Вы находитесь на странице: 1из 14

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/5657573

Sexual Orientation and Childhood Gender Nonconformity: Evidence From


Home Videos

Article  in  Developmental Psychology · February 2008


DOI: 10.1037/0012-1649.44.1.46 · Source: PubMed

CITATIONS READS
187 4,635

4 authors, including:

Gerulf Rieger J. Michael Bailey


University of Essex Northwestern University
71 PUBLICATIONS   1,641 CITATIONS    147 PUBLICATIONS   7,707 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by J. Michael Bailey on 02 June 2014.

The user has requested enhancement of the downloaded file.


Developmental Psychology Copyright 2008 by the American Psychological Association
2008, Vol. 44, No. 1, 46 –58 0012-1649/08/$12.00 DOI: 10.1037/0012-1649.44.1.46

Sexual Orientation and Childhood Gender Nonconformity: Evidence From


Home Videos

Gerulf Rieger and Joan A. W. Linsenmeier Lorenz Gygax


Northwestern University Swiss Federal Veterinary Office

J. Michael Bailey
Northwestern University

Homosexual adults tend to be more gender nonconforming than heterosexual adults in some of their
behaviors, feelings, and interests. Retrospective studies have also shown large differences in childhood
gender nonconformity, but these studies have been criticized for possible memory biases. The authors
studied an indicator of childhood gender nonconformity not subject to such biases: childhood home
videos. They recruited homosexual and heterosexual men and women (targets) with videos from their
childhood and subsequently asked heterosexual and homosexual raters to judge the gender nonconfor-
mity of the targets from both the childhood videos and adult videos made for the study. Prehomosexual
children were judged more gender nonconforming, on average, than preheterosexual children, and this
pattern obtained for both men and women. This difference emerged early, carried into adulthood, and was
consistent with self-report. In addition, targets who were more gender nonconforming tended to recall
more childhood rejection.

Keywords: sexual orientation, development, sex-typed behavior, behavioral observation

On average, gay men tend to be somewhat more feminine, and subsequently examined whether ratings of videos were correlated
lesbians somewhat more masculine, compared with heterosexual with self-reports of gender nonconformity, the relation between
people of their own sex. That is, homosexual people tend to be childhood and adult gender nonconformity, and whether the ex-
relatively gender nonconforming. These tendencies do not occur pression of gender nonconformity was related to reports of past
for all possible gender-related traits (e.g., Bailey, Gaulin, Agyei, & social treatment.
Gladue, 1994), but they do occur for a range of behaviors, feelings,
and interests (e.g., Ambady, Hallahan, & Conner, 1999; Bailey,
2003; Bailey, Dunne, & Martin, 2000; Bailey, Finkel, Black-
Masculinity, Femininity, Gender Conformity, and Gender
welder, & Bailey, 1995; Gaudio, 1994; Johnson, Gill, Reichman, Nonconformity: Concepts and Terminology
& Tassinary, 2007; Lippa, 2000, 2002). Psychological study of masculinity and femininity has a long,
Many studies have also shown large differences between homo- controversial history (see Lippa, 2005a, Chapter 2). Some of this
sexual and heterosexual people in their memories of how gender controversy concerns terminology, and some of the controversy is
nonconforming they were as children (see the review by Bailey & conceptual. A brief review is useful for understanding both the
Zucker, 1995). However, retrospective reports may be subject to conceptual background and the methodological limitations of our
various biases, and this has provoked skepticism regarding the study.
validity of the association between sexual orientation and child- Suppose that one assessed men and women on all psychological
hood gender nonconformity (e.g., Gottschalk, 2003). We investi- traits that differ, on average, between the sexes. If one subse-
gated the difference in childhood gender nonconformity between quently conducted factor analyses of these traits within each sex,
homosexual and heterosexual people using a method not suscep- one would not obtain a general factor of Masculinity–Femininity.
tible to memory bias: blind ratings of childhood home videos. We This is so because some traits that differ between the sexes are
uncorrelated within the sexes. This fact motivated interest in
two-factor conceptions of masculinity and femininity by Bem
Gerulf Rieger, Joan A. W. Linsenmeier, and J. Michael Bailey, Depart- (1974) and others (e.g., Spence & Helmreich, 1978). As Spence
ment of Psychology, Northwestern University; Lorenz Gygax, Swiss Fed- (1993) and others have noted, however, there is no reason to limit
eral Veterinary Office, Agroscope Tänikon, Ettenhausen, Switzerland. gender to just two dimensions; it is multifactorial.
We thank Shannon Summerfield, Debbie Zutty, Noah Metz, Stella
Masculine and feminine are, of course, words commonly used
Fayman, Sarah Kate McGowan, Tamar Carmel, Tom Grant, Ashley Ma-
son, and Evan Drake for their assistance in this research.
not only in the psychological literature but also in everyday dis-
Correspondence concerning this article should be addressed to Gerulf course. Although each term represents a complex composite of
Rieger, Department of Psychology, Northwestern University, 2029 Sheri- characteristics and those characteristics most closely associated
dan Road, Swift Hall #102, Evanston IL 60208. E-mail: with masculinity may not always be exact opposites of those
gerulf@northwestern.edu associated with femininity, people tend to employ these terms in a

46
SPECIAL SECTION: SEXUAL ORIENTATION AND GENDER NONCONFORMITY 47

way that suggests a single dimension. That is, when laypeople rate formerly gender-nonconforming boys but none of the typical boys
others on the traits masculine and feminine, they tend to use these reported a bisexual or homosexual orientation. Similar findings
terms as opposite poles of the same dimension (e.g., Pedhazur & were recently reported for 25 very gender-nonconforming girls,
Tetenbaum, 1979). In the study described herein, we asked some who were more likely than the typical girls to report a bisexual or
of our participants to compare other participants to the average homosexual orientation in adulthood (Drummond, Peterson-
person of the same sex using the terms more masculine and more Badali, & Zucker, 2008). However, even though prospective stud-
feminine as endpoints on a single dimension. Ratings were based ies have shown that children with objectively pronounced gender-
on 10- to 30-s video excerpts. Thus, we did not obtain sophisti- nonconforming behaviors were likely to become homosexual, it is
cated assessments of targets’ personality based on rich samples of unclear how much the findings generalize to the development of
their attitudes, interests, and behaviors but rather measured quick most homosexual people. Children with extreme gender noncon-
impressions based on very brief behavior samples. Our index of formity are rare, and it is at least conceivable that most homosex-
interrater reliability (see the Results section) suggests that raters’ ual people showed minimal gender nonconformity during child-
quick impressions had a high degree of agreement. hood.
In this article, rather than using the words masculine and femi- Although both retrospective and prospective studies have limi-
nine, we often use the terms gender nonconforming and gender tations, they converge on the conclusion that there is a strong link
conforming. By gender nonconforming we mean men and boys between sexual orientation and childhood gender nonconformity.
who are relatively feminine or relatively unmasculine compared However, it is clearly desirable to investigate this association in
with other men and boys, as well as women and girls who are ways that avoid the aforementioned limitations. Home videos of
relatively masculine or relatively unfeminine compared with other prehomosexual children (i.e., children who later identified as ho-
women and girls. By gender conforming we mean men and boys mosexual in adulthood) provide a relatively objective behavior
who are relatively masculine or relatively unfeminine compared sample that can be rated blindly with respect to adult sexual
with other men and boys, as well as women and girls who are orientation. Such videos can provide a measure of childhood
relatively feminine or relatively unmasculine compared with other gender nonconformity not subject to the recall bias that may affect
women and girls. By our usage, then, an individual who is ex- retrospective studies based on self-report, and they can also be
tremely gender conforming, like one who is extremely gender obtained from a more representative sample compared with the
nonconforming, is unusual for his or her sex. Although we write prospective studies of very gender-nonconforming children.
most often about the origins, development, and measurement of
gender nonconformity, our study is just as much about the origins, Consistency Among Measures of Gender Nonconformity
development, and measurement of gender conformity.
Homosexual people not only tend to report more gender non-
Childhood Gender Nonconformity and Sexual Orientation conformity in their childhood; they also tend to be more gender
nonconforming as adults in some respects, including their self-
Compared with heterosexual men, homosexual men tend to concepts (Bailey et al., 1995, 2000), their occupational and recre-
recall that during childhood they were more feminine with respect ational interests (see the review by Lippa, 2005b), their move-
to their interests, behavior, and self-concepts (see the review by ments (Ambady et al., 1999; Johnson et al., 2007), and their speech
Bailey & Zucker, 1995). Analogously, homosexual women recall patterns (e.g., Gaudio, 1994). However, the degree to which child-
that they were relatively masculine girls. These differences are hood gender nonconformity predicts adult gender nonconformity
large; effect sizes (Cohen’s ds) are approximately 1.3 for men and is unclear, and in the current study we tried to estimate this
1.0 for women. Specific childhood characteristics that tend to be relationship.
gender nonconforming in homosexual people include frequency of A second question involving consistency of measures has to do
rough play, toy and activity preference, clothing preference, and with the use of self-reports versus ratings by observers. Most
role-playing behavior. Retrospective studies have often defined studies of gender nonconformity and sexual orientation have used
childhood as before puberty, so their results suggest that these self-report measures of gender nonconformity. Because behaviors
childhood differences in gender nonconformity emerge before displayed in videos can be rated by others, they allow investigation
individuals have developed an adult sexual orientation. However, into the reliability and validity of such self-reports. This, too, was
an important limitation of these studies is their reliance on self- addressed in our study.
report. Memories can be biased or distorted. For example, homo-
sexual adults may be more likely than heterosexual adults to Social Correlates of Childhood Gender Nonconformity
acknowledge past gender nonconformity. It is thus uncertain how
much researchers can trust these recollections of childhood behav- Reactions of others to gender-nonconforming children tend to
iors (Gottschalk, 2003; but see Zucker, 2005). be negative. Parents do not generally encourage gender noncon-
A few prospective studies have also assessed the link between formity in their children (e.g., Caldera, Huston, & O’Brien, 1989;
childhood gender nonconformity and sexual orientation (see Jacklin, DiPietro, & Maccoby, 1984; Langlois & Downs, 1980;
Bailey & Zucker, 1995). The most extensive study to date fol- Tauber, 1979). Peers are especially discouraging of gender-
lowed two groups of boys into adulthood (Green, 1987). One nonconforming behaviors; in particular, boys discourage such be-
group included 66 boys who showed pronounced gender- havior in other boys (e.g., Carter & McCloskey, 1983; Fagot,
nonconforming behaviors, such as cross-dressing. The other group 1977, 1985; Langlois & Downs, 1980). Consequently, homosexual
consisted of 56 typical boys. For two thirds of each group, sexual men who were more gender-nonconforming boys recall more
orientation was assessed in or after adolescence. About 75% of the rejection by their parents and peers (Landolt et al., 2004) and recall
48 RIEGER, LINSENMEIER, GYGAX, AND BAILEY

