You are on page 1of 100

P1: FOM

Archives of Sexual Behavior [asb]

PL113-227171

September 8, 2000

11:42

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 29, No. 6, 2000

Editors Introduction

Two years ago, I invited members of the Editorial Board to guest edit an issue
on a topic of their choosing. The first to accept was Gilbert Herdt. Along with
his invited coeditor, Martha McClintock, he has assembled a provocative series
of papers on sexual attraction and desire, a complex, enigmatic core of human
sexuality.
Richard Green
Editor

525
C 2000 Plenum Publishing Corporation
0004-0002/00/1200-0525$18.00/0

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227172

September 8, 2000

12:7

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 29, No. 6, 2000

Guest Editors Introduction

On the Development of Sexual Attraction

This special issue opens a series of new and important questions regarding
the development of attraction and desire in human sexuality studies. The articles
span the range of high theory to high empiricism, and have a satisfying scope
that includes developmental and biopsychology, anthropology, psychoanalysis,
and social survey and social psychological analysis. The authors respond to the
questions: What is the development and form of attraction in earlier development?
How shall we define attraction and its subsequent effects upon sexual behavior
in adulthood? Moreover, several of the papers consider the earliest attraction and
how we should understand the variation within or between the genders.
In the modern period, it was customary to understand that one gender desires
only the opposite gender, because of their differences in anatomy and hence of
sexual nature, making the opposite sex an object of attraction. This old-fashioned
view is being increasingly questioned in the biological and social sciences, and
certainly, its status as a universal of human development is subjected to critical
investigation in the following studies. Increasingly it appears that the range of
attractions may be more diverse or fuzzy than this binary once implied for crosscultural human development. How do societies stimulate and regulate sexual
desires, in the broad sense of what is socially valued and may be expressed in
public? The sexual and the emotional or social appear to contextually interact.
Such a view suggests that we cannot reduce attraction either to pure biology or
purely to culture, as theorists in the past were wont to do. Conceptions of attraction
are closer to what anthropologists have called valuesthe general disposition
to desire and merit an ontological state or a culturally valued social role (to be a
warrior, a husband, a chief) or aim in life (to be brave, virtuous, powerful, etc.). To
desire the toys of Santa Claus as a child or the muscles of Arnold Scwartzanegger
as an adult is to participate in a collective system of valued objects that lies halfway
between the individual and culture. Growing up and being socialized into families,
villages, or schools; becoming playmates and friends with age-mates; and having
the company of a favorite peer, or a circle of buddies, are all means and social ways
of forming sexual subjectivity, of objectifying the demands made upon children
as they become habituated to systems of erotic and social desires. We thus have
527
C 2000 Plenum Publishing Corporation
0004-0002/00/1200-0527$18.00/0

P1: FLF
Archives of Sexual Behavior [asb]

528

PL113-227172

September 8, 2000

12:7

Style file version Nov. 19th, 1999

Herdt

inscribed into our beingour private parts, and even in the whispers of the self
sexual and psychosexual markers that signify membership in society.
Daryl Bem summarizes his significant theory (Exotic Becomes Erotic, or
EBE) on how erotic attraction and subjective sexual excitement emerges out of
perceived difference, as children feel generally psychophysiologically aroused by
those who are exotic to them. Desire, in turn, stimulates fantasy, and in childhood development, Bem suggests, intrinsic personality temperament (and gender
typicality or atypicality) serves to mediate between aspects of the self and the
perception of the exotic other. In subsequent development these attractions and
arousal, which may differ by gender, are likely to be expressed in sexual behavior and stable adult sexual orientations. Bem reviews the critical responses to his
theory, including its gender bias, and answers his critics through a new conceptualization of the issues surrounding what arousal means to each gender.
Michael P. Dunne, J. Michael Bailey, Katerine M. Kirk, and Nicholas G.
Martin describe the variation within a survey of a national sample of Australian
adult twins (N + 4, 901, age range: 1952 years) to understand the emergence of
attractions in development. The authors report a high percentage of individuals,
15.2% of the men, who report ever having homosexual behavior, while 11.5%
report same-sex attractions, and 6.4% report that they are not heterosexual. These
important findings also background their data on the emergence of first sexual
attractions at ages 13.7 years for men and 16.3 years for women. The authors note
that this is significantly later than other studies reported in the literature, including McClintock and Herdts (McClintock and Herdt, 1996). Their study provides
much food for thought.
However, it should be noted that Dunnes study is a more representative
sample from another society (Australia) than the research reports reviewed by
McClintock and Herdt, and later by Herdt and McClintock. We do not know how
the recruitment bias in these studies may have effected the age of first attraction.
Perhaps, even more important, Dunne et al. herein defined attraction explicitly
as sexual, whereas the studies reviewed by Herdt and McClintock, as well as
Savin-Williams and Diamonds reported later, used measures that were more vague
and general regarding the meaning of attraction. In particular, this difference in
measurement may account for the later Australian age of attraction reported by
Dunne et al., and the gender disparity as well, particularly if Savin-Williams and
Diamond are correct in arguing that females tend to associate their first attraction
with emotional, rather than sexual, contexts.
Richard C. Friedman and Jennifer I. Downey review psychoanalytic theorizing on sexual development, desire, and sexual fantasy. By focusing upon deep
psychological structures, the meanings of arousal to men and women, and the nature of sexual fantasy in erotic experience, the authors review Freuds prior theories,
especially his oedipal and bisexual theories, which they critique, by contrast with
contemporary ideas and psychoanalytic practice, particularly the understanding of

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227172

On the Development of Sexual Attraction

September 8, 2000

12:7

Style file version Nov. 19th, 1999

529

homosexuality. Their linking of sexual development to narratives is most timely:


Sexual fantasies are stories told to Ourselves that are embedded in sexual feelings.
By linking this work to hormonal, pschophysiological, and sexual orientation, the
authors open possibilities for the intersection of qualitative and quantitative study.
Gilbert Herdt and Martha McClintock theorize that the development of attraction begins during adrenarche (ages 610) and reaches a memorable, and stable,
representation by age 10. Reviewing the evidence for hormonal development, they
build upon their previous study (McClintock and Herdt, 1996), and suggest that
adrenal puberty, followed by gonadal puberty, expand the puberty process in human
development upto gonadarche. Females and males, heterosexuals and homosexuals, in the United States, in the studies reviewed, were found to experience first
attraction by age 10. In the comparison of New Guinea societies, the authors found
that many cultures mark age 10 as being highly significant, especially for males,
in their sexual transition or ritual initiation into adulthood. Precocious pubertal
development in some minority children in the United States is also reviewed. The
authors conclude that the age of 10 is not magicalonly a convenience marker in
the cultural reasoning of societies about powerful hormonal processes.
Ritch C. Savin-Williams and Lisa M. Diamond review the literature on gender
differences in the development of sexual identity. Their predominantly white, middle class, college sample reveals a young age of first same-sex attraction: a mean
of 7.7 years for males and 9.0 years for females. By dividing their sample into
those attracted in emotional versus sexual contexts, they were able to determine
that males are significantly more likely to experience attraction in a sexual context, whereas females are significantly more likely to experience attraction in an
emotional context. Likewise, this contextual difference seems to carry through in
subsequent sexual behavior and identity changes, as females emphasize the emotional and relational elements of romantic and sexual bonds. Overlap is evident in
the developmental pathways of both genders too. But as the authors summarize
the evidence, the pathways leading from sexual desire to sexual behavior are not
equivalent for males and females, when sexual orientation is held constant.
Gilbert Herdt
Guest Editor
Program in Human Sexuality Studies
San Francisco State University
San Francisco, California 91432

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 29, No. 6, 2000

Exotic Becomes Erotic: Interpreting the Biological


Correlates of Sexual Orientation
Daryl J. Bem, Ph.D.1

Although biological findings currently dominate the research literature on the determinants of sexual orientation, biological theorizing has not yet spelled out a
developmental path by which any of the various biological correlates so far identified might lead to a particular sexual orientation. The Exotic-Becomes-Erotic
(EBE) theory of sexual orientation (Bem, 1996) attempts to do just that, by suggesting how biological variables might interact with experiential and sociocultural
factors to influence an individuals sexual orientation. Evidence for the theory is
reviewed, and a path analysis of data from a large sample of twins is presented
which yields preliminary support for the theorys claim that correlations between
genetic variables and sexual orientation are mediated by childhood gender nonconformity.
KEY WORDS: sexual orientation; homosexuality; heterosexuality; erotic orientation; sexuality; path
analysis; genetic correlates.

INTRODUCTION
Biological findings currently dominate the research literature on the determinants of sexual orientation. Reports of correlations between various biological
variables and homosexuality appear regularly in the professional journals and, just
as regularly, receive instant replay in the mass media. As a result, some researchers,
many journalists, and sizable segments of the lesbian/gay/bisexual community
have rushed to embrace the conclusion that a homosexual orientation is coded in
the genes, caused by prenatal hormones, or determined by brain neuroanatomy.
Except for the reparative therapists, most of the personality, clinical, and developmental psychologists and psychiatrists who once dominated the discourse on
1 Department

of Psychology, Uris Hall, Cornell University, Ithaca, New York 14853; e-mail: d.bem@

cornell.edu.
531
C 2000 Plenum Publishing Corporation
0004-0002/00/1200-0531$18.00/0

P1: FLF
Archives of Sexual Behavior [asb]

532

PL113-227173

September 8, 2000

12:37

Style file version Nov. 19th, 1999

Bem

this topic have fallen conspicuously silent. Many have probably become closet
converts to biology because they cannot point to a coherent body of evidence that
supports a developmental, experience-based account of sexual orientation. The
general public is not far behind: In 1983, only 16% of Americans believed that
homosexuality is something that people are born with (Moore, 1993); by 2000,
that figure had more than doubled to 35% (Reuters, 2000).
I find at least two aspects of the current zeitgeist scientifically problematic.
The first is the premature rush to interpret correlation as causation. In the absence
of any theory oflet alone, evidence fora developmental pathway from the
biological markers to sexual orientation, such an interpretation is still a leap of
faith. At best, there seems to be an implicit, primitive gender-inversion theory of
homosexuality: If, for example, a biological characteristic that is more prevalent
in gay men than in heterosexual men happens also to be more prevalent in women
than in men, then, ipso facto, that is somehow deemed to explain the homosexual
orientation. It was my dissatisfaction with this default theory that challenged me
to spell out a specific developmental process in which biological variables would
interact with experiential and sociocultural factors to determine an individuals
sexual orientation. My Exotic-Becomes-Erotic (EBE) theory of sexual orientation
(Bem, 1996) was the result of that effort.
The second problematic aspect of the current zeitgeist is that it narrowly focuses on the question What causes homosexuality? This framing of the inquiry
implicitly presumes that heterosexuality is so well understood, so obviously the
natural evolutionary consequence of reproductive advantage, that only deviations
from it require explanation. Freud himself did not so presume: [Heterosexuality]
is also a problem that needs elucidation and is not a self-evident fact based
upon an attraction that is ultimately of a chemical nature (Freud, 1905/1962,
pp. 11, 12).
I agree with Freud. In fact, I would go further and assert that even the use of
gender as the basis for choosing a sexual partner is a problem that needs elucidation. Accordingly, EBE theory attempts to account for three major observations:
First, most men and women in our culture have an exclusive and enduring erotic
preference for either males or females; gender is, in fact, the overriding criterion
for most peoples erotic choices. Second, most men and women in our culture have
an exclusive and enduring erotic preference for persons of the opposite sex. And
third, a substantial minority of men and women have an exclusive and enduring
erotic preference for persons of the same sex. In seeking to account for these observations, EBE theory provides a single unitary explanation for both opposite-sex
and same-sex desireand for both men and women. In addition, the theory seeks
to account for sex differences in sexual orientation and for departures from the
modal patterns, such as bisexual orientations, orientations that are not enduring
but fluid and changeable, and sexual orientations that are not even based on the
gender of potential partners.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

Interpreting Biological Correlates

Style file version Nov. 19th, 1999

533

OVERVIEW OF EBE THEORY


The central proposition of EBE theory is that individuals can become erotically attracted to a class of individuals from whom they felt different during
childhood. Figure 1 shows how this phenomenon is embedded in the overall sequence of events that leads to an individuals erotic attractionsthe component of
sexual orientation addressed by the theory. The sequence begins at the top of the
figure with Biological Variables (labeled A) and ends at the bottom with Erotic
Attraction (labeled F).
A B. According to the theory, biological variables such as genes or prenatal
hormones do not code for sexual orientation per se but for childhood temperaments, such as aggression and activity level.
B C. A childs temperaments predispose him or her to enjoy some activities
more than other activities. One child will enjoy rough-and-tumble play and
competitive team sports (male-typical activities); another will prefer to socialize
quietly or play jacks or hopscotch (female-typical activities). Children will also
prefer to play with peers who share their activity preferences; for example,
the child who enjoys baseball or football will selectively seek out boys as
playmates. Children who prefer sex-typical activities and same-sex playmates
are referred to as gender conforming; children who prefer sex-atypical activities
and opposite-sex playmates are referred to as gender nonconforming.
C D. Gender-conforming children will feel different from opposite-sex peers,
and gender-nonconforming children will feel different from same-sex peers.
D E. These feelings of being different produce heightened physiological
arousal. For the male-typical child, it may be felt as antipathy or contempt
in the presence of girls (girls are yucky); for the female-typical child, it
may be felt as timidity or apprehension in the presence of boys. A particularly
clear example is the sissy boy who is taunted by male peers for his gender
nonconformity and, as a result, is likely to experience the strong physiological arousal of fear and anger in their presence. However, the theory claims that
every childconforming or nonconformingexperiences heightened, nonspecific physiological arousal in the presence of peers from whom he or she feels
different. For most children, this arousal in neither affectively toned nor consciously experienced.
E F. Regardless of the specific source or affective tone of the childhood arousal,
it is subsequently transformed into erotic attraction. Steps D E and E F
thus encompass specific psychological mechanisms that transform exotic into
erotic (D F).
It is important to emphasize that Fig. 1 is not intended to describe an inevitable,
universal path to sexual orientation but the modal path followed by most men and

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

Style file version Nov. 19th, 1999

534

Bem

Fig. 1. The temporal sequence of events leading to sexual


orientation for most men and women in a gender-polarizing
culture.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

Interpreting Biological Correlates

Style file version Nov. 19th, 1999

535

women in a gender-polarizing culture like ours, a culture that emphasizes the


differences between the sexes by pervasively organizing both the perceptions and
realities of communal life around the malefemale dichotomy (Bem, 1993).
EVIDENCE FOR THE THEORY
Exotic Becomes Erotic (D F)
The central proposition that individuals can become erotically attracted to
a class of individuals from whom they felt different during childhood is very
general and transcends erotic orientations that are based on gender. For example,
a light-skinned person could come to eroticize dark-skinned persons through one
or more of the processes described by the theory. To produce a differential erotic
attraction to one sex or the other, however, requires that the basis for feeling
different must itself differentiate between the sexes; that is, to arrive at a sex-based
erotic orientation, an individual must feel different for sex-based or gender-related
reasons. Simply being lighter-skinned, poorer, more intelligent, or more introverted
than ones childhood peers does not produce the kind of feeling different that
produces differential homoerotic or heteroerotic attraction.
Data consistent with this analysis comes from an intensive, large-scale interview study conducted in the San Francisco Bay Area by the Kinsey Institute
for Sex Research (Bell et al., 1981a). Using retrospective reports from adult respondents, the investigators compared approximately 1,000 gay men and lesbians
with 500 heterosexual men and women to test several hypotheses about the development of sexual orientation. The study (hereinafter, the San Francisco study)
yielded virtually no support for current experience-based theories of sexual orientation, including those based on processes of learning or conditioning or on family
psychodynamics.
The study did find, however, that 71% of the gay men and 70% of the lesbians
in the sample reported that they had felt different from their same-sex peers during
childhood, a feeling that was sustained throughout childhood and adolescence
for most respondents. When asked in what ways they had felt different, they
overwhelmingly cited gender-related reasons. Gay men were most likely to say
that they had not liked boys sports; lesbians were most likely to say that they had
been more masculine than other girls were and had been more interested in sports
than other girls. In contrast, fewer than 8% of heterosexual men or women said that
they had felt different from same-sex childhood peers for gender-related reasons.
Those who had felt different from their peers tended to cite such reasons as having
been poorer, more intelligent, or more introverted. (All statistical comparisons
between gay and heterosexual respondents were significant at p < 0.0005.)
Several other studies have also reported that gay men and lesbians recall
having felt different from same-sex peers on gender-related characteristics during

