the level of participation in a religious organization may be a better indicator of religious influence than religious affiliation because it reveals a part of the social network in which individuals carry out their daily lives. A social network that is inclined to more restrictive views of sex behavior is likely to offer fewer partners willing to engage in high-risk sex behavior. Our expectations regarding the influence of social roles were only slightly supported. With the exception of a significant bivariate relationship with ever having had casual sex, current work status was not associated with the high-risk sex behaviors. One family role indicator, the number of children in the home, did diminish the likelihood of having multiple vaginal sex partners over the short term. Future research on the relationship between social roles and sex behavior might explore more specific characteristics of social roles. For example, number of hours worked or type of employment may better identify opportunities for high-risk sex behavior than simply whether or not someone is employed. Similarly, a more thorough understanding of the relationship between family roles and high-risk sex behavior might be found by looking at specific family influences, such as the ages of children in the household and the duration of marriage. The results of this study are subject to certain limitations in the research methodology. In our analysis of lifetime sex experience, we were unable to establish the causal order of influence of some of our control variables. For example, we found that high church attendance reduced the likelihood of ever having had casual sex. It is possible that individuals who engaged in casual sex were less likely to frequent a religious organization because the prescripts and values expressed were inconsistent with their own behaviors. In this case, certain sex behaviors would have influenced the level of religious activity. Second, the data analyzed may underor overestimate the level of sex activity due to nonresponse and misreporting. For example, if, compared to others, potential respondents with any homosexual activity in the past 10 years were systematically less likely to participate in the survey, the analysis would be biased in terms of the effects of factors associated with this group. Finally, our results may have underestimated the role of gender given the individual-level analysis of sex behavior. Past studies provide evidence that gender differences in sex behavior may be more pronounced when considering couple activity rather than just individual behavior (Blumstein and Schwartz, 1983; Peplau et al., 1977). In particular, decisions regarding the type of sex activity (e.g., anal versus vaginal intercourse) may be highly influenced by gender (Byers and Heinlein, 1989; Gerrard et al., 1990). Qualitative studies on the interactional process leading to certain sex behaviors may be helpful in this regard. Overall our study showed that gender effects on high-risk sex behaviors are due in part to how men and women differ on other social and cultural characteristics. Most importantly, this research points to the way gender determines how race and marital status affect high-risk sex behaviors. This suggests that gender
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
April 19, 2000
12:8
The Influence of Gender on Sex
Style file version Nov. 19th, 1999
255
may condition the effects of social and cultural factors on other risk-taking and preventative behaviors. It also raises the possibility that other important status characteristics (e.g., race or ethnicity) may determine the effects of structural and cultural conditions on sex behavior. As such, future studies should consider the interactive influences of gender and other status characteristics in determining sex behavior. ACKNOWLEDGMENTS The authors gratefully acknowledge support through National Institute of Child Health and Human Development (NICHD) Grants HD-26288 and HD-26631 and National Institute of Allergy and Infectious Diseases (NIAID) Grant AI-34360. The authors express their gratitude to John O. G. Billy, Tom Smith, and Chien Liu for helpful comments and suggestions. REFERENCES Baldwin, J., and Baldwin, J. (1988). Factors affecting AIDS-related sexual risk-taking behavior among college students. J. Sex Res. 25: 181–196. Batson, C., Schoenrade, P., and Ventis, W. (1993). Religion and the Individual: A Social-Psychological Perspective, Oxford University Press, New York. Benson, P., Donahue, M., and Erickson, J. (1989). Adolescence and religion: A review of the literature from 1970–1986. Res. Soc. Sci. Study Religion 1: 153–181. Billy, J., Tanfer, K., Grady, W., and Klepinger, D. (1993). The sexual behavior of men in the United States. Family Plan. Perspect. 25: 52–60. Bishop, Y., Fienberg, S., and Holland, P. (1975). Discrete Multivariate Analysis: Theory and Practice, MIT Press, Cambridge, MA. Blumstein, P., and Schwartz, P. (1983). American Couples: Money, Work, Sex, Morrow, New York. Brines, J. (1994). Economic dependency, gender, and division of labor at home. Am. J. Sociol. 100: 652–660. Byers, E., and Heinlein, L. (1989). Predicting initiations and refusals of sexual activities in married and cohabiting heterosexual couples. J. Sex Res. 26: 210–231. Catania, J., Coates, T., Stall, R., Turner, H., Peterson, J., Hearst, N., Dolcini, M., Hudes, E., Gagnon, J., Wiley, J., and Groves, R. (1992). Prevalence of AIDS-related risk factors and condom use in the United States. Science 258: 1101–1106. Clark, R. (1990). The impact of AIDS on gender differences in willingness to engage in casual sex. J. Appl. Soc. Psychol. 20: 771–782. Clement, U. (1989). Profile analysis as a method of comparing intergenerational differences in sexual behavior. Arch. Sex. Behav. 18: 229–237. Cochran, S., and Peplau, L. (1991). Sexual risk reduction behaviors among young heterosexual adults. Soc. Sci. Med. 33: 25–36. Cohen, L. L., and Shotland, R. L. (1996). Timing of first sexual intercourse in a relationship: Expectations, experiences, and perceptions of other. J. Sex Res. 33: 291–299. Cornwall, M. (1988). The influence of three agents of religious socialization: Family, church, and peers. In Thomas, D. (ed.), The Religion and Family Connection: Social Science Perspectives, Religious Studies Center, Brigham Young University, Provo, UT, pp. 207–231. Darabi, K. (1987). Childbearing Among Hispanics in the United States, Greenwood Press, New York. Darnton, J. (1994). The extent of monogamy in Britain. New York Times Feb. 1: B8. Davidson, J. K., Darling, C. A., and Norton, L. (1995). Religiosity and the sexuality of women: Sexual behavior and sexual satisfaction revisited. J. Sex Res. 32: 235–243.
P1: FMN Archives of Sexual Behavior [asb]
256
PL110-329
April 19, 2000
12:8
Style file version Nov. 19th, 1999
Cubbins and Tanfer
Davis, J., and Smith, T. (1991). General Social Surveys, 1972–1991: Cumulative Codebook, National Opinion Research Center, Chicago. Forrest, J., and Singh, S. (1990). The sexual and reproductive behavior of American women. Family Plan. Perspect. 22: 206–214. Gerrard, M., Breda, C., and Gibbons, F. (1990). Gender effects in couples’ sexual decision making and contraceptive use. J. Appl. Soc. Psychol. 20: 449–464. Gagnon, J., and Simon, W. (1973). Sexual Conduct, Aldine, Chicago. Glass, S., and Wright, T. (1992). Justifications for extramarital relationships: The association between attitudes, behaviors, and gender. J. Sex Res. 29: 361–387. Grauerholz, E., and Serpe, R. (1985). Initiation and response: The dynamics of sexual interaction. Sex Roles 12: 1041–1059. Hanushek, E., and Jackson, J. (1977). Statistical Methods for Social Scientists, Academic Press, New York. Hendrick, S., Hendrick, C., Slapion-Foote, H., and Foote, F. (1985). Gender differences in sexual attitudes. J. Pers. Soc. Psychol. 48: 1630–1642. Herold, E., and Mewhinney, D. K. (1993). Gender differences in casual sex and AIDS prevention: A survey of dating bars. J. Sex Res. 30: 36–42. Jasso, G. (1985). Marital coital frequency and the passage of time: Estimating the separate effects of spouses’ ages and marital duration, birth and marriage cohorts, and period influences. Am. Sociol. Rev. 50: 224–241. Keller, J., Elliott, S., and Gunberg, E. (1982). Premarital sexual intercourse among single college students: A discriminant analysis. Sex Roles 8: 21–32. Kost, K., and Forrest, J. (1992). American women’s sexual behavior and exposure to risk of sexually transmitted diseases. Family Plan. Perspect. 24: 244–254. Laumann, E., Gagnon, J., Michael, R., and Michaels, S. (1994). The Social Organization of Sexuality: Sexual Practices in the United States, University of Chicago Press, Chicago. Leigh, J. P. (1983). Direct and indirect effects of education on health. Soc. Sci. Med. 17: 227–234. McCormick, N., Brannigan, G., and LaPlante, M. (1984). Social desirability in the bedroom: Role of approval motivation in sexual relationships. Sex Roles 11: 303–314. Miller, A., and Hoffmann, J. (1995). Risk and religion: An explanation of gender differences in religiosity. J. Sci. Study Religion 34: 63–75. O’Sullivan, L., and Byers, E. (1993). Eroding stereotypes: College women’s attempts to influence reluctant male sexual partners. J. Sex Res. 30: 270–282. Peplau, L., Rubin, Z., and Hill, C. (1977). Sexual intimacy in dating relationships. J. Soc. Issues 33: 86–109. Phillis, D., and Gromko, M. (1985). Sex differences in sexual activity: Reality or illusion? J. Sex Res. 21: 437–443. Reiss, I. (1986a). A sociological journey into sexuality. J. Marriage Family 48: 233–242. Reiss, I. (1986b). Journey into Sexuality: An Exploratory Voyage, Prentice-Hall, Englewood Cliffs, NJ. Ross, C., and Van Willigen, M. (1997). Education and the subjective quality of life. J. Health Soc. Behav. 38: 275–297. Schwartz, P., and Rutter, V. (1998). The Gender of Sexuality, Pine Forge, Thousand Oaks, CA. Simon, W., and Gagnon, J. (1984). Sexual scripts: Permanence and change. Society 22: 53–60. Smith, T. (1992). Discrepancies between men and women in reporting number of sexual partners: A summary from four countries. Soc. Biol. 39: 203–211. Smith, T. (1994). Attitudes toward sexual permissiveness: Trends, correlates, and behavioral connections. In Rossi, A. (ed.), Sexuality Across the Life Course, University of Chicago Press, Chicago, pp. 63–97. Sprecher, S. (1989). Premarital sexual standards for different categories of individuals. J. Sex Res. 26: 232–248. Tanfer, K., and Cubbins, L. (1992). Coital frequency among single women: Normative constraints and situational opportunities. J. Sex Res. 29: 221–250. Tanfer, K., Cubbins, L., and Billy, J. (1995). Gender, race, class, and sexually transmitted diseases. Family Plan. Perspect. 27: 196–102. Taris, T. W., and Semin, G. R. (1997). Gender as a moderator of the effects of the love motive and relational context on sexual experience. Arch. Sex. Behav. 26: 159–180.
P1: FMN Archives of Sexual Behavior [asb]
PL110-329
The Influence of Gender on Sex
April 19, 2000
12:8
Style file version Nov. 19th, 1999
257
The Gallup Report (1987). Number 259 (April). Trussell, J., and Westoff, C. (1980). Contraceptive practice and trends in coital frequency. Family Plan. Perspect. 12: 246–249. Tucker, M. B., and Mitchell-Kernan, C. (1995). The Decline in Marriage Among African Americans: Causes, Consequences, and Policy Implications, Russell Sage, New York. Udry, J., Deven, F., and Coleman, S. (1982). A cross-national comparison of the relative influence of male and female age on the frequency of marital intercourse. J. Biosoc. Sci. 14: 1–6. U.S. Department of Health and Human Services (1991). Healthy People 2000: National Health Promotion and Disease Prevention Objectives, DHHS Publication No. (PHS) 91-50212, U.S. Government Printing Office, Washington, DC. U.S. Department of Health and Human Services (1997). Sexually Transmitted Disease Surveillance, 1996, Division of STD Prevention, Public Health Service, Centers for Disease Control, Atlanta, GA. U.S. Women’s Bureau (1993). Facts on Working Women, Report No. 93-2, U.S. Government Printing Office, Washington, DC. Verbrugge, L. M. (1989). The twain meet: Empirical evidence of sex differences in health and mortality. J. Health Soc. Behav. 30: 282–304. Weinberg, M., and Williams, C. (1988). Black sexuality: A test of two theories. J. Sex Res. 25: 197–218. Wiederman, M. W. (1997). The truth must be in here somewhere: Examining the gender discrepancy in self-reported lifetime number of sex partners. J. Sex Res. 34: 375–386. Wilson, S., and Medora, N. (1990). Gender comparisons of college students’ attitudes toward sexual behavior. Adolescence 25: 615–627. Wilson, W. J. (1978). The Declining Signficance of Race, University of Chicago Press, Chicago.
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
Archives of Sexual Behavior, Vol. 29, No. 3, 2000
Sexual Orientation of Female-to-Male Transsexuals: A Comparison of Homosexual and Nonhomosexual Types Meredith L. Chivers, B.Sc.,1,2 and J. Michael Bailey, Ph.D.1
Homosexual and nonhomosexual (relative to genetic sex) female-to-male transsexuals (FTMs) were compared on a number of theoretically or empirically derived variables. Compared to nonhomosexual FTMs, homosexual FTMs reported greater childhood gender nonconformity, preferred more feminine partners, experienced greater sexual rather than emotional jealousy, were more sexually assertive, had more sexual partners, had a greater desire for phalloplasty, and had more interest in visual sexual stimuli. Homosexual and nonhomosexual FTMs did not differ in their overall desire for masculinizing body modifications, adult gender identity, or importance of partner social status, attractiveness, or youth. These findings indicate that FTMs are not a homogeneous group and vary in ways that may be useful in understanding the relation between sexual orientation and gender identity. KEY WORDS: transsexual; female-to-male; sexual orientation; gender identity; mating psychology.