being treated more negatively by their peers (Saghir & Robins, Mean ages were 28.6 years (SD ⫽ 7.9), 23.8 years (SD ⫽ 5.3),
1973). Perhaps because of this social pressure, many gender- 25.0 years (SD ⫽ 5.9), and 24.6 years (SD ⫽ 7.4), respectively.
nonconforming boys who become homosexual men seem to sup- Ages did not significantly differ between groups ( p ⫽ .13; ␤ ⫽
press or lose their gender nonconformity as they grow up (Bell, ⫺.16, confidence interval [CI] ⫽ ⫺.37 to .05). (The effect size
Weinberg, & Hammersmith, 1981; Harry, 1982). beta is, throughout this article, the standardized regression coeffi-
Childhood rejection may have long-term psychological conse- cient as derived from multiple regression analyses. Also given is
quences. For example, family problems because of childhood the 95% CI of the effect. We estimated all group differences by
gender nonconformity may persist into adulthood; these include calculating the main effects of sex, sexual orientation, and their
problematic relationships with parents, particularly with one’s interaction effect—the combination of sex and sexual orientation.)
father (Beard & Bakeman, 2000; Freund & Blanchard, 1983). Eighty-two percent of targets were Caucasian, and this percentage
Other studies found negative correlations of childhood gender did not significantly differ across groups, ␹2(3, N ⫽ 90) ⫽ 2.8, p ⫽
nonconformity with psychological well-being in adulthood (Harry, .42. No other demographic information besides age and ethnicity
1983a, 1983b; Skidmore, Linsenmeier, & Bailey, 2006; Weinrich was collected.
et al., 1992). Landolt et al. (2004) suggested that the link between Targets as children. Some targets provided home videos from
childhood gender nonconformity and psychological well-being in only one age period, but others had videos from several ages. For
adulthood is mediated by childhood rejection by parents and peers. each target we selected one childhood video clip from each age
The present study also explored the correlations of recalled child- represented, which resulted in one to six clips per target (details of
hood rejection with measures of gender nonconformity based on this selection process are given below). The average number of
both observer ratings of videos and self-reports. clips per child was 2.4 (SD ⫽ 1.5), 2.3 (SD ⫽ 1.5), 2.6 (SD ⫽ 1.6),
and 2.3 (SD ⫽ 1.2) for homosexual men, homosexual women,
The Present Study heterosexual men, and heterosexual women, respectively. Mean
ages represented in the clips were 4.6 years (SD ⫽ 3.6), 5.4 years
We recruited homosexual and heterosexual adults who had (SD ⫽ 3.9), 5.5 years (SD ⫽ 3.6), and 5.5 years (SD ⫽ 3.6),
videos from their childhood (i.e., from ages 0 to 15 years), and we respectively. These ages were not significantly different between
also videotaped them during an interview. We then asked partic- groups ( p ⫽ .44; ␤ ⫽ .09, CI ⫽ ⫺.13 to .30).
ipants to recall their degree of gender nonconformity during child- Raters. Twenty homosexual men and 20 homosexual women,
hood and during adulthood and to recall treatment by others during who were recruited via the same media as target participants, and
childhood. Subsequently, we recruited homosexual and heterosex- 20 heterosexual men and 22 heterosexual women, who were un-
ual adults to rate the degree of gender nonconformity from both the dergraduate students in an introductory psychology class, rated
childhood and the adult video clips. We focused on four main video clips from targets. Mean ages of raters were 28.3 years
questions. First, do videos reveal increased gender nonconformity (SD ⫽ 12.5), 26.4 years (SD ⫽ 7.1), 18.9 years (SD ⫽ 1.2), and
in the prehomosexual children? Second, does gender nonconfor- 18.4 years (SD ⫽ 0.7), respectively. Homosexual raters were older
mity persist from childhood into adulthood? Third, do self-reports than the heterosexual raters ( p ⬍ .0001; ␤ ⫽ .53, CI ⫽ .38 to .69).
of childhood and adult gender nonconformity correspond with The proportion of Caucasians was higher among women (81%)
observer ratings? Finally, do individuals who are more gender than among men (50%), ␹2(1, N ⫽ 82) ⫽ 8.4, p ⫽ .004. Because
nonconforming, both in their childhood and in adulthood, recall judgments of the four groups of raters correlated highly with each
that others treated them more negatively as children? other and barely differed (see below), differences in age and
ethnicity did not seem to have important influences on ratings.
Method
Recruitment and Participants Self-Report Measures
Targets as adults. We placed advertisements in an urban al- Gender nonconformity. Childhood gender nonconformity was
ternative newspaper (The Chicago Reader) and an alternative Web assessed with the Childhood Gender Nonconformity Scale (Bailey
site (chicago.craigslist.org), which are read by both heterosexual et al., 1995), with 7 statements for each sex. Sample items are “As
and homosexual people in the greater Chicago area. Advertise- a child I often felt that I had more in common with [the opposite
ments were placed in the classified sections under miscellaneous sex]” and “I preferred playing with girls rather than boys.” Adult
job opportunities. Four different advertisements were placed, one gender nonconformity was measured with the Continuous Gender
for gay men, one for lesbians, one for straight men, and one for Identity Scale (Bailey et al., 1995), with 10 items for each sex.
straight women. For example, one ad read, “Gay men wanted who Examples include “I feel as if part of me is male and part of me is
have home videos from their childhood and are willing to be female” and “It would be fun to go to a costume party dressed as
interviewed for a study on child development and sexual orienta- [the opposite sex].” Participants rated their agreement with each
tion.” Participants who contacted the lab were encouraged to bring item using a 7-point scale ranging from strongly disagree to
into the lab all available home videos of their childhood, regardless strongly agree. Item reliability (Cronbach’s alpha) of the child-
of their context and number of recordings. They were also told that hood scale was .89 and .91 for men and women, respectively. For
we would like to videotape them during a casual interview to have the adult scale the respective numbers were .78 and .83. Responses
some record of how they were as adults. Sexual orientation was to different items were averaged, and higher scores represented
assessed by self-identification as gay or straight. greater gender nonconformity. Items differed somewhat between
Recruited participants included 21 homosexual men, 20 homo- the sexes. Hence, targets’ scores for childhood and adult gender
sexual women, 23 heterosexual men, and 26 heterosexual women. nonconformity were standardized within sex to allow correlational
SPECIAL SECTION: SEXUAL ORIENTATION AND GENDER NONCONFORMITY 49

