***From FEMINET, Felton CA 408-335-4387 or 408-335-7888
Childhood Sexual Identity, Childhood Religiosity, and 'Homophobia' as Influences in the Development of Transsexualism, Homosexuality, and Heterosexuality
Ronald E. Hellman, MD; Richard Green, MD; James L. Gray, PhD; Katherine Williams, MA
We have studied the interaction between boyhood cross-gender behavior, "homophobia," and religiosity in men raised as Catholics who, during adulthood, considered themselves transsexual, homosexual, or heterosexual. The sample consisted of 43 transsexual, 78 homosexual, and 43 heterosexual subjects matched for age, race, educational level, and economic status. The transsexual men recalled the most "feminine" behavior during boyhood, followed by the homosexual men. The heterosexual group was most "homophobic," followed by the transsexuals. "Homophobia" positively correlated with religiosity among adult transsexuals and heterosexuals. Transsexuals, recalling childhood, perceived their parents as being more religious than did the homosexuals. At intermediate levels of "femininity," greater "homophobia" scores were associated with more transsexual characteristics. These data support the thesis that early developmental aspects of sexual identity, and later concerns over homosexuality that are partly of a religious derivative, may contribute to a transsexual outcome.
(Arch Gen Psychiatry 1981;38:910-915) Accepted for publication April 6, 1981
From South Beach Psychiatric Center Brooklyn, NY (Dr. Hellman), the Departments of Psychiatry and Behavioral Science and Psychology (Dr. Green), the Computing Center (Dr. Gray), and the Long Island Research Institute (Ms Williams), State University of New York at Stony Brook.
Reprint requests to Department of Psychiatry and Behavioral Science, Health Sciences Center, State University of New York at Stony Brook, Stony Brook, NY 11794 (Dr. Green).
Of the research evidence supporting which variables are most important in determining sexual identity, little attention has been given to the effect of "homophobia" and religiosity. Some psychoanalysts have stated that transsexualism is a defense against homosexuality, allowing the transsexual to escape responsibility for homosexuality. This mechanism allows the emphasis to be shifted from an unacceptable object choice (male) to the more acceptable issue of core-morphologic identity. The transsexual, by identifying as a woman, is able to label the sexual experience with an anatomical male as "heterosexual." According to this view, "masculine" development implies an associated fear of "feminine" identity that is equated with homosexuality by the heterosexual.1
Weinberg2 developed these conceptualizations further, introducing the term "homophobia," which he defined as an irrational fear of homosexuality. He discussed the following five motives that sustain "homophobia" (1) religiosity, (2) reaction formation, (3) repressed envy, (4) threat to social values, and (5) fear of death. Of these variables, religiosity seems noteworthy, since some clinicians have observed that a majority of transsexuals they have studied are Catholic and were actively religious when young (W. C. Pomeroy, personal communication, March 1, 1976).3 Catholic prohibitions against homosexuality are considered severe and may be the reason why an underrepresentation of Catholics has been observed among homosexuals.4
Related to the apparent Catholic majority noted among transsexuals are the statements of some transsexuals that homosexuality is so immoral that the only way they could become erotically engaged with a member of their own sex was to become a member of the opposite sex.3 Thus, transsexualism may be a way of resolving the problem of a sexual orientation that is unacceptable because of religious belief. We have explored the possibility that a resolution of this conflict may occur by a metamorphosis, or change, in one dimension of sexual identity, the change in core-morphologic identity from male to female. We hypothesized that a religious individual with an emerging cross-gender behavioral repertoire and a later same-sex orientation that conflicts with religious belief may resolve the conflict by becoming a member of the opposite sex, ie, becoming a "heterosexual," in conformity with religious approval.