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

Style file version Nov. 19th, 1999

536

Bem

childhood (e.g., Newman and Muzzonigro,1993; Savin-Williams,1998; Telljohann


and Price, 1993; Troiden, 1979). The major weakness in all these studies, including the San Francisco study, is that they rely on adults retrospective reports of
childhood feelings. On the other hand, the respondents in some of the studies
were relatively close in time to their childhood years; in one study, for example, 88% of gay male youths as young as 14 years reported having felt different
from other boys on gender-related characteristics throughout their childhood years
(Savin-Williams, 1998). Moreover, the link between childhood gender nonconformity and sexual orientation (described in the next section) has been confirmed in
over 50 studies, including prospective ones (Bailey and Zucker, 1995; they also
discuss the retrospective problem at length).
Gender Conformity and Nonconformity: The Antecedents
of Feeling Different (C D)
Feeling different from ones childhood peers can have any of several antecedents, some common, some idiosyncratic. The most common antecedent is
gender polarization. Virtually all human societies polarize the sexes to some extent,
setting up a sex-based division of labor and power, emphasizing or exaggerating sex
differences, and, in general, superimposing the malefemale dichotomy on virtually every aspect of communal life (Bem, 1993). These gender-polarizing practices
ensure that most boys and girls will grow up feeling different from opposite-sex
peers and, hence, will come to be erotically attracted to them later in life. This,
according to the theory, is why gender becomes the most salient category and,
hence, the most common criterion for selecting sexual partners in the first place
and why heteroeroticism is the modal preference across time and culture. Thus,
the theory provides a culturally based alternative to the assumption that evolution
must necessarily have programmed heterosexuality into the species for reasons of
reproductive advantage.
Obviously heterosexual behavior is reproductively advantageous, but it does
not follow that it must therefore be sustained through genetic transmission. As
long as prevailing environments support or promote a reproductively successful
behavior sufficiently often, it will not necessarily get programmed into the genes
by evolution. This is true even in species whose sexual choices are far more hardwired than our own. For example, it is presumably reproductively advantageous
for ducks to mate with other ducks, but as long as most baby ducklings encounter
other ducks before they encounter a member of some other species (including
ethologists), evolution can simply implant the imprinting process itself into the
species rather than the specific content of what, reproductively speaking, needs
to be imprinted (Hess and Petrovich, 1977). Analogously, because most cultures
ensure that boys and girls will see each other as exotic, it would be sufficient for
evolution to implant an exotic-becomes-erotic process into our species rather than

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

Style file version Nov. 19th, 1999

Interpreting Biological Correlates

537

Table I. Percentage of Respondents Reporting Gender-Nonconforming Preferences and Behaviors


During Childhood
Men
Response
Had not enjoyed sex-typical
activities
Had enjoyed sex-atypical
activities
Atypically sex-typed
(masculinity/femininity)
Most childhood friends were
opposite sex

Women

Gay
(n = 686)

Heterosexual
(n = 337)

Lesbian
(n = 293)

Heterosexual
(n = 140)

63

10

63

15

48

11

81

61

56

80

24

42

13

60

40

Note. Percentages have been calculated from the data given in Bell et al. (1981b, pp. 74, 75, 77). All
chi-square comparisons between gay and heterosexual subgroups are significant at p < 0.0001.

heterosexuality per se. In fact, an exotic-becomes-erotic process is actually a builtin component of sexual imprinting in some species. For example, Japanese quail
reared with their siblings later prefer their slightly different-appearing cousins to
their own siblings (Bateson, 1978). This has been interpreted as a mechanism that
prevents inbreedinga biologically promoted incest taboo.
How, then, does a child come to feel different from same-sex peers? As
cited earlier, the most common reasons given by gay men and lesbians in the San
Fransisco study for having felt different from same-sex peers in childhood were
sex-atypical preferences and behaviors in childhoodgender nonconformity. In
fact, in the path analyses of the San Francisco study, childhood gender conformity
or nonconformity was not only the strongest but the only significant childhood
predictor of later sexual orientation for both men and women (Bell et al., 1981a).
As Table I shows, the effects are large and significant. For example, compared
with heterosexual men, gay men were significantly less likely to have enjoyed
boys activities (e.g., baseball and football) during childhood, more likely to have
enjoyed girls activities (e.g., hopscotch, playing house, and jacks), and less likely
to rate themselves as having been masculine. These were the three variables that
defined gender nonconformity in the study. Additionally, gay men were more
likely than heterosexual men to have had girls as childhood friends. The corresponding comparisons between lesbian and heterosexual women are also large and
significant.
It is also clear from the table that relatively more women than men reported
enjoying sex-atypical activities and having opposite-sex friends during childhood.
As these data confirm, enjoying male-typical activities is common for a girl in our
society, implying that being a tomboy is not sufficient by itself to cause her to feel
different from other girls. In fact, we see in the table that the difference between
the percentages of lesbians and heterosexual women who report having enjoyed

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

Style file version Nov. 19th, 1999

538

Bem

boys activities during childhood (81% vs. 61%, respectively) is less than half the
size of the difference between them in their aversion to girls activities (63% vs.
15%). Moreover, this latter difference is virtually identical to that between gay
men and heterosexual men in their reported childhood aversions to boys activities
(63% vs. 10%).
As noted in the previous section, the San Francisco study does not stand alone.
A meta-analysis of 48 studies confirmed that gay men and lesbians are more likely
than heterosexual men and women to recall gender-nonconforming behaviors and
interests in childhood (Bailey and Zucker, 1995, p. 49). As the authors observed,
these are among the largest effect sizes ever reported in the realm of sex-dimorphic
behaviors. Prospective longitudinal studies come to the same conclusion. In the
largest of these, 75% of gender-nonconforming boys became bisexual or homosexual in later years compared with only 4% of gender-conforming boys (Green,
1987). In six other prospective studies, 63% of gender-nonconforming boys later
had homosexual orientations (Zucker, 1990). Currently there are no prospective
studies of gender-nonconforming girls.
How Does Exotic Become Erotic? (D E F)
EBE theory proposes that exotic becomes erotic because feeling different from
a class of peers in childhood produces heightened nonspecific physiological arousal
(D E), which is subsequently transformed into erotic attraction (E F). To
my knowledge, there is no direct evidence for the first step in this sequence beyond
the well-documented observation that novel (exotic) stimuli produce heightened
physiological arousal in many species, including our own (Mook, 1999); filling in
this empirical gap in EBE theory must await future research. In contrast, there are
at least three mechanisms that can potentially effect the second step, transforming
generalized arousal into erotic attraction (Bem, 1996). Only one of these, the
extrinsic arousal effect, is discussed here.
In his first-century Roman handbook, The Art of Love, Ovid advised any
man who was interested in sexual seduction to take the woman in whom he was
interested to a gladiatorial tournament, where she would more easily be aroused to
passion. However, he did not say why this should be so. A contemporary version
of Ovids claim was introduced by Walster (Berscheid and Walster, 1974; Walster,
1971), who suggested that it constitutes a special case of the 2-factor theory of
emotion by Schachter and Singer (1962). This theory states that the physiological
arousal of our autonomic nervous system provides the cues that we feel emotional
but that the more subtle judgment of which emotion we are feeling often depends
on our cognitive appraisal of the surrounding circumstances. According to Walster,
then, the experience of erotic desire results from the conjunction of physiological
arousal and the cognitive causal attribution (or misattribution) that the arousal is
elicited by a potential sexual partner.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

Style file version Nov. 19th, 1999

Interpreting Biological Correlates

539

Although not all investigators agree that it arises from a cognitive attribution
process, there is now extensive experimental evidence that an individual who has
been physiologically aroused will show heightened sexual responsiveness to an
appropriate target person. In one set of studies, male participants were physiologically aroused by running in place, by hearing an audio tape of a comedy routine, or
by hearing an audio tape of a grisly killing (White et al., 1981). No matter how they
had been aroused, these men reported more erotic interest in a physically attractive
woman than did men who had not been aroused. This effect has also been observed
physiologically. In two studies, preexposure to a disturbing (nonsexual) videotape
subsequently produced greater penile tumescence in men and greater vaginal blood
volume increases in women when they watched an erotic videotape than did preexposure to a nondisturbing videotape (Hoon et al., 1977; Wolchik et al., 1980).
In other words, generalized physiological arousal, regardless of its source or
affective tone, can subsequently be experienced as erotic desire. At that point, it
is erotic desire. My proposal, then, is that an individuals protracted and sustained
experience of feeling different from same- or opposite-sex peers throughout childhood and adolescence produces a correspondingly sustained physiological arousal
that gets eroticized when the maturational, cognitive, and situational factors coalesce to provide the critical defining moment.
The precise timing of this moment, however, is influenced by several factors,
including actual sexual experience with opposite- and same-sex peers. A recent review suggests that, in general, men and women recall their first sexual attractions,
whether same-sex or opposite-sex, as occurring when they were between 10 and
10.5 years of age (McClintock and Herdt, 1996). Nevertheless, social norms and
expectations inevitably influence an individuals awareness and interpretation of
early arousal. Most individuals in our culture are primed to anticipate, recognize,
and interpret opposite-sex arousal as erotic or romantic attraction and to ignore, repress, or differently interpret comparable same-sex arousal. We should also expect
to see secular changes and cohort effects. For example, the heightened visibility
of gay men and lesbians in our society is now prompting individuals who experience same-sex arousal to recognize it, label it, and act on it at earlier ages than in
previous years (Dube, 1997; Fox, 1995; Savin-Williams, 1995, 1998).
The Biological Connection (A F) versus (A B)
As outlined in Fig. 1, EBE theory proposes that to the extent biological factors
such as the genotype, prenatal hormones, or brain neuroanatomy influence an individuals later sexual orientation, they do so only indirectly, by intervening earlier in
the chain of events to influence a childs preference for sex-typical or sex-atypical
activity and peer preferenceshis or her gender conformity or nonconformity.
More specifically, the theory specifies that any link between, say, the genotype and gender nonconformity (A C) is composed of two parts: a link between

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

540

Style file version Nov. 19th, 1999

Bem

the genotype and childhood temperaments (A B) and a link between those temperaments and gender nonconformity (B C). This implies that the mediating
temperaments should possess three characteristics: First, they should be plausibly
related to those childhood activities that define gender conformity and nonconformity. Second, because they manifest themselves in sex-typed preferences, they
should show sex differences. And third, because they are hypothesized to derive from the genotype, they should have significant heritabilities. (For general
discussions and reviews of childhood temperaments, see Goldsmith et al., 1987;
Kohnstamm et al., 1989.)
One likely candidate is aggression and its benign cousin, rough-and-tumble
play. Gay men score lower than heterosexual men on measures of childhood aggression (Blanchard et al., 1983), and parents of gender-nonconforming boys
specifically rate them as having less interest in rough-and-tumble play than do
parents of gender-conforming boys (Green, 1976). Second, the sex difference in
aggression during childhood is one of the largest psychological sex differences
known (Hyde, 1984). Rough-and-tumble play in particular is more common in
boys than in girls (DiPietro, 1981; Fry, 1990; Moller et al., 1992). And third, individual differences in aggression have a large heritable component (Rushton et al.,
1986).
Another likely candidate is activity level, considered to be one of the basic
childhood temperaments (Buss and Plomin, 1975, 1984). Like aggression, differences in activity level would also seem to characterize the differences between
male-typical and female-typical play activities in childhood. Moreover, gendernonconforming boys and girls are lower and higher on activity level, respectively,
than are control children of the same sex (Bates et al., 1973, 1979; Zucker and
Green, 1993). Second, the sex difference in activity level is as large as it is for
aggression. Even before birth, boys in utero are more active than girls are (Eaton
and Enns, 1986). And third, individual differences in activity level have a large
heritable component (Plomin, 1986; Rowe, 1997).

A Test of the EBE Model


There have now been several studies showing a correlation between an individuals sexual orientation and his or her genotype. In one, a sample of 115 gay
men who had male twins, 52% of identical twin brothers were also gay compared
with only 22% of fraternal twin brothers and 11% of adopted brothers (Bailey
and Pillard, 1991). In a comparable sample of 115 lesbians, 48% of identical
twin sisters were also lesbians compared with only 16% of fraternal twin sisters
and 6% of adopted sisters (Bailey et al., 1993). A subsequent study of nearly
5,000 twins who had been systematically drawn from a twin registry confirmed
the significant heritability of sexual orientation for men but not for women (Bailey

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

Interpreting Biological Correlates

September 8, 2000

12:37

Style file version Nov. 19th, 1999

541

and Martin, 1995). Finally, an analysis of families in which there were two gay
brothers, suggested a correlation between a homosexual orientation and the inheritance of genetic markers on the X chromosome (Hamer and Copeland, 1994;
Hamer et al., 1993).
But these same studies also provided evidence for the link proposed by EBE
theory between an individuals genotype and his or her childhood gender nonconformity. For example, in the 1991 study of male twins, the correlation on
gender nonconformity between gay identical twins was as high as the reliability
of the nonconformity measure would permit, 0.76, ( p < 0.0001), compared with
a nonsignificant correlation of only 0.43 between gay fraternal twins (Bailey and
Pillard, 1991). This implies that even when sexual orientation is held constant,
there is a significant correlation between the genotype and gender nonconformity.
Similarly, the 1993 family study found that gay brothers who shared the same
genetic markers on the X chromosome were more alike on gender nonconformity than were gay brothers who did not (Hamer and Copeland, 1994; Hamer
et al., 1993). Finally, childhood gender nonconformity was significantly heritable for both men and women in the large twin registry studyeven though sexual orientation itself was not significantly heritable for the women (Bailey and
Martin, 1995).
Because this twin registry study is based on a very large sample and includes
heterosexual as well as bisexual and homosexual individuals, the data can be used
in a path analysis to test the EBE model against the competing default model that
the genotype is more directly linked to sexual orientation or is linked via some
alternative but unspecified path. In particular, the EBE model predicts that any
correlation between the genotype and sexual orientation is mediated by gender
nonconformity and, hence, should vanish when gender nonconformity is entered
into the path model. In contrast, the default model predicts that the correlation between the genotype and sexual orientation should remain unaffected when gender
nonconformity is entered into the path model.
The path analysis presented here is based on the fact that monozygotic twins
will be more similar than dizygotic twins on any trait with nonzero heritability. This
is equivalent to saying that zygosity is itself correlated with trait similarity across
pairs of twins; the higher the heritability of the trait, the higher the correlation.
Accordingly, the unit of analysis here is the twin pair, and each variable is a measure
of the pairs similarity on the three variables at issue. (The variables are actually all
coded in the direction of dissimilarity.) Genetic similarity (zygosity) is coded as 0
for monozygotic twin pairs and as 1 for dizygotic pairs. The similarity of a pairs
childhood gender nonconformity is the absolute value of the difference between
their scores on a multi-item scale of childhood gender nonconformity; and, the
similarity of their sexual orientations is the absolute value of the difference between
their scores on the 7-point Kinsey scale of sexual orientation, which ranges from
0 = exclusively heterosexual to 6 = exclusively homosexual. A full description of

P1: FLF
Archives of Sexual Behavior [asb]

542

PL113-227173

September 8, 2000

12:37

Style file version Nov. 19th, 1999

Bem

Fig. 2. Path coefficients between genetic similarity (zygosity), childhood gender nonconformity similarity, and sexual orientation similarity for male and female twin pairs. p < 0.001.

the twin sample and the methodology of the study appears elsewhere in this issue
(Dunne et al., 2000).2
As shown in Fig. 2, the pattern of path coefficients is consistent with the EBE
model for both male and female twin pairs: For both sexes, there is a significant
path between the genotype and childhood gender nonconformity and a further
significant path between childhood gender nonconformity and sexual orientation,
but there is no remaining, direct link between the genotype and sexual orientation.3
2 Michael