analyses across the sexes. Similar scales of childhood and adult activities, this would contribute to orientation differences in gender
gender nonconformity have demonstrated substantial sexual ori- nonconformity ratings. However, behavioral style (i.e., manner of
entation differences (see Bailey et al., 2000). Please see the Ap- performing a given activity) could affect ratings, too, and we
pendix for a complete list of scale items. attempted to distinguish these effects from effects of activity
Past parental and peer treatment. A 40-item version of the differences in our analyses.
Recollection of Early Childrearing Scale (Ross, Campbell, & Editing videos. Childhood videos came in different formats—
Clayer, 1982), which is the English version of the Swedish EMBU for example, VHS tapes, Video 8 tapes, and DVDs of original
(Egna Minnen Beträffande Uppfostran; see Gerlsma, Emmelkamp, 8-mm films. All selected footage was converted into digital files
& Arrindell, 1990), was used to assess past rejection versus ac- via the Apple program iMovie.
ceptance by mothers and by fathers. Sample statements are “My Rating videos. Equal numbers of raters from each group (het-
[father/mother] would punish me hard, even for little things” and erosexual and homosexual men and women) watched each of four
“I think that my [father/mother] wished I had been different in video sets (female child targets, male child targets, female adult
some way.” Ten items of the Mother–Father–Peer Scale were used targets, and male adult targets). Each rater first watched children of
to assess former peer reactions (Epstein, 1983). Sample statements one sex and then watched adults of the other sex. Within each set,
are “When I was a child, other children picked on me and teased videos were presented in random order. Raters indicated their
me” and “Other children were often unfair to me.” These items impression of each target’s appearance or behavior on a 7-point
were rated on a 7-point scale, and higher numbers indicated more scale. For example, after each video clip of a boy, the rater was
rejection. Across targets, item reliability (Cronbach’s alpha) ex- asked to indicate “whether this boy appeared or behaved in
ceeded .90 for all three scales. These scales have been used in a more masculine or feminine way.” The score of 1 was
other research assessing correlations between self-reported child- labeled more masculine, 4 was labeled average, and 7 was
hood gender nonconformity and recalled childhood treatment labeled more feminine. A reversed scale was used for ratings
(Landolt et al., 2004). of women and girls. Thus, for both sexes 7 represented maxi-
mum gender nonconformity.
Raters received no explicit information about the targets’ sexual
Procedure
orientation. We asked them to compare each target to the average
Target interviews. We videotaped targets for 20 min, during a person of the same sex and age. We encouraged raters to rate each
casual interview. Targets sat in a chair in front of a digital camera, video independently, even if they recognized the same child in
which captured their entire body. For analyses, we used targets’ more than one video. Each rater was limited to 1 hr of rating
responses to a question asking them to describe winters in the videos.
Midwest. We used the first complete sentences produced by the Descriptions and ratings of childhood activities. Two research
targets within the first 30 s of their answer. Selected videos of all assistants, who had no explicit information about the targets’
targets were at least 10 s long. Previous research indicates that sexual orientation, independently watched the selected childhood
from such short video clips raters can, on average, very reliably videos and wrote down descriptions of each child’s activity. They
distinguish heterosexual from homosexual targets (Ambady et al., were instructed to focus on the activity and ignore physical ap-
1999). Thus, for the purpose of the current study, 10-s videos pearance, mannerisms, and voice patterns. They then compared
seemed a sufficient minimum length. After the interview, targets their transcripts; with one exception, they had described identical
completed the questionnaire battery. activities for each clip. The exception was resolved by discussion
Selecting childhood videos. Targets watched their childhood of the two raters. Examples of activities performed by boys are
videos together with the experimenter and identified themselves on “5-year-old boy imitating fighting moves with small swords” and
the footage. For each age for which a target had a video sample, we “11-year-old boy dancing to 80’s music, female vocalist.” Exam-
selected the first scene in which the target was seen for at least ples of activities performed by girls are “4-year-old girl in tutu
10 s. When possible, we selected clips in which the target child outfit, demonstrating ballet moves” and “7-year-old girl putting
was alone and spoke or made some other sound. We excerpted a together electric train.”
maximum of 30 s. We then forwarded to the next available scene Six graduate students read the written activity descriptions and
in which the child was least a year older and excerpted another 10 rated them for gender nonconformity using a 7-point scale. Ratings
to 30 s of material, and so forth until all included ages (0 to 15 were completed for one sex of target and then for the other. The
years) were covered. order of sex was random, and, within each sex activity, descrip-
The context of the selected clips varied. The most popular tions were presented in random order.
scenes were simple close-ups of the child, followed by holiday and
birthday parties and family trips to amusement parks. Specific
Results
activities of children also varied across clips, including singing,
dancing, acting, playing, walking, biking, and playing various Interrater Consistency
sports. The activities depicted in the video, specifically whether
they were activities more commonly performed by boys or by Within each Target Sex ⫻ Age (i.e., childhood vs. adulthood)
girls, seemed likely to affect ratings of gender nonconformity. This combination, video ratings of the four groups (heterosexual and
was expected because interest in gender-typical activities repre- homosexual men and women) were highly correlated, with all
sents an important aspect of gender conformity and nonconformity correlations exceeding .80. Across all raters, interrater reliability
(Lippa, 2005b). Thus, if, for example, prehomosexual boys were (Cronbach’s alpha) exceeded .90 within each combination. For
more likely than preheterosexual boys to engage in female-typical childhood gender nonconformity, the four groups of raters did not
50 RIEGER, LINSENMEIER, GYGAX, AND BAILEY

differ significantly in how closely their ratings were related to preheterosexual children were rated as somewhat less gender non-
target’s sexual orientation ( p ⫽ .56; ␤ ⫽ ⫺.02, CI ⫽ ⫺.09 to .05). conforming than average (a score of 4) across most ages, while
For adulthood gender nonconformity, judgments of both male and prehomosexual children increased in their gender nonconformity
female homosexual raters were more closely linked to the targets’ with age. Thus, as children grew older, the sexual orientation
sexual orientation than were judgments of male and female het- difference in gender nonconformity became more pronounced.
erosexual raters ( p ⫽ .008; ␤ ⫽ .12, CI ⫽ .09 to .15). However, This interaction did not significantly differ between the sexes ( p ⫽
this effect was relatively small, and the correlations between .57; ␤ ⫽ ⫺.04, CI ⫽ ⫺.17 to .09). Note in Figure 1 that for both
groups of raters were strong (according to the criteria suggested by sexes the prehomosexual and preheterosexual groups’ CIs sepa-
Cohen, 1988). Thus, in subsequent analyses we used each target’s rated between ages 3 and 4, and thus the sexual orientation
average score (averaged across all raters) for childhood and adult effect became significant during that interval. The interaction
gender nonconformity. between sexual orientation and the curvilinear effect of age was
Regarding descriptions of childhood activities, interrater reli- not significant.
ability (Cronbach’s alpha) of gender nonconformity ratings ex- The finding that preheterosexual children were somewhat less
ceeded .88 for both male and female children. For analyses, we gender nonconforming than average needs elaboration (see Figure
used each target’s average score (across raters) for gender non- 1). Ratings were done within each target sex, and raters were asked
conformity of childhood activities. to compare each child to an average child of his or her age and sex.
The fact that the average child was very likely a preheterosexual
Are Prehomosexual Children More Gender child might explain why ratings of preheterosexual children were
Nonconforming? close to the average on the scale. Nevertheless, both prehetero-
We compared the gender nonconformity ratings of videos of sexual boys and preheterosexual girls tended to be less gender
preheterosexual and prehomosexual children. Some children were nonconforming (or more gender conforming) than the average
repeatedly rated at different ages, and we accounted for these score of 4, as suggested by one-sample t tests ( p ⫽ .01, d ⫽ ⫺0.8,
repeated judgments by using a mixed-effects model that treated CI ⫽ ⫺1.0 to ⫺0.5; and p ⬍ .0001, d ⫽ ⫺1.2, CI ⫽ ⫺1.4 to ⫺0.9,
children as a random effect (e.g., Zar, 1999). Other predictors of respectively).
childhood gender nonconformity included targets’ sexual orienta- Figure 1 also depicts one outlying data point, which came from
tion and sex. We also tested for a curvilinear effect of age to the oldest prehomosexual girl, who was also relatively gender
examine possible nonlinear effects. Finally, we included all inter- conforming. This data point had high residual and leverage value
actions between sex, sexual orientation, and age (and its curvilin- and relatively high influence on the shape of the estimated inter-
ear effect) as predictors to test whether any difference in gender action effects between sexual orientation and age. To avoid unre-
nonconformity between prehomosexual and preheterosexual chil- liable estimates, we have excluded this data point from analyses
dren depended on sex, age, or the combination of sex and age. (see Judd & McClelland, 1989). The significance of the main
For both sexes, prehomosexual children were significantly more effect of sexual orientation was barely influenced by inclusion or
gender nonconforming than preheterosexual children ( p ⬍ .0001; exclusion of this data point ( p ⬍ .0001, ␤ ⫽ .41, CI ⫽ .21 to .54,
␤ ⫽ .38, CI ⫽ .23 to .53). This effect was moderated by an vs. p ⬍ .0001, ␤ ⫽ .38, CI ⫽ .23 to .53, respectively).
interaction between sexual orientation and the linear effect of age We finally investigated whether the relationship between adult
( p ⬍ .0001; ␤ ⫽ .32, CI ⫽ .20 to .45). Figure 1 shows that sexual orientation and observer-rated childhood gender noncon-