INTRODUCTION Transsexualism in genetic females has previously been thought to occur predominantly in homosexual women. Clinical presentation by nonhomosexual female transsexuals (i.e., gender dysphoric genetic females who are sexually attracted to males) is extremely rare. Blanchard et al. (1987) reported that only 1 of 72 transsexual women seen at a Canadian gender identity clinic was primarily attracted to males. Because these individuals have been so infrequently seen by gender clinics, some researchers have thought that this form of female transsexualism was 1 Department
of Psychology, 2029 Sheridan Road, Swift Hall, Northwestern University, Evanston, Illinois 60208-2710. 2 To whom correspondence should be addressed. e-mail:
[email protected]. 259 C 2000 Plenum Publishing Corporation 0004-0002/00/0600-0259$18.00/0 °
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
260
Style file version Nov. 19th, 1999
Chivers and Bailey
nonexistent or was incorrectly diagnosed homosexual transsexualism (Blanchard et al., 1987). However, researchers and practitioners have begun to investigate nonhomosexual female transsexualism as a valid diagnostic entity (Blanchard, 1990; Blanchard et al., 1987; Clare and Tully, 1989; Coleman and Bockting, 1988; Coleman et al., 1993; Dickey and Stephens, 1995). These authors have typically described nonhomosexual female-to-male transsexuals (FTMs) as gender-dysphoric genetic females who describe themselves as gay or bisexual men and are attracted primarily to (most often gay) men. Dickey and Stephens (1995) synthesized findings from two case studies of nonhomosexual FTMs and available research data and concluded that nonhomosexual FTMs are characterized by the desire to be homosexual men, attraction to feminine men, interest in sexual activities performed by gay men, sexual fantasies of gay male sex during heterosexual intercourse, and a less extensive history of childhood or adolescent cross-gender identification compared with homosexual FTMs. Contrary to Dickey and Stephen’s second conclusion, Blanchard (1989) reported the case of a nonhomosexual FTM who was attracted to masculine as well as feminine men. Coleman and Bockting (1988) argued that gender identity and sexual orientation are discordant in the case of nonhomosexual FTMs because they have masculine gender identities and role behavior but have a “feminine” sexual orientation (toward men). If this assertion is correct, it would seem useful to examine variables known to differ between nongender dysphoric lesbian and heterosexual women to identify other similarities and differences between homosexual and nonhomosexual FTMs. In the research reported herein, we gathered information about aspects of sexuality and gender identity that have been empirically related to female sexual orientation. We also studied other traits that past research has suggested may be fruitful in understanding differences between homosexual and nonhomosexual FTMs. Specifically, this study examined FTMs with respect to variables pertaining to gender identity, partner preferences, sexual activities and interests, and body modifications. We begin by briefly reviewing the research examining these variables in heterosexual and lesbian women. We also describe any relevant research on FTMs. Gender Identity “Gender identity” refers to one’s psychological sense of being male or female, masculine or feminine (Money, 1972). Typically, retrospective accounts of sextypical behaviors and feelings of masculinity/femininity have served as indicators of childhood gender identity. Lesbians score as substantially more masculine than heterosexual women on such measures (Bailey and Zucker, 1995). Furthermore, “masculine” lesbians report greater childhood gender nonconforming behavior than nonmasculine lesbians (Bell et al., 1981; Singh et al., 1999).
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Sexual Orientation of Female-to-Male Transsexuals
Style file version Nov. 19th, 1999
261
Few studies have examined childhood gender nonconformity systematically in an FTM sample. Ehrhardt et al. (1979) found no difference in the frequency of sex-atypical behaviors (e.g., tomboyish behavior) between lesbians and homosexual FTMs. However, gender identity confusion in childhood, adolescence, and adulthood was absent in the lesbian sample and almost-unanimously reported by the FTM sample. If these retrospective findings are accurate, then cross-gender behavior is not synonymous with cross-gender identity. Steiner and Bernstein (1981) found that all 41 homosexual FTMs in their study reported high levels of childhood gender nonconformity. Coleman et al.’s (1993) sample of nine nonhomosexual FTMs reported, during interviews, that their experience of gender dysphoria began in childhood. Coleman and Bockting’s (1988) case report of a nonhomosexual FTM also found that this individual had gender atypical interests and activities during childhood. Unfortunately, none of these studies employed controls or compared nonhomosexual and homosexual FTMs. Investigators have hypothesized that nonhomosexual FTMs would report higher levels of childhood gender nonconformity compared with most genetic females but lower levels compared with homosexual FTMs (Blanchard, 1989; Dickey and Stephens, 1995). Thus we predicted that homosexual FTMs would report significantly higher levels of childhood gender nonconformity than nonhomosexual FTMs. Because gender identity and behavior are not perfectly correlated (Ehrhardt et al., 1979), we also examined the relationship between childhood behavior and identity (the two components of childhood gender nonconformity) and sexual orientation. Bailey et al. (1999) examined adult gender identity in lesbian and heterosexual women using the Continuous Gender Identity Scale, which assesses subjective feelings of masculinity and femininity, and found that lesbians reported more cross-gender identity feelings than heterosexual women. To our knowledge, there have been no empirical studies of adult gender identity comparing homosexual and nonhomosexual FTMs. We expected that FTMs would report high adult cross-gender identity because FTMs identify themselves as male. However, homosexual FTMs should report more masculine feelings than nonhomosexual FTMs. We also expected that a significant positive relationship would exist between childhood gender nonconformity and adult gender identity, replicating the findings of Bailey et al. (1999) with lesbian and heterosexual women. Partner Preferences Homosexual FTMs prefer feminine women (Fleming et al., 1984; Steiner and Bernstein, 1981). In contrast, research examining preferences for masculine versus feminine partners in nonhomosexual FTMs is inconsistent. Regarding this issue, Blanchard (1989) asked whether FTMs attracted to effeminate gay men
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
262
Chivers and Bailey
constitute a distinct subgroup and whether any female gender dysphorics strongly prefer masculine men. The present study examined the partner preferences (i.e., for masculine or feminine partners) of homosexual and nonhomosexual FTMs. We also investigated whether FTMs of both types prefer homosexual or heterosexual partners. Devor (1997) discussed the partner preference histories of several FTMs: a consistent theme was the interest of these FTMs in partners who desired them as males once they had established a transsexual identity. By definition, the desired partners of FTMs would not include lesbian women or heterosexual men. Based on this definition, we would predict that homosexual FTMs would be more interested in heterosexual versus lesbian women and that nonhomosexual FTMs would express greater interest in gay men versus heterosexual men. Researchers using an evolutionary perspective have identified several sexually dimorphic partner preferences, including the importance of a partner’s physical attractiveness, youth, and status, and, less directly related, sexual versus emotional jealousy. Sexual and emotional jealousy refers to an individual’s tendency to experience greater distress at the prospect of a partner being sexually or emotionally unfaithful, respectively. Although this is not a partner preference per se, it does indirectly refer to a preference for a certain type of partner behavior. Bailey et al. (1994) reported that both heterosexual and lesbian women rated partner’s physical attractiveness and youth as relatively unimportant and reported a tendency toward greater emotional than sexual jealousy. Lesbians were more masculine with regard to partner status, because this was relatively unimportant to them. If these partner preferences are related to gender identity, then FTMs should be similar to men. However, we hypothesized that preferences distinguishing homosexual and heterosexual (nontranssexual) women would also distinguish homosexual and nonhomosexual FTMs. Sexual Activities/Interests Lesbian sexual relationships often involve a differentiation of partners’ roles as either “top” (active) or “bottom” (passive). “. . . The top is the person who conducts and orchestrates the episode. . . . The bottom is the one who responds, acts out, makes visible or interprets the sexual initiatives and language of the top” (Newton and Walton, 1984, p. 246). Preference for the active or passive sexual role in lesbians has been related to adult gender identity; “butch” lesbians tend to prefer the active sexual role, while “femme” lesbians preferred the passive sexual role (Bailey et al., 1999; Singh et al., 1999). Furthermore, one study found that lesbians who recalled gender conforming behavior in childhood reported a preference for the passive sexual role (Singh and Vidaurri, 1999). We hypothesized that preference for the passive role would be significantly related to childhood nonconforming behavior and to a continuous measure of adult gender identity. We
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Sexual Orientation of Female-to-Male Transsexuals
Style file version Nov. 19th, 1999
263
also hypothesized that nonhomosexual FTMs would report a passive preference, while homosexual FTMs would report an active preference. Evolutionary psychologists have also studied sexual interests, such as interest in uncommitted sex and interest in visual sexual stimuli, with respect to both gender and sexual orientation. On average, men are more interested in both casual sex and visual sexual stimuli (Symons, 1979). Bailey et al. (1994) found that heterosexual and lesbian women described low levels of interest in uncommitted sex but that lesbians were more masculine in their interest in visual sexual stimuli, which was higher than that of heterosexual women. Furthermore, masculine lesbians express greater enjoyment of visual erotica than feminine lesbians (Singh and Vidaurri, 1999). Masculine lesbians have reported greater numbers of sexual partners than feminine lesbians or heterosexual women (Singh et al., 1999). Because these traits are sexually dimorphic and because of the suggestive findings with respect to some of them among nontranssexual women, we predicted that homosexual and nonhomosexual FTMs would report sexual interests that are analogous to those of lesbian and nontranssexual heterosexual women. Specifically, we hypothesized that homosexual FTMs, compared with nonhomosexual FTMs, would report equivalent interest in uncommitted sex, higher interest in visual sexual stimuli, and a greater number of sexual partners. Body Modifications FTMs may engage in a variety of procedures to become more physically masculine, which vary in complexity and permanence (e.g., wearing short hair or building muscle mass through exercise versus surgical treatments such as bilateral mastectomy or phalloplasty). Desire for masculinizing body modifications is potentially related to the degree of cross-gender identification. If nonhomosexual FTMs have not experienced as much gender identity confusion as homosexual FTMs have, then they may have less desire for physical masculinization. We thus predicted that homosexual FTMs would have a greater desire for masculinizing body modifications than nonhomosexual FTMs do. More specifically, many FTMs appear to have a strong desire to have a penis. Steiner and Bernstein (1981) reported that all 41 of the homosexual FTMs they studied had this wish. However, because of high cost and current surgical limitations, many opt not to have this surgery. It is possible that the desire for a penis would be stronger among nonhomosexual FTMs because their lack of a penis is very obvious during sexual interactions with genetic men and this absence may reinforce their feelings of not being truly male (Devor, 1993). Alternatively, desire for a penis may be related to cross-gender identification in childhood and adulthood. If nonhomosexual FTMs are less cross-gender identified than homosexual FTMs, as we have hypothesized, then nonhomosexual FTMs might express less desire for phalloplasty than homosexual FTMs.
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
264
Style file version Nov. 19th, 1999
Chivers and Bailey
Thus, the intent of this study is to characterize the similarities and differences between homosexual and nonhomosexual FTMs. Very generally, we predicted greater sex atypicality among homosexual FTMs; we expected that if the two groups differed on sexually dimorphic traits, homosexual FTMs would appear more masculine than nonhomosexual FTMs. We particularly expected differences on sexually dimorphic traits that prior research found related to sexual orientation among nontranssexual women. METHOD Participants We recruited female-to-male transsexuals (FTMs) using advertisements posted on several Internet web pages and news groups for female-to-male transsexuals or, more generally, for transgendered people. One of these web pages was specifically designed for nonhomosexual FTMs. The advertisements stated that female-to-male transsexuals at any stage of transition were desired for a study of the development of sex differences. Those who responded received questionnaires through the mail. As the study progressed, we also took advantage of snowball sampling opportunities. The final sample included 39 FTMs. Due to incomplete questionnaires, the sample size available for different measures fluctuated slightly from 35 to 39 FTMs. Measures Demographics Demographic information collected included age, level of education, and ethnicity. Level of education attained ranged from 1 (no high school) to 7 (graduate degree completed). Sexual Orientation A modified Kinsey scale was administered (Kinsey et al., 1953). The scale assessed sexual fantasy and behavior during the past year using a self-report, 7-point scale format: a score of 0 indicates exclusive sexual feelings toward, or sexual behavior with, men, and a score of 6 indicates exclusive sexual feelings toward, or behavior with, women. Items for the Passive Sexual Role scale, the Preference for Partner Masculinity scale, and the Body Modification scale and two items concerning Sexual versus Emotional Jealousy [subsequently added to those written by Buss et al. (1992)] were written by the second author and are included in the Appendix. Characteristics of all scales included in the study are given in Table I.
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
Sexual Orientation of Female-to-Male Transsexuals
265
Table I. Description of Study Measures Name of scale
Number of items
Childhood gender nonconformity Childhood behavior
7
Childhood identity
4
Continuous gender identity
10
Preference for partner masculinity
Importance of partner physical attractiveness
3
7
11
Interest in younger partners
9
Low concern with partner status
12
Sexual vs. emotional jealousy
4
Passive sexual role
5
Interest in uncommitted sex
10
Interest in visual sexual stimuli
Body modification
8
11
Rating scale 1 (strongly disagree) to 7 (strongly agree) 1 (strongly disagree) to 7 (strongly agree) 1 (strongly disagree) to 7 (strongly agree)
Sample item “I was a masculine girl”
“As a child, I preferred playing with boys” “As a child, I sometimes wished I had been born a boy rather than a girl” 1 (strongly disagree) “In many ways, I feel more to 7 (strongly agree) similar to men than to women” 7-point scale: Very “Would your ideal partner (masculine be: very hairy, somewhat characteristic) to very hairy, slightly hairy, neither, (feminine equivalent slightly unhairy, somewhat of characteristic) unhairy, very unhairy” 1 (strongly disagree) “It is more important to to 7 (strongly agree) me how nice a potential romantic partner is than how good looking they are” 1 (strongly disagree) “I am most sexually to 7 (strongly agree) attracted to younger adults (aged 18–25)” 1 (strongly disagree) “I would not want to get to 7 (strongly agree) romantically involved with someone who did not have a job” 1 (strongly disagree) “I would end my to 7 (strongly agree) relationship if I discovered that my partner had been sexually unfaithful” 1 (strongly disagree) “I am more sexually to 7 (strongly agree) aggressive than my sex partners” 1 (strongly disagree) “I could easily imagine to 7 (strongly agree) myself enjoying one night of sex with someone I would never see again” 1 (strongly disagree) “Whether or not I approve to 7 (strongly agree) of them, I find films of attractive people having sex to be very sexually exciting” 1 (I’ve never See Table IV considered it) to 5 (I’ve done it)
α .84 .65 .82 .62 .93
.77
.83 .64
.78
.83 .91
.81
.81
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
266
Chivers and Bailey
Gender Identity Four scales were included to assess both childhood and adulthood gender identity. The Childhood Gender Nonconformity scale measured participants’ retrospective concepts of self as masculine or feminine in childhood and crossgender behavior. We divided this scale into two subscales, Childhood Behavior and Childhood Identity, to assess retrospective reports of childhood behavior and self concepts separately. The Continuous Gender Identity scale assessed participants’ current self-concepts as masculine or feminine. High scores on all scales indicate sex-atypical responses. Partner Preferences Six scales were included to assess aspects of partner preference. The Preference for Partner Masculinity scale assessed preference for masculine physical and behavioral characteristics in a partner. Items for scales assessing concern with partner status, partner attractiveness, and youth were written by Bailey et al. (1994): Low Concern with Partner Status, Importance of Partner’s Physical Attractiveness, and Interest in Younger Partners. The Sexual vs. Emotional Jealousy Scale included the original items written by Buss et al. (1994), and items written by the second author. High scores on this scale indicate a tendency toward sexual jealousy. Sexual Interest/Activity Preferences Three scales and one item were included to assess aspects of preferences for certain sexual activities and interests. The Passive Sexual Role scale assessed preferences for the active (top) or passive (bottom) role in sexual encounters. Interest in Uncommitted Sex and Interest in Visual Sexual Stimuli scales and items were written by Bailey et al. (1994). One item asked participants to estimate the total number of sex partners they have. Body Modification The Body Modification Scale asked FTMs whether they had considered various physical alterations or procedures to appear masculine (e.g., short hair, breast binding, phalloplasty, see Tables I and IV for details). Analyses We have suggested that homosexual and nonhomosexual FTMs are not a uniform group. Consistent with this, we divided our sample into two groups and used
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Sexual Orientation of Female-to-Male Transsexuals
Style file version Nov. 19th, 1999
267
t tests to compare them on relevant variables; details about our classification system are provided. The Kinsey scale is a quasi-continuous scale rather than dichotomous, and not all FTMs were easily classified as homosexual or nonhomosexual. Therefore, we also performed a parallel set of analyses consisting of Pearson correlations between variables and present Kinsey Sexual Fantasy score. Although the correlation and t-test analyses are somewhat statistically redundant, they are not equivalent, and in some cases, the statistical significance of results differed between them. Because of the relatively small sample size and the exploratory nature of this study, we used a type 1 error rate (α) of .10 (two-tailed).