Bailey has generously provided the relevant data for these path analyses and, even more
generously, given me permission to publish them here even though he and his collaborators have not
yet published their own genetic analyses of these data.
3 To ensure that this pattern of results is not simply an artifact of differing distributions of the two
continuous variables (childhood gender nonconformity similarity and sexual orientation similarity),
a logistic analysis (Darlington, 1990) was also performed in which these two variables were first
transformed into dichotomous variables with identical distributions. Following Dunne et al. (1999),
a twin pair was considered concordant for sexual orientation if both twins were either exclusively
heterosexual (Kinsey scores of 0) or not (Kinsey scores greater than 0). The difference scores on
childhood gender nonconformity were then dichotomized so that the number of concordant pairs
on this variable equalled the number of pairs who were concordant on sexual orientation. In this
way, both variables were given an equal chance of being correlated with genetic similarity. This
alternative analysis yielded the same correlational patterns and the same significance levels as the
analysis depicted in Fig. 2.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

Interpreting Biological Correlates

Style file version Nov. 19th, 1999

543

Other Biological Correlates


In addition to the genotype, prenatal hormones and brain neuroanatomy have
also been correlated with sexual orientation in some studies (for summaries, reviews, and critiques see Bailey, 1995; Bem, 1996; Byne and Parsons, 1993; Zucker
and Bradley, 1995). But these correlationseven if they turn out to be replicable
and not artifactualdo not necessarily controvert the EBE account. Any biological
factor that correlates with one or more of the intervening processes proposed by
EBE theory could also emerge as a correlate of sexual orientation. For example,
any neuroanatomical feature of the brain that correlates with childhood aggression
or activity level is likely to emerge as a difference between gay men and heterosexual men, between women and men, and between heterosexual women and
lesbians. Even if EBE theory turns out to be wrong, the more general pointthat
a mediating personality variable could account for observed correlations between
biological variables and sexual orientationstill holds.
INDIVIDUAL DIFFERENCES
As noted earlier, Fig. 1 is not intended to describe an inevitable, universal
path to sexual orientation but only the modal path followed by most men and
women in a gender-polarizing culture like ours. Individual differences, including
apparent exceptions to the theory, can arise in a number of ways. First, of course,
the theory could simply be wrong or incomplete in fundamental ways. But some
of the apparent exceptions to the sequence of events laid out in Fig. 1 are arguably
theory-consistent variations.
One such possibility is that some individuals enter the EBE path in the middle
of the sequence rather than at the beginning. For example, some children may come
to feel different from same-sex peers not because of a temperamentally induced
preference for sex-atypical activities but because of more idiosyncratic factors,
such as a physical disability, an illness, or an atypical lack of contact with samesex peers. Some of the gay men and lesbians in the San Francisco study reported
that although they had been gender conforming in their childhood behaviors, they
still felt different from their same-sex peers for gender-related reasons. Moreover, even the sex-typical lesbians in the study were more likely than heterosexual
women to report that most of their friends in grade school had been boys. And,
consistent with the subsequent steps in the EBE path, this was the strongest predictor of homosexual involvements in adolescence and their homosexual orientation
in adulthood.
Cultural factors can also enter to create individual differences that appear
to be exceptions to the EBE model. For example, some children might have an
activity preference that is gender neutral or even sex typical in the wider culture
but gender deviant in their own peer subculture. A contemporary example is the

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

544

Style file version Nov. 19th, 1999

Bem

boy who is a clever computer hacker: He would be considered a regular guy


or even a hero in some male subcultures but a gender-deviant nerd in others.
Similarly, a child can be permissibly gender nonconforming in some waysand
hence not feel different from same-sex peersif he or she is gender conforming in other ways that are more gender-defining in his or her subculture. And
finally, changes in the wider culture can produce cohort effects; behaviors that
are gender nonconforming in one cohort can become more or less so in a later
cohort.
For some individuals, the erotic attractions predicted by EBE theory might be
supplemented or even superseded by erotic attractions acquired after adolescence.
For example, the same-sex eroticism of most of the bisexual men and women in the
San Francisco study appeared to be a socially learned, post-adolescent add-on
to an already established heterosexual orientation. Not surprisingly, these bisexual
respondents differed from their exclusively homosexual counterparts on some of
the major antecedent variables as well. For example, the path correlation between
gender nonconformity and same-sex eroticism was only 0.18 for the bisexual
women, but it was 0.62 for the exclusively homosexual women. In fact, 80% of
the bisexual women and 75% of the bisexual men in the study reported that as
children they had been sex-typically feminine or masculine, respectively.
Finally, some women who would otherwise be predicted by the EBE model
to have a heterosexual orientation might choose for social or political reasons to
center their lives around other women. This could lead them to avoid seeking out
men for sexual or romantic relationships, to develop affectional and erotic ties to
other women, and to self-identify as lesbians or bisexuals (Kitzinger, 1987), which
in turn leads to the topic of sex differences.
SEX DIFFERENCES
One of the more audacious claims made for EBE is that it provides a single
unitary explanation for both opposite-sex and same-sex desireand for both men
and women. Not everyone is convinced, however, and I have been challenged to
defend the theory against the charge that it is androcentric: valid for men, perhaps,
but not for women (Bem, 1998; Peplau et al., 1998).
To be sure, there is now substantial evidence that men and women differ from
one another on several aspects of sexuality, irrespective of their sexual orientations
(Peplau et al., 1998). As I tell my students, if you want to understand the sexuality
of gay men, think of them as men; if you want to understand the sexuality of
lesbians, think of them as women. But most of these differences have to do with
the primacy or intensity of erotic desire, the relative emphasis on the physical
attributes of potential partners, and the willingness to engage in impersonal sex
without romantic involvement. Such differences are not pertinent to EBE theorys
account of how erotic orientations develop.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

12:37

Interpreting Biological Correlates

Style file version Nov. 19th, 1999

545

There is, however, one sex difference that is pertinent to EBE theory: Womens
sexual orientations are more fluid than mens. Many studies, including a national
random survey of Americans (Laumann et al., 1994), have found that women are
more likely to be bisexual than exclusively homosexual, whereas the reverse is
true for men. Nonheterosexual women are also more likely to see their sexual
orientations as flexible, even chosen, whereas men are more likely to view their
sexual orientations in essentialist terms, as inborn and unchangeable (Whisman,
1996). For example, men who come out as gay after leaving heterosexual marriages
or relationships often describe themselves as having finally realized their true
sexual orientation. Lesbians in similar situation, however, are more likely to reject
the implication that their previous heterosexual relationships were inauthentic or at
odds with who they really were: Thats who I was then, and this is who I am now.
The greater fluidity of womens sexual orientations is consistent with EBE
theory. As noted earlier, Fig. 1 describes the path to sexual orientation in a genderpolarizing culture. But in our society, women grow up in a (phenomenologically)
less gender-polarized culture than do men. Compared with boys, girls are punished
less for being gender nonconforming, and, as the data in Table I reveal, they are
more likely than boys to engage in both sex-typical and sex-atypical activities and
are more likely to have childhood friends of both sexes. This implies that girls are
less likely than boys to feel differentially different from opposite-sex and same-sex
peers and, hence, are less likely to develop exclusively heteroerotic or homoerotic
orientations.
It is even possible that some of todays nonheterosexual women may be giving
a preview of what sexual orientations would look like in a less gender-polarized
future. It is possible that we might even begin to see more men and women who,
instead of using gender as the overriding criterion for selecting a partner, might
base their erotic and romantic choices on a more diverse and idiosyncratic variety
of attributes. As I remarked in my original article, Gentlemen might still prefer
blonds, but some of those gentlemen (and some ladies) might prefer blonds of any
sex (Bem, 1996, p. 332).
EBE THEORY VERSUS WHAT?
Many of my biologically oriented friends and colleagues tell me that they think
EBE theory is very cleverand very wrong. They may be right. The existing data
are far from decisive, and I am genuinely open to the possibility that biological
factors influence sexual orientation more directly than EBE theory would have it.
But as much as I prefer being right to being wrong, I will be content if
EBE theory does no more than provoke some affirmative competition. To my
knowledge, there is no competing theory for a more direct or alternative path
between the genotype and sexual orientation. It is not that such a theory has been
advanced, tested, and found wanting, but that it has not yet been made.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

September 8, 2000

546

12:37

Style file version Nov. 19th, 1999

Bem

In their public statements and published articles, my biologically oriented


colleagues dutifully point out that correlation is not cause. But, as I have commented elsewhere (Bem, 1996), the reductive temptation of biological causation
is so seductive that the caveat cannot possibly compete with the excitement of
discovering yet another link between the anatomy of our brains and the anatomy
of our lovers genitalia. Unfortunately, the caveat vanishes completely as word of
the latest discovery moves from Science to Newsweek. Surely the public can be
forgiven for believing that we are but one NIH grant away from pinpointing the
penis preference gene.
REFERENCES
Bailey, J. M. (1995). Biological perspectives on sexual orientation. In DAugelli, A. R., and Patterson,
C. J. (Eds.), Lesbian, Gay and Bisexual Identities over the Lifespan, Oxford University Press,
New York, pp. 102135.
Bailey, J. M., and Martin, N. G. (1995). A twin registry study of sexual orientation. Paper presented
at the twenty-first annual meeting of the International Academy of Sex Research, Provincetown,
MA.
Bailey, J. M., and Pillard, R. C. (1991). A genetic study of male sexual orientation. Arch. Gen. Psychiat.
48: 10891096.
Bailey, J. M., Pillard, R. C., Neale, M. C., and Agyei, Y. (1993). Heritable factors influence sexual
orientation in women. Arch. Gen. Psychiat. 50: 217223.
Bailey, J. M., and Zucker, K. J. (1995). Childhood sex-typed behavior and sexual orientation: A
conceptual analysis and quantitative review. Dev. Psychol. 31: 4355.
Bates, J. E., Bentler, P. M., and Thompson, S. K. (1973). Measurement of deviant gender development
in boys. Child Dev. 44: 591598.
Bates, J. E., Bentler, P. M., and Thompson, S. K. (1979). Gender-deviant boys compared with normal
and clinical controls boys. J. Abnorm. Child Psychol. 7: 243259.
Bateson, P. P. G. (1978). Sexual imprinting and optimal outbreeding. Nature 273: 659660.
Bell, A. P., Weinberg, M. S., and Hammersmith, S. K. (1981a). Sexual Preference: Its Development in
Men and Women, Indiana University Press, Bloomington.
Bell, A. P., Weinberg, M. S., and Hammersmith, S. K. (1981b). Sexual Preference: Its Development in
Men and Women. Statistical Appendix, Indiana University Press, Bloomington.
Bem, D. J. (1996). Exotic becomes erotic: A developmental theory of sexual orientation. Psychol. Rev.
103: 320335.
Bem, D. J. (1998). Is EBE theory supported by the evidence? Is it androcentric? A reply to Peplau
et al. Psychol. Rev. 105: 395398.
Bem, S. L. (1993). The Lenses of Gender: Transforming the Debate on Sexual Inequality, Yale
University Press, New Haven, CT.
Berscheid, E., and Walster, E. (1974). A little bit about love. In Huston, T. (Ed.), Foundations of
Interpersonal Attraction, Academic Press, New York, pp. 355381.
Blanchard, R., McConkey, J. G., Roper, V., and Steiner, B. W. (1983). Measuring physical aggressiveness in heterosexual, homosexual, and transsexual males. Arch. Sex. Behav. 12: 511524.
Buss, A. H., and Plomin, R. (1975). A Temperament Theory of Personality Development, Wiley, New
York.
Buss, A. H., and Plomin, R. (1984). Temperament: Early Developing Personality Traits, Erlbaum,
Hillsdale, NJ.
Byne, W., and Parsons, B. (1993). Human sexual orientation: The biologic theories reappraised. Arch.
Gen Psychiat. 50: 228239.
Darlington, R. B. (1990). Regression and Linear Models, McGraw-Hill, New York.
DiPietro, J. A. (1981). Rough and tumble play: A function of gender. Dev. Psychol. 17: 5058.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227173

Interpreting Biological Correlates

September 8, 2000

12:37

Style file version Nov. 19th, 1999

547

Dube, E. M. (Artist). (1997). Sexual Identity and Intimacy: Development Among Two Cohorts of Gay
and Bisexual Men. Unpublished Masters Thesis, Cornell University, Ithaca, NY.
Dunne, M. P., Bailey, J. M., Kirk, K. M., and Martin, N. G. (2000). The subtlety of sex-atypicality.
Arch. Sex. Behav. 29: 549565.
Eaton, W. O., and Enns, L. R. (1986). Sex differences in human motor activity level. Psychol. Bull.
100: 1928.
Fox, R. C. (1995). Bisexual identities. In DAugelli, A. R., and Patterson, C. J. (Eds.), Lesbian, Gay
and Bisexual Identities over the lifespan, Oxford University Press, New York, pp. 4886.
Freud, S. (1905/1962). Three Essays on the Theory of Sexuality, Basic Books, New York.
Fry, D. P. (1990). Play aggression among Zapotec children: Implications for the practice hypothesis.
Aggressive Behav. 17: 321340.
Goldsmith, H. H., Buss, A. H., Plomin, R., Rothbart, M. K., Thomas, A., Chess, S., Hinde, R. A.,
and McCall, R. B. (1987). Roundtable: What is temperament? Four approaches. Child Dev. 58:
505529.
Green, R. (1976). One-hundred ten feminine and masculine boys: Behavioral contrasts and demographic
similarities. Arch. Sex. Behav. 5: 425426.
Green, R. (1987). The Sissy Boy Syndrome and the Development of Homosexuality, Yale University
Press, New Haven, CT.
Hamer, D., and Copeland, P. (1994). The Science of Desire: The Search for the Gay Gene and the
Biology of Behavior, Simon and Schuster, New York.
Hamer, D. H., Hu, S., Magnuson, V. L., Hu, N., and Pattatucci, A. M. L. (1993). A linkage between
DNA markers on the X chromosome and male sexual orientation. Science 261: 321327.
Hess, E. H., and Petrovich, S. B. (1977). Imprinting, Dowden, Hutchinson and Ross, Stroudsburg, PA.
Hoon, P. W., Wincze, J. P., and Hoon, E. F. (1977). A test of reciprocal inhibition: Are anxiety and
sexual arousal in women mutually inhibitory? J. Abnorm. Psychol. 86: 6574.
Hyde, J. S. (1984). How large are gender differences in aggression? A developmental meta-analysis.
Dev. Psychol. 20: 722736.
Kitzinger, C. (1987). The Social Construction of Lesbianism, Sage, London.
Kohnstamm, G. A., Bates, J. E., and Rothbart, M. K. (1989). Temperament in Childhood, Wiley, New
York.
Laumann, E. O., Gagnon, J. H., Michael, R. T., and Michaels, S. (1994). The Social Organization of
Sexuality: Sexual Practices in the United States, University of Chicago Press, Chicago.
McClintock, M. K., and Herdt, G. (1996). Rethinking puberty: The development of sexual attraction.
Current Directions in Psychol. Sci. 5: 178183.
McGue, M., and Lykken, D. T. (1992). Genetic influence on risk of divorce. Psychol. Sci. 3(6): 368
373.
Moller, L. C., Hymel, S., and Rubin, K. H. (1992). Sex typing in play and popularity in middle
childhood. Sex Roles. 26: 331353.
Mook, D. G. (1999). Motivation: The Organization of Animal and Human Action, Norton, New York.
Moore, D. W. (1993). Public polarized on gay issue. The Gallup Poll Monthly, 3034.
Newman, B. S., and Muzzonigro, P. G. (1993). The effects of traditional family values on the coming
out process of gay male adolescents. Adolescence 28: 213226.
Peplau, L. A., Garnets, L. D., Spalding, L. R., Conley, T. D., and Veniegas, R. C. (1998). A critique of
Bems Exotic Becomes Erotic theory of sexual orientation. Psychol. Rev. 105: 387394.
Plomin, R. (1986). Development, Genetics, and Psychology, Erlbaum, Hillsdale, NJ.
Reuters. (2000). Most in U. S. favor laws barring gay discrimination. Reuters Limited, February 8.
Rowe, D. C. (1997). Genetics, temperament, and personality. In Hogan, R., Johnson, J., and Briggs, S.
(Eds.), Handbook of Personality Psychology, Academic Press, San Diego, CA, pp. 367386.
Rushton, J. P., Fulker, D. W., Neale, M. C., Nias, D. K. B., and Eysenck, H. J. (1986). Altruism and
aggression: The heritability of individual differences. J. Pers. Soc. Psychol. 50: 11921198.
Savin-Williams, R. C. (1995). Lesbian, gay male, and bisexual adolescents. In DAugelli, A. R., and
Patterson C. J. (Eds.), Lesbian, Gay, and Bisexual Identities over the Lifespan, Oxford University
Press, New York, pp. 165189.
Savin-Williams, R. C. (1998). . . . And then I Became Gay, Routledge, New York.
Schachter, S., and Singer, J. E. (1962). Cognitive, social, and physiological determinants of emotional
state. Psychol. Rev. 69: 379399.