7
Gender Nonconformity
Observed

0 5 10 15 0 5 10 15
Age of Boys Age of Girls

Figure 1. Observer ratings of gender nonconformity from childhood videos of 23 heterosexual and 21
homosexual male targets and 26 heterosexual and 20 homosexual female targets. Target scores were averaged
across raters. The higher lines represent homosexual targets. Dashed lines are the curves’ 95% confidence
intervals. The asterisk indicates an outlying data point from a homosexual woman that was excluded from
analyses. The x-axis represents children’s age. On the y-axis, 1 is the least gender-nonconforming score, and 7
is the most gender-nonconforming score.
SPECIAL SECTION: SEXUAL ORIENTATION AND GENDER NONCONFORMITY 51

formity was influenced by the children’s activities (as opposed to lated with sexual orientation and with each other. To simplify
other aspects of the video, e.g., the manner in which the children analyses, we used each child’s average gender nonconformity
moved and spoke). Descriptions of childhood activities were rated score from videos across all available ages. This strategy is useful
as significantly more gender nonconforming in homosexual targets to the extent that gender nonconformity is stable across the ages.
( p ⬍ .0001; ␤ ⫽ .36, CI ⫽ .22 to .49), regardless of their sex ( p ⫽ In this case, averaging ratings across ages yielded a more reliable
.17; ␤ ⫽ .03, ⫺.10 to .17). We then computed the same mixed- estimate compared with ratings at a single age. To validate this
effects model as described earlier, except that we now controlled decision, we examined the degree of correlation between gender
for the degree of gender nonconformity in the children’s activities. nonconformity across three age categories: 0 to 3 years, 4 to 6
In this model, the relationship between sexual orientation and years, and 7 to 15 years. Within each age category, gender non-
observer-rated childhood gender nonconformity was somewhat conformity was averaged within children if they were shown in
smaller than in the original model ( p ⫽ .0002, ␤ ⫽ .27, CI ⫽ .13 more than one clip. Not every target contributed a video clip from
to .41, vs. p ⬍ .0001, ␤ ⫽ .38, CI ⫽ .23 to .53, respectively). In each age category. Gender nonconformity was positively, although
addition, gender nonconformity of described activities was posi- not always significantly, correlated across age categories; p ⫽ .23,
tively related to observer-rated childhood gender nonconformity r(27) ⫽ .23 (CI ⫽ ⫺.22 to .68) for the first and second category;
( p ⬍ .0001; ␤ ⫽ .33, CI ⫽ .22 to .44). Differences between
p ⫽ .01, r(15) ⫽ .62 (CI ⫽ .26 to .97) for the first and third
prehomosexual and preheterosexual children in the gender non-
category; and p ⫽ .003, r(17) ⫽ .67 (CI ⫽ .35 to .98) for the
conformity of specific activities did thus contribute to the finding
second and third category. It seemed that children displayed de-
that prehomosexual children were more gender nonconforming
grees of gender nonconformity that persisted, on average, over
than preheterosexual children. Nevertheless, the orientation differ-
time. Thus, estimating an overall child gender nonconformity
ences in rated gender nonconformity did not appear to reflect
score by averaging gender nonconformity across ages seemed
exclusively those activity differences and might also have de-
pended on more stylistic aspects of behavior, such as motor be- reasonable.
havior. Next we correlated sexual orientation with observer-rated
and self-reported childhood and adult gender nonconformity.
On all measures, homosexual targets were more gender non-
How Closely Related Are Different Measures of Gender
conforming than heterosexual targets (see Figure 2). None of
Nonconformity?
these sexual orientation effects significantly differed between
We investigated the degree to which observer ratings and self- the sexes (significance levels between .15 and .78, betas be-
reports of childhood and adult gender nonconformity were corre- tween ⫺.03 and ⫺.12). This suggests that increased gender

7
Gender Nonconformity
Male

1 1.0 (.45)* 1.5 (.61)*** 2.3 (.74)*** 0.9 (.42)*

Childhood Adulthood Childhood Self- Adulthood Self-


Observed Observed Report Report
7
Gender Nonconformity
Female

1 1.2 (.53)** 1.6 (.61)*** 1.1 (.49)** 0.8 (.36)*


hetero- homo- hetero- homo- hetero- homo- hetero- homo-
sexual sexual sexual sexual

Figure 2. Observer ratings and self-reports of the gender nonconformity of 23 heterosexual and 21 homosexual
male targets and 26 heterosexual and 20 homosexual female targets. Points represent the scores of individual
targets, averaged across raters. Lines are the means’ 95% confidence intervals. On the y-axis, 1 is the least
gender-nonconforming and 7 the most gender-nonconforming score. Numbers are effect sizes expressed as
Cohen’s ds and, in parentheses, as correlation coefficients. *p ⬍ .05. **p ⬍ .001. ***p ⬍ .0001.
52 RIEGER, LINSENMEIER, GYGAX, AND BAILEY