RESULTS Sample Characteristics FTMs were classified as homosexual or nonhomosexual according to their present sexual fantasies, measured by the Kinsey Sexual Fantasy Scale. We used the Kinsey Sexual Fantasy rather than the Sexual Behavior Scale, because sexual behavior is potentially influenced by opportunity. Sexual fantasy provides a clearer picture of whom an individual wishes to have sexual relations with regardless of opportunity. FTMs reporting Kinsey Sexual Fantasy Scores of 4 (most sexual feelings toward females, but some definite fantasy about males) or higher were designated homosexual (relative to the genetic sex of the subject) and FTMs reporting Kinsey Sexual Fantasy Scores of 3 (sexual feelings about equally divided between males and females) or lower were designated nonhomosexual. This method of classification yielded 21 homosexual FTMs and 17 nonhomosexual FTMs. Descriptive statistics for the subsamples are given in Table II. The nonhomosexual FTMs were significantly older [t(36) = 1.77, p < .1] and attained a higher level of education [t(36) = −2.08, p < .05] than the homosexual FTMs, the former having completed some graduate work and the latter having graduated college, on average.
Gender Identity Means and standard deviations for scales related to gender identity, partner preferences, and sexual activities/interests are presented in Table III. Consistent with our predictions, homosexual FTMs reported higher Childhood Gender Nonconformity than nonhomosexual FTMs [t(36) = 1.67, p = .10]. Sexual orientation, as a continuous variable (Kinsey Sexual Fantasy Scale), was significantly correlated with Childhood Gender Nonconformity in the predicted direction [r (36) = .37, p < .05], with FTMs who were more attracted to men tending to report less childhood cross-gender identification.
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
268
Chivers and Bailey Table II. Sample Group Characteristics (Standard Deviations in Parentheses) Homosexual (N = 21) Demographics Mean age (yr)∗ Ethnicity (No.) Caucasian Other Education∗∗ Kinsey fantasy score Frequency (No.) 6 5 4 3 2 1 0 Kinsey present sexual Fantasy (mean score)∗∗∗
Nonhomosexual (N = 17)
36 (8)
32 (8)
16 5 5.1 (1.2)
16 1 6.1 (1.6)
13 4 4 0 0 0 0
0 0 0 4 10 0 3
5.4 (.8)
1.9 (1.0)
∗ p < .1. ∗∗ p < .05. ∗∗∗ p < .001.
Homosexual FTMs recalled significantly more masculine behaviors in childhood (Childhood Behavior: M = 6.5, SD = 1.0) than nonhomosexual FTMs (M = 5.6, SD = 1.6) [t(36) = 2.15, p = .05] but reported equivalent feelings of masculinity in childhood (Childhood Identity: M = 6.4, SD = 1.2) as nonhomosexual FTMs (M = 6.3, SD = 1.0) [t(36) = .99 n.s.]. Childhood Behavior was significantly correlated with sexual orientation [r (37) = .36, p < .05], but Childhood Identity was not [r (37) = .2, n.s.]. However, these correlations were not significantly different from each other. To test more rigorously whether the difference in recalled masculine behavior was independent of (even nonsignificant) differences in recalled gender identity, we performed the following multiple regression. The dependent variable, Masculine Childhood Behavior, was regressed on both continuously measured Sexual Orientation and Childhood Identity. Both Childhood Identity [t(1) = 5.47, p < .001] and Sexual Orientation [t(1) = −1.96, p < .1] were significant predictors of Childhood Behavior (adj. r 2 = .495). Continuous Gender Identity was significantly correlated with childhood gender nonconformity [r (36) = .34, p < .05], but homosexual and nonhomosexual FTMs did not differ significantly on this variable, or was it significantly correlated with sexual orientation score. Partner Preferences Not surprisingly, homosexual FTMs found lesbians and heterosexual women more sexually appealing than nonhomosexual FTMs did. In contrast,
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
Sexual Orientation of Female-to-Male Transsexuals
269
Table III. Group Means and Standard Deviations for Study Scales
Gender identity Childhood gender nonconformity Continuous gender identity Partner preferences Preference for partner masculinity Sexual desirability of Heterosexual women Heterosexual men Lesbians Gay men Importance of partner physical attractiveness Interest in younger partners Low concern with partner status Sexual vs. emotional jealousy Sexual activities/interests Passive sexual role Number of sexual partners Interest in uncommitted sex Interest in visual sexual stimuli
Homosexuala
Nonhomosexualb
d
Correlation with Kinsey score (r )
6.6 (1.1)
6.0 (1.2)
.56∗
.37∗∗
6.0 (.9)
5.5 (.9)
.21
.12
2.7 (1.2)
5.1 (1.0)
−2.26∗∗∗
−.81∗∗∗
5.6 (1.0) 1.3 (.7) 4.5 (1.3) 2.7 (1.7) 3.9 (.8)
3.6 (1.8) 3.2 (1.8) 3.2 (1.8) 5.5 (1.2) 3.8 (1.2)
−1.8∗∗∗ 1.8∗∗∗ .83∗∗ 1.95∗∗∗ .07
.77∗∗∗ −.56∗∗∗ .44∗∗∗ −.77∗∗∗ .08
3.3 (1.2) 4.3 (.9)
3.2 (1.3) 4.3 (.7)
.09 −.09
4.1 (1.7)
2.6 (1.2)
−1.2∗∗∗
2.2 (1.6) 22.0 (16.0) 3.7 (1.6) 5.8 (.7)
3.9 (1.0) 11.9 (11.6) 4.2 (1.6) 5.0 (1.1)
−1.24∗∗∗ −.73∗∗ −.35 1.06∗∗
.13 −.06 .58∗∗∗ −.61∗∗∗ .42∗∗∗ −.27 .33∗∗
Note. Significance of t tests: ∗ p < .1; ∗∗ p < .05; ∗∗∗ p < .01. a n for homosexual FTM group ranged from 19 to 21. b n for nonhomosexual FTM group ranged from 16 to 17.
nonhomosexual FTMs found gay and heterosexual men more appealing than homosexual FTMs did. Paired t tests revealed that homosexual FTMs rated the sexual desirability of heterosexual women higher than that of lesbians [t(18) = 2.6, p < .05], and nonhomosexual FTMs found gay men more appealing than heterosexual men [t(15) = 2.5, p < .01]. Homosexual FTMs preferred very feminine characteristics in their partners while nonhomosexual FTMs indicated a preference for a partner with masculine characteristics [t(32) = 3.49, p < .01]. Sexual Orientation was significantly related to Preference for Partner Masculinity [r (33) = −.81, p < .001], as was Childhood Gender Nonconformity [r (33) = −.34, p < .05]. In contrast to findings from nontranssexual women (Bailey et al., 1994), there was a sexual orientation effect on jealousy: homosexual FTMs reported greater sexual jealousy than nonhomosexual FTMs [t(36) = −3.57, p < .001]. There was a significant relation between Sexual Orientation and Sexual Jealousy [r (37) = .43, p <.01]. All other comparisons (Importance of Partner’s Physical Attractiveness, Low Concern with Partner Status, and Interest in Younger Partners) were not significant.
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
270
Chivers and Bailey
Sexual Activities/Interests Homosexual FTMs reported a preference for an assertive sexual role, while nonhomosexual FTMs seemed to prefer a more neutral sexual role (neither dominant/top nor submissive/bottom) [t(32) = 3.49, p < .01]. Sexual Orientation was significantly related to Passive Sexual Role [r (34) = −.61, p < .001], as was Childhood Gender Nonconformity [r (34) = −.45, p < .01]. Contrary to our prediction, Continuous Gender Identity was not significantly correlated with preference for the passive sexual role [r (34) = −.16, p = .34]. Thus, FTMs reporting less childhood gender nonconformity and sexual fantasies featuring predominantly males also reported a preference for the passive sexual role and a preference for masculine partners. As predicted, based on prior research with nontranssexual females (Bailey et al., 1994), homosexual FTMs reported a significantly greater interest in visual sexual stimuli than nonhomosexual FTMs. Sexual Orientation was significantly correlated with Interest in Visual Sexual Stimuli [r (36) = .33, p < .05]. As predicted, no significant differences were found between homosexual and nonhomosexual FTMs on Interest in Uncommitted Sex. Homosexual FTMs reported having significantly more sexual partners than nonhomosexual FTMs [t(33) = 2.09, p < .05]. The t tests for this comparison were performed with one outlier removed, a homosexual FTM who reported 107 partners (over 2 SD from the next highest score). Subsequent analysis was performed with 19 homosexual FTMs and 16 nonhomosexual FTMs. Sexual orientation was significantly correlated with number of sexual partners [r (33) = .42, p < .01]. Body Modifications Percentages of body modifications completed by each group are given in Table IV. The groups did not differ in total desired body modifications [t(34) = .19, n.s.], nor was total body modification score significantly correlated with Sexual Orientation [r (34) = −.016, n.s.]. Childhood Gender Nonconformity was significantly related to desired body modification [r (34) = .46, p < .01]; FTMs reporting higher cross-gender identification in childhood reported a stronger interest in masculinizing body modifications. The t tests performed on each body modification item revealed that the homosexual FTMs reported a greater desire for phalloplasty (M = 2.4, SD = .9) than the nonhomosexual FTMs (M = 2.0, SD = .6) [t(36) = 1.6, p < .1, d = .54] as predicted. Correlations between each body modification item and Sexual Orientation were all less than .24 ( p > .15) except for the first item (“wearing makeup to appear that you have facial hair” [r (36) = −.59, p < .1]. The correlation between desire for phalloplasty and sexual orientation was in the predicted direction [r (36) = .24, p = .15]. Childhood Gender
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
Sexual Orientation of Female-to-Male Transsexuals
271
Table IV. Body Modifications Homosexual (n = 21) Body modifications completed (%) Wearing makeup to give the appearance of facial hair Shaving to promote growth of facial hair Wearing short hair Lifting weights to become bulkier Breast binding Padding pants to give the appearance of having a penis Bilateral mastectomy Hysterectomy Phalloplasty Metoidioplasty Testosterone injections Body modification scale (mean score)
Nonhomosexual (n = 17)
48
29
71
76
91 81 86 76
82 82 88 82
38 33 0 5 62 3.9 (0.9)
41 12 0 6 65 3.9 (0.5)
Non-conformity was not significantly related to desire for phalloplasty [r (36) = .12, n.s.], but Continuous Gender Identity was [r (36) = .29, p < .1]. All other tests were nonsignificant. DISCUSSION The results of this study suggest that FTMs are not a homogeneous group. Though similar in many respects, homosexual and nonhomosexual FTMs differed in ways that were generally consistent with our predictions and those of other researchers. Homosexual FTMs reported greater childhood gender nonconformity, preferred more feminine partners, were more sexually assertive, had more sexual partners, had a greater desire for phalloplasty, and reported sexual interests which are analogous to those of nontranssexual lesbians (higher interest in visual sexual stimuli). Contrary to our predictions, however, homosexual FTMs did not report a greater desire for masculinizing body modifications, greater adult masculinity, or less importance of partner social status. Homosexual FTMs indicated a tendency toward sexual jealousy, whereas nonhomosexual FTMs reported an inclination toward emotional jealousy; this finding was not predicted because research on nontranssexual women has not found an analogous difference. The homosexual FTMs reported more masculinity in childhood than our nonhomosexual FTMs, but the two groups did not differ in degree of cross-gender identification in adulthood. It is noteworthy that both groups of FTMs reported high gender nonconformity in childhood, but homosexual FTMs reported significantly higher memories of childhood masculine behavior compared with nonhomosexual FTMs. The effect size for this finding was moderate (d = .56) and is analogous
P1: FQJ Archives of Sexual Behavior [asb]
272
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
Chivers and Bailey
to findings in nontranssexual women (Bailey and Zucker, 1995), with lesbians reporting more masculine childhoods than heterosexual women. Homosexual FTMs differed from nonhomosexual FTMs in recalled crossgender childhood behavior but not recalled cross-gender identification. There are at least two explanations why nonhomosexual FTMs might report more atypicality in childhood gender identity than in sex-atypical behavior. The first is that they indeed had strong feelings of masculinity in childhood, but those feelings were not expressed in overt behavior. The second is that their memories of cross-gender feelings are magnified by retrospective bias to a greater extent than their memories of cross-gender behavior. This finding is also interesting with regard to the etiology of homosexual and nonhomosexual transsexualism in females. Gender dysphoria was reported in childhood and adulthood by both homosexual and nonhomosexual groups, suggesting that cross-gender identity is not solely the result of same-sex attraction. Both groups of FTMs were, however, very similar in their reports of adult feelings of masculinity. This was contrary to our prediction, but not entirely surprising, as both homosexual and nonhomosexual FTMs identify as men. FTMs preferred partners who are attracted to males (heterosexual women and gay men) and who thus regard their FTM partner as male. This supports the impressions of Devor (1997) that FTMs are very interested in those individuals who will eroticize them as males. Preferences for partner masculinity differed for homosexual and nonhomosexual FTMs; homosexual FTMs reported a preference for “slightly” to “somewhat” feminine heterosexual women and nonhomosexual FTMs reported a preference for “slightly” masculine partners. It would be interesting to explore how homosexual and nonhomosexual FTMs’ preferences compare with nontranssexual females with similar orientations. Homosexual and nonhomosexual FTMs also differed in their past number of sex partners. This pattern was analogous to the results of Singh and Vidaurri’s study of nontranssexual “butch” and “femme” lesbians. It is noteworthy that nonhomosexual FTMs had fewer sex partners despite the fact that their preferred partners were gay men, who are generally more interested in casual sex than the heterosexual women whom homosexual FTMs pursue sexually (Bailey et al., 1994). There are at least two possible explanations. The first is that homosexual FTMs are more successful in finding partners than nonhomosexual FTMs. The second explanation is that homosexual FTMs are more motivated than nonhomosexual FTMs to engage in casual sex. Contradicting the latter interpretation, our two groups did not differ significantly with respect to interest in casual sex. Alternatively, perhaps nonhomosexual FTMs are less desirable to gay men than homosexual FTMs are to heterosexual women. This might reflect a male tendency to be discriminating with respect to the genitals of a potential partner. Alternatively, perhaps nonhomosexual FTMs are more feminine than the average gay man, and this femininity is not desired by gay men; Bailey et al. (1997) have shown that gay men typically value masculinity in partners. It is also possible that heterosexual
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
Sexual Orientation of Female-to-Male Transsexuals
9:13
Style file version Nov. 19th, 1999
273
women are less selective with respect to femininity in partners and might even value this in a homosexual FTM partner (Fleming et al., 1984). Sexual role preference of FTMs was related to sexual orientation, with homosexual transsexuals more likely to be active and nonhomosexual transsexuals to be passive. Our findings are analogous to those of Bailey et al. (1999) and Singh et al. (1999), who studied nontranssexual women. One case study of a nonhomosexual FTM (Coleman and Bockting, 1988) provided a detailed account of the sexual activities this individual engaged in with his gay male partner, and these included receptive anal intercourse and penile–vaginal intercourse (this FTM still had a vagina). This nonhomosexual FTM apparently preferred a “bottom” or passive sexual role, consistent with our results. Homosexual and nonhomosexual FTMs did not differ in their desire for masculinizing body modifications. Contrary to the speculation of Devor (1993), nonhomosexual FTMs were less interested in phalloplasty than FTMs. Having a penis allows an individual to assume the insertive sexual role that homosexual FTMs tended to prefer. The lack of a penis would not necessarily compromise nonhomosexual FTMs’ sexual interactions with gay men, as they had no role preference. The sexual interests of homosexual and nonhomosexual FTMs appear to be analogous to those of nontranssexual females of the same sexual orientation. Both groups of FTMs reported interest in visual sexual stimuli but homosexual FTMs reported a higher interest that nonhomosexual FTMs, paralleling the results of Bailey et al. (1994) and Singh et al. (1999) studies of nontranssexual women. It seems implausible that this result is attributable to FTMs mimicking a more masculine sexual interest. For example, this could not explain the difference between homosexual and nonhomosexual FTMs’ interest in visual sexual stimuli. Biological explanations of masculinized sexual behaviors in women seem more plausible. For example, Money and Ehrhardt (1972) found that prenatally androgenized women (women with congenital adrenal hyperplasia [CAH]) were more responsive to visual sexual imagery than nonandrogenized women. Women with CAH have also been shown to exhibit more masculine childhood behavior (Money and Schwartz, 1977) and to show markedly greater rates of bisexuality and lesbianism (Money et al., 1984). Therefore, a relationship may exist between exposure to masculinizing agents during development, masculine childhood behavior, homosexuality, and interest in visual sexual stimuli. If so, nonhomosexual FTMs may have had less exposure to masculinizing agents. Thus, despite their gender dysphoria, their sexual orientation and sexual psychology remain unaffected and femaletypical. Sexual jealousy was more intense for homosexual FTMs than for nonhomosexual FTMs, a result that has not been observed with nontranssexual women. In one study, lesbian women reported levels of emotional jealousy similar to those of heterosexual women (Bailey et al., 1994). It is possible to interpret these results as related to the gay male culture that nonhomosexual FTMs would enter as gay men.