P1: FLF
Archives of Sexual Behavior [asb]

548

PL113-227173

September 8, 2000

12:37

Style file version Nov. 19th, 1999

Bem

Telljohann, S. K., and Price, J. P. (1993). A qualitative examination of adolescent homosexuals life
experiences: Ramifications for secondary school personnel. J. Homosexual. 26: 4156.
Troiden, R. R. (1979). Becoming homosexual: A model of gay identity formation. Psychiatry 42:
362373.
Walster, E. (1971). Passionate love. In Murstein, B. I. (Ed.), Theories of Attraction and Love, Springer,
New York, pp. 8599.
Whisman, V. (1996). Queer by choice, Routledge, New York.
White, G. L., Fishbein, S., and Rutstein, J. (1981). Passionate love and the misattribution of arousal.
J. Pers. Soc. Psychol. 41: 5662.
Wolchik, S. A., Beggs, V. E., Wincze, J. P., Sakheim, D. K., Barlow, D. H., and Mavissakalian, M.
(1980). The effect of emotional arousal on subsequent sexual arousal in men. J. Abnorm. Psychol.
89: 595598.
Zucker, K. J. (Ed.). (1990). Gender Identity Disorders in Children: Clinical Descriptions and Natural
History, American Psychiatric Press, Washington, DC.
Zucker, K. J., and Bradley, S. J. (1995). Gender Identity Disorder and Psychosexual Problems in
Children and Adolescents, Guilford Press, New York.
Zucker, K. J., and Green, R. (1993). Psychological and familial aspects of gender identity disorder.
Child Adolesc. Psychiat. Clinics North Am. 2: 513542.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

10:54

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 29, No. 6, 2000

The Subtlety of Sex-Atypicality


Michael P. Dunne, Ph.D.,1,4 J. Michael Bailey, Ph.D.,2
Katherine M. Kirk, Ph.D.,3 and Nicholas G. Martin, Ph.D.3

Memories of sex-atypical behavior and interests in childhood usually differ between homosexual and heterosexual people. However, variation within these broad
groups has not previously been explored in detail, especially among women. We
utilized data from a postal survey of a nationwide sample of Australian adult twins
(n = 4,901, age range: 1952 years). Among men, 15.2% reported homosexual
behavior (ever), 11.5% said they had been sexually attracted to the same sex,
and 6.4% said they were not heterosexual; the corresponding figures for women
were 7.9, 10.6, and 3.5%. A continuous measure of childhood gender nonconformity (CGN) was sensitive to slight variations in homosexual attraction and
behavior. In particular, among both men and women who identified as heterosexual, there were significant differences between complete heterosexuals and
those who admitted to only one or a few same-sex behaviors but no homosexual
attraction. Among men, CGN scores distinguished between heterosexuals who admitted to same-sex behavior only and those who admitted to some homosexual
attraction. The sexual subgroups also differed on a measure of gender atypicality
in adulthood. Implications for developmental theories of sexuality are discussed.
KEY WORDS: sexual orientation; attraction; childhood gender nonconformity.

INTRODUCTION
The link between sex-typicality in childhood play and interests, and sexual
orientation in adolescence and later life, is a fascinating aspect of human sexual development. Although noted many times over the years, the strength of the
1 School

of Public Health, Queensland University of Technology, Brisbane, Australia.


Department, Northwestern University, Evanston, Illinois 60208-2710.
3 Queensland Institute of Medical Research, Brisbane, Australia.
4 To whom correspondence should be addressed at School of Public Health, Queensland University
of Technology, Victoria Park Road, Kelvin Grove, 4059, Queensland, Australia; e-mail: m.dunne@
qut.edu.au.
2 Psychology

549
C 2000 Plenum Publishing Corporation
0004-0002/00/1200-0549$18.00/0

P1: FLF
Archives of Sexual Behavior [asb]

550

PL113-227174

September 19, 2000

10:54

Style file version Nov. 19th, 1999

Dunne, Bailey, Kirk, and Martin

association was illustrated most clearly by Bailey and Zucker (1995) in a metaanalysis of data from 41 studies. When retrospective reports of homosexual and
heterosexual adults were compared, the average effect size across a range of childhood behaviors was substantial (1.19 standard deviation units). Prospective studies,
although fewer in number and limited mainly to clinical samples of males, similarly suggest that sex-atypicality of childhood behavior is an important precursor
of homosexual orientation (Green, 1987; Zucker, 1990).
This observation is a central element of Bems developmental theory of sexual
orientation (Exotic Becomes Erotic, or EBE), where biologically mediated aspects of personality in childhood (temperament and sex-typical/atypical behavior)
are thought to underlie feelings that the self is similar to, or different from, people
of the same sex (Bem, 1996). According to Bem, those from whom we feel most
different as children are exotic, and subsequently become eroticized during
early puberty. These attractions may be expressed in early sexual behaviors and
stable sexual orientation in later life.
Enthusiasm for this model must be tempered by at least two serious questions.
First, recalled childhood sex-typed behavior may vary more within, rather than
between, individuals grouped on the basis of their sexual orientation, and hence
the true predictive value of sex-atypical behavior in childhood for adult sexual
orientation may be low. In Bailey and Zuckers meta-analysis, the great majority of retrospective individual studies had recruited people who self-identified as
either homosexual or heterosexual (Bailey and Zucker, 1995). Within these broad
groups, there was substantial variation in recalled childhood behaviors, and sexual
orientation per se accounted for less than 40% of the variance in sex-atypicality.
Bailey and Zucker (1995) suggested that the variance in childhood sex-atypicality
within groups could reflect the existence of subtypes, and sex-atypicality may be
etiologically important for only some of these.
To date, relatively few studies have explored childhood sex-typicality among
people with different degrees of homosexual or heterosexual attractions, behavior,
and orientation. Among 392 male Australian twins who rated themselves as heterosexual, McConaghy et al. (1994) found low but significant correlations between
degree of adult homosexual attraction and dislike of rough-and-tumble play, outdoor and contact sports, and a childhood desire to be of the opposite sex. Also in
Australia, Phillips and Over (1992) found bisexual men to be intermediate between
gay and straight men on ten measures of childhood sex-typed behavior. In contrast, males in the USA interviewed by Bailey (1989; cited in Bailey and Zucker,
1995), female twins in the USA (Bailey et al., 1993), and primarily heterosexual
women interviewed in Australia by McConaghy and Silove (1991) did not show a
continuous relationship between the degree of homosexual feelings and childhood
sex-atypicality. There is clearly a need for more data, especially from women.
A second, important question in this type of research is whether the apparent association between childhood behavior and later orientation arises from
intentional or unintentional distortion of recall; for example, heterosexuals might

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

Childhood Sex-Atypicality and Sexual Behavior

10:54

Style file version Nov. 19th, 1999

551

underestimate the true level of sex-atypicality (Risman and Schwartz, 1988; Ross,
1980). This might arise if there are strong social expectancies for, or against, reporting childhood cross-gender interests and behaviors. On this basis, one might
predict a difference in retrospective reports between polar opposites (such as gays
and straights). However, what if there are significant differences between the recall
of people who identify as heterosexual but who vary only slightly in their degree
of homosexual attraction or experience? A social expectancy explanation would
be less plausible.
In this study, we examine self-reports of sexual orientation, behaviors and
attractions, and childhood sex-atypicality, from a nationwide sample of 4,901
Australian adult twins. More than 95% of these people could be placed in one of
five groups: a nonheterosexual group (who said they were bisexual or gay/lesbian
and who reported both homosexual attractions and behaviors), and four groups of
people who said they were heterosexual, but who differed in their degree of homosexual attractions and behaviors. The primary aim of this paper was to examine
the sensitivity of a retrospective measure of sex-atypicality before the age of 13
(childhood gender nonconformity, or CGN) to subtle variations in homosexual
attraction and experience.

METHODS
Subjects
Participants were drawn from the Australian National Health and Medical
Research Council Twin Register (ATR). The ATR is a volunteer register that was
begun in 1978 and has about 25,000 twin pairs of zygosity types and all ages enrolled and in various stages of active contact. Participants for this sexuality study
were recruited from two phases of a large twin-family study of alcohol use and
abuse (Heath et al., 1994). The twins were residing in eight states and territories
of Australia. There is a disproportionate number of young women and people with
higher than average levels of education (Baker et al., 1996). In relation to psychological factors, comparisons with normative data indicate that participants are generally representative of the Australian population in terms of personality, depression, and alcohol consumption (Dunne et al., 1997a). Diversity within this primary
cohort in terms of religious affiliation, social attitudes, and age at first sexual intercourse has been documented elsewhere (Dunne et al., 1997b,c; Martin et al., 1986).
During 1992, we asked all ATR twins aged between 17 and 50 years who
had completed a postal Health and Lifestyle survey between 1988 and 1990
(N = 9,112) about their willingness to receive a questionnaire regarding sex.
Specifically, they were asked: We have applied for funding to carry out an anonymous study of sexual behavior and attitudes. Would you be willing to receive a questionnaire with explicit questions on these topics? All those who said Yes were

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

552

10:54

Style file version Nov. 19th, 1999

Dunne, Bailey, Kirk, and Martin

mailed the sex questionnaire. When subjects received the sex questionnaire, they
were asked to complete a consent form with their name, date of birth, and signature,
which they had to return separately to indicate whether or not they had consented
to complete the sex questionnaire. Anonymity was assured, and we asked cotwins
privately to choose the same 10-digit identification number so that we could match
their questionnaires. Approximately two weeks after initial mailing of the sex
questionnaire, all twins were sent a reminder letter. Consent forms were logged as
they were returned and subsequently all twins who had not returned a consent form
were followed up once by telephone. Because we received many queries from twins
asking whether they should complete the questionnaire if their cotwin had decided
not to participate, we sent a further letter urging such singles to cooperate.
Twenty-eight percent explicitly refused to participate, and 54% (4,901) completed questionnaires. The remainder (18%) initially agreed to participate but did
not respond when contacted (following one letter or one phone call). Our response
rate was not substantially lower than that of other large-scale mail sex surveys,
which have typically achieved responses from between 55 and 65% (Biggar and
Melbye, 1992; Sundet et al., 1992). In recent analyses, we have compared the
individuals who returned the sex survey consent form with those who did not, on a
range of psychological and behavioral characteristics derived from data collected
in other thematically unrelated research interviews carried out with these twins
between 1988 and 1995 (Dunne et al., 1997a, 1998). There was some indication
of a modest participation bias; people who returned consent forms and those who
initially agreed to participate but could not subsequently be contacted had generally more liberal sexual attitudes, more novelty-seeking and less harm-avoidant
personalities, had an earlier age of first sexual intercourse and a greater likelihood
of childhood sexual abuse than people who explicitly refused to participate in the
sex survey. However, the effect sizes were small, suggesting that the behavioral
data in the mailed sex survey probably do not seriously misrepresent sexual activity
and attitudes.
Measures
Childhood Gender Nonconformity
The male and female measures of CGN included items retrospectively assessing childhood sex-typed behavior (i.e., participation in sex-stereotypic games and
activities) and gender identity (i.e., internal feelings of maleness or femaleness).
Childhood was defined as being before the age of 12.
Our CGN measures were adapted from several published scales, by taking relevant items (e.g., those related to childhood rather than to adulthood) and
in some cases, rewriting the items so that they were appropriate for Australian
participants (e.g., cricket rather than baseball). For males, items were taken

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

10:54

Childhood Sex-Atypicality and Sexual Behavior

Style file version Nov. 19th, 1999

553

from the Gender Identity Scale for Males (Freund et al., 1977), the Childhood
Play Activities Checklist (Grellert et al., 1982), the Recalled Childhood Gender
Behaviors Questionnaire (Mitchell and Zucker, 1992), and the Physical Aggressiveness Scale (Blanchard et al., 1983). For females, items were taken from the
Childhood Play Activities Checklist, the Recalled Childhood Gender Behaviors
Questionnaire, and the Masculine Gender Identity Scale (Blanchard and Freund,
1983). All of these scales have been shown to differ reliably between homosexual
and same-sex heterosexual individuals, as indeed have all scales of similar content
(Bailey and Zucker, 1995).
Both the male and female questionnaires contained 24 items, but the items
differed between the two versions, and therefore male and female scores are not
comparable. Items varied in their response format, and included both dichotomously rated items and rating scales. Scree tests of the principal components
suggested that for each sex, one general factor primarily accounted for the item
intercorrelations. Items were standardized within sexes and summed to yield a
total CGN score. Coefficient alpha was 0.79 for both male and female CGN.
Continuous Gender Identity (CGI)
This scale consisted of seven items taken from Finn (1987). The items assessed participants self-concepts as masculine or feminine (e.g., In many ways
I feel more similar to women/men than to men/women.) using 7-point rating
scales. A subscale including these items (as well as some other items, primarily concerning childhood gender nonconformity) distinguished homosexual from
same-sex heterosexual individuals (Finn, 1987). Separate scree tests for each sex
were both consistent with a single factor underlying CGI item intercorrelations.
Items were summed to yield total scores, and coefficient alpha was 0.52 for men
and 0.57 for women.
Self-Identification of Sexual Orientation
This was determined with the question Do you consider yourself to be heterosexual (straight), bisexual, or homosexual (lesbian/gay). Respondents were
asked to choose one option. We provided a definition, which stated (for females)
Heterosexual means that sexually, you desire contact only with men; bisexual
means that you desire contact with both men and women; homosexual means that
you desire contact only with women.
Sexual Attraction
We used a Kinsey-type rating for the question Which of the following best
describes your sexual feelings at present? There were seven response options,

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

10:54

554

Style file version Nov. 19th, 1999

Dunne, Bailey, Kirk, and Martin

ranging from (for women) I am attracted to men only, never to women (scored 0),
I am almost always attracted to men, but on rare occasions I am attracted to
women (scored 1), through to I am attracted to women only, never to men
(scored 6). Following this, we asked
Have you ever been sexually attracted to a female?
Have you ever been sexually attracted to a male?

Yes
Yes

No
No

In this paper, we have considered a person as positive for same-sex attraction


if he/she scored 1 or more for the Kinsey-type question about current attractions,
and/or ever answered Yes to the single question about same-sex attractions.

Sexual Behavior
Two questions were used to estimate numbers of same- and opposite-sex
partners. For women, we asked During your entire life, approximately how many
men have you had sexual contact with? Response options included none, 1 only, 2,
35, 610, 1120, 2150, over 50. This was followed by a question about female
partners. Sexual contact was defined as any activity which made you sexually
excited and in which your genitals (for women, vagina) made contact with any
part of the other person. Same-sex experience (ever) was scored positive if the
person chose any except the first response option.