nonconformity of homosexual targets is general across age and exceeded zero, the expected chance value ( p ⫽ .05, d ⫽ 1.1, CI ⫽
type of measure. 0.0 to 2.0, for homosexual men; p ⫽ .11, d ⫽ 0.8, CI ⫽ ⫺0.3 to
We note in Figure 2 the high variability in some of the groups, 1.9, for homosexual women; p ⫽ .0007, d ⫽ 3.1, CI ⫽ 2.0 to 4.3,
which shows that not all homosexual individuals were gender for heterosexual men; and p ⫽ .03, d ⫽ 1.2, CI ⫽ 0.1 to 2.2, for
nonconforming, nor were all heterosexual individuals gender con- homosexual women). Furthermore, there were no significant group
forming. For two measures, significant group differences in vari- differences for these correlations ( p ⫽ .52; ␤ ⫽ .14, CI ⫽ ⫺.29 to
ability were found. A Levene test for unequal variance (which .56), and across all groups the average correlation, .32, also sig-
compares the magnitude of absolute residuals between groups) nificantly exceeded zero ( p ⬍ .0001; d ⫽ 1.3, CI ⫽ 0.8 to 1.7).
indicated that for observer-rated adult gender nonconformity, ho- These additional results support the hypothesis that within each
mosexual targets of either sex showed more variability than did sex and sexual orientation, measures of gender nonconformity tend
heterosexual targets of either sex ( p ⫽ .0003; ␤ ⫽ .37, CI ⫽ .18 to be correlated.
to .56). Furthermore, for self-reported childhood gender noncon- Note in Tables 2 and 3 that the magnitude of some correlations
formity, heterosexual male targets showed less variability than all differed considerably between the groups. For example, the cor-
other groups ( p ⫽ .02; ␤ ⫽ ⫺.24, CI ⫽ ⫺.33 to ⫺.14). Across all relations of observer-rated childhood and observer-rated adult gen-
measures (i.e., by taking their average), we found no significant der nonconformity were stronger for heterosexual targets than for
group differences in variability ( p ⫽ .59; ␤ ⫽ ⫺.06, CI ⫽ ⫺.43 homosexual targets. However, a subsequent regression analysis
to .32). Still, the aforementioned results may indicate that, for did not reveal a significant sexual orientation difference for this
some indexes of gender nonconformity, variation could be related effect ( p ⫽ .25; ␤ ⫽ ⫺.11, CI ⫽ ⫺.33 to .86). The relationship of
to sex and sexual orientation. observer-rated childhood gender nonconformity and self-reported
We then computed correlations among the various measures of adult gender nonconformity was, though, stronger for male targets
childhood and adult gender nonconformity. To give our analyses than for female targets ( p ⫽ .02; ␤ ⫽ .28, CI ⫽ .07 to .53). In
more power, we first computed correlations between measures of particular, this relationship was positive and significant for male
gender nonconformity across all targets. Specifically, we com- targets ( p ⫽ .004; ␤ ⫽ .56, CI ⫽ .25 to .87), while it was
puted the partial correlations controlling for both sexual orienta- nonsignificant for female targets ( p ⫽ .95; ␤ ⫽ .01, CI ⫽ ⫺.35 to
tion and sex (both converted into a numeric 2-point scale). All of .34). The meaning of this sex difference is unclear, however. For
the partial correlations were significant (see Table 1, below the all other relationships between measures of gender nonconformity,
diagonal). Thus, for both sexes and sexual orientations, individuals their moderations by the targets’ sex and sexual orientation (and
who were relatively gender-nonconforming children tended to their combination) were insignificant (significance levels between
become relatively gender-nonconforming adults, and those who .13 and .98, betas between ⫺.01 and .18). Hence, for the majority
rated themselves as relatively gender nonconforming were rated by of the correlations as shown in Table 1, heterosexual and homo-
others as being relatively gender nonconforming as well. Table 1 sexual men and women did not significantly differ.
also shows the first-order correlations between measures of gender
nonconformity across all targets. These were generally stronger How Is Gender Nonconformity Related to Past Treatment
than the respective partial correlations because sexual orientation by Parents and Peers?
had an additional influence on their magnitude (homosexual tar-
gets were generally more gender nonconforming; see Figure 2). For each target, we computed three average scores, representing
We then computed correlations between measures of gender degree of childhood rejection by father, mother, and peers. Across
nonconformity separately for heterosexual and homosexual men all targets, parental and peer rejection were significantly correlated
and women. As shown in Tables 2 and 3, in each group the large ( ps ⬍ .01, correlations between .35 and .41). Next we computed
majority of these correlations were positive, although many of correlations of parental and peer rejection with self-rated and
them were not significant (because of a loss of power). However, observer-rated childhood and adult gender nonconformity. Only 4
the average of these correlations was positive in all groups: .32 out of 12 effects were significant (see Table 4). With respect to
(SD ⫽ .30), .24 (SD ⫽ .30), .46 (SD ⫽ .15), and .25 (SD ⫽ .21) parental rejection, we found only one significant correlation out of
for all homosexual men, homosexual women, heterosexual men, six. Nevertheless, past rejections from peers tended to correlate
and heterosexual women, respectively. Single-sample t tests sug- significantly with both childhood and adult gender nonconformity.
gested that in most groups these average correlations reliably If this peer rejection experienced by more sex-atypical targets was,

Table 1
Intercorrelations Among Measures of Gender Nonconformity Across All Targets (N ⫽ 90).

Measure Childhood observed Adulthood observed Childhood self-report Adulthood self-report

Childhood observed .48*** (.30, .67) .43*** (.24, .63) .40** (.21, .60)
Adulthood observed .25* (.05, .46) .49*** (.30, .68) .47*** (.29, .66)
Childhood self-report .20* (.01, .39) .22* (.01, .43) .64*** (.47, .80)
Adulthood self-report .25* (.04, .46) .31* (.11, .51) .56*** (.38, .73)

Note. Numbers above the diagonal are simple correlations. Numbers below the diagonal are partial correlations, after the effects of sexual orientation and
sex were partialled out. Numbers in parentheses are 95% confidence intervals of the correlation coefficients.
*
p ⬍ .05. ** p ⬍ .001. *** p ⬍ .0001.
SPECIAL SECTION: SEXUAL ORIENTATION AND GENDER NONCONFORMITY 53

Table 2
Intercorrelations Among Measures of Gender Nonconformity Across Heterosexual Male Targets (n ⫽ 23) and Homosexual Male
Targets (n ⫽ 21)

Measure Childhood observed Adulthood observed Childhood self-report Adulthood self-report

Childhood observed .57* (.20, .94) .31 (⫺.12, .74) .53* (.15, .92)
Adulthood observed .08 (⫺.40, .56) .26 (⫺.18, .70) .64* (.29, .99)
Childhood self-report .60* (.22, .98) ⫺.10 (⫺.57, .38) .46* (.06, .86)
Adulthood self-report .51* (.09, .92) .18 (⫺.29, .65) .62* (.25, .99)

Note. Heterosexual male targets are shown above the diagonal, and homosexual male targets are shown below the diagonal. Numbers in parentheses are
95% confidence intervals of the correlation coefficients.
*
p ⬍ .05.

in fact, elicited by their gender nonconformity, it would suggest in sexual orientation as adults are already influential in child-
that gender nonconformity persisted into adulthood despite past hood and contribute to a corresponding difference in gender
negative social reactions. nonconformity.
Measures of gender nonconformity tended to be correlated The current study yielded three other main findings. More
across all targets (see Table 1). Thus, we subsequently performed gender-nonconforming children tended to become more gender-
multiple regressions, using the targets’ average gender nonconfor- nonconforming adults. Further, self-report of gender nonconfor-
mity score across all four measures as a predictor of parental mity was generally confirmed by observer ratings. In addition,
rejection, maternal rejection, and peer rejection. Other variables gender-nonconforming individuals did not report that their parents
included as predictors were the targets’ sex and sexual orientation or peers were especially tolerant; to the contrary, they recalled
and their interactions with average gender nonconformity. With increased childhood rejection, probably especially from peers,
these interactions, we tested whether the relationship between compared with gender-conforming individuals.
gender nonconformity and rejection depended on the target’s sex,
sexual orientation, or their combination. Targets who were more Convergence Among Measures of Gender Nonconformity
gender nonconforming tended to report more paternal, maternal, and Sexual Orientation
and peer rejection ( p ⫽ .09, ␤ ⫽ .25, CI ⫽ ⫺.04 to .55; p ⫽ .04,
␤ ⫽ .30, CI ⫽ .01 to .59; and p ⫽ .02, ␤ ⫽ .34, CI ⫽ .06 to .61, In the current study, most measures of gender nonconformity
respectively). These effects were not significantly moderated by were significantly correlated with sexual orientation and each
targets’ sex, sexual orientation, or their combination (significance other (see Figure 2 and Table 1). Overall, these correlations
levels between .20 and .93, betas between ⫺.19 and .12). support the validity of all measures in the sense that they seemed
to measure a general factor of gender nonconformity, which was
more common in homosexual people. Furthermore, relatively high
Discussion
levels of gender nonconformity during childhood tended to predict
Raters who were blind to the adult sexual orientation of targets high levels of gender nonconformity during adulthood, suggesting
evaluated brief samples of their childhood behavior. Those targets that gender nonconformity can be a stable trait. We admit that one
who, as adults, identified themselves as homosexual were judged of these correlations (between observer-rated childhood gender
to be more gender nonconforming as children. Thus, our analysis nonconformity and self-reported adulthood gender nonconformity)
of childhood videos from homosexual and heterosexual individu- was only significant in men, not in women. However, all other
als confirmed the findings of many studies based on retrospective correlations between measures of childhood and adult gender
self-reports: Children who subsequently identified as homosexual nonconformity were not substantially influenced by the targets’
in adulthood tended to be more gender nonconforming than chil- sex or sexual orientation, suggesting that, in general, the relation
dren who subsequently identified as heterosexual adults. This is between childhood and adult behavior was similar for all groups.
consistent with the possibility that factors causing people to differ Finally, the finding that self-reports were substantially related to

Table 3
Intercorrelations Among Measures of Gender Nonconformity Across Heterosexual Female Targets (n ⫽ 26) and Homosexual Female
Targets (n ⫽ 20)

Measure Childhood observed Adulthood observed Childhood self-report Adulthood self-report