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
274
Chivers and Bailey
Gay men are less sexually jealous than heterosexual men (Bailey et al., 1994) and some authors have suggested that sexual exclusivity is less valued in gay male culture (Hawkins, 1990). Enculturated nonhomosexual FTMs would also subscribe to this set of values. Our results indicate, however, that nonhomosexual FTMs actually report significantly fewer sexual partners than homosexual FTMs. Methodological Limitations The results of this study are limited by three methodological concerns: nonrandom recruitment of participants, limited information about the validity of some of our measures, and small sample size. FTMs were recruited via computer-based media, which limits recruitment to participants who are computer-literate and who have the resources to gain access to such technology. This method of recruitment restricts the education level and socioeconomic status of participants. Future studies should attempt to vary recruitment methods to avoid these and other potential sampling biases. A second potential limitation of this study concerns the validity of our measures. Some of the measures we used are new and have not yet been rigorously validated with nontranssexual populations. Even so, these measures demonstrated a relatively high internal consistency reliability and high face validity. As well, some of these measures have been used in previous studies (Bailey et al., 1994; Bailey et al., 1999) and have shown consistent sex and sexual orientation differences. Our hypotheses related to sexual psychology were based on the results of these studies and were supported with data from, in some ways, a very different population. A third limitation of this study is our small sample size. Given the rarity of our target populations, the number of subjects we did recruit is quite remarkable. Even so, larger numbers of individuals would allow for greater generalizability and statistical power. Replication of our results with larger and more representative samples is therefore desirable. Future Directions This study represents, to our knowledge, the first systematic investigation of a typology of female gender dysphoria. Our results illuminate the differences between homosexual and nonhomosexual FTMs, but there are many unanswered questions. For example, it would seem beneficial to ascertain whether nonhomosexual FTMs exhibit “autoandrophilia,” the female analogue of autogynephilia, which appears to be a core component of nonhomosexual male-to-female transsexualism (Blanchard, 1989). Authors reporting on female-to-male transsexualism have noted a childhood genesis of cross-dressing in their sample of FTMs (Roback and Lothstein, 1986), but only one specified that cross-dressing (in a singular
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
Sexual Orientation of Female-to-Male Transsexuals
275
heterosexual FTM) was not accompanied by sexual excitement (Dickey and Stephens, 1995). Given that the incidence of true paraphilia among genetic females is extremely low (DSM-IV, 1994), we would expect that “autoandrophilia” would be rare or nonexistent among nonhomosexual FTMs. It would also be beneficial to examine whether nonhomosexual FTMs exhibit hormonal abnormalities and medical conditions which have been reported in samples of, presumably, homosexual FTMs: increased levels of testosterone and differential incidences of polycystic ovarian disease, a medical condition associated with hormonal abnormalities (Futterweit et al., 1986). The investigation of etiological variables such as gender identity would be highly illuminating not only for the understanding of the development of nonhomosexual FTMs but for the understanding of the relationship between sexual orientation and gender identity. Longitudinal studies of tomboys (masculine girls) may reveal whether either form of female transsexualism is associated with marked childhood masculinity in girls. The hypothesis that different brain modules are implicated in different aspects of sexual psychology may be applicable to our results (see Freund, 1990; Quinsey and Lalumiere, 1995). Modules that control for sexual orientation (target preference) may be closely related to those that control for interest in feminine appearing partners, interest in visual sexual stimuli, the tendency for sexual jealousy, and dominant sexual roles. If the neurohormonal hypothesis of sexual orientation applies to homosexual females with gender dysphoria, these related sexual psychology modules may have become “masculinized” during a critical developmental period. Other sexual psychology modules would remain unaffected, the result being a mixture of typically feminine and masculine traits and preferences in homosexual FTMs. From this hypothesis, nonhomosexual FTMs would not experience extensive masculinization during development and would thus retain sexual “modules” congruent with their genetic sex and sexual orientation. Other modules, such as gender identity, would be affected, hence the “independence” of effects observed in this study. ACKNOWLEDGMENTS We thank Amanda Schwegler for her assistance in data collection. We also thank Joan Linsenmeier and Michael Seto for their helpful comments on an early version of this article. REFERENCES American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th ed., APA, Washington, DC.
P1: FQJ Archives of Sexual Behavior [asb]
276
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
Chivers and Bailey
Bell, A. P., Weinberg, M. S., and Hammersmith, S. K. (1981). Sexual Preference: Its Development in Men and Women, Alfred C. Kinsey Institute of Sex Research, Bloomington, IN. Bailey, J. M., and Zucker, K. (1995). Childhood sex-typed behavior and sexual orientation: A conceptual analysis and quantitative review. Dev. Psychol. 31(1): 43–55. Bailey, J. M., Gaulin, S., Agyei, Y., and Gladue, B. A. (1994). Effects of sexual orientation on evolutionarily relevant aspects of mating psychology. J. Person. Soc. Psychol. 66(6): 1081–1093. Bailey, J. M., Kim, P., Hills, A., and Linsenmeier, J. A. W. (1997). Butch, femme, or straight-acting? Partner preferences of gay men and lesbians. J. Person. Soc. Psychol. 73(5): 960–973. Bailey, J. M., Finkel, E., Blackwelder, K., and Bailey, T. (1999). Masculinity, femininity and sexual orientation (in preparation). Blanchard, R. (1989). The classification and labeling of nonhomosexual gender dysphorias. Arch. Sex. Behav. 18(4): 315–334. Blanchard, R. (1990). Gender identity disorders in adult women. In Blanchard, R., Steiner, B. W., et al. (eds.), Clinical Management of Gender Identity Disorders in Children and Adults. The Clinical Practice Series, No. 14, American Psychiatric Press, Washington, DC, pp. 79–91. Blanchard, R., and Freund, K. (1983). Measuring masculine gender identity in females. J. Consult. Clin. Psychol. 51(2): 205–214. Blanchard, R., Clemmensen, L., and Steiner, B. W. (1987). Heterosexual and homosexual gender dysphoria. Arch. Sex. Behav. 16(2): 139–152. Buss, D. M., Larsen, R. J., Westen, D., and Semmelroth, J. (1992). Sex differences in jealousy: Evolution, physiology, and psychology. Psychol. Sci. 3: 251–255. Clare, D., and Tully, B. (1989). Transhomosexuality or the dissociation of sexual orientation and sex object choice. Arch. Sex. Behav. 18(6): 531–536. Coleman, E., and Bockting, W. U. (1989). Heterosexual prior to sex reassignment surgery, homosexual afterwards. J. Psychol. Hum. Sex. 1(2): 69–81. Coleman, E., Bockting, W. O., and Gooren, L. (1993). Homosexual and bisexual identity in sexreassigned female-to-male transsexuals. Arch. Sex. Behav. 22(1): 37–50. Devor, H. (1993). Sexual orientation identities, attractions, and practices of female-to-male transsexuals. J. Sex Res. 30(4): 303–315. Devor, H. (1997). FTM: Female-to-Male Transsexuals in Society, Indiana University Press, Bloomington. Dickey, R., and Stephens, J. (1995). Female-to-male transsexualism, heterosexual types: Two cases. Arch. Sex. Behav. 24(4): 439–445. Ehrhardt, A., Gridanti, G., and McCauley, E. A. (1979). Female-to-male transsexuals compared to lesbians: Behavioral patterns of childhood and adolescent development. Arch. Sex. Behav. 8(6): 481–490. Fleming, M. Z., MacGowan, B. R., and Salt, P. (1984). Female-to-male transsexualism and sex roles: Self and spouse ratings on the PAQ. Arch. Sex. Behav. 13(1): 51–57. Fleming, M., MacGowan, B., and Costos, D. (1985). The dyadic adjustment of female-to-male transsexuals. Arch. Sex. Behav. 14(1): 47–55. Freund, K. (1990). Courtship disorder. In Marshall, W. L., Laws, D. R., et al. (eds.), Handbook of Sexual Assault: Issues, Theories, and Treatment of the Offender, Plenum Press, New York, pp. 195– 207. Futterweit, W., Weiss, R. A., and Fagerstrom, R. M. (1986). Endocrine evaluation of forty femaleto-male transsexuals: Increase frequency of polycystic ovarian disease in female transsexualism. Arch. Sex. Behav. 15(1): 69–78. Hawkins, R. O. (1990). The relationship between culture, personality, and sexual jealousy in men in heterosexual and homosexual relationships. J. Homosex. 19: 67–84. Kinsey, A. C., Pomeroy, W. B., Martin, C. E., and Gebhard, P. H. (1953). Sexual Behavior in the Human Female, Saunders, Philadelphia. Money, J., and Ehrhardt, A. A. (1972). Man and Woman, Boy and Girl: The Differentiation and Dimorphism of Gender Identity from Conception to Maturity, Johns Hopkins Press, Baltimore. Money, J., and Schwartz, M. (1976). Fetal androgens in the early treated adrenogenital syndrome of 46 XX hermaphroditism: Influence on assertive and aggressive types of behavior. Aggress. Behav. 2(1): 19–30.
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
9:13
Style file version Nov. 19th, 1999
Sexual Orientation of Female-to-Male Transsexuals
277
Money, J., Schwartz, M., and Lewis, V. G. (1984). Adult erotosexual status and fetal hormone masculinization and demasculinization: 46, XX congenital virilizing adrenal hyperplasia and 46, XY androgen-insensitivity syndrome compared. Psychoneuroendocrinology 9: 405–414. Newton, E., and Walton, S. (1984). The misunderstanding: Toward a more precise sexual vocabulary. In Vance, C. (ed.), Pleasure and Danger: Exploring Female Sexuality, Routledge and Kegan Paul, Boston, pp. 242–250. Pillard, R. C. (1991). Masculinity and femininity in homosexuality: “Inversion” revisited. In Gonsiorek, J. D. W. J. C. (ed.), Homosexuality: Research Implications for Public Policy, Sage, Newbury Park, CA, pp. 32–43. Quinsey, V. L., and Lalumiere, M. L. (1995). Evolutionary perspectives on sexual offending. Sex. Abuse J. Res. Treat. 7(4): 301–315. Roback, H. B., and Lothstein, L. M. (1986). The female mid-life sex change applicant: A comparison with younger female transsexuals and older male sex change applicants. Arch. Sex. Behav. 15(5): 401–415. Singh, D., Vidaurri, M., Zambarano, R. J., and Dabbs, J. M. Jr. (1999). Lesbian erotic role identification: Behavioral, morphological, and hormonal correlates. Journal of Personality and Social Psychology. 76: 1035–1049. Steiner, B. W., and Bernstein, S. M. (1981). Female-to-male transsexuals and their partners. Can. J. Psychiatry 26: 178–182. Symons, D. (1979). The Evolution of Human Sexuality, Oxford University Press, New York.
APPENDIX Passive Sexual Role Scale 1. I am more sexually aggressive than my sex partners. 2. I consider myself a “top.” 3. My partners have tended to initiate sex or sexual activities more than I have.∗ 4. My sex partners have tended to be “bottoms.” 5. I would enjoy my partner using a dildo on me.∗ Preference for Partner Masculinity Scale The first item is written as it appeared in the questionnaire. The same 7-point scale was used for the rest of the items. Would your ideal partner be 1. Very muscular: Somewhat: Slightly: Neither: Slightly: Somewhat: Very unmuscular.∗ 2. Very hairy to Very unhairy.∗ 3. Very tall to Very short.∗ 4. Very strong to Very weak.∗ 5. Very rugged looking to Very delicate looking.∗ 6. Very feminine looking to Very masculine looking. 7. Very feminine acting to Very masculine acting.