Furthermore, the probability of a transsexual outcome in adult life was hypothesized to be increased when, in childhood, there is considerable crossgender role behavior and a devout early religious upbringing. If conflicting religious beliefs form a basis for the individual's gender "dysphoria" and transsexualism emerges as a defense against homosexuality, we additionally hypothesized that a transsexual would demonstrate notable "homophobia." In individuals with cross-gender role behavior in childhood but without a devout religious background, we hypothesized that homosexuality would be the outcome. This is based on reports indicating that cross-gender behavior in the boy is associated with a higher than average probability of adult homosexuality.5-7 Finally, we hypothesized that "masculine" role behavior in childhood is predictive of a heterosexual outcome, irrespective of religious upbringing.
SUBJECTS AND METHODS
Subjects
All subjects were born into Catholic families. The sample consisted of
homosexual men actively practicing Catholicism, homosexual men not actively
practicing Catholicism, male-to-female transsexuals before and after
operation, and heterosexual men. Subjects were categorized into one of the
following three groups: (1) transsexuals, (2) homosexuals, and (3)
heterosexuals. Those subjects who had had a male-to-female sex-change
operation or were currently taking "female" hormones and seriously considering
a sex change were included in the transsexual group, even if they defined
themselves as homosexual or heterosexual. All other subjects were categorized
according to their self-definition as homosexual or heterosexual.
Subjects were recruited through special population groups, newspaper advertisements, professional and personal contacts, and word of mouth. The groups contacted included Dignity, Inc, an organization that attempts to help members integrate active Catholic religious practice with a homosexual life style, and Confide, Inc, a transsexual-transvestite group. Advertisements were placed in local newspapers and Transition, a publication for transsexuals and transvestites.
The three groups of subjects were matched for age, race, educational level, and economic status by selecting a subsample of each group from a larger sample so that the three groups would be similar on these four variables. Selection was blind with regard to all other variables. A x2 test showed no significant difference between groups for these demographic variables (age, x2 = 3.53, df = 8, P = .90; race, x2 = 3.08, df = 8, P = .93; educational level, x2 = 8.99, df = 10, P = .53; and average income in the last five years, x2 = 15.78, df = 10, P = .11).
Materials and Procedure
A packet consisting of a multiple-choice questionnaire and consent and
reimbursement forms was sent to each subject. Subjects were informed that all
information would be kept confidential. They were asked not to identify
themselves on the questionnaires, which were labeled with a code number.
Consent forms were kept under private lock and key.
The questionnaire inquired about religious background, sexual-identity development, and attitudes toward homosexuality. Childhood, adolescence, and adult life were examined retrospectively. Religiosity scales were adapted from Shaw and Wright.8 These items explored the degree of church attendance, other religious group activity, personal commitment to religion, religious conviction, intellectual understanding of religious issues, and parental influence on religiosity. Since no established scales to measure sexualidentity development were available, to our knowledge, at the time of the study, these questions were developed from our extensive research experience in this area. Questions on sexual-identity development inquired about recreational, dress, and companionship preferences, male sex-typed behaviors, self-rating of "masculinity"; sexual orientation; and possible transsexualism. Questions on attitudes toward homosexuality were adapted from Smith's "homophobia" scale.9 (A third alternative, "not sure," was added to each question.) Inquiries were made about the belief that homosexuality is a "sin"; that society should be protected from homosexuals; whether homosexuals should be allowed to hold government, teaching, and organizational positions; and emotional reactions to homosexuality.
Several subjects from each group (not part of the final test group) were used to pretest the questionnaire. As a result, some items were modified to improve clarity.
The items on the questionnaire were divided into the following three subgroups: those related to sexual identity, those related to religiosity, and those adapted from Smith's "homophobia" scale. The items from each subgroup were factor analyzed separately. As a result of the factor analysis, ten indexes of "femininity," five indexes of religiosity, and an index of "homophobia" were constructed. The indexes were constructed to separate independent sources of variance observed in the factor analysis and to facilitate testing of the initial hypotheses. The ten indexes of "femininity" have a significant amount of common variance. There is also significant intercorrelation among the five indexes of religiosity. However, parental religiosity has a low correlation with the other four indexes of religiosity, and childhood religiosity and current religiosity do not correlate with each other. The range of possible scores for each was 0 to 100, with 100 representing most "feminine," most religious, or most "homophobic." The indexes of "femininity" and religiosity were divided into three developmental periods. In addition, the indexes of "femininity" were divided into the following two types: gender-role behavior and transsexual characteristics. The latter were composed of questions about the wish to be a girl, the wish to have female anatomy, and the desire to have a sex-change operation (Table 1).