Openness and Accuracy of the Respondents Self-Reports


This survey raised many issues that may be particularly sensitive to response
bias (Wiederman, 1993). It is conceivable, for instance, that people who identify
as heterosexual but who have some same-sex feelings and behaviors may find
it difficult to be completely open and accurate when answering questions about
their sex lives, even within an anonymous questionnaire. This would introduce
a serious bias in the present study if such people were less self-disclosing than
complete heterosexuals or people who identified as gay or bisexual. We included
two questions that may indicate willingness to self-disclose:
1. To what extent do you feel you were able to be completely open in
answering this questionnaire?
2. How accurately do you believe your answers to the above questions reflect
your true feelings and behavior?
For each question, the response options were completely, moderately, not
very, and not at all.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

10:54

Childhood Sex-Atypicality and Sexual Behavior

Style file version Nov. 19th, 1999

555

RESULTS
Demographic Characteristics
There were 1,824 males (mean age: 30.5 years, range: 1952 years) and 3,077
females (mean age: 31.1 years, range: 1952 years). Among the males, 30% were
41% were married, 22% lived in de facto relationships, 30% were single, and
8% were divorced, separated, or widowed. Among women, 40% were married,
22% in de facto relationships, 27% were single, and 10% divorced, separated,
or widowed. In comparison to average Australian adults, these volunteers were
quite highly educated, with 31% of males and 24% of females having completed a
university degree. Eighty-three percent of men and 69% of women were employed
in a paid job. We have compared the volunteers for this sex survey with twins in
the longitudinal Australian research registry who refused to participate in this
particular study: there were no major differences in demographic background,
although volunteers were slightly more likely to be female, married, employed,
and highly educated (Dunne et al., 1997a).
Sexual Diversity Within the Sample
Figure 1 shows the percentages of participants who considered themselves
to be nonheterosexual, those who admitted to ever having been sexually attracted
to someone of the same sex, and those who reported any sexual contact (which

Fig. 1. Same-sex attractions, behaviors, and orientation.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

10:54

556

Style file version Nov. 19th, 1999

Dunne, Bailey, Kirk, and Martin


Table I. Numbers of Opposite-Sex and Same-Sex Partners (Lifetime Estimates)

Males
Opp-Sex (%)
Same-Sex (%)
Females
Opp-Sex (%)
Same-Sex (%)

35

610

1120

2150

50+

5.1
84.7

12.6
6.9

8.4
2.0

21.6
2.2

18.2
1.1

17.5
1.0

12.6
0.9

3.9
1.3

3.7
92.2

20.8
4.3

12.8
1.1

28.6
1.4

19.1
0.6

10.0
0.3

3.9
0.1

1.1
0.0

included sexual excitement and genital contact). It is clear that although the prevalence of homosexual attraction was similar among men and women, men were
approximately twice as likely to have had any same-sex contact, and to consider
themselves to be bisexual or gay. Of the 112 men who said that they were not heterosexual, 57 (50.9%) were bisexual whereas 55 (49.1%) were gay. Among 103
women who considered themselves nonheterosexual, 82 (79.6%) were bisexual
whereas 21 (20.4%) were lesbian. The great majority of both men (97.2%) and
women (96.3%) said that they had been sexually attracted to someone of the opposite sex at some time in their lives; this included 96.5% of bisexual men, 44.4%
of gay men, 98.8% of bisexual women, and 71.4% of lesbians.
Respondents estimated the number of men and women with whom they had
had sex during their lifetime. The figures in Table I show a predictable pattern,
with men reporting a higher number of partners than women did. Interestingly,
men and women had the same modal category for heterosexual partners (35),
with the excess for men clearly evident in the 1120 and 2150 categories.
Among both men and women who experienced any sexual behavior with the same
sex, the modal number of partners was 1; this included 45.1% (121/268) of men
and 54.1% (125/231) of women who reported same-sex contact.

Classification of Sexual Subgroups


There was little evidence for true bipolarity in sexual orientation. For example,
the majority of men (64.3%) and women (67.7%) who admitted to at least some
sexual behavior with the same sex (defined by us to include sexual excitement
as well as genital contact) identified as heterosexual. Similarly, large proportions
of this sample of men (46.3%) and women (69.6%) who admitted to same-sex
attraction (ever) also preferred to see themselves as heterosexual.
We grouped people on the basis of three variables: self-identification as
heterosexual/nonheterosexual, the presence or absence of same-sex attractions,
and the presence or absence of same-sex partners. This produced eight groups, as
shown in Table II. Three of these groups (C, E, and G) contained few people (less
than 1% of the sample).

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

10:54

Style file version Nov. 19th, 1999

Childhood Sex-Atypicality and Sexual Behavior

557

Table II. Sexual Subgroups


Characteristics
Group
A
B
C
D
E
F
G
H

Male

Female

Same-Sex
attraction?

Non-Hetero
identity?

Same-Sex
behavior?

Yes
Yes
Yes
Yes
No
No
No
No

Yes
No
Yes
No
Yes
No
Yes
No

Yes
Yes
No
No
Yes
Yes
No
No

95
57
13
84
1
108
0
1363

5.5
3.3
0.8
4.9
0.1
6.3
0.0
79.2

74
92
25
293
0
63
2
2352

2.6
3.2
0.9
10.1
0.0
2.2
0.1
81.1

Approximately four in every five men and women appeared to be completely heterosexual; they self-identified as such and reported no same-sex attractions or behaviors. There was no sex difference in the prevalence of complete
heterosexuality ( p = 0.130). However, self-identification as nonheterosexual was
twice as common among men than it was among women ( 2 = 26.2, p < 0.0001).
Women who said that they had been sexually attracted to the same sex, but who
identified as heterosexual and did not report same-sex behavior, were twice as
common as men in this category (10.1% vs. 4.9%; 2 = 38.6, p < 0.0001). In
contrast, same-sex behavior in the absence of same-sex attraction or nonheterosexual identification was significantly more common among men than women
(6.3% vs. 2.2%; 2 = 49.9; p < 0.0001).
Childhood Gender Nonconformity (CGN)
The composition of our CGN measure is described in detail in the Methods
section. Scores for men ranged from 0 to 37.5, with a mean of 13.3 (SD = 6.4); high
scores indicate greater sex-atypicality in childhood. Among women, CGN scores
ranged from 0.5 to 38.5, with a mean of 13.4 (SD = 6.6). Figure 2 summarizes
data from men and women in each of five sexual subgroups (excluding groups C,
E, and G because of small numbers).
There was a clear linear trend for greater childhood gender nonconformity
as people moved away from complete heterosexuality. Two-way analysis of
variance included sexual subgroup, age-category, and their interaction, with analyses conducted separately for men and women. Among men, the main effect of
sexual subgroup was significant (F(4,1661) = 65.98, p < 0.0001), and there was
an interesting pattern in the results of Duncans multiple range tests, with post
hoc contrasts revealing significant differences between three of the four subgroups of men who identified as heterosexual and the one group who identified as bisexual/homosexual. The CGN measure appears to be sensitive to fairly
subtle variations, which is evident in the significant difference between complete

P1: FLF
Archives of Sexual Behavior [asb]

558

PL113-227174

September 19, 2000

10:54

Style file version Nov. 19th, 1999

Dunne, Bailey, Kirk, and Martin

Fig. 2. Childhood gender nonconformity among five sexual subgroups.

heterosexual men and men who had some same-sex behavior but who otherwise
identified as heterosexual and reported no same-sex attraction. In turn, this latter group had significantly lower CGN scores than did men who reported some
same-sex attraction (with or without same-sex behaviour). Finally, all groups were
significantly different from the men who identified as bisexual or homosexual.
A somewhat similar pattern occurred among women. The main effect of
sexual subgroup was significant (F(4,2811) = 47.12; p < 0.0001). Post hoc comparisons revealed that complete heterosexual women differed from women who
reported some same-sex behavior but no homosexual attraction or orientation.
In contrast to men, however, CGN scores did not differentiate between selfidentifying heterosexual women with same-sex behavior who did, or did not, report
any same-sex attraction. All four nominally heterosexual groups had significantly
lower CGN scores than the bisexual/lesbian women had.
These findings should be seen in the context of reported numbers of partners.
The great majority of Group F men (94/108, or 87%) and Group F women
(46/63, or 73%) reported having had sex with only one or two people of the same
sex in their lifetime. It seems, therefore, that CGN is able to differentiate between

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

10:54

Style file version Nov. 19th, 1999

Childhood Sex-Atypicality and Sexual Behavior

559

complete heterosexuals and heterosexuals who (probably) engaged in either


single or sporadic homosexual contact. Another interesting finding is that, among
men and women who said they were heterosexual but who admitted to some samesex attraction, CGN scores did not differentiate between those who had, or had
not, experienced homosexual behavior.
In these analyses of variance, subjects were grouped in four age bands (19
24 years, 2534 years, 3544 years and 4552 years). Among men, there was a
significant main effect of age on CGN scores (F(3,1673) = 4.57; p = 0.0034). Post
hoc contrasts revealed that the lowest CGN scores were reported by the youngestage group (mean = 12.59), and these were significantly lower than CGN scores
reported by men aged 3544 years (mean = 14.1). Among women, the main effect
of age was also significant (F(3,2792) = 8.81; p < 0.0001), although a different
age-cohort effect emerged: it was the women from the youngest-age group who
reported the most sex-atypicality in childhood (mean = 13.86), and this differed
significantly from the those in the oldest-age group (mean = 12.34).
Importantly, there was no significant interaction between age and sexual subgroup for either men ( p = 0.685) or women ( p = 0.355), indicating that the magnitude of the association between childhood sex-atypicality and sexual subgroup
was fairly constant across all ages.

Continuous Gender Identity (CGI)


The participants also rated how masculine or feminine they felt at the time of
the study. Data for each of the five main sexual subgroups are shown in Table III.
For both men and women, there were significant main effects of subgroup, reflecting greater femininity among males, or masculinity among females, with increasing
distance from complete heterosexuality (males: F(4,1661) = 44.71, p < 0.0001;
females: F(4,2778) = 74.45, p < 0.0001). Duncans post hoc comparison means revealed that complete heterosexuals (Group H) differed from all other subgroups,
Table III. Continuous Gender Identity
Male
Groupa
A
B
D
F
H
a Group

Female

Mean

SD

Contrastb

Mean

SD

Contrastb

93
55
82
106
1326

11.3
10.4
10.7
9.5
8.7

3.4
3.0
3.5
2.5
2.0

b,f,h
a,f,h
a,f,h
a,b,d
a,b,d,f

73
89
283
63
2290

11.3
9.1
9.6
9.1
8.1

3.2
2.3
2.8
2.4
1.8

b,d,f,h
a,h
a,h
a,h
a,b,d,f

A: Bisexual/homosexual attractions, identity and behavior; B: Hetero identity, with some


same-sex attractions and same-sex behavior; D: Hetero identity, some same-sex attraction, but
no same-sex behavior; F: Hetero identity, no same-sex attraction, but some same-sex behavior;
H: Hetero identity, no same-sex attraction or behavior.
b Duncans multiple range post-hoc comparison significant at p < 0.05.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

560

10:54

Style file version Nov. 19th, 1999

Dunne, Bailey, Kirk, and Martin

and that people who identified as bisexuals/homosexuals differed from each group.
In general, the intermediate heterosexual groups did not differ significantly from
each other on this measure.
The main effects of age on CGI scores were significant both for men (F(3,1642) =
4.44; p = 0.0041) and women (F(3,2778) = 10.83; p < 0.0001). Post hoc contrast
revealed a fairly linear age-cohort effect, with the oldest men and women reporting
the least gender nonconformity as adults. The interaction between age and sexual
subgroup in CGI scores was not significant for men ( p = 0.08) and was modest
for women (F(12,2778) = 1.92; p = 0.021).
Openness and Accuracy of Self-Reports
The majority of respondents believed that they had been completely open in
answering the questionnaire (see Table IV). In the case of men in the five different
sexual subgroups, the percentage of men who said that they were completely open
ranged from 83.5% of complete heterosexuals (Group H) to 87.7% of those
who said that they were heterosexual but who reported some same-sex attraction
and behavior. There was slightly more variance in this measure among the female
subgroups, with the percentages of those who were completely open ranging from
81% in Group D to 90% in Group B.
Fewer people believed that their answers provided a completely accurate
reflection of their true feelings and behaviors (see Table IV). Among males, there
was little variance between sexual subgroups (ranging from 57% in Group C to
65% in Group B). The groups of women differed somewhat, from 51% of group
F to 64% of Group B.
We assume that self-reports of heterosexuality are least sensitive to reporting
biases. Using the complete heterosexual group (Group H) as a reference point
in chi-square tests, we found that the percentages of respondents in each of the
smaller subgroups who said that they were completely open, or that their answers
Table IV. Percentages Who Reported That They Had Been Completely Open and That
Answers Were a Completely Accurate Reflection of True Feelings, by Sexual Subgroup
Males

Females

Groupa

Open

Accurate

Open

Accurate

A
B
D
F
H

87.4
87.7
84.2
87.0
83.5

64.2
64.9
57.3
64.5
61.2

87.8
90.1
81.0
87.1
82.6

55.4
64.4
54.4
50.8
59.4

a Group

A: Bisexual/homosexual attractions, identity and behavior; B: Hetero identity,


with some same-sex attractions and same-sex behavior; D: Hetero identity, some samesex attraction, but no same-sex behavior; F: Hetero identity, no same-sex attraction, but
some same-sex behavior; H: Hetero identity, no same-sex attraction or behavior.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

Childhood Sex-Atypicality and Sexual Behavior

10:54

Style file version Nov. 19th, 1999

561

provided a completely accurate reflection of their true experiences, did not differ
significantly from Group H on either measure for both males and females.
DISCUSSION
Consistent with previous research, we found substantial differences between
self-identifying homosexuals/bisexuals and heterosexuals in their recall of sexatypical play and interests during childhood. This basic demarcation of two groups
yielded the single largest difference in means (see Fig. 2). This is a robust finding,
which seems entirely predictable (Bailey and Zucker, 1995; Bell et al., 1981).
In addition, though, we have found the measure of CGN to be very sensitive
to subtle variations in distance away from complete heterosexuality. Variation
within groups of people who identify as heterosexual has previously been observed
among males (McConaghy and Silove, 1991; McConaghy et al., 1994), but not
among females.
Among both men and women, most notable was the observation that CGN
scores differed significantly between complete heterosexuals and self-identifying
heterosexuals who admitted no same-sex attractions but some homosexual behavior. Further, this effect was not limited to recollections of childhood. Self-rated
adult feelings of masculinity and femininity also were sensitive to subtle variation in homosexual attractions and behaviors among people who identified as
heterosexual.
Are these findings artefactual? Differences between homosexuals/bisexuals
and heterosexuals in social norms for or against recall of gender atypicality could
influence their inclination to report feelings truthfully. However, this seems an unlikely explanation for the relatively linear increase in CGN across the five sexual
subgroups observed here: what social script could plausibly dictate, for example,
greater willingness to report childhood feminine interest, friendships, and avoidance of rough-and-tumble play among men who have one or a few same-sex contacts ever, but who otherwise dont classify themselves as bisexual or homosexual?
Also, consider that we found very similar associations between CGN and sexuality in both men and women. Would we expect social norms to exert equivalent
influence on recall of CGN in both sexes, when most social prescriptions regarding
gender roles and stereotypes have quite different effects on men and women?
It remains possible that the difference in CGN between self-identifying heterosexual and homosexual/bisexual groups is real, but that the trend within the
heterosexual group is artefactual, and arises simply because of greater openness
among a minority of heterosexuals to talk about their true feelings. We examined
this indirectly by gauging the extent to which respondents felt totally open and
whether, on reflection, their answers provided an accurate picture. Assuming that
people who identify as completely heterosexual have the least reason to dissimulate, we compared them to all other groups. On neither measure was there a

P1: FLF
Archives of Sexual Behavior [asb]

562

PL113-227174

September 19, 2000

10:54

Style file version Nov. 19th, 1999

Dunne, Bailey, Kirk, and Martin

significant difference between complete heterosexuals and heterosexuals who


admitted to some same-sex feelings or behavior.
Implications for Developmental Theory
It is clear from this and other works (McConaghy et al., 1994) that the majority of people who admit to ever having homosexual attractions or behavior also
consider themselves to be heterosexual. This group is more than twice as large as
the number of people who self-identify as bisexual or homosexual. At one level,
it may be possible to dismiss the (perhaps) transient homosexual attraction and
occasional behavioral experimentation by heterosexuals as being uninformative
as far as essential impulses are concerned; variation within heterosexual groups
could arise mainly because of social and interpersonal influences experienced by
some individuals but not others. However, the apparent sensitivity of our measures
of sex-atypicality to subtle variations in sexual experience suggests otherwise.
Higher than average sex-atypicality among some heterosexuals could indicate that
these people are more likely than most to reach one of the lower thresholds of
a homosexual continuum. In terms of Bems EBE theory, slight sex-atypicality
might engender slight homoeroticism, and this could explain why some heterosexual people respond to opportunities for homosexual experimentation, whereas
most do not (Bem, 1996).
Unfortunately, variation within heterosexual groups has not been examined
in most etiologic studies of homosexuality, such as family studies of genetic linkage and sibling sex ratios, which rely heavily on self-selected samples of overtly
homosexual people (Bailey and Pillard, 1991; Bailey et al., 1993; Blanchard and
Bogaert, 1996; Hamer et al., 1993). In a separate analysis of these twin data,
we found substantial familial aggregation of homosexual orientation (defined in
terms of Kinsey-type measures of sexual attractions and fantasy), but our confidence intervals for heritability estimates were so wide that we could not reject the
null hypothesis of no genetic influence (Bailey et al., 2000). Importantly, though,
childhood gender nonconformity was significantly heritable for both men and
women, which indicates that this precursor of homoeroticism may have a genetic
basis (Bem, 1996).
Adequate empirical tests of etiologic theory must include data on multiple
thresholds along a continuum of homosexuality. As Pattatucci (1998) recently said
Definitions set parameters and thus constrain possibilities (p. 370). Estimates of
the variance explained by genetic, hormonal, and other biological factors will
always differ depending upon how narrowly the trait is defined. We suggest that
future developmental studies should include measurement of sex-atypicality as part
of multidimensional assessments of attractions, fantasies, and overt orientation.
One final point which now seems justifiable to make is that apparently sporadic