P1: FQJ Archives of Sexual Behavior [asb]
PL110-330
May 3, 2000
278
9:13
Style file version Nov. 19th, 1999
Chivers and Bailey
Sexual Versus Emotional Jealousy—Items Added to Buss et al. (1992) 1. Even if my partner were sexually faithful, I would feel terrible if s/he confided more in another person than me.∗ 2. I could tolerate my partner “straying” sexually with someone else, as long as I remained the most important person in his/her life.∗ ∗
Indicates reverse scoring for these items.
P1: FLF Archives of Sexual Behavior [asb]
PL110-331
May 9, 2000
10:7
Style file version Nov. 19th, 1999
Archives of Sexual Behavior, Vol. 29, No. 3, 2000
Group Psychotherapy with Female-to-Male Transsexuals in Turkey S¸ahika Yuksel, ¨ M.D.,1,3 I¸sin Baral Kulaksizo˘glu, M.D.,1 Nuray Turksoy, ¨ M.D.,2 and Do˘gan S¸ahin, M.D.1
The change in Turkish law to allow local sex reassignment surgery was passed in May 1988. By law, a candidate for such surgery must obtain a medical certificate attesting that the operation is necessary. However, the law does not specify conditions for granting such a certificate, so any physician can give a certificate based on his own criteria. Sex reassignment surgery can therefore be performed without preoperative psychiatric evaluation and preparation. This is a report of 40 female-to-male transsexuals. They had completed psychiatric assessment and participated in group therapy. These meetings provided a valuable setting for getting to know transsexuals and their families. Participants’ long, regular attendance and low dropout rate demonstrate high group cohesion. The aim of the study is to report characteristics of a group of transsexuals living in a different cultural setting from other studies of this patient population in order to identify problems of Turkish transsexuals and advocate changes in current Turkish laws for sex reassignment. KEY WORDS: transsexuals; sex reassignment; group therapy; law.
INTRODUCTION The change in Turkish laws to allow local sex reassignment surgery was passed in May 1988 (Resmi Gazete). The National Health Service does not reimburse costs for this operation and sex reassignment surgeries are performed based on patients’ request. Transsexuals who have undergone surgery achieve the right to change their birth certificate and get married. According to the Turkish laws, 1 Department
of Psychiatry, I.U. Istanbul Medical School. specialist. 3 To whom correspondence should be addressed at Istanbul Medical School, Psychiatry Dept. Topkapi, Istanbul Turkey. Fax: (90) (212) 631 2400. e-mail:
[email protected]. 2 Psychiatry
279 C 2000 Plenum Publishing Corporation 0004-0002/00/0600-0279$18.00/0 °
P1: FLF Archives of Sexual Behavior [asb]
280
PL110-331
May 9, 2000
10:7
Style file version Nov. 19th, 1999
Yuksel, ¨ Kulaksizo˘glu, Turksoy, ¨ and S¸ahin
a candidate for sex reassignment surgery must obtain medical clearance attesting that the operation is necessary. The law does not specify the requisites for granting ¨ this request (Will and Oztan, 1994). Any physician therefore can provide medical clearance based on entirely subjective criteria. It is well attested that diagnostic procedures for transsexualism are timeconsuming and require many counseling sessions. The procedure is even more complex in cases of female-to-male sex reassignment surgery, since recognition of the existence of female transsexuals in Turkey is of recent origin. The required surgery calls for a high level of expertise not possessed by surgeons in general. This can cause more psychiatric referrals to be made for femaleto-male transsexuals. In contrast, male patients may more readily undergo sex reassignment surgery on demand without sufficient psychological and psychiatric evaluation. Since this is a procedure paid for by the patients, many surgeons prefer to operate on males in Turkey, while female-to-male transsexuals travel abroad for the surgery, i.e., England (Walker et al., 1985; Y¨uksel et al., 1994). In 1987, a special unit with an associated clinic was founded at the Psychiatry Department of Istanbul School of Medicine to evaluate gender identity problems. The unit offers pre- and postoperative counseling to transsexuals about sex reassignment surgery. The unit also provides counseling to patients’ family members (Y¨uksel et al., 1992). This study reports the outcomes of psychological assessment and other evaluation procedures of female-to-male transsexuals admitted to this specialized clinic. SUBJECTS AND METHOD Diagnosis and Evaluation The group included biological females older than 16 years of age who had requested sex reassignment surgery. They had a preliminary diagnosis of transsexualism according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders, 3rd ed., revised (DSM-III-R) (American Psychiatric Association, 1987) and were neither psychotic nor mentally retarded. Participants agreed that they would not undergo hormone therapy or surgery before obtaining permission of their assigned therapists. Endocrinological, genetic, and gynecological assessments were carried out by specialists working in coordination with the clinic. Sexual and gender identity history starting from childhood were obtained through a semistructured interview. Patients’ professional, social, and family relationships were evaluated pre- and posttherapy. Special attention was given to sexual experiences. Family members and partners were interviewed with the patients’ permission.
P1: FLF Archives of Sexual Behavior [asb]
PL110-331
May 9, 2000
10:7
Style file version Nov. 19th, 1999
Group Psychotherapy with Transsexuals in Turkey
281
Therapy Members were assigned to group sessions after 1 to 4 months of individual evaluation. Clients with other psychopathologies were treated individually before being invited to join the group. For instance, an agoraphobic patient had behavioral therapy for 2 months and joined the group after improvement of this phobia. All patients attended group therapy for a minimum of 2 years. Individual patients were reevaluated once every year. Additional individual therapy sessions were provided on per need in response to crises situations. Group sessions were held monthly, facilitated by two therapists who also received supervision from a more experienced therapist. Groups consisted of 8–14 members and meetings were carried out as open groups (Yalom, 1985). At the beginning of each session, members briefly recounted important positive and negative events of the prior month. After sharing their problems, the therapists and group members determined the agenda together. Endocrinologists and surgeons were invited to provide information on relevant medical issues. This allowed group members to receive direct answers to their problems, which were then discussed among themselves. Changes were assessed by self-ratings as well as therapists’ observations.
RESULTS Sociodemographic Characteristics The group included 40 members, all of whom were unmarried except for 2 widows. Age at referral ranged from 16 to 38 years, with a mean of 25.3 years (SD ± 5.2 years). The majority (85%) joined the group at an age of less than 30 years, and more than half (60%) before 25 years. Only a few were illiterate and approximately 25% had only primary school education. One-third were highschool or university graduates. Half were brought up in metropolitan areas, some in shanty towns. Half had stable professional lives, while one-quarter frequently changed jobs. One-quarter were self-employed (Table I).
Family Characteristics More than half of the patients (25/40) were brought up by parents until the age of 16. One-fourth were raised without a father. The mean number of sisters (2.2; SD, 1.89) was greater than the mean number of brothers (1.47; SD, 1.98), but the difference was not statistically significant. Of the 40 cases, 27 reported having a close relationship with female family members, i.e., mother (45%) or sister (22%).
P1: FLF Archives of Sexual Behavior [asb]
PL110-331
May 9, 2000
282
10:7
Style file version Nov. 19th, 1999
Yuksel, ¨ Kulaksizo˘glu, Turksoy, ¨ and S¸ahin Table I. Sociodemographic Characteristics
Age at admission 16–20 21–25 26–30 >31 Marital status Divorced Single Education None Primary school High school University Background Metropolitan Rural Place of settlement City Village/small town Profession Worker Civil officer Small business Business owner Not working Regularity of work Regular Part-time jobs Frequently changing jobs Not working
No. of people
%
11 13 10 6
27.5 32.5 25 15
2 38
5 95
4 15 15 6
10 37.5 37.5 15
29 11
72.5 27.5
37 3
92.5 7.5
21 1 10 5 3
52.5 2.5 25 12.5 7.5
21 5 11 3
52.5 12.5 27.5 7.5
More than half were still living with their families and only 7.5% were living with a partner. One-third were raised in traditionally structured families, while only a small proportion was raised either in a liberal environment or in a rigid, deeply religious family (Tables II and III).
Cross-Gender Role A great majority of patients (80%) had realized that they were different from their same-sex peers before puberty. Half preferred not to wear girls’ clothes before adolescence and fantasized about becoming a boy. At first interviews, most presented looking like males and more than half (N = 22) wore typical men’s outfits. The rest wore unisex clothes. None wore makeup and all had short hair. Although aware of being “different” from others since early childhood, their families were reluctant to recognize their predicament and were typically in denial.
P1: FLF Archives of Sexual Behavior [asb]
PL110-331
May 9, 2000
10:7
Style file version Nov. 19th, 1999
Group Psychotherapy with Transsexuals in Turkey
283
Table II. Family Characteristics No. of people
%
26 1 8 5
65 2.5 20 12.5
18 3 9 2 4
45 7.5 22.5 5 10
11 4 10 5 6 3 1
27.5 10 25 12.5 15 7.5 2.5
10 11 11 6 2
25 27.5 27.5 15 5
Family integrity Intact Single parent (father) Single parent (mother) No parent Closest (most intimate) person Mother Father Sister or female relative Brother or male relative Siblings of the same sex (no) 0 1 2 3 4 5 8 Siblings of the opposite sex (no) 0 1 2 3 4
More than half first disclosed their gender identity differences to close family members but even after disclosure some families continued to deny the condition. Other families accepted the situation with relief after a detailed explanation. Some family members found it comforting to talk to their children about these matters. Some reacted with strong rejection and tried to impose an explicit female identity on their child (Tables III and IV).
Sexual History Mean menstruation age was 13.8 ± 1.87 years. A quarter experienced hormonal and/or menstrual irregularities. Half reported having masturbated with the imagery of themselves as men. The rest stated they could not masturbate without a penis. The two widows, who had a history of sexual intercourse with men, stated that these experiences were unsatisfactory. Female transsexuals considered their sexual identity to be male and objected strongly to being referred to as “lesbians.” The average duration of stable relationships with an intimate partner was 3.9 ± 1.2 years, and 27.5% had emotional relations without genital intercourse. Seventy percent reported sexual experiences of some sort which were preferred to
P1: FLF Archives of Sexual Behavior [asb]
284
PL110-331
May 9, 2000
10:7
Style file version Nov. 19th, 1999
Yuksel, ¨ Kulaksizo˘glu, Turksoy, ¨ and S¸ahin Table III. Family Characteristics as Regards Sexual Identity
Family reaction to sexual preferencea Acceptance as if it is a trivial thing Oppressive Denial A strongly disapproving person in the family A very understanding person in the family Characteristics of the milieu in which the person was reared Very traditional and religious Not very oppressive despite practicing beliefs Liberal Person’s own attitude toward religion Very traditional and religious Practicing believer but not too strong beliefs Liberal Not answered Family attitude toward sexuality Traditional and oppressive Not too traditional Liberal Age at which family became aware of the person’s sexual preference Before elementary school During elementary school or before puberty After puberty a Some
No. of people
%
11 20 5 14 13
17.5 50 12.5 35 32.5
13 22 5
32.5 55 12.5
4 20 6 10
10 50 15 25
14 21 5
35 52.5 12.5
15 4 21
37.5 10 52.5
of the participants have chosen more than one option for this category.
Table IV. Recognition of Sexual Identity
Age at which sexual choice first became apparent Before elementary school During elementary school (between age 7 and age 11) Adolescence Age at which a change in dressing style first occurred ≤12 12–15 16–20 ≥20 Age at which a desire to be of opposite sex first felt Before puberty (≤12 years of age) Adolescence ≥16 years of age
No. of people
%
23 8 9
57.5 20 22.5
18 13 5 2
45 32.5 12.5 5
24 12 4
60 30 10
take place in darkness, not to undress completely, and strictly avoided sex during menstrual bleeding periods. They did not let their partners touch their genitals or breasts. Menstrual periods were considered a great burden for those living with a partner (Table V).
P1: FLF Archives of Sexual Behavior [asb]
PL110-331
May 9, 2000
10:7
Style file version Nov. 19th, 1999
Group Psychotherapy with Transsexuals in Turkey
285
Table V. Sexual Attitudes and Characteristics No. of people Reaction to developing breasts Bandaging the breasts Wearing loose clothes, adopting an arched gait, etc. Bandaging the breasts and wearing loose clothes, etc. Hormonal delaying Absent Present Relations to the opposite sex None Temporary relations under pressure from the family Temporary, voluntary relations to give it a try Relations to the same sex None Emotional relations only Emotional and physical contact Reaction to menstruationa Feeling unbearable and disgusting Would rather not have it Adopting a name of the opposite sex Using her own name Name suitable for both sexes Using a male name Outward appearance Looking like a male Wearing unisex clothes Masturbation Not masturbating Fantasies containing members of the same sex Not answered a The
%
6 10 24
15 25 60
32 8
80 20
32 6 2
80 15 5
1 11 28
2.5 27.57 70
35 5
87.5 12.5
2 2 36
5 5 90
23 17
57.5 42.5
16 19 5
40 47.5 12.5
mean age of menarche in our group was 13.8 years (SD, ±1.9 years).
Table VI. Additional Psychiatric Diagnoses
None Anxiety disorders Depressive disorder Mental retardation Psychotic disorders Personality disorders
No. of people
%
25 3 7 3 0 2
62.5 7.5 17.5 7.5 0 5
Psychological Status The participants’ intellectual capacities (IQ) were normal. Half had a psychiatric diagnosis according to the DSM-III-R criteria, but none was psychotic. Depressive disorders were the most frequent comorbidity (Table VI).