RESULTS
Sexual Identity
Transsexuals scored more "feminine" than homosexuals and heterosexuals
during all developmental periods (Table 2). The transsexuals had
significantly more "feminine" scores than the other two groups of subjects for
all but one index; here there was no difference between the transsexuals and
homosexuals, since both groups showed a lack of interest in "typical
masculine" games. The relationship among the three groups of subjects was
different for the two different types of indexes of "femininity." The
homosexuals had scores between those of the transsexuals and heterosexuals on
indexes of general sexual identity. However, the homosexuals and
heterosexuals scored virtually the same on indexes of transsexualism.
Table 1. List of Indexes*
Index Description
General "femininity" All sexual identity and transsexual
characteristic items for all age levels Childhood gender-role Childhood peer group composition,
behavior participation in male sex-typed games,
which parent the subject preferred to help
with chores, play with "Barbie"-type
dolls, use of cosmetics, frequency of
being called "sissy," crossdressing,
self-rating of childhood "masculinity"
Adolescent gender-role Participation in male sex-typed games,
behavior frequency of being called "sissy," self-
rating of "masculinity" in early and late
adolescence
Adult gender-role behavior Participation in male sex-typed games,
frequency of being called "fag" or
considered "effeminate," etc., self-rating
of "masculinity" during twenties and
thirties
Nonparticipation in typical Participation in male sex-typed games in
"masculine" games childhood, adolescence, and adulthood
Childhood transsexual Use of cosmetics, crossdressing, wish to
be a girl, did not want to have a penis,
wanted to have breasts, wanted to have a
vagina, seriously considered a sex-
change operation during childhood
Adolescent transsexual Use of cosmetics, crossdressing, wish
to be a woman in early and late
adolescence, did not want to have a penis
in early and late adolescence, wanted to
have female breasts in early and late
adolescence, wanted to have a vagina in
early and late adolescence, seriously
considered a sex-change operation in
early and late adolescence
Adult transsexual Use of cosmetics, crossdressing, wish to
be a woman in early adulthood and
currently, did not want a penis in early
adulthood and currently, wanted to have
female breasts in early adulthood and
currently, wanted to have a vagina in
early adulthood and currently seriously
considered a sex-change operation in
early adulthood and currently, taking
"female" sex hormones, has had a sex-
change operation
Did not want penis Did not want a penis at all developmental
stages
Wanted breasts and vagina Wanted female breasts and vagina at all
developmental stages
Childhood, adolescent, Questions are the same at each
adult, and current developmental stage: church attendance,
religiosity importance of religion, belief in God,
active in church activities, prayer outside
of church, attendance at church school,
worked at understanding his religion,
sureness of religious beliefs, belief in
happiness and security from religion
Parental religiosity Were parents religious during
respondent's childhood, adolescence,
early adulthood, and currently?
"Homophobia" "Homophobia" scale adapted from
Smith,9 with question about laws
against homosexuality eliminated:
belief that homosexuality is a sin,
homosexuals should be locked up to
protect society, it would be upsetting for
one to find out one was alone with a
homosexual, homosexuals should be
allowed to hold government positions,
membership in organizations with
homosexuals, thought of homosexual
acts disgusting, a homosexual could be a
good President of the United States, fear
of one's child having a homosexual
teacher, nervousness about having a
homosexual sit next to oneself on a bus
or subway
*All items in all indexes are weighted equally.