Childhood observed .43* (.09, .83) .09 (⫺.35, .53) .09 (⫺.36, .53)
Adulthood observed .07 (⫺.42, .55) .11 (⫺.32, .54) .20 (⫺.22, .63)
Childhood self-report ⫺.03 (⫺.52, .47) *
.52 (.10, .95) .55* (.19, .91)
Adulthood self-report ⫺.12 (⫺.61, .37) .40† (⫺.05, .85) .59* (.20, .99)

Note. Heterosexual female targets are shown above the diagonal, and homosexual female targets are shown below the diagonal. Numbers in parentheses
are 95% confidence intervals of the correlation coefficients.

p ⬍ .10. * p ⬍ .05.
54 RIEGER, LINSENMEIER, GYGAX, AND BAILEY

Table 4
Correlations Between Measures of Gender Nonconformity and Measures of Rejection Across All Targets (N ⫽ 90)

Measure Childhood observed Adulthood observed Childhood self-report Adulthood self-report

Paternal rejection .16 (⫺.05, .38) ⫺.08 (⫺.29, .14) .04 (⫺.17, .26) .12 (⫺.09, .34)
Maternal rejection ⫺.01 (⫺.23, .21) ⫺.09 (⫺.30, .12) .10 (⫺.11, .32) .25* (.04, .45)
Peer rejection .32* (.12, .53) .13 (⫺.09, .34) .35* (.11, .32) .29* (.09, .50)

Note. Numbers in parentheses are 95% confidence intervals of the correlation coefficients.
*
p ⬍ .05.

observer ratings supports the use of self-report measures in re- First, as described above, both homosexual and heterosexual sam-
search on gender nonconformity. This is valuable because self- ples varied substantially in their degree of gender nonconformity;
reports are easier to obtain than other people’s ratings of videos. there was considerable overlap in the distributions of the two
Our findings of convergence among various measures of gender groups.
nonconformity, and between these measures and sexual orienta- Second, sexual orientation differences in gender nonconformity
tion, are consistent with those of other researchers. For example, became pronounced between ages 3 and 4 and subsequently in-
Lippa (1991, 1998a, 1998b, 2002, 2005a, 2005b) has shown that creased (see Figure 1). This result mirrors previous findings re-
gender-related occupational and recreational interests show large garding important sex differences (based primarily on studies of
sex differences and are also correlated with other aspects of children who grew up to be heterosexual). Some key differences
personality and behavior often thought to reflect masculinity and between boys and girls in gender-typical behavior, including pref-
femininity, such as sexual orientation. He wrote, erences for gender-typed play activities and for male versus female
playmates, become pronounced between ages 2 and 4 (e.g., Fagot,
Is there in fact a core to masculinity and femininity? My hunch is that 1985; Fagot & Leinbach, 1989; Maccoby, 1990), and the magni-
there is and that it is to be found in gender-related interests (occupa-
tude of these sex differences tended to increase with age (e.g.,
tional preferences, hobbies, and everyday activities), gender-related
appearances (nonverbal mannerisms, dress, grooming), and perhaps
Maccoby, 1998; Trautner, 1992). This general observation pro-
sexuality (sexual orientation). (Lippa, 2005a, p. 67) vides some support for the notion that the causes of gender
nonconformity are also the causes of gender conformity—for
This is the view of masculinity and femininity that motivated our example, the expression of feminine behaviors in prehomosexual
research, and our findings provide further confirmation for it. boys and femininity in preheterosexual girls might reflect the same
processes. We speculate below about the nature of these causes.
Sexual Orientation Differences in Variability of Gender Other similarities between the sexual orientations concern the
Nonconformity computed correlations. We found no substantial sexual orientation
differences for the correlations between measures of gender non-
For one measure, observer-rated adult gender nonconformity, conformity. Thus, the finding that targets who were more gender
homosexual targets showed greater variability than heterosexual nonconforming (or more gender conforming) on one measure
targets (see Figure 2). The pattern of greater variation for homo- tended to be more gender nonconforming (or more gender con-
sexual participants was not found for other measures. Neverthe- forming) on others applied generally to both heterosexual and
less, some previous research has shown that homosexual adults homosexual participants. Similarly, the correlations between past
show greater variability in gender nonconformity than heterosex- childhood rejection and gender nonconformity were not signifi-
uals, when measured both by self-report and by observer ratings cantly different for the sexual orientations. Thus, if gender-
(Lippa, 2005b; Rieger, Linsenmeier, Gygax, Garcia, & Bailey, nonconforming individuals experienced more negative past social
2007). Lippa (2005b) argued that both biological and social factors treatment, this effect may be independent of whether they became
might contribute to this difference; for example, the predominant heterosexual or homosexual.
pressure toward typical gender-role socialization may make some
homosexual individuals behave in a gender-conforming manner. Possible Developmental Implications
At the same time, biological predispositions and counteractive
influences against stereotypical gender-role behaviors within the The present study was not specifically designed to provide
gay and lesbian community may make other individuals behave in evidence concerning the development of gender-nonconforming
a gender-nonconforming manner. Whatever the causes, these re- and gender-conforming behavior, but, arguably, its results are
sults demonstrate that not all homosexual people are more gender more consistent with some possibilities than with others. One
nonconforming than all heterosexual people. In fact, Figure 2 hypothesis is that gender development is caused by socialization
suggests that for some measures a few of the most gender- (e.g., Bem, 1983; Bussey & Bandura, 2004; Lott & Maluso, 2001).
conforming targets were homosexual. Socialization may explain not only differences between men and
women but also gender-related differences within each sex. One
Similarities Between Heterosexual and Homosexual possible socialization mechanism leading to gender nonconformity
is the encouragement of gender-nonconforming behavior. Our
People
results, however, suggest otherwise; gender nonconformity
Results of our study suggest not only differences between peo- emerged early and persisted despite the fact that gender-
ple with different sexual orientations but also certain similarities. nonconforming individuals tended to recall more rejection, prob-
SPECIAL SECTION: SEXUAL ORIENTATION AND GENDER NONCONFORMITY 55