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

Childhood Sex-Atypicality and Sexual Behavior

10:54

Style file version Nov. 19th, 1999

563

homosexual behaviors should not be dismissed as uninformative, but rather should


be seen as a potential signal of true psychosexual variation.
Limitations
Despite the large size of this nationwide sample, and the relatively broad
assessment of sexuality, this study has some significant limitations. It is based on
a postal survey of twins in a longitudinal research cohort, and thus cant be seen as
a true reflection of the population; indeed, there were some differences in sexual
behaviors between the twins who volunteered and those who refused (Dunne et al.,
1997c, 1998). Although we have no reason to believe that the distribution of sexual
experiences of twins is unusual (McConaghy et al., 1994), we have no normative
data on Australians against which to compare.
The use of a twin sample would probably minimize the total variance in
the data if, as appears likely, there are significant correlations between twins in
many aspects of sexuality (Bailey et al., 1993, 2000; Dunne et al., 1997b). One
effect would be to minimize the denominator in the calculation of F statistics.
It is uncertain, therefore, whether the statistically significant association between
recalled childhood sex-atypicality and degree of homosexual orientation found
here would emerge to the same degree in nontwin samples. Another possible
complication is that the present sample is very large, and smaller studies might not
have sufficient power to detect subtle effects. These patterns should be examined
in future surveys with various groups, including random probability samples from
otherwise unselected populations.
ACKNOWLEDGMENTS
Funded by a grant to JMB from the US National Institute of Mental Health
(USA) and a Commonwealth AIDS Research Grant (Australia) to NGM and MPD.
The authors thank Ann Eldridge, John Pearson, Olivia Zheng and Dr Gu Zhu for
assistance, and the twins for their cooperation.
REFERENCES
Bailey, J. M., Dunne, M. P., and Martin, N. G. (2000). Genetic and environmental influences on sexual
orientation and its correlates in an Australian twin sample. J. Pers. Soc. Psychol. 78: 524536.
Bailey, J. M., and Pillard, R. C. (1991). A genetic study of male sexual orientation. Arch. Gen. Psychiat.
48: 10891096.
Bailey, J. M., Pillard, R. C., Neale, M. C., and Agyei, Y. (1993). Heritable factors influence sexual
orientation in women. Arch. Gen. Psychiat. 50: 217223.
Bailey, J. M., and Zucker, K. J. (1995). Childhood sex-typed behavior and sexual orientation: A
conceptual analysis and quantitative review. Dev. Psychol. 31: 4355.

P1: FLF
Archives of Sexual Behavior [asb]

564

PL113-227174

September 19, 2000

10:54

Style file version Nov. 19th, 1999

Dunne, Bailey, Kirk, and Martin

Baker, L. A., Treloar, S. A., Reynolds, C., Heath, A. C., and Martin, N. G. (1996). Genetics of educational
attainment in Australian twins: Sex differences and secular changes. Behav. Genet. 26: 89102.
Bell, A. P., Weinberg, M. S., and Hammersmith, S. K. (1981). Sexual Preference: Its Development in
Men and Women, Indiana University Press, Bloomington.
Bem, D. (1996). Exotic becomes Erotic: A developmental theory of sexual orientation. Psych. Bull.
103: 320355.
Biggar, R. J., and Melbye, M. (1992). Responses to anonymous questionnaires concerning sexual
behavior: A method to examine potential biases. Am. J. Public Health 82: 15061512.
Blanchard, R., and Bogaert, A. F. (1996). Homosexuality in men and number of older brothers. Am. J.
Psychiat. 153: 2731.
Blanchard, R., and Freund, K. (1983). Measuring masculine gender identity in females. J. Cons. Clin.
Psychol. 51: 205214.
Blanchard, R., McConkey, J. G., Roper, V., and Steiner, B. W. (1983). Measuring physical aggressiveness in heterosexual, homosexual, and transsexual males. Arch. Sex. Behav. 12: 511524.
Dunne, M. P., Martin, N. G., Bailey, J. M., Heath, A. C., Bucholz, K. K., Madden, P. A. F., and Statham,
D. J. (1997a). Participation bias in a sexuality survey: Psychological and behavioral characteristics
of responders and nonresponders. Int. J. Epidemiol. 26(4): 844854.
Dunne, M. P., Martin, N. G., Statham, D. J., Slutske, W. S., Dinwiddie, S.H., Madden, P. A. F., Bucholz,
K. K., and Heath, A. C. (1997b). Genetic and environmental contributions to variance in age at
first sexual intercourse. Psychol. Sci. 8: 211216.
Dunne, M. P., Martin, N. G., Pangan, T., and Heath, A. C. (1997c). Personality and change in the
frequency of religious observance. Pers. Indiv. Diff. 23: 527530.
Dunne, M. P., Bailey, J. M., Martin, N. G., and Heath, A. C. (1998). Self-disclosure of sensitive
personal experiences among people who refused to volunteer for a sex survey. Paper presented at:
International Academy of Sex Research Conference, Sirmione, Italy (June, 1998).
Finn, S. E. (1987). The structure of masculinity and femininity self ratings. Unpublished manuscript.
Freund, K., Langevin, R., Satterborg, J., and Steiner, B. (1977). Extension of the gender identity scale
for males. Arch. Sex. Behav. 6: 507519.
Green, R. (1987). The Sissy Boy Syndrome and the development of homosexuality, Yale University
Press, New Haven.
Grellert, E. A., Newconb, M. D., and Bentler, P. M. (1982). Childhood play activities of male and
female homosexuals and heterosexuals. Arch. Sex. Behav. 11: 451478.
Hamer, D. H., Hu, S., Magnuson, V. L., Hu, N., and Pattatucci, A. M. (1993). A linkage between DNA
markers on the X chromosome and male sexual orientation. Science 261: 321327.
Heath, A. C., Bucholz, K. K., Slutske, W. S., Madden, P. A. F., Dinwiddie, S. H., Dunne, M. P., Statham,
D. J., Whitfield, J. B., Martin, N. G., and Eaves, L. J. (1994). The assessment of alcoholism in
surveys of the general community: What are we measuring? Some insights from the Australian
twin panel interview survey. Int. Rev. Psychiat. 6: 295307.
Martin, N. G., Eaves, L. J., Heath, A. C., Jardine, R., Feingold, L. M., and Eysenck, H. J. (1986).
Transmission of social attitudes. Proc. Nat. Acad. Sci. 83: 43644368.
McConaghy, N. (1987). Heterosexuality/Homosexuality: Dichotomy or continuum? Arch. Sex. Behav.
16: 411424.
McConaghy, N., Buhrich, N., and Silove, D. (1994). Opposite sex-linked behaviors and homosexual
feelings in the predominantly heterosexual male majority. Arch. Sex. Behav. 34: 565577.
McConaghy, N., and Silove, D. (1991). Opposite sex behaviours correlate with degree of homosexual
feelings in the predominantly heterosexual. Aust. N. Z. J. Psychiat. 25: 7783.
Mitchell, J., and Zucker, K. J. (1992). The Recalled Childhood Gender Behaviors Questionnaire:
Psychometric properties. Paper presented at: International Academy of Sex Research Conference,
Barrie, Ontario.
Pattatucci, A. M. (1998). Molecular investigations into complex behavior: Lessons from sexual orientation studies. Hum. Biol. 70: 367386.
Phillips, G., and Over, R. (1992). Adult sexual orientation in relation to memories of childhood genderconforming and gender nonconforming behaviors. Arch. Sex. Behav. 21: 543558.
Risman, B., and Schwartz, P. (1988). Sociological research on male and female homosexuality.
Ann. Rev. Sociol. 14: 125147.
Ross, M. W. (1980). Retrospective distortion in homosexual research. Arch. Sex. Behav. 9: 523531.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227174

September 19, 2000

Childhood Sex-Atypicality and Sexual Behavior

10:54

Style file version Nov. 19th, 1999

565

Sundet, J. M., Magnus, P., Kvalem, I. L., Samuelson, S. O., and Bakketeig, L. S. (1992). Secular trends
and sociodemographic regularities in coital debut age in Norway. Arch. Sex. Behav. 21: 241252.
Wiederman, M. W. (1993). Demographic and sexual characteristics of nonresponders to sexual experience items in a national survey. J. Sex. Res. 30: 2735.
Zucker, K. J. (1990). Gender identity disorders in children: clinical descriptions and natural history. In
Blanchard, R., and Steiner, B. W. (Eds.), Clinical Management of Gender Identity Disorders in
Children and Adults, American Psychiatric Press, Washington, DC, pp. 123.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 29, No. 6, 2000

Psychoanalysis and Sexual Fantasies


Richard C. Friedman, M.D.,1,3 and Jennifer I. Downey, M.D.2

Psychoanalysis began as a depth psychology, heavily based on the sexual experiences and memories of patients. A long-term treatment, utilizing a free association
method, psychoanalysis has provided a window onto the meanings and functions
of fantasy, including sexual fantasy. Although psychoanalysis has produced some
scientific research, the field has tended to rely on observational data collected
from individuals studied in depth. Sex research on the other hand, carried out by
investigators from different disciplines is based on empirical investigation. Each
field has made contributions fundamentally important to the other. In this article,
we review psychoanalytic ideas about human sexuality and distinguish those that
have been invalidated by systematic research from those that remain useful. Perhaps, the single most important revision of psychoanalytic theory during the past
century was concerned with the psychological development of girls and women.
We separately discuss the development of the sexes, and stress the need for bridge
building between psychoanalysis and sex research.
KEY WORDS: psychoanalysis; fantasy; sexual fantasy; ambivalence.

INTRODUCTION
Psychoanalysis is a depth psychology that originated from the study of sexually tinged fantasy. A problem present in the field from its inception, however,
has been the lack of a definition of erotic fantasy or even a general sense of agreement about its specific attributes. One reason for this may be that Freud blurred
the distinction between the sexual and that which nonpsychoanalytically oriented
laypersons might consider nonsexual. He believed that stimulation of various zones
1 Department of Psychiatry, and Payne Whitney Clinic, Cornell University Medical College, New York,

New York.
2 New York State Psychiatric Institute, Columbia University College of Physicians and Surgeons, New

York, New York.


whom correspondence should be addressed at 225 Central Park West, #103, New York, New
York 10024.

3 To

567
C 2000 Plenum Publishing Corporation
0004-0002/00/1200-0567$18.00/0

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

September 11, 2000

14:2

568

Style file version Nov. 19th, 1999

Friedman and Downey

of the body, oral, anal, and phallic, led to pleasureful sensations, which he considered sexual (Freud, 1940). The psychoanalytic literature, therefore, contains
countless articles, nominally about sexuality, discussing bodily zones invested
with the so-called libidinal energy. The libido theory, placed sexual energy at
the center of all human motivation and all psychopathology, is outdated and only
of historical interest.
The topic of sexual fantasy is more complex in women than men, and we
reserve consideration of its meaning(s) in women until later in this article. In men,
by sexual fantasy we mean subjectively experienced narratives that are associated
with psychophysiological changes that occur during sexual excitement; and which
always include a consciously experienced emotion that is explicitly sexual or lustful. These fantasies are experienced during masturbation, and are mobilized by
pornography and many other stimuli of daily life (Freund et al., 1974; Masters and
Johnson, 1966; Paredes and Baum, 1997; Stoller, 1979).
THE FUNCTION OF FANTASIES
Freud suggested that daydreams were a continuation of childhood play and
were the product of wishes that compensated for lifes frustrations. Person has
pointed out that fantasies are a type of imaginative thought that serve many different
functions (Person, 1995). As Freud observed, they may represent wishes evoked in
response to frustration in order to convert negative feelings into pleasurable ones.
They may soothe, enhance security, and bolster self-esteem, or repair a sense
of having been abandoned or rejected. Fantasies may (temporarily) repair more
profound damage to the sense of self that occurs as a result of severe trauma. They
also frequently serve role rehearsal functions, as occurs, for example, in little girls
who consider dolls to be their babies and who play house as preparation for
becoming adult mothers. Organized as images, metaphors, and dramatic action,
fantasies in the form of artistic productions and mythology have been part of the
human heritage probably for the entire life of our species. Freud provided a new
framework for understanding these universal forms of human expression by noting
that they could be critically analyzed in a similar fashion as all other products of
the mind.
THE MEANINGS OF FANTASY: CONSCIOUS
AND UNCONSCIOUS
The story line of a fantasy, meaningful in itself, also symbolically expresses
additional hidden meanings. Underneath one narrative is another, and under that
yet another, arranged in layers as is the mind itself. A fundamental discovery of
Freuds, and one that remains valid today (unlike many of his ideas about human
sexuality), is that some aspects of mental functioning are not subject to conscious

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

September 11, 2000

Psychoanalysis and Sexual Fantasies

14:2

Style file version Nov. 19th, 1999

569

recall. Even when unconscious, they may influence motivation. Connections between conscious and unconscious thoughts, feelings, and memories occur in the
form of associations. Just as neural networks exist in the brain, so do psychic
networks in the mind, although the precise correspondences between the two have
not yet been clarified (Olds, 1994). Freud termed his unique method of exploring
the pathways of these connections as psychoanalysis. His initial explorations
in his self-analysis, carried out at the end of the nineteenth century, led to the
awareness that the closer one comes to mental processes that are out of awareness,
the more the rules governing mental organization appear to change. Although the
thinking processes of ordinary daily life are more or less logical, unconscious
mentation seems to be organized more like dreams. In dreams, many ideas, memories, and feelings may be represented by a single image. The narrative line of a
dream consists of strings of such symbols and emotions that may or may not be
ordinarily connected with the images as usually experienced during waking life, all
arranged without regard to ordinary time/space rules of the physical universe. In
dreams everyone can be anyoneman, woman, or child, or even nonhumanand
all is possible. Freud termed the organization of the unconscious part of the mind
the primary process and contrasted it with the secondary processthe system
of organization of ordinary, everyday thinking (Freud, 1940). A perspective about
fantasy, unique to psychoanalytic psychology, is that beneath the immediately
coherent narrative line of waking fantasy are disguised stories that carry hidden
meanings. These latent narratives consist of memories of real and imagined events
linked in the imagination of the present. Since single symbols can represent multiple meanings, the amount of information carried by sequences of symbols is vast.
Another core psychoanalytic idea is that one reason why some story lines are
unconscious is because they contain wishes that are unacceptable to the conscience.
The mind has the capacity to erase from its awareness certain unpleasant ideas,
but not the power to completely eliminate their motivational force.
SEXUAL FANTASIES
Sexual fantasies are stories told to ourselves that are embedded in sexual
feelings. These depend in both sexes on adequate enough blood levels of androgen
(Money and Ehrhardt, 1972). Pharmacological blockade of the effects of androgen
eliminates sexual feelings and thus the motivation to participate in sexual activity. The reason this is important from a psychoanalytic perspective is that psychoanalysts specialize in illuminating unconscious motivation. A psychoanalyst
would immediately ask whether elimination of sexual feelings from the conscious
mind actually meant that they were totally absent; could they be unconscious? Of
course, there is no definitive way of proving that they are not. Total loss of sexual interest, however, and of erotic activity without positive evidence that sexual
motivation is present, suggests otherwise (Bradford, 1995). Thus, when sexual

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

September 11, 2000

14:2

570

Style file version Nov. 19th, 1999

Friedman and Downey

fantasy is eliminated by antiandrogens, it is best conceptualized as absent from the


mind. Since personality functioning is preserved when this occurs, it is an empirical demonstration that Freuds ideas about the central importance of the role of
sexuality in total personality functioning were incorrect.

MALE SEXUAL FANTASY


In physiologically normal adults, the most important influence on the contents
of sexual fantasy is probably gender. Male sexual fantasies are best understood
as part of the psychology of men; female sexual fantasies in the context of the
psychology of women.