Table 2. Means on Indexes of 'Femininity'*
Index TS t (TS vs. HS)+ HS HT F+
General "femininity" 76 15.65 34 15 249.45
Childhood gender-role
behavior 64 6.53 38 18 77.30
Adolescent gender-role
behavior 76 2.79 64 32 54.31
Adult gender-role behavior 71 5.85 47 29 50.10
Nonparticipation in typical
"masculine" games 76 1.47 69 25 42.94
Childhood transsexual 68 13.26 7 2 219.92
Adolescent transsexual 79 19.92 4 3 377.32
Adult transsexual 84 39.56 2 3 1,113.72
Did not want penis 71 10.50 3 8 100.81
Wanted breasts and vagina 90 26.39 2 2 589.95
*TS indicates transsexuals; HS, homosexuals; and HT, heterosexuals. +All values are significant at P < .05, except the t value for nonparticipation in typical "masculine" games.
The three groups were compared on the following three individual items (Table 3): childhood peer group composition; which parent the subject, as a child, preferred helping with chores; and frequency of play with "Barbie"-type dolls. These three items which were also components of the childhood genderrole behavior index, were selected because they had more independent variance than the other items that comprise the childhood gender-role behavior index. There was no significant difference between the transsexuals and homosexuals in the composition of the childhood peer group (x2 = .82, df = 2, P = .7, r = -.08). However, heterosexuals were significantly different from the other two groups (x2 = 23.1, df = 4, P = .0001, r = -.31). Twelve transsexuals (28%), 28 homosexuals (36%), and 32 heterosexuals (74%) reported an all-male childhood peer group.
Significantly more transsexuals than homosexuals preferred helping mother with chores x2 = 9.04, df = 2, P = 0.1, r = -.28). The heterosexuals showed less interest in helping mother than did the homosexuals. Twenty-six transsexuals (60%) preferred helping mother, compared with 29 homosexuals (37%) and ten heterosexuals (23%).
There was a significantly higher frequency of play as a child with "Barbie"-type dolls among the transsexuals than among the homosexuals (x2 = 27.64, df = 2, P < .0001, r = -.43). The heterosexuals showed less interest in doll play than the homosexuals. Frequency of play with "Barbie"-type dolls was as follows: transsexuals answering either "most of the time" or "occasionally," 26 (60%); homosexuals answering "most of the time" or "occasionally," 18 (23%); and heterosexuals answering "most of the time" or "occasionally," two (5%).
'Homophobia'
Each of the three groups scored differently on the "homophobia" index. In
addition, all three groups scored low. Transsexuals had a mean score of 13;
homosexuals, 2; and heterosexuals, 28. Kendall's r was used as a
nonparametric test to make comparisons between groups of subjects. Comparing
transsexuals and homosexuals on the "homophobia" index resulted in r = -.40 (P
= .001). Comparing transsexuals and heterosexuals on "homophobia" resulted in
r = .29 (P = .001). The values of r indicate greater difference between
transsexuals and homosexuals than between transsexuals and heterosexuals.
'Homophobia' and 'Femininity' Interaction At intermediate levels of "femininity" there was an interaction with "homophobia" to increase transsexual characteristics. A stepwise regression analysis was done using the adult transsexual index score as the dependent variable. There was very little overlap between the transsexuals and other subjects on this index. Only one transsexual had a score lower than the highest score of a nontranssexual. The initial variable list consisted of the following indexes: childhood "femininity," adolescent "femininity," childhood religiosity, parental religiosity, and the "homophobia" index, as well as the three individual items discussed earlier. Also included were the orthogonal quadratic-interaction terms corresponding to childhood "femininity" and adolescent "femininity needed to test the prediction, the linear interaction terms, and the simple quadratic terms for childhood and adolescent "femininity". The only variables allowed to enter the equation were the ones whose, B values were significantly greater than zero (P < .05). The results are given in list 1 of Table 4. They indicate increasing transsexual tendencies with increasing childhood "femininity" and with increasing "homophobia." The significant quadratic term indicates that the slope of the regression line predicting the adult transsexual index score from childhood "femininity" is increasing as "femininity" increases. The significance of the interaction term supports the hypothesis that at intermediate values of "femininity," greater "homophobia" scores are associated with more transsexual characteristics.