ably especially from peers. This suggests, then, that gender non- reported gender nonconformity might have been underestimated
conformity emerged or persisted not because of encouragement by our method.
but because of other factors. Representativeness of samples. Targets were a self-selected
We emphasize, however, that our argument is speculative. Our group, which might have resulted in some sampling bias. For
relevant data are based on a crude self-report measure of childhood example, the majority of homosexual people were openly homo-
rejection (i.e., we used measures of general rejection by parents sexual. It is possible that more gender-nonconforming homosexual
and peers across the several years of childhood and adolescence). people are more likely to be “out,” perhaps because their presen-
Furthermore, the data are correlational, and not all correlations tation makes it harder for them to be closeted about their sexual
were significant (see Table 4). Additionally, although gender- orientation. If this assumption is true, our sample of homosexual
nonconforming people reported greater childhood rejection, we do people might have been generally biased toward gender noncon-
not know for certain that they actually experienced more rejection, formity.
nor that any such childhood rejection was due to gender noncon- Some evidence consistent with this possibility comes from
formity. It is also possible that social influences other than accep-
inspection of effect sizes in our data. For self-reported childhood
tance versus rejection of gender nonconformity influence the de-
gender nonconformity (see Figure 2), the effect size of sexual
velopment of these behaviors. For example, degree of gender
orientation for male targets (d ⫽ 2.3) was considerably larger than
nonconformity could reflect, in part, the availability of gender-
the effect size reported in Bailey and Zucker’s (1995) review (d ⫽
nonconforming role models.
1.3), whereas the effect sizes for female targets were comparable
The relative early onset of gender nonconformity in prehomo-
sexual children (and of gender conformity in preheterosexual chil- (respective ds ⫽ 1.1 and 1.0). If the homosexual male targets in the
dren) found in our study could, in part, be explained by innate current study recalled higher levels of childhood gender noncon-
influences. Much past research suggests that early exposure to formity than most homosexual men report, this would make our
different levels of gonadal hormones can influence the develop- sample unusual. It is also possible, however, that homosexual
ment of gender-typed behaviors (Cohen-Bendahan, van de Beek, participants have become more likely or willing to report higher
& Berenbaum, 2005; Collaer & Hines, 1995) and sexual orienta- levels of childhood gender nonconformity since the review was
tion (Ellis & Ames, 1987) and can account for the covariation of published in 1995. Some indication for this comes from a recent
gender nonconformity and sexual orientation (Bailey & Zucker, study on twins, including thousands of participants (Bailey et al.,
1995). Bailey and Zucker also speculated that the variation of 2000). In this study, the respective effect sizes were approximately
gender nonconformity within each sexual orientation could be 1.8 for both men and women.
influenced by prenatal hormonal factors. However, they also ac- Our findings could also be biased if participants who were
knowledged that within-orientation differences in gender noncon- interested in our study were trying to confirm the stereotype of
formity are consistent with the hypothesis of multiple etiological their group. Although we told participants to bring all available
routes to a homosexual orientation, including noninnate factors. videotapes from childhood, and although we did not explicitly tell
them that the study was about sexual orientation and gender
Limitations nonconformity, they might have guessed our main hypothesis.
This could have encouraged some participants to select childhood
This study had important methodological limitations, some of videos and present themselves in a way that would be consistent
which we have already mentioned. These include lack of standard- with the hypothesis.
ization of childhood videos; the brevity of video clips, possibly Finally, our sample was restricted to heterosexual and homo-
leading to unreliable measures of the underlying trait of gender sexual participants. Future research should address the level of
nonconformity; possible volunteer biases of a small, self-selected, childhood gender nonconformity among other groups—for exam-
and restricted sample; self-report measures of childhood treatment; ple, among people who identify as bisexual, questioning, or unla-
and the correlational nature of the data.
beled. Such research could provide insights into how similar—and
Unstandardized videos. Childhood videos varied in context, in
different—these groups are from people who use the more com-
contrast to the video samples of adults. For example, the activities
mon self-labels straight and gay.
engaged in by different children were not the same. It is unlikely
Self-report of past treatment. Measures of childhood rejection
that differences in context spuriously enhanced the effect of sexual
all depended on self-report. Such retrospective measures may be
orientation, however. Clips were chosen in the same manner for
both homosexual and heterosexual targets. This is not to say that subject to both unsystematic and systematic measurement error
variation in context was irrelevant to the association between (see Gerlsma et al., 1990).
childhood gender nonconformity and sexual orientation. Our anal- Correlational results. We emphasize that ours was a correla-
yses suggest that raters relied partially on the children’s activities tional study, with inherent limitations in establishing causation.
when they assessed gender nonconformity, although other infor- For example, as mentioned above, even if we accept that correla-
mation also seemed relevant. tions between past rejection and gender nonconformity were valid,
Brief behavioral samples. We used video clips that were at we cannot be sure that children experienced this rejection because
most 30 s long to assess gender nonconformity. Ratings based on of their gender nonconformity. Future research should therefore
such short samples are likely to contain measurement error. Thus, develop more rigorous operationalizations of rejection due to
the true gender nonconformity differences between heterosexual gender nonconformity. It is also possible that unexplored mediat-
and homosexual people may be larger than those reported here. ing factors influenced this association, and these, too, should be
Furthermore, the associations between observer-rated and self- addressed in future research.
56 RIEGER, LINSENMEIER, GYGAX, AND BAILEY

Conclusion Drummond, Bradley, Peterson-Badali, & Zucker (2008). A follow-up


study of girls with gender identity disorder. Developmental Psychology,
One should take methodological limitations into account when 44, 34 – 45.
considering the results of this study. However, these limitations are Ellis, L., & Ames, M. (1987). Neurohormonal functioning and sexual
different from the biases that may affect retrospective studies of orientation: A theory of homosexuality-heterosexuality. Psychological
heterosexual and homosexual adults using only self-reports or Bulletin, 101, 233–258.
prospective studies of very atypical children. The convergence of Epstein, S. (1983). The Mother–Father–Peer Scale. Unpublished manu-
results from these three research approaches increases our confi- script.
dence in the validity of their shared conclusion that degree of Fagot, B. I. (1977). Consequences of moderate cross-gender behavior in
preschool children. Child Development, 48, 902–907.
gender nonconformity versus gender conformity is a predictor of
Fagot, B. I. (1985). Beyond the reinforcement principle: Another step
future sexual orientation and that gender-nonconforming children
toward understanding sex role development. Developmental Psychology,
tend to become homosexual adults. 21, 1097–1104.
Fagot, B. I., & Leinbach, M. D. (1989). The young child’s gender schema:
References Environmental input, internal organization. Child Development, 60,
663– 672.
Ambady, N., Hallahan, M., & Conner, B. (1999). Accuracy of judgments Freund, K., & Blanchard, R. (1983). Is the distant relationship of fathers
of sexual orientation from thin slices of behavior. Journal of Personality and homosexual sons related to the sons’ erotic preference for male
and Social Psychology, 77, 538 –547. partners, or to the sons’ atypical gender identity, or to both? Journal of
Bailey, J. M. (2003). The man who would be queen: The science of
Homosexuality, 9, 7–25.
gender-bending and transsexualism. Washington, DC: Joseph Henry
Gaudio, R. (1994). Sounding gay: Pitch properties in the speech of gay and
Press.
straight men. American Speech, 69, 30 –57.
Bailey, J. M., Dunne, M. P., & Martin, N. G. (2000). Genetic and envi-
Gerlsma, C., Emmelkamp, P. M., & Arrindell, W. A. (1990). Anxiety,
ronmental influences on sexual orientation and its correlates in an
depression, and perception of early parenting: A meta-analysis. Clinical
Australian twin sample. Journal of Personality and Social Psychology,
Psychology Review, 10, 251–277.
78, 524 –536.
Gottschalk, L. (2003). Same-sex sexuality and childhood gender non-
Bailey, J. M., Finkel, E., Blackwelder, K., & Bailey, T. (1995). Masculin-
conformity: A spurious connection. Journal of Gender Studies, 12,
ity, femininity, and sexual orientation. Unpublished manuscript.
35–50.
Bailey, J. M., Gaulin, S., Agyei, Y., & Gladue, B. A. (1994). Effects of
Green, R. (1987). The “sissy boy syndrome” and the development of
gender and sexual orientation on evolutionarily relevant aspects of
homosexuality. New Haven, CT: Yale University Press.
human mating psychology. Journal of Personality and Social Psychol-
Harry, J. (1982). Gay children grown up: Gender culture and gender
ogy, 66, 1081–1093.
deviance. New York: Praeger.
Bailey, J. M., & Zucker, K. J. (1995). Childhood sex-typed behavior and
Harry, J. (1983a). Defeminization and adult psychological well-being
sexual orientation: A conceptual analysis and quantitative review. De-
among male homosexuals. Archives of Sexual Behavior, 12, 1–19.
velopmental Psychology, 31, 43–55.
Beard, A., & Bakeman, R. (2000). Boyhood gender nonconformity: Re- Harry, J. (1983b). Parasuicide, gender, and gender deviance. Journal of
ported parental behavior and the development of narcissistic issues. Health and Social Behavior, 24, 350 –361.
Journal of Gay & Lesbian Psychotherapy, 4, 81–97. Jacklin, C. N., DiPietro, J. A., & Maccoby, E. E. (1984). Sex-typing
Bell, A. P., Weinberg, M. S., & Hammersmith, S. K. (1981). Sexual behavior and sex-typing pressure in child/parent interaction. Archives of
preference: Its development in men and women. Bloomington: Indiana Sexual Behavior, 13, 413– 425.
University Press. Johnson, K. L., Gill, S., Reichman, V., & Tassinary, L. G. (2007). Swag-
Bem, S. L. (1974). The measurement of psychological androgyny. Journal ger, sway, and sexuality: Judging sexual orientation from body motion
of Consulting and Clinical Psychology, 42, 155–162. and morphology. Journal of Personality and Social Psychology, 93,
Bem, S. L. (1983). Gender schema theory and its implications for child 321–334.
development: Raising gender-aschematic children in a gender-schematic Judd, C. M., & McClelland, G. H. (1989). Data analysis: A model-
society. Signs, 8, 598 – 616. comparison approach. San Diego, CA: Harcourt Brace Jovanovich.
Bussey, K., & Bandura, A. (2004). Social cognitive theory of gender Landolt, M. A., Bartholomew, K., Saffrey, C., Oram, D., Perlman, D., &
development and functioning. In A. H. Eagly, A. E. Beall, & R. Stern- Bartholomew, K. (2004). Gender nonconformity, childhood rejection,
berg (Eds.), The psychology of gender (2nd ed., pp. 92–119). New York: and adult attachment: A study of gay men. Archives of Sexual Behavior,
Guilford Press. 33, 117–128.
Caldera, Y. M., Huston, A. C., & O’Brien, M. (1989). Social interactions Langlois, J. H., & Downs, A. (1980). Mothers, fathers, and peers as
and play patterns of parents and toddlers with feminine, masculine, and socialization agents of sex-typed play behaviors in young children. Child
neutral toys. Child Development, 60, 70 –76. Development, 51, 1237–1247.
Carter, D., & McCloskey, L. A. (1983). Peers and the maintenance of Lippa, R. (1991). Some psychometric characteristics of gender diagnostic-
sex-typed behavior: The development of children’s conceptions of cross- ity measures: Reliability, validity, consistency across domains, and
gender behavior in their peers. Social Cognition, 2, 294 –314. relationship to the Big Five. Journal of Personality and Social Psychol-
Cohen, J. (1988). Statistical power analysis for the behavioral sciences ogy, 61, 1000 –1011.
(2nd ed.). Hillsdale, NJ: Erlbaum. Lippa, R. (1998a). Gender-related individual differences and the structure
Cohen-Bendahan, C. C. C., van de Beek, C., & Berenbaum, S. A. (2005). of vocational interests: The importance of the people–things dimension.
Prenatal sex hormone effects on child and adult sex-typed behavior: Journal of Personality and Social Psychology, 74, 996 –1009.
Methods and findings. Neuroscience & Biobehavioral Reviews, 29, Lippa, R. (1998b). The nonverbal display and judgment of extraversion,
353–384. masculinity, femininity, and gender diagnosticity: A lens model analy-
Collaer, M. L., & Hines, M. (1995). Human behavioral sex differences: A sis. Journal of Research in Personality, 32, 80 –107.
role for gonadal hormones during early development? Psychological Lippa, R. A. (2000). Gender-related traits in gay men, lesbian women, and
Bulletin, 118, 55–107. heterosexual men and women: The virtual identity of homosexual-
SPECIAL SECTION: SEXUAL ORIENTATION AND GENDER NONCONFORMITY 57