P1: FLF Archives of Sexual Behavior [asb]
286
PL110-331
May 9, 2000
10:7
Style file version Nov. 19th, 1999
Yuksel, ¨ Kulaksizo˘glu, Turksoy, ¨ and S¸ahin
Treatment Process After preliminary evaluation, 25 patients were invited to participate in group therapy without prior individual therapy. Five had a course of independent individual therapy or were prescribed antidepressant medications. Two patients with personality disorders, two female homosexuals, and one with learning difficulty did not progress with the group therapy. Four dropped out in the course of group treatment. Average attendance for group work was nearly 3 years (33 ± 23.3 months). Group Themes At the commencement of the group experience, the members’ only expectation was to present themselves as true transsexuals and at an appropriate time secure medical approval to proceed with surgery. During the sessions some common themes emerged with time, including relationship difficulties with their families, work, and partners. The group also provided an opportunity for participants to meet other transsexuals. They had typically tried to prove themselves to be male. They found that they could share the burdens and hardships of being a transsexual in a friendly, supportive environment. The group quickly established itself as a self-help resource. Decision About Surgery Twenty-two of the participants received medical clearance for reassignment surgery after attending group therapy (mean time, 28 ± 4 months). The certificate obtained for the patients is that of a transsexual for whom sex reassignment surgery can be provided. For those not certified, evaluation and counseling continues. At the end of 1 year of group therapy, two identified as female homosexuals rather than transsexuals; they had not recognized at the start of the therapy and both withdrew their applications for sex reassignment surgery. Of the 22, 12 had surgery without major complications and applied to the court to have their birth certificate changed . Five were later married and one couple adopted a child. No obvious psychological problems have been encountered in the operated transsexuals, either during the surgery or in their the later life. DISCUSSION This study provides information about female-to-male transsexuals living in Turkey, a country that is characterized by marked influences of both Western and Islamic cultures. Discussion of sexuality and gender identity problems is
P1: FLF Archives of Sexual Behavior [asb]
PL110-331
May 9, 2000
Group Psychotherapy with Transsexuals in Turkey
10:7
Style file version Nov. 19th, 1999
287
of recent origin and has led to the recognition of a group of people whose existence has been denied for a long time. There is a shortage of clinicians with experience in assessment and advising on appropriate management plans for the individuals concerned. Although this study summarized results of the largest series in Turkey; the database has limitations. Since male-to-female transsexuals rarely apply to our unit, a male-to-female transsexual group could not be formed or studied. The reason for this may be that surgeons do not request a psychiatric evaluation of their patients before operating on male-to-female transsexuals. In consequence, males are not required to spend a long time waiting for the psychiatric decision-making process. A similar report has been published from Poland (Godlewski, 1988) which accounts a ratio of 5.5 females to 1 male transsexual admitted to psychiatric clinics. In contrast to that report, surveys reported from the other countries (Van Kesteren et al., 1996) are contradictory in sex ratio, which may be the result of legal or social differences. The study design does not include a control group and the number of patients included in this study is small. Growing up in a traditional, conservative cultural background did not prevent our patients from acting in accordance with their perceived gender identity even though experiencing many difficulties, for example, in religious practices. Islam demands different prayer practices for men and women; our patients who held deep religious beliefs faced difficult conflicts. For example, one of the basic forms of praying in Islam, “namaz,” involves different body movements in different compartments of mosques for each sex. All women must cover their heads in their daily life and there is some strict discrimination in life styles of women. Accordingly patients had to make a decision totally to give up religious practices or adopt male gender-assigned forms of religious behaviors. Homosexuality and transsexualism are strictly unacceptable to Islamic laws. The majority of femaleto-male transsexuals in our group wanted to be considered male during their daily religious life and, in time, at their own funeral. Feelings and presentations as males were clear and frank since childhood. Their self images had been male and this was reflected in their appearance. Verchoor and Poortinga (1988) reported female-to-male transsexuals to have better parental relationships than male-to-female transsexuals. Female transsexuals in this sample were generally close to at least one female family member, typically their mothers. This finding differs from the Tsoi (1990) report of female-to-male transsexuals having unsatisfactory relationships with mothers. Interpretation of these findings could be an account of different child raising practices in different countries. In Turkey mothers tend to assume a highly influential traditional role. Mothers carry most of the responsibilities of child rearing and fathers are rarely involved in problems during this period. In this group, there is a low male:female sibling sex ratio (59/86 × 100 = 69). Homosexual males have been found to have the opposite: an excess of brothers
P1: FLF Archives of Sexual Behavior [asb]
288
PL110-331
May 9, 2000
10:7
Style file version Nov. 19th, 1999
Yuksel, ¨ Kulaksizo˘glu, Turksoy, ¨ and S¸ahin
(Blanchard et al., 1995). This finding may indicate a sibling dominance exerted by the same sex or may be a cultural effect specific to Turkey. Having a son is deemed important for Turkish families to inherit the family name and traditions. Some families choose to continue having more children until a boy is born. In the studied group there was a member who had eight sisters. According to traditional Turkish norms a sexual relationship before marriage is strictly forbidden for girls but not for boys. Some of our patients reported that their families were “allowing” their transsexual girls to have sexual relationships and some arranged unofficial, but nevertheless religious, wedding ceremonies for their “female-son.” The average age of menarche for this group was slightly above the national norm reported for Turkey (T¨umerdem et al., 1984). Menstrual irregularities and ovarian pathologies seemed frequent among the group. This has also been described by Futterweit et al. (1986) and Heresova et al. (1986). As emphasized by some authors, transsexuals are prone to depression. Althof and Keller (1980) stated that during group sessions participants realized that they are not unique and alone and that they can support each other to overcome their hopelessness. They use the strategies emphasized by other group members to solve problems. There were no suicide attempts during group therapy. During therapy, the rate of oversensitivity, aggression, and withdrawal reactions reported prior to joining the group decreased. Unstable behavior and lack of assertiveness reduced during the course of group therapy and tolerance for frustration increased as they developed realistic expectations for their own future. Long, regular attendance, with a low dropout rate of 10%, compared favorably with a 30–40% dropout ratio reported in other therapy groups (Yalom, 1985). This demonstrates the high value attached to group relations, group cohesion, and solidarity. This finding indicates that individuals who are stigmatized and excluded by society in general have a need to share their experiences and problems with others. Meeting in a place where they were not considered “rotten,” “perverted,” or “harmful to society” allowed group members to speak of their experiences in the company of other transsexuals for the first time. Many started to make distinctions based on the criteria of “before the group” and “after the group,” thus revealing its importance to them. Patients showed repeated, stable, and consistent attitudes toward their bodies from an early age and avoidance and disgust felt toward their bodies during masturbation and other sexual activities. Many had scars caused by wearing a tight flattening girdle around their breasts so as not to accentuate this sex characteristic (Blanchard, 1990; Walter and Ross, 1986). However, in spite of having had long-term relationships, some could not tell their partners of their situation. This was especially so for those who preferred having nonsexual emotional affairs and avoided physical contact. A frequently stated excuse presented to their partners was having urogenital anomalies. Most group members were hoping to get
P1: FLF Archives of Sexual Behavior [asb]
PL110-331
May 9, 2000
10:7
Style file version Nov. 19th, 1999
Group Psychotherapy with Transsexuals in Turkey
289
married after the operation. This might be explained by female transsexuals being monogamous with tendencies toward nonpromiscuous relations. When working with female transsexuals, family issues should be given prominent consideration. In the Turkish culture, family approval is expected when someone is effecting a major lifestyle change. For instance, before an operation, a recognition of the facts of the situation by family members is extremely important. Sometimes subgroups visited families to demonstrate that transsexuals belong to nondeviant families. The family meetings proved to be of benefit both to transsexuals and their families. Interfamilial support systems have also been formed. During the group therapy period, many families changed their attitudes, behaviors, and knowledge about transexualism. Their denial usually ended as well. An important and recurring topic discussed by the group was “the realization of the limits of surgery.” When they talked about gender reassignment surgery, they spoke of their breasts as “a tumor” to be removed from their bodies. Some seemed to believe as if their female past would not exist after the operation. After appropriate education, patients denied unrealistic expectations. Furthermore, seven members introduced other transsexuals to the group who in time became regular attendees.
CONCLUSION This report summarizes what has been learned from transsexual group therapy in Turkey. Working with them in group therapy created a valuable opportunity for the therapists to get to know each of them and their families. The participants come from “ordinary,” “nondeviant” families. They have regular jobs, friends, partners, and family lives. In a society where sexual taboos are strong and prominent, where differences in sexual orientation are not readily accepted, and sexual education is not given properly, group therapy is a valuable resource beyond that of individual therapy. Problems centered on gender identity differences are not only medical and psychological but also sociopolitical. Recognition of the structure of the social setting in which female transsexuals live is essential for understanding their behaviors and reactions. Based on this, an ethical committee within the Turkish Medical Association and the Forensic Psychiatry Department will be organized to determine the rules for sex reassignment surgery.
ACKNOWLEDGMENTS Special thanks go to Dr. Roderick Orner and Mrs. Aliza Marcus for their contributions on the linguistic check and kind remarks for this article.
P1: FLF Archives of Sexual Behavior [asb]
290
PL110-331
May 9, 2000
10:7
Style file version Nov. 19th, 1999
Yuksel, ¨ Kulaksizo˘glu, Turksoy, ¨ and S¸ahin
REFERENCES Althof, S., and Keller, A. (1980). A group therapy with gender identity patients. Int. J. Group Psychother. 30: 481–489. American Psychiatric Association (1987). Diagnostic and Statistic Manual of Mental Disorders, 3rd ed. rev., APA, Washington, DC. Blanchard, R. (1990). Gender identity disorders in adult women. In Blanchard, R., and Steiner, B. W. (eds.), Clinical Management of Gender Identity Disorders in Children and Adults, American Psychiatric Press, Washington, DC, pp. 77–91. Blanchard, R., and Sheridan, P. M. (1992). Sibship size, sibling sex ratio, birth order and parental age in homosexual and nonhomosexual gender dysphorics. J. Nerv. Ment. Dis. 180: 40–47. Blanchard, R., Steiner, B. W., and Clemmenson, L. H. (1985). Gender dysphoria, gender reorientation, and the clinical management of transsexualism. J. Consult. Clin. Psychol. 53: 295–304. Blanchard, R., Clemmenson, L. H., and Steiner, B. W. (1987). Heterosexual and homosexual gender dysphoria. Arch. Sex. Behav. 16: 139–152. Blanchard, R., Zucker, K. J., Bradley, S. J., and Hume, C. (1995). Birth order and sibling sex ratio in homosexual male adolescents and probably prehomosexual feminine boys. Dev. Psychol. 31: 22–30. Dixen, J. M., Maddever, H., Van Maasdam, J., and Edwards, P. W. (1984). Psychosocial characteristics of applicants evaluated for surgical reassignment. Arch. Sex. Behav. 13: 269–276. Futterweit, W., Weiss, R. A., and Fagerstram, R. M. (1986). Endocrine evaluation of forty female to male. Increased frequency of polycystic ovarian disease in female transsexual women. Exp. Clin. Endocrinol. 88: 219–223. Godlewski, L.,van Kesteeren, P., and Megens, J. (1988). Transsexualism and anatomic sex reversal in Poland. Arch. Sex. Behav. 17: 547–548. Heresova, J., Pobisova, Z., Hampl, R., and Starka, L. (1986). Androgen administration to transsexual women. Exp. Clin. Endocrinol. 88: 219–223. Resmi Gazete (1988). T¨urkish Civil Law number 743, article 29, May 12. Tsoi, W. F. (1990). Parental influence in Singapore. Med. J. 31(5): 443–446. T¨umerdem, Y., Co¸skun, A., and Ayhan, B. (1984). Menarch Phenomenon, XII T¨urk Pediatri Kongre Kitabi, Istanbul. Van Kesteren, P. J., Gooren, L. J., and Megens, J. A. (1996). An epidemiological and demographic study of transsexuals in the Nederlands. Arch. Sex. Behav. 25(6): 589–600. Verschoor, A. M., and Poortinga, J. (1988). Psychosocial differences between Dutch male and female transsexuals. Arch. Sex. Behav. 17(2): 173–178. Walker, P. A., Berger, J. C., and Green, R. (1985). Standarts of care: The hormonal and surgical sex reassignment of gender dysphoric persons. Arch. Sex. Behav. 14: 79–90. Walters, W., and Ross, M. (1986). Transsexualism and Sex Reassignment, Oxford University Press, London. ¨ Will, M. R., and Oztan, B. (1994). Hukukun Sebebiyet Verdigi Bir Aci-Transseks¨uellerin Hukuki Durumu, Journal of Law School, University of Ankara, Ankara, pp. 227–268. Yalom, I. (1985). The Theory and Practice of Group Psychotherapy, 3rd ed., Basic Books, New York. Y¨uksel, S. (1994). Indications of sex reassignment surgery. Presented at the National Forensic Medicine Congress, Istanbul, Nov. 1–4. Y¨uksel, S., Y¨ucel, B., T¨ukel, R., and Motavalli, N. (1992). Assessment of 21 transsexual cases in group psychotherapy, admitted to hospital. Nordisk Sex. 10: 227–235.
P1: FOM/FLK
P2: FQJ
Archives of Sexual Behavior [asb]
PL110-332
March 13, 2000
12:17
Style file version Nov. 19th, 1999
Archives of Sexual Behavior, Vol. 29, No. 3, 2000
BOOK REVIEWS Against My Better Judgment: An Intimate Memoir of an Eminent Gay Psychologist. By Roger Brown. Haworth Press, Binghamton, New York, 1996, 253 pp., $36.00 (hardback), $22.00 (paperback). Reviewed by Edward Stein, Ph.D.1
Brown was an eminent psychologist known for his work on how children develop language (e.g., Brown, 1958) and for his widely read textbook on social psychology, with its noteworthy chapters discussing “issues in sexual liberation” (Brown, 1986). Brown’s teaching and work were foundational to the development of cognitive science. Against My Better Judgment, written after Brown’s retirement from Harvard University and published just a year before his death in 1997, is a very different book from anything he had written before. There are no general theories of human thought or nature offered here. Rather, the book contains a thoughtful and moving autobiography focusing on Brown’s sex life and psychosexual wandering after the death of his lover of some 40 years. After several months of mourning, Brown, aged 65, began paying young men to have sex with him. Against My Better Judgment is primarily the story of how Brown “fell in love” with three such young men and how he developed romantic relationships with them. Many gay men and some lesbians have written autobiographies. Some of them are surely of interest to many lesbians, gay men, bisexual, and transgendered people who are looking to read the stories of other sexual minorities, stories that mainstream culture, in various ways, hides. A much smaller number of these queer autobiographies are of interest to sex researchers. Surveys can tell us something about a person’s sexual desires, but even lengthy questionnaires are going to simplify and cubbyhole the complexities of individuals, their experiences, and their characters. An autobiography allows a person to tell her own story. Although a person is likely to be able to provide an accurate report of her life (she was there for all of it, after all), she may not be the most reliable reporter of her sexual motivations and the source of her sexual desires. 1 Departments
of Philosophy and Law, Yale University, P.O. Box 208306, New Haven, Connecticut
06520-8306. 291 C 2000 Plenum Publishing Corporation 0004-0002/00/0600-0291$18.00/0 °
P1: FOM/FLK
P2: FQJ
Archives of Sexual Behavior [asb]
PL110-332
March 13, 2000
12:17
Style file version Nov. 19th, 1999
292
Book Reviews
Brown describes his feelings and desires for these young men and his anxieties about paying for their attention. I was not always convinced that Brown understood himself or the young men with whom he had sex. But I learned a lot from reading his accounts of his experiences and emotions. There is much in these pages that spoke to me as a gay man, even though I am about 40 years younger than Brown would have been were he still alive and my sexual interests are significantly different from what his were. More importantly, there was much in these pages that spoke to me as a theorist of human sexual desire. The sexual desires and activities of people over 50 and of lesbians and gay men who survive their long-term companions have not been the subject of much scientific or psychological research. Although Brown was not attempting to conduct such research on himself, as Against My Better Judgment suggests, though not directly, he provided some indications about the ways that we might begin to think about these interesting aspects of human sexuality. Along the way, it makes for fun, interesting, and engaging reading.