Table 3. Responses to Three Selected Items Concerning Sex-Typed Childhood Behaviors
% of Subjects*
TS HS HT
(N = 43) (N = 78) (N = 43) x2 df
P
All male
peer group 28 36 74 23.1 4 .0001
Preferred helping
mother with
chores 60 37 23 21.2 4 .003
Played with "Barbie"-type dolls
Most of the time 25 0 2
48.4 4 .0001
Occasionally 40 24 2
*TS indicates transsexuals; HS, homosexuals; and HT, heterosexuals.
Table 4. Standardized Regression Coefficients of VariablesPredicting Transsexualism
B F
List 1 (multiple r2 = .45)
Childhood "femininity" .64 102.34
Childhood "femininity"
quadratic term .15 5.65
"Homophobia" .13 3.86
Childhood femininity
quadratic term by
"homophobia" interaction term -.16 6.63
List 2 (multiple r2 = .75)
Childhood transsexual .79 260.77
Childhood femininity
quadratic term by
"homophobia" interaction term -.11 6.62
Play with "Barbie"-type dolls .12 5.96
List 3 (multiple r2 = .89)
Adolescent transsexual .93 1,104.77
Parental religiosity .05 3.79
Two other stepwise regression analyses, with additions to the list of variables, were conducted. The childhood transsexual index score was added to the first list. The results are given in list 2 of Table 4. The interaction hypothesis just discussed was supported again. The adolescent transsexual index score was substituted for the childhood transsexual index score in the second list, and the results are given in list 3 of Table 4. The childhood "femininity" quadratic term-by-"homophobia" score interaction was not statistically significant. This suggests that during adolescence, individuals who had intermediate scores on "femininity" during child hood already had high scores on transsexual traits if they also had high "homophobia" scores.
The correlation coefficient between the "homophobia" index score and the adult transsexual index score was calculated for 92 subjects with intermediate values of childhood "femininity." The result was r = .27 (P < .01).
Religiosity
Comparing transsexuals and homosexuals on parental religiosity resulted in
r = -.17 (P = .02), indicating the transsexuals perceived their parents as
more religious than did the homosexuals. The same comparison was made on
childhood religiosity resulting in r = .08 (P = .2). This nonsignificant
trend is in the direction of greater childhood religiosity among the
homosexuals. Table 5 gives the means for each of the groups on the indexes of
religiosity. As can be seen, the transsexuals are not different from the
homosexuals on any other index of religiosity. Even with an analysis of
covariance controlling for "femininity," there is no significant difference
between the transsexuals and homosexuals on childhood religiosity (eg, with
childhood "femininity" as covariate, F = 1.32, df = 1, 118, P = .25). Crosstabulations
controlling for "femininity" and using nonparametric tests also
show no difference between transsexuals and homosexuals on childhood
religiosity.