THE OEDIPUS COMPLEX


Psychoanalytic ideas about the development of sexual fantasy have been
greatly influenced by Freuds speculations about the Oedipus complex. In keeping
with scientific progress, these have required extensive revision since his death.
Here we consider his thoughts about males, of which only some have withstood the test of time.
Freud, a physician and neurologist, was fascinated with the Oedipal myth long
before he ever made systematic observations about human psychology (Sulloway,
1979). His ideas about the role of the Oedipus complex were formulated in the
late 1890s and remained unchanged throughout his life.
According to Freud, the sexual development of children is biologically determined to occur in two waves of intensity; or, in Freuds terms, diphasically.
As he put it:
. . . It is further found that these phenomena which emerge in early childhood form part of
a regular process of development, that they undergo a steady increase and reach a climax
toward the end of the fifth year after which there follows a lull. During this lull, progress
is at a standstill and much is unlearned and undone. After the end of this period of latency
as it is called, sexual life is resumed with puberty, or, as we might say, it has a second
efflorescence. Here we come upon the fact that the onset of sexual life is diphasic; that it
occurs in two waves, this is unknown except in man. . . . (Freud, 1905)

Freud believed that human beings are biologically determined to be erotically


attracted to their opposite-sexed parents. He considered this incestuous wish to be
a part of the biological heritage of all people and to increase in intensity during
early childhood, reaching a peak between ages 4 and 5. As a result of his incestuous
desire, the boy finds himself in a competitive relationship with his father. As part
of a complex of linked feelings, associated with the erotic desire for his mother,
is rage at his father. The child fears retaliation from his father for his incestuous
and parricidal wishes. Freud termed the boys fear castration anxiety, although
he actually meant fear of penectomy. Freud viewed castration fear as the third

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

Psychoanalysis and Sexual Fantasies

September 11, 2000

14:2

Style file version Nov. 19th, 1999

571

component of the Oedipus complex (after desire for the mother and competition
with the father).
The childs next developmental stage was characterized by two central psychological processesidentification with the father and repression of the Oedipus
complex. These processes were associated with the formation of the superego,
a name given by Freud for what is generally termed as the conscience. Repression
of the Oedipus complex and identification by the son with his father leads to the
capacity to experience guilt, to regulate impulses, and to develop normal morality.
Freud also proposed that as a result of the repression of the Oedipus complex, children went through a period of latency, which gets terminated by the hormonal
surge of puberty.
We have criticized Freuds ideas about the Oedipus complex elsewhere and
refer the interested reader to Friedman and Downey (1995) for a more extensive
discussion of this issue. The major aspects of the theory that have become obsolete
are as follows:
1. Freuds ideas about the timing of developmental processes affecting or
influencing psychosexuality were found to be invalid. Freud reviewed the
sparse database available to him for possible effects of prenatal hormones
on neuroembryology and behavior, and rejected the likelihood that they
had an important effect. This has proved to be erroneous. Thus, his exclusive emphasis on postnatal biopsychological events was misguided
(Gorski, 1991; Money, 1998; Money and Ehrhardt, 1972).
2. Also invalid was his belief about the chronology of sexual desire, e.g., his
discussion of the biphasic nature of childhood sexuality. The concept
of a universal norm consisting of an intense surge of sexual arousal at
age 45 followed by latency has little empirical support. There is actually
substantial variability in the way sexuality is experienced and expressed,
depending on constitutional as well as sociocultural influences. Early genital exploration, including masturbation, commonly occurs prior to age 45
(Galenson and Roiphe, 1981), although there is no reason to believe that
such activity is normative. Throughout childhood, boys participate in
sexual activity more frequently than girls, a sex difference that is preserved
later in life. Childhood sexual activity is likely to consist of genital inspection and manual exploration, but not sexual intercourse unless there has
been a history of sexual abuse. Among many diverse developmental pathways is one in which there is a linear increase in sexual interest and activity
over time without evidence of a quiescent phase of latency (Yates, 1993).
3. The hypothesis (expressed by Freud as a discovery) that there is a universal
childhood incestuous wish has not only failed to be validated, but there is
considerable evidence indicating that it is erroneous (Erickson, 1993).
4. Although there is no evidence of universal parricidal wishes, it is possible
to conceptualize boys aggressive/competitive feelings about their fathers
(and vice versa) as falling on a developmental line separate from sexual

P1: FLF
Archives of Sexual Behavior [asb]

572

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Friedman and Downey

development. This line reflects the influence of prenatal androgens on aggression, especially as expressed in childhood (Friedman and Downey,
1995).
5. The universality of castration fear during early childhood also requires
additional empirical validation. Although fears of bodily injury, including
genital injury, are common, their role in child development remains to be
clarified (Schrut, 1993).
6. The development of conscience does not follow the pathway outlined by
Freud. Moral development has been shown to be the result of complex interactions between many genetic, cognitive, and psychosocial influences
that occur throughout childhood (Kohlberg, 1981).
PSYCHOANALYTIC PERSPECTIVES ABOUT THE OEDIPUS
COMPLEX AND SEXUAL FANTASY
Freuds ideas about the role of the Oedipus complex were quite specific and
rooted in his ideas about the role of biology in psychological development. Subsequent psychoanalytic scholars, however, have frequently discussed the Oedipus
complex in a much less precise fashion (Greenberg, 1988). A tendency emerged
to blur distinctions between erotic and nonerotic, parents and their symbolic representations, and castration anxiety and its symbolic representations. In addition,
psychoanalysts frequently viewed the timing of childhood events, retrospectively
reported during psychoanalysis, as being accurate. For example, consider a 40year-old man who was being treated with psychoanalysis because of work inhibitions. Such a patient might have a dream that he was to deliver a speech before
a committee of older men. Following it, he stubbed his toe and it bled and then
fell off. Psychoanalysts throughout the world, during Freuds life and even today,
might interpret the patients bleeding toe as a symbolic representation of castration
anxiety even though no history might ever have been uncovered of the man actually fearing damage to his genitals. It is also credible enough to assume that the
older men represented father figures and that, having been frightened of his father
as a child, he remained so unconsciously even though his actual father was long
dead. Many psychoanalysts would assume that the anxiety symbolically expressed
in his dream and causing his vocational inhibition was triggered by unconscious
incestuous desire. This assumption might be made, regardless of the facts of the
patients sexual history, in light of the Oedipus complex described by Freud as
occurring during early childhood. Using this case as an example of more general
phenomena, we speculate that this person may never have desired to possess (or
even be stimulated by) his mother erotically. He may, however, have felt competitive towards his father for his mothers attention. The fear of competition with
his rival may have been experienced in imagery with symbolic connotations of
genital damage. As an adult psychoanalytic patient, the man may have in addition

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Psychoanalysis and Sexual Fantasies

573

expressed sexual conflicts that seemed to be connected in his mind to his fear of
older men. It is likely that these sexual components of his wishes and fears were
integrated into his life narrative, however, years after what classical psychoanalysts
consider the Oedipal phase of development.
We do not believe that it is valid to infer the existence of an Oedipus complex
as a universal norm, as described by Freud, either from the narratives of adults
or even children in the psychoanalytic situation. We do believe that subgroups of
people experience such an Oedipus complex and that Oedipal fears influence
the form of sexual psychopathology in many patients.
PSYCHOANALYSIS, THE OEDIPUS COMPLEX, AND THE
PATHOLOGICAL MODEL OF HOMOSEXUALITY
There are many ways of categorizing the stimuli that elicit sexual excitement
in men, and any method selected must necessarily be somewhat arbitrary. We, therefore, find it helpful to think of three large categorieshomosexual, heterosexual,
and bisexualwithin which are many subgroups, depending on the specific characteristics of objects and situations that are sexually stimulating. Men of homosexual
orientation are predominantly or exclusively sexually excited by males; those of
heterosexual orientation by females; and those of bisexual orientation by both.
Although Freud (1905) believed that the origins of heterosexual orientation
were obscure, psychoanalysts in the next generation tended to regard heterosexuality in all of its manifestations as a biological norm and other forms of sexual orientation as derailments produced by psychosocial traumata (American Psychiatric
Association, DSM-I, 1957; DSM-II, 1968; Socarides, 1978). Freud also believed in
universal bisexuality. By this he meant that all people have members of both sex as
sexual objects in fantasy. The attraction to one or the other sex may be more or less
conscious. In Freuds view (Freud, 1905), a man who consciously was exclusively
heterosexual nonetheless had unconsciously made a homosexual object selection.
Similarly a man who was exclusively homosexual harbored in his unconscious
mind a heterosexual object. Freuds theory of bisexuality has been extensively
criticized (Friedman, 1988; Friedman and Downey, 1993a,b). There is virtually
no evidence that this theory is valid, and we do not use the term bisexuality
in the Freudian sense. Modern gender psychology indicates that all people make
various types of bigender identifications. These have conscious and unconscious
components but do not influence gender identity once it has differentiated.
The psychoanalytic community viewed homosexual orientation to be pathological until quite recently, lagging behind the general psychiatric community. The
reasons for this have been extensively discussed by Lewes (1988), Isay (1989),
Friedman (1988), and Friedman and Downey (1998).
Perhaps the most compelling reason that psychoanalysts adhered to the pathological model of homosexuality, however, was widespread conviction that Freuds

P1: FLF
Archives of Sexual Behavior [asb]

574

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Friedman and Downey

ideas about the Oedipus complex were accurate. Believing that boys were destined
to be erotically attracted to their mothers, psychoanalysts could see no pathway
except a pathological one for the development of homosexual orientation. Thus,
a (simplified) idea about homosexual orientation was widely accepted. It was believed that homosexual orientation was defensively motivated as a response to
unconscious anxieties about heterosexuality. Thus, each time a man experienced
a sexual desire for a man, he enacted in his mind a pathological scenario. It was
also widely believed that men of homosexual orientation had defective superego
functioning because of inadequate identification with their fathers. It was generally
held that psychoanalysis could and should convert sexual orientation from homosexual to heterosexual (Socarides, 1978). These ideas are formally viewed as being
outdated and invalid by the American Psychoanalytic Association. Psychoanalysis
is an international enterprise, however, attitudes towards homosexuality in some
European and Latin American countries tend to be those that were in vogue in
North America in the 1950s.
PSYCHOANALYTIC PERSPECTIVES CONCERNING
EROTIC IMAGERY
Freud (1905) considered young children to be innately polymorphously perverse and that the capacity to be excited by so called perverse stimuli remained
during adulthood as part of the normal sexual response. Freud was probably more
interested in disorders that he termed neuroses than those which would now be
called paraphilias (American Psychiatric Association, DSM-IV, 1994). He was
also interested in sexual experience as it occurred in day-to-day living among people without psychiatric disabilities. The interests of subsequent psychoanalysts
have been similar. Much more attention has been devoted to the so called perverse sexual fantasy and activity among normal and neurotic people than among
those with paraphilias, although some psychoanalysts have made important contributions to understanding the psychology of the latter (Stoller, 1975a,b). This is
probably because patients with paraphilias tend to have limitations in social skill
and capacity for insight, and respond poorly to insight-oriented psychotherapy
including psychoanalysis (Abel et al., 1992).
Psychoanalysts as well as nonpsychoanalysts have speculated on the reason
that paraphilias and paraphiliac-like erotic imagery should occur so much more
commonly among males than among females. An extremely abbreviated summary
of their views is that the experiences with abusive and neglectful caretakers, generally women during early childhood, beginning well prior to the Oedipal developmental phase, are of great etiological importance. The specific paraphiliac image
selected by a given individual is often a function of actual experiences he has had
during childhood. For example, Stoller discussed a case of a fetishist cross-dresser
who was forcibly cross-dressed during childhood (Stoller, 1975a,b). Sometimes,
however, the determinants of specific fantasy remain obscure. In many of these

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Psychoanalysis and Sexual Fantasies

575

instances, psychoanalysts assume that the image was generated in fantasy. The sex
difference in frequency of paraphilias is compatible with the sex difference in frequency of nonsexual violence and with the theory that hostility and sexual arousal
are fused in these disorders. Stoller (1975a,b) stressed that paraphiliacs need to
experience the sexual object in a dehumanized way, as a so-called part-object or
a fetish. Themes of revenge (usually directed at the symbolic representation of
the mother) lead to fantasies of torture, control, and domination. The need to soil
and devalue the object is in keeping with anal influences persistent in unconscious
mental life (Kernberg, 1995). Regardless of the particular fantasy, there is a general tendency to experience the sense of danger overcome by the powerful sense
of self in the sexually arousing situation. A number of psychoanalytic writers have
also conceptualized the diverse objects and situations that psychoanalysts have traditionally considered perverse as defensive efforts to master traumata (Stoller,
1975a,b). A generally accepted idea is that a range exists with regard to intensity and exclusivity with which paraphiliac-like imagery are experienced. Most
men experience such imagery at least to some degree. The determinants of intensity, frequency, and exclusivity vary depending on constitutional and psychosocial
developmental influences. We return to the topic of so-called perverse sexual
fantasies in nonparaphiliac people later in this article.
MALE SEXUAL FANTASY: FURTHER DEVELOPMENTAL
CONSIDERATIONS
The contents of erotic fantasy in males appear to be usually formed prior
to puberty, more or less at the age of adrenarche (see Herdt and McClintock,
2000). Once experienced with the full erotic intensity that occurs with complete
androgenization, they tend to define the limits of erotic arousal during the entire life
span. Stimuli that fall outside a persons pattern are experienced with indifference.
For that reason, some authors have referred to individually specific sexual pattern
as sexprints (Person, 1995) or lovemaps (Money, 1988). We find it helpful to
think of male sexual fantasy in terms of two major dimensions. We call the first,
orienting sexual fantasy by which we mean whether the person responds erotically
to stimuli that are same gender, opposite gender, or both.
The developmental differences between individuals who are homosexual,
heterosexual, and bisexual remain to be ascertained. From a psychoanalytic perspective, boys and men who have the capacity for bisexual arousal are of unusual
interest. Theoretically, such people might repress or amplify either the homo- or
the hetero-erotic component of their erotic imagery depending on life stress and
psychodynamic influences. Such men might, for example, experience themselves
as changing sexual orientation during various life phases and attribute such change
to life circumstances (such as falling in love), or psychotherapeutic interventions
of one type or another. However, it is important that the experiences of men in this
group not be taken as valid for all men.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

September 11, 2000

14:2

576

Style file version Nov. 19th, 1999

Friedman and Downey

The second dimension of sexual fantasy that we just alluded to concerns


the situations and objects that characterize erotic imagery. We hypothesize that
both dimensions of sexual experience become more or less fixed during later
childhood, although the outer limits of such differentiation remain to be established
by research. We qualify this, however, by leaving room for unusual men whose
sexual fantasy life seems to follow idiosyncratic rules.
FEMALE SEXUAL FANTASY
Although male erotic fantasy functions as a limit defining the realm of that
which is erotically possible, the situation with respect to sexual excitement in
women is more variable. In men, sexual arousal is generally equated with an awareness of a specific affect (e.g., lust), erection, and intense desire to achieve orgasm.
Although the female equivalent of this occurs in some women, it is not universal. There is far more variability in the multiple behavioral dimensions constituting
sexual experience. In order to understand the significance of developmental aspects
of female sexuality, it is helpful to first consider aspects of the sexual experience
and activity of adult women.
SOMATIC AND SUBJECTIVELY EXPERIENCED ASPECTS
OF FEMALE SEXUAL RESPONSE
In laboratory situations, there is a good correlation between a mans awareness
of sexual excitement/arousal and objective measurements of erection. This is not
so for women. A number of investigators have confirmed that the relationship
between what has been termed genital arousal (more or less the equivalent of
male erection), and the subjective sense of feeling sexually aroused, is inconstant
at best. In fact, at least in experimental situations, women may report little or no
sexual arousal even when objective signs of genital arousal were unmistakable
(Laan and Everaerd, 1995).
Moreover, unlike men, only a minority of women consider orgasm the most
important source of sexual satisfaction with a partner. In addition, whereas erection
is necessary for men to achieve intromission, women can participate in heterosexual intercourse without sexual arousal. Common sense, clinical experience, and
research data indicate that they frequently do so. Although most men who engage
in intercourse experience sexual orgasm, a substantial number of women do not.
These data indicate that women tend to participate in sexual activity, including
intercourse, for many reasons, only some of which would be considered erotic
by standards applied by men to their own behavior (Everaerd and Laan, 1994;
Hatfield and Rapson, 1993).
Thus, although men tend to experience sexual arousal in a simple unitary
way, women experience it more contextually in terms of combinations of emotions.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Psychoanalysis and Sexual Fantasies