REFERENCES Brown, R. (1958). Words and Things, Free Press, New York. Brown, R. (1986). Social Psychology: The Second Edition, Free Press, New York.
Fetish: Fashion, Sex & Power. By Valerie Steele. Oxford University Press, New York, 1995, 243 pp., $35.00. Reviewed by Albert Wong, M.D.2
Steele is a cultural historian and this book is an analysis of the relation between fetish clothing items and their sexual meanings both to the wearer and in the context of the fetish subculture. Steele’s focus is on the history of fetish fashions rather than the nature of fetishes themselves, or on fetishism as a broader cultural discourse. Well researched and documented, this book contains hundreds of references to the psychological and historical literature as well as information from pornography and interviews with people involved in sadomasochism, cross-dressing, and fetishism. In a clear and intelligent style, Steele presents ideas of interest to scholars of history, anthropology, fashion, and psychology. The book begins with a brief discussion of fetishism and reviews the salient psychiatric, psychoanalytic, sociological, and anthropological perspectives. Steele 2 Centre for Addiction and Mental Health—Clarke Division, 250 College Street, Toronto, Ontario M5T
1R8, Canada.
P1: FOM/FLK
P2: FQJ
Archives of Sexual Behavior [asb]
Book Reviews
PL110-332
March 13, 2000
12:17
Style file version Nov. 19th, 1999
293
addresses some controversial topics such as why most paraphiliacs are men, the “essentialism/constructivism debate” on the nature of sexuality, evolutionary theories of sexuality, and the psychiatric debate on sexually normative behavior, fetishism, and psychopathology. Steele draws from a wide range of theorists including Freud, Foucault, Stoller, Kunzle, and Krafft-Ebbing. The analysis is cursory at times but summarizing this area in one chapter necessitates a brief overview. Chapter 2, entitled “Fashion and Fetishism,” considers the emergence of fetish clothing, mainstream fashion, and popular culture. Steele explains this as a combination of several factors: the sexual liberation of the 1960s and 1970s, the anarchistic punk movement, the “undercurrent of sex and violence” in subculture styles, the influence of fashion photographers and stylists such as Helmut Newton, and the “sex appeal of the commodity.” She refutes the notion that our era is different in its overt sexuality, and convincingly argues that the asexual stereotype of the Victorian age is inaccurate. The conflicting feminist viewpoints on fashion imagery as degrading and objectifying women versus a liberating reclamation of sexually powerful roles are presented. The chapter ends with an insightful discussion of Marxist and neo-Marxist theories on fetish objects as commodities and fashion in general as “capitalism’s favorite child.” The rest of the book is divided into chapters that concentrate on specific types of fetish clothing: “The Corset,” “Shoes,” “Underwear” and “Second Skin,” as well as related topics such as body piercing and tattooing. These chapters are mostly a descriptive chronology of fetish clothing in relation to mainstream fashions, with some discussion of the symbolism of dominance, control, and submission inherent in the clothes. She uses examples of fetish tastes from pornography, catalogs of fetish clothing suppliers, literature, photography, and fashion and couture shows. Twenty-four pages of well-chosen color photographs supplement these chapters. In the concluding chapter, “Fashion, Fetish, Fantasy,” Steele analyzes the various archetypes that have been used in both fashion and fetish clothing: the dominatrix, the uniform, the biker, leathermen, and the French maid. This book is an excellent overview of the history of fetish clothing from the perspective of fashion culture. Steele is often witty and entertaining, and she manages to integrate a diverse set of discourses: “the postmodern, the politicized, the psychiatric, the popular, and the pornographic,” into a thoughtful and balanced book. She remains objective and nonjudgmental without being distant. Although parts of the book are not directly relevant to sexologists, it provides a fascinating background to the often mysterious world of fetish subculture and clothing.
P1: FOM/FLK
P2: FQJ
Archives of Sexual Behavior [asb]
PL110-332
March 13, 2000
294
12:17
Style file version Nov. 19th, 1999
Book Reviews
Women and New Reproductive Technologies: Medical, Psychosocial, Legal, and Ethical Dilemmas. Edited by Judith Rodin and Aila Collins. Lawrence Erlbaum, Hillsdale, NJ, 1991, 171 pp., $36.00. Reviewed by Norma L. McCoy, Ph.D.3
This book is based on a conference sponsored by the John D. and Catherine T. MacArthur Foundation Network on the Determinants and Consequences of HealthPromoting and Health-Damaging Behaviors. It consists of 10 chapters, 8 chapters by different contributors and an introduction and concluding chapter by the editors. In the Preface, the editors state that this volume focuses on reproductive technologies because it is an area in which the profound ethical, legal, social, and psychological issues surrounding the use of medical technology are easily discerned. Chapter 2 (“The History of the Relationship Between Women’s Health and Technology”) by Schrom Dye focuses almost exclusively on the history of medical practice in childbirth and the role that technology played during the 19th and early 20th century. Although caesarean section and the use of forceps were possible in the 19th century, doctors had respect for the simple ways of nature and technological interventions were relatively rare. After 1880, Schrom Dye argues that this view broke down and was supplanted by one that supported considerable technological intervention even though infection associated with its use remained the major cause of maternal death until the 1930s. Chapter 3 (“Pregnancy-Inducing Technologies: Biological and Medical Implications”) by Thatcher and DeCherney consists mainly of a detailed description of the technologies involved in in vitro fertilization and embryo transfer (IVF/ET). Major topics in laymen’s terms are inducing ovulation, retrieval of eggs, fertilization and growth of the fertilized egg in the laboratory, and transfer of the fertilized egg to the uterus. The authors report statistics from the 1988 United States Registry of IVF/ET, indicating that 16% of such attempts to impregnate were successful and 12% of them resulted in live births. Chapter 4 (“Autonomy, Choice, and the New Reproductive Technologies: The Role of Informed Consent in Prenatal Genetic Diagnosis”), by Faden, begins with a brief history of “informed consent” in American medicine and a discussion of its meaning. This serves as the backdrop for the issue of consent in genetic diagnosis. Currently, this technology is most commonly employed to diagnose disease in fetuses and identify carrier status in adults. Genetic technology increasingly will confront parents with tough decisions about the use of abortion in their pursuit of perfect offspring. Other issues include whether mothers should be compelled to undergo testing and whether maternal consent will be required when the fetal condition tested for is treatable. 3 Department
of Psychology, San Francisco State University, 1600 Holloway Avenue, San Francisco, California 94132.
P1: FOM/FLK
P2: FQJ
Archives of Sexual Behavior [asb]
Book Reviews
PL110-332
March 13, 2000
12:17
Style file version Nov. 19th, 1999
295
In Chapter 5 (“Ethical Issues Raised by the New Medical Technologies”), Whitbeck argues that applied ethics has been shaped by economic ideology that sanctions treating everything as a resource and utilizes cost–benefit analysis, restricting consideration to those consequences that can be quantified. Loss of integrity is difficult to quantify and thus is not considered as a cost. Whitbeck discusses the effects of medical technologies “on human relationships, on character and moral integrity, and on families and communities” in the context of HIV/AIDS, IVF/ET, and contraception. Chapter 6 (“Women’s Reproductive Rights: The Impact of Technology”), by Ruzek, is a well-referenced and excellent discussion of the social and ethical issues surrounding current birth technologies, such as electronic fetal monitoring, Caesarian section, episiotomy, out-of-hospital birth, and prenatal care. Ruzek points out the lack of critical evaluation of adopted surgical technologies and makes the case that even when sufficient evaluation exists, it does not necessarily affect practice. The United States is dominated by fee-for-service medicine where services are rationed on the basis of ability to pay. Ruzek concludes that “[t]he social and economic consequences of supporting unjustifiable medical tinkering and failing to provide a ‘floor of equity’ for birth are . . . enormous and . . . must be changed.” Chapter 7 (“Women and Advances in Medical Technologies: The Legal Issues”), by Clayton, deals with reproductive technologies. Two new forces with legal ramifications have emerged: procreation is increasingly viewed as a medical/health issue in which physicians can intervene and there is the increasing tendency to view the fetus as having interests separate from the pregnant woman. Clayton argues that both have contributed to decreasing women’s freedom of choice. Chapter 8 (“Psychological Issues in New Reproductive Technologies: Pregnancy-Inducing Technology and Diagnostic Screening”), by Adler, Keyes, and Robertson, focuses on the psychological effects of pregnancy-inducing technologies and prenatal diagnostic screening, particularly amniocentesis and chorionic villus sampling. Relevant studies on psychological effects are reviewed but the authors stress their limitations given that women with the most negative experiences rarely cooperate in research. Chapter 9 (“Communicating About the New Reproductive Technologies: Cultural, Interpersonal, and Linguistic Determinants of Understanding”), by Rapp, explores social and cultural aspects of prenatal diagnosis and genetic counseling. This interesting contribution is based on two years of fieldwork in New York City observing and interviewing genetic counselors, observing a cytogenetics laboratory, and interviewing pregnant women and their families. This volume contains a discussion of many thought-provoking issues surrounding the use of reproductive technologies, but like most collections of conference papers, the contributions are not integrated with each other in terms of content to form a coherent whole, and the introduction and concluding chapter do little if anything to remedy the problem. The format of chapters varies and two
P1: FOM/FLK
P2: FQJ
Archives of Sexual Behavior [asb]
PL110-332
March 13, 2000
296
12:17
Style file version Nov. 19th, 1999
Book Reviews
of six contributions have no headings or subheadings. These failings are minor and not unusual. What is not excusable is the failure even to name, much less discuss, the major reason for the rise to power of these reproductive technologies! The fruits of the feminist movement—the escape from early marriage and childbearing in the pursuit of a career—have had their costs. As women age, fertility declines. With increasing age, women have a greater chance of having experienced a sexually transmitted disease or contracted another disease that impairs fertility. Moreover, increased age of the mother is associated with an increase in genetic aberrations in the fetus (e.g., Down’s syndrome) as well as with increased difficulties in childbirth. Clearly, late childbearing has contributed to the development of reproductive technologies and the many medical, psychosocial, legal, and ethical dilemmas they create. The failure to consider this issue is a major weakness of this volume.
Counselling for Fertility Problems. By Jane Read. Sage, London, 1995, 204 pp., $39.95 (hardback), $17.95 (paperback). Reviewed by Jules Black, M.D.4
This is the ninth volume in the Sage “Counselling in Practice” Series. Read is an accredited counselor based in London. Her book is aimed at those who counsel the 1:6 couples who present to an infertility clinic with a perceived difficulty in achieving a successful pregnancy or those who have difficulty deciding whether or not to continue with a pregnancy, normal or otherwise. Since the mid-1980s, the British health authorities have recognized the need for providing this counseling facility for those undergoing fertility treatments. Sufficient time has passed for considerable experience to have been amassed in this area. Four distinct types of counseling are defined: (1) information counseling, (2) implications counseling, (3) support counseling, and (4) therapeutic counseling. No one is obliged to accept counseling, but it is generally recognized to be beneficial. Interestingly, contrary to popular beliefs, there is still no proven association between a couples’ psychodynamics and infertility. Much is anecdotal and not supported by research—for example, that adoption will lead to a subsequent spontaneous conception. It is stressed that the counselor must know about the various procedures involved in infertility treatment. It should not be overlooked that abortion counseling also falls within the ambit of fertility problems and is discussed at length. There are some 14 text boxes throughout the book containing useful questions from which the counselor can select to consider asking the client. The book is 42
Rae Street, Randwick, NSW 2031, Australia.
P1: FOM/FLK
P2: FQJ
Archives of Sexual Behavior [asb]
Book Reviews
PL110-332
March 13, 2000
12:17
Style file version Nov. 19th, 1999
297
clearly written in an easily absorbed style. Case examples are given throughout covering a comprehensive range of situations. These include special cases such as the “appropriateness” of treatment of women with HIV positive men or who are HIV positive themselves, infertility and adoption, and genetic counseling. Regarding genetic counseling, this includes management of selective abortion or refers to “micromanipulation” through in vitro fertilization to guarantee genetically intact offspring where defects occurred previously in the couple itself or in previous offspring. Four counseling modalities are explored, the fourth being developed by the author. The first uses the Kubler–Ross death and dying model. The second is the Worden tasks of mourning model, also for grief counseling and grief therapy. The third is Egan’s helping model, followed by Read’s Infertility Counseling Model (ICM). The ICM Model has five phases: (1) diagnosis, (2) managing feelings, (3) planning action, (4) having treatment, and (5) awaiting outcomes. At the end of the chapter is a good comparative table of the four models. The ICM Model is then applied to abortion counseling. Ample strategies are discussed to help the therapist with the client’s decision-making process. The raison d’ˆetre for this review to appear in these pages is a good chapter addressing the issues of fertility and sexual problems. A 1988 study is quoted which found that sexual dysfunction was the primary cause of infertility in 5% of the cases seen. The author also writes about her experiences with cases where the converse pertained, i.e., where infertility problems led to sexual dysfunction. After all, the infertility therapist often creates demands, more correctly, dictates that the couple “do it” at a certain time, on certain dates, and a certain number of times. This creates sometimes enormous performance problems for both, and certainly is likely to remove any modicum of pleasure left in the sexual act for the couple. At the end of the road, there can be a complex mix of issues which will possibly need further attention and resolution postpartum. Finally, the author has sections on couples counseling, gender issues, donation issues, and counseling for the donors themselves. There are useful appendices. Pertinent British acts are included so that the counselor can familiarize himself or herself with what the letter of the law is on issues such as the Human Fertilisation and Embryology Act of 1990 or the Abortion Act of 1967 and its latest amendment revision of 1991. A chapter on resources and organizations in the United Kingdom follows. This is one of the few criticisms I need to make about the book. It is clearly for British conditions, and for this book to “travel” overseas, local publishers would need to insert their equivalent list of local resources and organizations. Separately, conception technologies and assisted conception methods are galloping along so quickly that the glossary given at the start of the book is bound to be out-of-date some five minutes after publication. In summary, this is a useful, blessedly concise book, and if one is starting out in the field of counseling infertility clients, this is an ideal starting point.