Table 5. Means of Indexes of Religiosity*
TS t (TS vs. HS) HS HT F
Childhood 79 -0.93 82 73 3.83 +
Adolescent 64 -0.56 67 55 3.57 +
Adult 44 -0.74 47 44 0.33
Current 45 -0.59 48 47 0.15
Parental 67 2.18 + 55 57 2.48
*TS indicates transsexuals; HS, homosexuals, HT, heterosexuals
+P < .05
Table 6. Religiosity Scores of Transsexuals
and Low-Religious Homosexuals
Low-Religious
Transsexuals Homosexuals
(N = 43) (N = 39) t P
Childhood 79 81 -0.44 .66
Adolescent 64 59 1.12 .27
Parental 67 50 2.54 .01
The data in this study indicate that an individual's religiosity per se is not a predictor of transsexualism. With respect to childhood religiosity, the possibility of a relationship between a transsexual outcome and childhood religiosity was investigated further by examining six of the individual items that comprise the childhood religiosity index. The values of r for four of these items are positive, indicating that homosexuals were more religious than transsexuals. The values of r for the other two items are very close to zero (-.01 and -.05). In addition, the magnitude of the difference between the two groups is very small. Homosexuals and transsexuals were compared on childhood religiosity using a x2 analysis. First the childhood religiosity scale was divided into eight discrete categories with approximately equal numbers of subjects in each category. The result was x = 3.3, df = 7, P = .86. Two other x2 tests were done, showing no relationship between childhood religiosity and transsexual characteristics. The relation ship between the adult transsexual index score and childhood religiosity was evaluated; the result was x2 = 17.37, df = 21, P = .69. The desire for female anatomy was also compared with childhood religiosity, which resulted in x2 = 23.36, df = 21, P = .33. In addition, the transsexuals were compared with the less religious half of the homosexuals on several indexes. The less religious homosexuals all had current religiosity scores of less than 50. Table 6 gives the results of the comparison.
Kendall's r showed a negative correlation (r = -.24, P = .01) between "homophobia" and childhood religiosity among the homosexuals but not in any of the other groups. This negative relation for only the homosexuals extends into adolescence (r = -.26, P = .01). In early adulthood and currently, the homosexuals did not have a significant correlation between "homophobia" and religiosity (r = -.04 and .03, respectively). However, both the transsexuals and the heterosexuals had a positive correlation between "homophobia" and religiosity in early adulthood and currently (transsexuals, r = .22 and .22, P = .05; heterosexuals, r = .23, P = .05, and pi = .26, P = .01, respectively), suggesting an indirect link between religiosity and the transsexual outcome.
Religious Change
There was no significant difference among the groups for the proportion who
reported a change in religion. Among transsexuals, 30% reported having
changed religion, compared with 15% of high-religious homosexuals, 41% of lowreligious
homosexuals, and 28% of heterosexuals (x2 = 6.34, df = 3, P = .10).
Crossdressing and Sexual Identity
In the transsexual group, arousal to crossdressing during adolescence was
related to a less "feminine" childhood than for those who reported
crossdressing without sexual arousal (t = 1.69, df = 36, P = .05). The sexual
arousal group (N = 17) had a mean childhood "femininity" score 58, and the no
arousal group (N = 21) had a mean score of 70. Among the homosexuals, the
effect was in the opposite direction (t = 2.27, df = 18, P = .04, two-tailed
test). The sexual-arousal group (N = 6) had a mean childhood "femininity"
score of 54, and the no-arousal group (N = 14) a mean score of 39. Table 7
gives the means and t test results of the differences on the various indexes
for the transsexuals and homosexuals reporting sexual arousal to crossdressing
during adolescence.