heterosexual diagnosticity and gender diagnosticity. Journal of Person- Saghir, M. T., & Robins, E. (1973). Male and female homosexuality: A
ality, 68, 899 –926. comprehensive investigation. Baltimore: Williams & Wilkins.
Lippa, R. A. (2002). Gender-related traits of heterosexual and homosexual Skidmore, C. W., Linsenmeier, J. A. W., & Bailey, J. M. (2006). Gender
men and women. Archives of Sexual Behavior, 31, 83–98. nonconformity and psychological distress in lesbians and gay men.
Lippa, R. A. (2005a). Gender, nature, and nurture (2nd ed.). Mahwah, NJ: Archives of Sexual Behavior, 35, 685– 697.
Erlbaum. Spence, J. T. (1993). Gender-related traits and gender ideology: Evidence
Lippa, R. A. (2005b). Sexual orientation and personality. Annual Review of for a multifactorial theory. Journal of Personality and Social Psychol-
Sex Research, 16, 119 –153. ogy, 64, 624 – 635.
Lott, B., & Maluso, D. (2001). Gender development: Social learning. In J. Spence, J. T., & Helmreich, R. (1978). Masculinity and femininity: Their
Worrell (Ed.), Encyclopedia of gender (pp. 537–549). San Diego, CA: psychological dimensions, correlates and antecedents. Austin: Univer-
Academic Press. sity of Texas Press.
Maccoby, E. E. (1990). Gender and relationships: A developmental ac- Tauber, M. A. (1979). Sex differences in parent-child interaction styles
count. American Psychologist, 45, 513–520. during a free-play session. Child Development, 50, 981–988.
Maccoby, E. E. (1998). The two sexes: Growing up apart, coming together. Trautner, H. M. (1992). The development of sex-typing in children: A
Cambridge, MA: Belknap Press. longitudinal analysis. German Journal of Psychology, 16, 183–199.
Pedhazur, E. J., & Tetenbaum, T. J. (1979). Bem Sex Role Inventory: A Weinrich, J. D., Grant, I., Jacobson, D. L., Robinson, S. R., McCutchan,
theoretical and methodological critique. Journal of Personality and J. A., & the HNRC Group. (1992). Effects of recalled childhood gender
Social Psychology, 37, 996 –1016. nonconformity on adult genitoerotic role and AIDS exposure. Archives
Rieger, G., Linsenmeier, J. A. W., Gygax, L., Garcia, S., & Bailey, J. M. of Sexual Behavior, 21, 559 –585.
(2007). Gaydar: Thin slices of sexual orientation and masculinity- Zar, J. H. (1999). Biostatistical analysis (4th ed.). Englewood Cliffs, NJ:
femininity. Unpublished manuscript. Prentice Hall.
Ross, M. W., Campbell, R. L., & Clayer, J. R. (1982). New inventory for Zucker, K. J. (2005). Commentary on Gottschalk’s (2003) “Same-sex
measurement of parental rearing patterns: An English form of the sexuality and childhood gender non-conformity: A spurious connec-
EMBU. Acta Psychiatrica Scandinavica, 66, 499 –507. tion.” Journal of Gender Studies, 14, 55– 60.

Appendix

Gender Nonconformity Scales

Scale Items: Childhood Gender Nonconformity Scale Several times I have dreamed I was a woman.
(Men)
I feel like part of me is male and part of me is female.
As a child I was called a sissy by my peers.
I don’t feel very feminine.
As a child I sometimes wished I had been born a girl
In many ways I feel more similar to women than to men.
rather than a boy.
People think I should act more masculine than I do.
As a child I preferred playing with girls rather than boys.
It would be fun to go to a costume party dressed as a
As a child I often felt that I had more in common with woman.
girls than boys.
I pride myself on being masculine.
As a child I sometimes wore feminine clothing (such as
dresses), makeup or jewelry. I am much less masculine than the average straight man.

As a child I disliked competitive sports such as foot- I am much less feminine than the average gay man.
ball, baseball and basketball.
Scale Items: Childhood Gender Nonconformity Scale
I was a feminine boy.
(Women)

Scale Items: Adulthood Continuous Gender Identity Scale As a child I was called a tomboy by my peers.
(Men)
As a child I sometimes wished I had been born a boy
I often wonder what it would be like to be a woman. rather than a girl.

(Appendix continues)
58 RIEGER, LINSENMEIER, GYGAX, AND BAILEY

As a child I preferred playing with boys rather than girls. I don’t feel very masculine.

As a child I often felt that I had more in common with In many ways I feel more similar to men than to
boys than girls. women.

As a child I usually avoided feminine clothing (such as People think I should act more feminine than I do.
dresses).
It would be fun to go to a costume party dressed as a
As a child I liked competitive sports such as football, man.
baseball and basketball.
I pride myself on being feminine.
I was a masculine girl.
I am much less feminine than the average straight
Scale Items: Adulthood Continuous Gender Identity Scale woman.
(Women)
I am much less masculine than the average lesbian.
I often wonder what it would be like to be a man.

Several times I have dreamed I was a man. Received August 28, 2006
Revision received May 22, 2007
I feel like part of me is male and part of me is female. Accepted May 24, 2007 䡲

Members of Underrepresented Groups:


Reviewers for Journal Manuscripts Wanted
If you are interested in reviewing manuscripts for APA journals, the APA Publications and
Communications Board would like to invite your participation. Manuscript reviewers are vital to the
publications process. As a reviewer, you would gain valuable experience in publishing. The P&C
Board is particularly interested in encouraging members of underrepresented groups to participate
more in this process.

If you are interested in reviewing manuscripts, please write to the address below. Please note the
following important points:

• To be selected as a reviewer, you must have published articles in peer-reviewed journals. The
experience of publishing provides a reviewer with the basis for preparing a thorough, objective
review.

• To be selected, it is critical to be a regular reader of the five to six empirical journals that are most
central to the area or journal for which you would like to review. Current knowledge of recently
published research provides a reviewer with the knowledge base to evaluate a new submission
within the context of existing research.

• To select the appropriate reviewers for each manuscript, the editor needs detailed information.
Please include with your letter your vita. In the letter, please identify which APA journal(s) you
are interested in, and describe your area of expertise. Be as specific as possible. For example,
“social psychology” is not sufficient—you would need to specify “social cognition” or “attitude
change” as well.

• Reviewing a manuscript takes time (1– 4 hours per manuscript reviewed). If you are selected to
review a manuscript, be prepared to invest the necessary time to evaluate the manuscript
thoroughly.

Write to Journals Office, American Psychological Association, 750 First Street, NE, Washington,
DC 20002-4242.

View publication stats

Вам также может понравиться