P1: FOM/FLK
P2: FQJ
Archives of Sexual Behavior [asb]
PL110-332
March 13, 2000
298
12:17
Style file version Nov. 19th, 1999
Book Reviews
The Janus Report on Sexual Behavior. By Samuel S. Janus and Cynthia L. Janus. John Wiley, New York, 1993, 430 pp., $29.95. Reviewed by Eugene E. Levitt, Ph.D.5,6
This book describes the results of a national survey of sexual behaviors, attitudes, and beliefs of a sample of 1347 men and 1418 women collected during the period 1988 to 1992. This number represents 60.8% of the questionnaires that were distributed. Apparently, the questionnaires were self-administered rather than providing structure for an interview, though this is not clearly stated. In addition, there were 125 in-depth interviews; again, it is not clear whether the data from the subsample are included with the main sample. In general, the methodology of the survey is inadequately described. As is always the case, a survey, especially a survey dealing with a sensitive topic, is evaluated primarily by its methodology rather than its findings. This applies even more clearly to the Janus report because so many of the survey items were designed to tap attitudes and beliefs rather than behaviors. A number of item wordings indicate that insufficient attention was given to this precaution. The questionnaire was composed of 105 items, 1 of which is actually 14 separate questions plus another that is open-ended and might yield another 14 individual items, making a total of 132 items which are presented in 280 tables. To avoid biasing responses, this kind of item must be even more carefully phrased than items dealing directly with sexual behaviors. Some of the stimulus items in the questionnaire contain salient words that require specialized knowledge, such as sexual surrogate, necrophilia, and “brown showers,” as well as terms that require interpretation, such as sexual molestation, traditional sex roles, and bisexual. A number of the target questions in the survey not only are suggestive, but are vulnerable to subjective interpretation, for example, “traditional sex roles have no place in modern society,” “it is better to love and be hurt than not to know love,” “abortion is murder.” A respondent might reasonably be puzzled as to how to respond on a four-step percentage scale ranging from 10 to 100 to the item, “How much below maximum sexual potential are you?” This fanciful, not easily comprehended, possible biasing language pervades the items in the Janus survey. Of course, a major avenue for assessing the value of a survey is its sampling. Like a number of its predecessors, the Janus sample is overweighted with bettereducated and more affluent respondents who are too often single. Compared to U.S. population norms, the Janus sample contained only a fifth of the required respondents with less than a high-school education, 12% more respondents with some college contact, 17% fewer low-income individuals, and a 20% shortage of Protestant respondents. In addition to these very possibly biasing sample characteristics, 5 Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana 46202-5200. 6 Deceased.
P1: FOM/FLK
P2: FQJ
Archives of Sexual Behavior [asb]
Book Reviews
PL110-332
March 13, 2000
12:17
Style file version Nov. 19th, 1999
299
the ethnic composition of the sample is unspecified. The authors state that the sample “includes respondents from most sizeable minority groups” but they intentionally ignored this aspect of the sample in presenting results. The reason was “to include them unidentified in our general sample and thereby reflect as much as possible heterogeneity of the American public” (p. 402), a most dubious motivation. Despite its methodological deficiencies, this report will probably be quoted from time to time like other methodologically weak surveys, including the various magazine polls such as Redbook, McCall’s, Psychology Today, etc. Janus is even more likely to be quoted because of the wide range of subject matter, which includes 10 forms of deviant practices, religion, politics, “money, power and sex,” and singles. Some of the findings are startling. More than 25% of the women in the sample report that they have had an abortion. Almost half of the married sample had lived together before marriage. Six percent of the men and 4% of the women reported personal experience with “golden showers,” an enormous number to be engaged in this esoteric practice assuming that the survey respondents understood that the item was not a reference to meteorological conditions. In the absence of contradictory data, it is not possible to infer with certainty but it appears at least likely that such remarkable results are a function of sampling or item wording shortcomings. The surveyors find that 23% of the women subjects and 11% of the males believe that they were sexually molested in childhood. The key term is not defined in any way. In a sizable minority of the cases, the molestation is allegedly “ongoing,” a peculiar finding in a group with a minimum age of 18 years. Another sizable minority reports that they were molested only once. If this subgroup is removed from the molested group on the grounds that it is a great deal easier to be mistaken about one incident than about many, the proportions will be 18% for women and 9% for men, a bit closer to expectation. Similarly, the authors analyzed their data to come to the conclusion that 9% of the men and 5% of the women are homosexuals. However, only 4% of the men and 2% of the women identify themselves as homosexuals. If the estimate is based on those who say that they are having homosexual experiences frequently or “ongoing,” the estimates are even lower, less than 3% for the men and less than 2% for the women. In summary, this survey shares the serious methodological shortcomings of most national surveys of human sexuality. Its findings may be considered to be grossly accurate within a considerable probable error. When frequencies are low, the error should preclude a definitive conclusion by the cautious reader. On the one hand, one might predict that the results of this survey will be quoted in future human sexuality texts because of the range of activities it examined. A sizable segment of its items will not be found in any other national survey. On the other hand, it is more likely that the Janus and Janus sex survey will take a back seat to the recent flurry of epidemiologically more rigorous sex surveys that appeared on the sexological scene shortly after the publication of this volume (see, e.g., Bozon
P1: FOM/FLK
P2: FQJ
Archives of Sexual Behavior [asb]
PL110-332
March 13, 2000
12:17
300
Style file version Nov. 19th, 1999
Book Reviews
and Leridon, 1996; Laumann et al., 1994; Wellings et al., 1994; cf. Schmidt, 1997; Wiederman, 1997). REFERENCES Bozon, M., and Leridon, H. (eds.) (1996). Sexuality and the Social Sciences: A French Survey on Sexual Behaviour (trans., G. Rogers), Dartmouth, Aldershot, England. Laumann, E. O., Gagnon, J. H., Michael, R. T., and Michaels, S. (1994). The Social Organization of Sexuality: Sexual Practices in the United States, University of Chicago Press, Chicago. Schmidt, G. (1997). Review of The Social Organization of Sexuality: Sexual Practices in the United States. Arch. Sex. Behav. 26: 327–332. Wellings, K., Field, J., Johnson, A. M., and Wadsworth, J. (1994). Sexual Behavior in Britain: The National Survey of Sexual Attitudes and Lifestyles, Penguin Books, London. Wiederman, M. W. (1997). Review of Sexual Behavior in Britain: The National Survey of Sexual Attitudes and Lifestyles. Arch. Sex. Behav. 26: 332–337.
A Guide to America’s Sex Laws. By Richard A. Posner and Katharine B. Silbaugh. University of Chicago Press, Chicago, 1996, 243 pp., $26.95. Reviewed by Kenneth J. Zucker, Ph.D.7,8
The senior author is chief judge on the United States Court of Appeals, Seventh Circuit, and a Senior Lecturer at the University of Chicago Law School. For those who work at the interface of sexology and the law, Posner is best known for his volume Sex and Reason (Posner, 1992), which received widespread attention and analysis (see, e.g., Reilly, 1996). In this volume, Posner and Silbaugh summarize the sex laws that exist in each of the 50 states and the District of Columbia for each of 17 sex crimes: rape and sexual assault, marital exemptions from rape and sexual assault, age of consent, sodomy, transmission of disease, public nudity and indecency, fornication, adultery, abuse of position of trust or authority, incest, bigamy, prostitution, possession of obscene materials, bestiality, necrophilia, obscene communications, and voyeurism. Each of the 17 chapters begins with a concise description of the offense, including whether it is a misdemeanor or a felony. For some of the putative sex crimes, some states do not have specific statutes, although it is likely that one might be charged under a nonsexual statute (e.g., with regard to necrophilia). The penalty for sex crimes varies markedly across states. For example, with regard to bestiality (“the abominable and detestable crime against nature with a beast”), Rhode Island 7 Book
Review Editor, Archives of Sexual Behavior. and Adolescent Gender Identity Clinic, Child Psychiatry Program, Centre for Addiction and Mental Health—Clarke Division, 250 College Street, Toronto, Ontario M5T 1R8, Canada.
8 Child
P1: FOM/FLK
P2: FQJ
Archives of Sexual Behavior [asb]
PL110-332
March 13, 2000
Book Reviews
12:17
Style file version Nov. 19th, 1999
301
imposes a penalty of not less than 7 years in prison, whereas in Minnesota the penalty is not more than 1 year. Either the remaining states do not have a specific statute for bestiality or, for those that do, the penalty is presumably left to the discretion of the judge. For those who are interested in sexology and the law, this is an extremely useful volume. One hopes that the authors will prepare a second volume comparing sex laws in the United States with those in other countries throughout the globe, as was recently done by West and Green (1997) with regard to homosexuality (cf. Grey, 1999).
REFERENCES Grey, A. (1999). Review of Sociolegal Control of Homosexuality: A Multi-Nation Comparison. Arch. Sex. Behav. 28: 271–276. Posner, R. A. (1992). Sex and Reason, Harvard University Press, Cambridge, MA. Reilly, M. T. (1996). Review of Sex and Reason. Arch. Sex. Behav. 25: 650–655. West, D. J., and Green, R. (eds.). (1997). Sociolegal Control of Homosexuality: A Multi-Nation Comparison, Plenum Press, New York.
Handbook of Sexuality-Related Measures (Second Edition). Edited by Clive M. Davis, William L. Yarber, Robert Bauserman, George Schreer, and Sandra L. Davis. Sage, Thousand Oaks, CA, 1998, 589 pp., $99.95. Reviewed by Kenneth J. Zucker, Ph.D.7,8
From abortion to vasectomy, this edited volume is an extremely useful compendium of extant measures relevant to the mission of sexual science. Each of the 200+ entries contains a description of the measure (including instructions for scoring and its psychometrics in the majority of cases), relevant references, and the measure itself. For those who engage in quantitative sexological research, this volume should be kept on one’s bottom shelf, close to the computer.
P1: FLF Archives of Sexual Behavior [asb]
PL110-333
April 19, 2000
19:1
Style file version Nov. 19th, 1999
Archives of Sexual Behavior, Vol. 29, No. 3, 2000
Letters to the Editor
Regarding “The Relation Between Sexual Orientation and Penile Size,” by A. F. Bogaert and S. Hershberger [Archives of Sexual Behavior, Vol. 28(3), pp. 213–221, 1999]. While the brevity of Bogaert’s and Hershberger’s article is to be commended, the authors’ exclusive reliance on data relating to average penile size in the two population groups seriously undermines the credibility of their far-reaching conclusions. While the average of 5, 6, and 7 in. and the average of 2, 6, and 10 in. may also be 6 in., surely the authors would agree that these two sets have significantly different characteristics that merit further explanation. I urge authors to complete their article by publishing charts showing the distribution of different units of penile length and circumference (e.g., in 0.25 or 0.5-in. increments) in the two populations. If the resulting curves are identical in their bell shape, with the homosexual curve slightly to the right of the heterosexual curve, the authors’ hypothesis would be strengthened. I suspect, however, that the data will in fact show something quite different: that the only statistically significant difference in the distribution of penile size increments in the two populations occurs in the highest tenth percentile of each group, and that it is this characteristic that explains the average differential between the two populations. A disproportionate representation of exceptionally large penises in the homosexual population would not support the authors’ broad conclusions. It would instead support a much narrower hypothesis that prenatal hormonal exposure may be a significant cause of homosexual behavior in a very small segment of the homosexual population. Alternatively, this phenomenon could be explained by any number of nongenetic hypotheses. One such explanation (that also puts into question the validity of the authors’ definition of heterosexuality) is the following: because of the importance of penile size in the sexual fantasies of many homosexuals, (a) an otherwise heterosexual man with an exceptionally large penis is probably exposed to a greater than average number of sexual advances from homosexual men, and is therefore (b) more likely than the average man to have engaged in more than five isolated homosexual incidents over the course of his life, and is therefore (c) arbitrarily classified by the authors as “homosexual,” with the result that (d) that the average penile size in the homosexual population as reported 303 C 2000 Plenum Publishing Corporation 0004-0002/00/0600-0303$18.00/0 °
P1: FLF Archives of Sexual Behavior [asb]
304
PL110-333
April 19, 2000
19:1
Style file version Nov. 19th, 1999
Letters to the Editor
by the authors is in fact overstated. While somewhat far-fetched, the plausibility of this explanation nonetheless illustrates the pitfalls and perils of trying to explain the causes of homosexuality on the basis of isolated statistical data. William Krisel, J.D. 68 rue de Faubourg Saint-Honore 75008 Paris, France
Variability, Sexual Orientation and Penile Size: A Reply to Krisel Krisel (see preceding Letter to the Editor) raises distribution/variability issues regarding our analyses of penile dimensions and sexual orientation in the Kinsey data, recently published in the Archives of Sexual Behavior (Bogaert and Hershberger, 1999). In particular, he suggests that the only difference in penile dimensions would occur in the highest tenth percentile of each group and that “that this is the characteristic that explains the average difference between the two populations.” We reanalyzed these data using only participants who had penile dimensions at the quarter-inch value closest to the 90th percentile or less. For example, for flaccid penis circumference, the quarter-inch value closest to the 90th percentile was 4.25 in., with 86.6% of the entire sample reporting this value or smaller penises. For flaccid penis length, the quarter-inch value closest to the 90th percentile was 4.75, with 90.7% of the entire sample reporting this value or smaller penes. (The Kinsey researchers recorded sizes to the nearest quarter-inch.) Thus, generally, the highest 10% of sample was eliminated for each of these analyses. Despite the truncated range and the loss of power because of a reduction in sample size, all five penile contrasts remained significant. It is of note as well that the distributions were generally normally distributed and were similar for both the homosexual and the heterosexual groups. In particular, the skew values were very similar for both the homosexual and the heterosexual distributions, with all values being between 0 and or ±1.25, suggesting fairly symmetric/balanced distributions. Kurtosis values were somewhat positive (and suggested a degree of “flatness” to the distributions), but they were very similar for both the homosexual and the heterosexual groups, with the same three penile measures having values of less than 2 for both groups and the same two penile measures exceeding 3 for both groups. Finally, contrary to Krisel’s (see above Letter) suggestion, providing data that may support the notion that prenatal hormones (or other biological mechanisms) affect sexual orientation development does not necessarily imply that this is the only influence on sexual orientation development. Thus, the Kinsey data on sexual
P1: FLF Archives of Sexual Behavior [asb]
PL110-333
April 19, 2000
19:1
Letters to the Editor
Style file version Nov. 19th, 1999
305
orientation and penis size (and our interpretation of them) are very compatible with multiinfluences on sexual orientation development.
REFERENCE Bogaert, A. F., and Hershberger, S. (1999). The relation between sexual orientation and penile size. Arch. Sex. Behav. 28: 213–221.
Anthony F. Bogaert, Ph.D. Departments of Community Health Sciences and Psychology Brock University St. Catharines, Ontario, Canada L2S 3A1 e-mail:
[email protected] Scott Hershberger, Ph.D. Department of Psychology California State University Long Beach, California 90840