Table 7. Scores of Transsexuals and Homosexuals Reporting Sexual arousal to Crossdressing During Adolescence
Transsexuals Homosexuals
Index (N = 17) (N = 6) t P
General femininity 75 51 3.13 .01
Childhood femininity 58 54 0.56 .58
Adolescent femininity 72 81 -0.74 .48
Adult femininity 65 56 0.76 .47
Nonparticipation in typical
"masculine" games 73 74 -0.06 .95
Childhood transsexual 67 15 5.77 < .001
Adolescent transsexual 81 25 4.15 .004
Adult transsexual 88 13 10.32 <.00l
Did not want penis 73 10 5.13 <.001
Wanted breasts and vagina 93 17 7.65 <.001
Childhood religiosity 80 81 -0.13 .90
Adolescent religiosity 64 63 0.13 .90
Adult religiosity 50 40 1.07 .31
Current religiosity 50 34 1.76 .12
Parental religiosity 62 50 0.81 .44
Homophobia 12 6 1.07 .30
Table 8. Scores of 'Transgenderists' and Remaining Transsexuals
Trans- Other Trans-
genderists sexuals
Index (N = 8) (N = 35) t P
General femininity 75 76 0.27 .79
Childhood femininity 57 65 1.02 .33
Adolescent femininity 74 76 0.21 .84
Adult femininity 65 72 1.07 .30
Nonparticipation in
typical masculine games 91 73 2.68 .01
Childhood transsexual 48 73 2.18 .06
Adolescent transsexual 72 80 1.41 .17
Adult transsexual 76 86 3.38 .002
Did not want penis 3 86 15.69 <.001
Wanted breasts and vagina 93 90 0.61 .55
Childhood religiosity 79 79 0.05 .96
Adolescent religiosity 69 63 0.72 .49
Adult religiosity 44 43 0.13 .90
Current religiosity 43 45 0.30 .77
Parental religiosity 63 68 0.45 .66
Homophobia 10 14 0.66 .52
'Transgenderists'
A subgroup of eight of the transsexuals was different from the other
transsexuals in one respect: they wanted to keep their penis. In other
respects they were very similar to the other transsexuals. This group of
subjects seems to be part of an emerging subgroup. They seem to be somewhat
more like transvestites than other transsexuals. They have been called
"transgenderists" (V. Prince, Ph.D., oral communication, March 1979). Some of
the ways in which these transgenderists differed are as follows. Five of
eight were Hispanic, while only two of the other 35 apparent transsexuals (6%)
were Hispanic (x2 = 17.94, df = 4, P = .001); the transgenderists had
significantly less education than the other apparent transsexuals (x2 = 17.69,
df = 5; P = .003); r = -.37, P = .002); and the transgenderists were more
likely to have had an all-male peer group as children (71%) than other
apparent transsexuals (23%) (x2 = 6.72, df = 2; P = .04); r = -.25, P = -.03).
The two groups of subjects showed no difference in preferences as a child for
helping mother or father and no difference in the proportion playing with
"Barbie"-type dolls. Table 8 gives the means of the transgenderists and
transsexuals on the various indexes and the t test results of the difference
between the means.
COMMENT
It should be recognized that we interpret these findings within the context
of a retrospective study and are limited necessarily by subjects' ability to
recall past events.
Our data indicate that childhood cross-gender behavior is most prevalent in the transsexual group, moderate among the homosexuals, and least evident in the heterosexual sample. The three groups fell on a continuum of "femininity." Our findings support the hypothesis that the probability of a transsexual outcome is greatest when there is a history of extensive crossgender behavior in childhood. The probability of a homosexual outcome was greatest for subjects with a history in the midrange of the "feminine" continuum, while the heterosexual group scored lowest on the scale. The homosexual group overlapped considerably with the transsexual and heterosexual groups.
Measures of childhood cross-gender peer group, "Barbie"-type doll play, together with chore preference most reliably predicted later sexual outcome. In addition, the early expressed wish to have a vagina and female breast, and not to have a penis, appears to be important in differentiating the pretranssexuals from the other groups. The transsexual index was able to distinguish the transsexual group as discontinuous from the other two, as recalled from childhood. This finding would seem to clarify the meaning of the wish to become a girl, which is reported by many male homosexuals who recall "feminine" behavior during childhood. The desire to become a girl, for these persons, appears to be a yearning for "feminine" gender-role behavior and not a desire for female sexual anatomy.
Our data provide the first comparison, to our knowledge between transsexuals, homosexuals, and heterosexuals on the parameter of "homophobia" and support the hypothesis that "homophobia" contributes to a transsexual outcome.
We found that heterosexuals were most "homophobic," followed by the transsexual group. Homosexuals were least "homophobic." Despite these differences, all groups scored in the lower range of the "homophobia" scale. This may be accounted for by recent shifts in attitude having an impact on attitudinal "homophobia,"10 inadvertent selection of samples with a low "homophobic" profile, or lack of sensitivity of the scale. Some of these factors suggest the need for more subtle measurements of this parameter. Our finding that the transsexuals were not as "homophobic" as the heterosexuals remains a subject for future research. One possibility is that since transsexuals view themselves as "deviant," they may come to view other "deviant" groups with more acceptance.11
Except for the transsexuals' perception of their parents as being more religious than those of homosexuals, no significant differences were found on the indexes of religiosity. Thus, our data did not confirm fully the hypothesis that transsexuals tend to have a devout religious background.
In addition, our data do not confirm most previous reports that homosexuals tend to be more alienated from religion. Greenberg,12 for example, reported that "conventionally religious homosexual males, especially those who were Roman Catholic" experience far more conflict over their homosexuality. Weinberg and Williams13 found that religiosity correlated with lower commitment to homosexuality, restricted sexual activity, less social involvement with other homosexuals, and concern in remaining covert. Bell and Weinbergl4 reported that homosexuals attend church less often than heterosexuals, and one third of homosexuals sampled said that their homosexuality tended to weaken their religious feelings. Half of their homosexual sample, however, reported that their homosexuality did not affect their religiosity. These authors cite two additional reports that are congruent with our findings, one in which there was no difference between homosexual and heterosexual samples on religious attitudes and another in which homosexual males seemed more committed to religion. Our findings would be expected, since half of our homosexual subjects were obtained through a gay, Catholic organization (Dignity, Inc).
Our findings also indicated no significant differences between groups in the proportion who changed religion. The highest percentage of persons changing religion were among the low-religious homosexuals; however, significant proportions from all groups reported a change in religion, probably reflecting a more generalized alienation from the Catholic church.
Our findings do not indicate that "homophobia" is primarily of religious origin. Our data do not confirm the view that transsexualism is simply motivated out of conflict between sexual orientation and religious belief. The "homophobic" response as a motivating factor in the development of transsexualism may reflect the need to maintain established social values and clear distinctions between the sexes." In this view, "homophobia" among transsexuals stems from the belief that heterosexuality is appropriate behavior and a desire for conformity with the established majority lessens the experience of "deviancy."
From our study, the overriding influence on transsexualism appears to be childhood "femininity" and a desire for cross-sex anatomy. However, we found that when the level of "femininity" is not extreme, "homophobia" does become significantly associated with a transsexual outcome. Thus, the two factors may interact to result in transsexualism.
We observed a positive correlation between religiosity and "homophobia" in adulthood among transsexuals and heterosexuals. The late developmental expression of this relationship suggests that although childhood religiosity is not a predictor of transsexualism, it is a late determinant of attitudes that may function to maintain the established mode of sexual identity.
Our findings confirm previous impressions that a subgroup of fetishistic crossdressers come to desire sex-change surgery and fulfill the criteria of some clinicians for a transsexual diagnosis. Our findings are consistent with previous observations that this group typically has a "masculine" background. However, a similar group of fetishistic crossdressers observed among our homosexual sample had higher "femininity" scores when compared with homosexuals not sexually aroused by crossdressing. This finding is not consistent with previous impressions.
An unexpected finding was the observation that 19% of our transsexual sample desired female breasts and male genitalia, so-called transgenderists. In contrast to other transsexuals, this group was largely Hispanic (possibly a geographic bias), less educated, and tended to have male companions in childhood. This suggests that cultural factors and some elements of "masculinity" contribute to the evolution of sexual identity in this group. Further study of this latter group may expand our understanding of the development of the various patterns of typical and atypical sexual identity.
This research was supported by National Institute of Mental Health postdoctoral training grant 5 T32 MH14621 (Drs. Hellman and Gray) and the Long Island Research Institute, Division of Psychobiology, State University of New York at Stony Brook. Dignity, Inc., Confide, Inc., and Leo Wollman, MD, assisted in this work.