577

Nuances of feelings, meanings, and attitudes towards real and imagined experience
are more characteristic of the consciously experienced female experience than the
male experience.
A woman presented with a stimulus that a man might find immediately erotic
(a picture of a nude member of the opposite sex) may or may not experience it
as sexually arousing. After all, what is traditionally considered pornography is
largely consumed by men. On the other hand, men do not consume romance novels in any great numbers. From a male perspective, endless attention is devoted in
these novels to a certain type of context and setting in which sexual arousal may
ultimately be mobilized. The stories which pique the interest of so many women
bore men (Stoller, 1975a,b). It seems evident that the characteristics and meanings
of erotic fantasy differ between the sexes (although, of course, some overlap
does exist).
At the beginning of this article, we put forth a definition of sexual fantasy for
men. The question naturally arises: Is it possible to put forth a definition for women
as well? Perhaps the most important sex difference that bears on the question of definition concerns the onset of the feeling of sexual arousal. Should the sexual situation for women be defined very broadly in a way that includes experiences that have
a high probability of leading erotic desire to be kindled? How does one deal with the
problem that genital arousal is often present without concurrent subjective arousal?
The second question is easier to deal with. Although it is true that genital
arousal might occur without subjective arousal, there is no evidence that the converse commonly occurs (although this area has been sparsely investigated). In
any case, we take the position that sexual arousal in women should be defined
subjectively, that is, in our view, subjective reports of arousal are necessary for
sexual fantasy to be so labeled. With regard to the question of kindling, we take
the narrow view that the term erotic fantasy should be restricted to fantasy that is
associated with the sense of being aroused. When circumstances that kindle are
discussedromantic fantasies, for examplethey should be so labeled.
The notion that female sexual fantasy depends more on relational context
than male sexual fantasy is also compatible with evolutionary psychology. It is
adaptive for women to seek partners who can provide resources and protect them
and their offspring. Men, on the other hand, should (according to an evolutionary
model) seek sexual encounters with as many different partners as possible (Buss
and Schmitt, 1993).
FURTHER THOUGHTS ABOUT SEX DIFFERENCES
IN PSYCHOLOGICAL DEVELOPMENT
A major theme that we have emphasized so far is the fact that males and
females are on different psychosexual developmental lines throughout their life
cycles. Their experiences of sexual fantasy, of boys and girls, men and women,

P1: FLF
Archives of Sexual Behavior [asb]

578

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Friedman and Downey

must be conceptualized within this context. Different prenatal neuroendocrine environments lead to temperamental differences during early childhood. It may well
be that the androgenization of the brain that occurs prenatally in males and not
females radically inhibits or extinguishes the interest of males in infants and doll
play during early childhood. Thus, even boys who are biologically normal but suffer from childhood Gender Identity Disorder and who enjoy playing with Barbie
Dolls as a way of expressing their fascination with the female body outline, and
its adornments, show little interest in maternal aspects of doll play (Zucker and
Bradley, 1995). As small children, girls who experience some degree of prenatal
androgenization as a result of genetically determined adrenal steroid metabolic
disorders show diminished interest in maternal aspects of doll play. Interestingly,
this phenomenon occurs independently of maternal rearing style, and among girls
who ultimately menstruate, and usually grow into heterosexual women and become
mothers (Collaer and Hines, 1995). The sex difference observed in all cultures during early childhood in rough-and-tumble play is also a result of prenatal androgen
differences between the sexes (Maccoby, 1998; Maccoby and Jacklin, 1974). The
themes of childhood play, different between the sexes, reflect the different organization of their self and object representations. From the perspective of cognitive
and social development, the sexes create different narratives or myths within which
sexual fantasy life develops.
In virtually all cultures, very young children are cared for primarily by women.
Similar exposure of boys and girls to mothers during early childhood has different
consequences for each. As separationindividuation proceeds, boys must struggle
against the tendency to identify with their mothers femaleness, whereas girls experience an increased sense of security by doing so. As Gilligan has pointed out,
this may be one reason for men to be threatened by intimacy, i.e., to feel less masculine, whereas women are more likely to be threatened by separation (Gilligan,
1982). This may also influence a need commonly experienced and expressed by
certain men defensively to devalue women in sexual situations and to engage in
exhibitionistic masculine display behaviors as a way of bolstering masculine selfesteem. Such behaviors begin during the latency-age phase when it is common for
boys to experience feminine traits in themselves, and find their same-sex peers as
negatively valued (Fine, 1987).
Sex-segregated play is characteristic of older latency-age children. Boys commonly organize themselves during free play into hierarchical groups, similar to
nonhuman primates. They are more apt than girls to be territorial and xenophobic
and less tolerant of nonstereotypical gender role behavior than girls are (Friedman
and Downey, 1993a).
Oedipal narrative themes enter the life cycle as early as children can describe
them and remain present throughout life. As we have discussed earlier with regard to men, when contemporary psychoanalysts refer to Oedipal themes, they
tend not to mean the concrete wish to participate in sexual relationships with the

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

Psychoanalysis and Sexual Fantasies

September 11, 2000

14:2

Style file version Nov. 19th, 1999

579

opposite-sex parent and the fear and retaliation because of incestuous desires from
the same-sex parent. We have earlier discussed Oedipal narrative themes with
regard to male development and turn now to females.
It is certainly common during childhood for girls to idealize their fathers in
a romantic way and to resent attentiveness to their mothers. It is also common for
girls to fear retaliation because of their jealous impulses, usually in the form of
disapproval and loss of love from their mothers, although themes of bodily damage
are also commonly experienced. These wishes and fears are usually kept secret.
It is easy to trace their elaboration and displacement into fantasies in which the
father figure is symbolically represented by a powerful older man who seduces
the woman, overcoming her protestations, forcing her to succumb to his will, even
raping her. This type of fantasy certainly is the bedrock of womens romance
novels, or so-called bodice rippers. It probably is experienced by many women
under diverse conditions and provides one type of imagined setting in which erotic
excitement can develop (Stoller, 1979).
In almost all societies, the rules regulating male and female sexual experience
differ, with male experience being valued much more positively. Abuse of women
by men is common, and overt physical abusesystematic rape, bodily mutilation,
sexual slaveryis still prevalent in many nations. This difference in power and
status interacts with the biological influences to lead to a final common pathway
with respect to many behavioral sex differences including sexual fantasy. For
example, the theme of being overwhelmed by a powerful male who stimulates but
also protects, need not refer exclusively to the father, but understandably probably
emerges commonly in response to the differential treatment of the sexes throughout
entire societies. This observation is more or less in keeping with a psychoanalytic
perspective first expressed by Horney (1924, 1926). She suggested that the Oedipus
complex in boys and girls is strongly influenced by sociocultural factors and varies
in intensity between children. She criticized Freuds biological determinism and
phallocentricism and pointed out that the Oedipal narratives of women are shaped
by sociocultural prejudices that today would be termed heterosexist.
In any case, the Oedipal fantasies of both sexes express themes of power
and dependency. We note that in our abbreviated discussion of Oedipal themes,
we did not review many of Freuds ideas about the role of Oedipus complex in
female development (anatomy is destiny, etc.) (Freud, 1925, 1933). These are
considered outdated and have been abandoned by the psychoanalytic mainstream,
at least in the United States. Freud also believed that penis envy was biologically
determined and part of the constitutional endowment of all women. This hypothesis
is also outdated (Downey and Friedman, 1998). Limitations of space do not allow
us to review how modern ideas about the development of gender identity led to
alteration of Freuds developmental model for girls. This topic has been discussed
extensively by others and interested readers are referred to the work of Zucker and
Bradley (1995), Tyson (1982, 1994), and Downey and Friedman (1998).

P1: FLF
Archives of Sexual Behavior [asb]

580

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Friedman and Downey

THE MENSTRUAL CYCLE AND SEXUAL FANTASY


In the late 1930s, a psychoanalyst, Benedek, and a gynecologist, Rubenstein,
began an astonishingly original research project on the relationship between the
hormonal fluctuations of the menstrual cycle and verbalized sexual fantasy during
the psychoanalytic process (Benedek, 1973).
The investigators began their work with the prospective investigation of a single woman in analysis with Benedek. The patient was instructed to take daily vaginal swabs and rectal temperatures. These data were analyzed blind by Rubenstein
who used them to date the physiological events of the cycle. After 10 cycle phases
had been completed, Benedek reviewed detailed notes that she made of each analytic session (the references to actual menstruation having been edited out) and
accurately predicted, on the basis of the patients verbalized sexual fantasies and
associations, the precise date of ovulation during every cycle. The investigators
went on to study 15 patients, for variable lengths of time, in whom Benedek accurately and blindly predicted cycle phase.
Benedek and Rubenstein (1942) described phasic alteration in the fantasies
of their patients. Early in the first part of the cycle, self-esteem was high and
erotic interest was primarily motivated by the desire for sexual stimulation. As
ovulation approached, the patients tended to experience increased sexual tension
and conflicts associated with this. Immediately following ovulation there was a
sense of relaxation following which during the luteal phase there was a pronounced
change in the quality of sexual fantasies. The women now tended to focus on
procreational aspects of sexuality: needs to be nurtured and protected were more
pronounced. Imagery of babies and mothers was more plentiful. This was followed
by a brief phase prior to menstruation when ego defenses appeared to weaken.
Negative affects increased, as did the patients level of regression.
Benedek and Rubenstein were aware that the phenomenon of estrus characteristic of lower animals did not regulate human sexual activity. They speculated,
however, that despite the freedom from regulation of sexual activity by the hormones of the cycle, woman might experience an equivalent of estrus in fantasy.
The investigation carried out by Benedek and Rubenstein, although achieving
instant acclaim, has been mostly ignored in subsequent psychoanalytic thought
about sexual fantasy. No attempt to replicate it has ever been carried out at a
psychoanalytic institute. Components of Benedeks observations and findings have
been validated in many nonpsychoanalytic studies (Hedricks, 1994; Severino and
Moline, 1989). Others still remain to be investigated.
A compressed way of summarizing a large interdisciplinary body of work in
this area would be as follows: There is substantial variability between women, regarding the timing of physiological events of the cycle and the experience and
meaning of psychological events. Diverse subgroups exist and more extreme
and repetitive fluctuations occur among some than among others. For example,
premenstrual regression as described by Benedek may have been a function

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

Psychoanalysis and Sexual Fantasies

September 11, 2000

14:2

Style file version Nov. 19th, 1999

581

of the neuroticism of her patients, and not of women in general. The sequential progression of fantasies from stimulatory to procreational may characterize a
subgroup of women rather than women as a whole. It is evident from extensive
research that many subgroups exist with regard to virtually all psychobiological
parameters of the menstrual cycle, and that substantial variability exists not only
between women but also may exist within individual women (Hedricks, 1994).
Across studies, and in many different societies, however, there is little question
that the inner experience of women is cyclic in many ways and that this is true of
the intensity and quality of sexual fantasies as well as other aspects of their mental
life. The notion of cyclicity does not suggest that any dimension of experience or
behavior is not influenced by psychosocial events or that any dimension of experience or behavior is somehow intrinsically more problematical (e.g., negatively
valued) than among men, nor does it suggest that women are under biological
influence to a greater degree than men are.
The complexity of womens sexual fantasy as discussed by Benedek is compatible with the earlier discussion of the psychophysiological area. Particularly
ignored in the psychoanalytic literature is the fact that the quality of sexual fantasies in women and their many meanings may change phasically. Psychoanalysis
as a whole treats sexual fantasy as if it is either trait-related or a result of reactions
to specific life events and traumata. The notion of innately influenced cyclicity imposes a requirement on clinicians to utilize more complex behavioral paradigms.
For example, cyclicity by no means implies the notion of biopsychological events
occurring in a vacuum without social context. In fact, the meanings attributed to
psychological events during the cycle, including sexual fantasies, may be greatly
influenced by interpersonal experience, including sexual experience, and may also
influence the way such experience is psychologically processed by the woman. It
might well be that the transference relationship with the analyst, a crucial aspect of
psychoanalytic treatment, is influenced by the menstrual cycle. Whether this is the
case, however, remains to be explored with empirical research. The onset of menstrual cyclicity during the life cycle creates a discontinuity in female development
which has no parallel among males. Benedeks point that procreative fantasies are
a crucial dimension of the sexual experience of women requires particular emphasis. We suspect that erotic fantasy is much more closely linked to procreative
fantasies in women than in men. These may include the wish to become pregnant,
to deliver, to raise children, or any component of this sequence.
Although the experience of cyclicity, not only with regard to physiological
events, but with feelings as well, is discontinuous with earlier female development
as well as female experience after menopause, many psychological continuities
are obviously retained. By this we mean that girls are not called upon to create
an entirely different representation of the self as a result of puberty. Rather, their
self-representations must adapt to new circumstances. One set of psychological
traits that we believe to be continuous throughout the life cycle is the one involving
maternal interest. Thus, the fantasies of girls and women, including their sexual

P1: FLF
Archives of Sexual Behavior [asb]

582

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Friedman and Downey

fantasies, tend to be influenced by broad narrative maternalistic themes whose


origins are in early childhood. From a psychoanalytic perspective, very little is actually known about the relationship between self- representation and sexual fantasy
during adolescence. Although the research design used by Benedek and Rubenstein is impractical to carry out with modern teenagers, other ways of studying their
fantasy life may be more practical. Longitudinal study of girls prior to and through
puberty, adolescence, and adulthood is needed to shed light on the multiple aspects
of the development of female sexual fantasies that even today remain a subject of
conjecture. Also requiring study are the ways in which girls and women create
unified narrative fantasy themes, some of whose components involve romance,
power, and Oedipal themes, whereas others seem to be basically procreational.
FEMALE HOMOSEXUAL ORIENTATION
Separate developmental lines for males and females are also indicated by sexspecific differences in homosexual orientation. Although some lesbians describe a
developmental pathway similar to that typical for gay men, many do not. Political
homosexuality, for example, is a phenomenon that is for all intents and purposes
confined to women. Political lesbians tend to be feminists who feel that heterosexual activity expresses unacceptable power differentials between women and men,
whether or not they have strong attractions to women (DeFries, 1979). A different
group of women experience homosexual desire for the first time during middle
adulthood. These women, many of whom are heterosocialized or even homophobic, nonetheless experience the emergence of erotic desire for another women in
the setting of an empathic, supportive relationship. The erotic component appears
as if kindled by the other aspects of the relationship. If this phenomenon occurs
among men, it is very rare and has never been seen by either of us. The psychoanalyst Kirkpatrick (1984, 1989) has pointed out that the need for intimacy seems to
be greater among women than men in sexual relationships and that genital release
may not be their primary organizing and motivating factor. She observed that even
though woman might have more intense orgasms with a man, she might nonetheless prefer sexual activity with another woman because of the greater degree of
intimacy in their relationship. In any case, the notion of sexual fantasy as a limit
in place by late adolescence may be true of only some women. Others may well
retain the capacity for plasticity at least with regard to sexual stimuli. We have
speculated elsewhere that if this is true with regard to sexual fantasy it might also
be true for other psychological characteristics (Friedman and Downey, 1998).
REFLECTIONS ON SEXUAL FANTASY AND PASSIONATE DESIRE
The data used by psychoanalytic theory have been provided by psychoanalytic treatment of patients predominately of Western European background. The

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Psychoanalysis and Sexual Fantasies

583

majority of these individuals were fluent, at least moderately well educated, and
from the middle and upper socioeconomic classes. Psychoanalytic theory has also
been applied to art and literature from diverse societies and historical periods.
However, inferences made from such scholarship should not be equated with
cross-cultural research. This qualification having been made, we now speculate
about sexual fantasy and passionate desire.
One of Freuds particularly important insights is that human relationships are
inherently ambivalent. Crystals of antipathy are always contained within the pattern
of our meaningful affections. Unambivalent friendliness, exuberance, expansive
playfulness, and joy are certainly part of our repertoire, but these feelings are
never unalloyed for long in ongoing meaningful interpersonal experience. Freud,
as have virtually all major psychoanalytic theorists, recognized that feelings of
fearanxiety and ragehostility are inevitably mobilized in ongoing relationships,
although they may be experienced in ways that are repressed or denied.
Thoughtful psychoanalytic scholars have realized that the concept of ambivalence is, in a condensed and often disguised way, an integral component of
sexual passion. Thus, during sexual arousal human beings tend to experience more
than a unitary sense of lust. Men as well as women experience feelings of anger
and anxiety that are usually associated with a sense of danger. Mixtures of these
feelings influence the attributions given to erotic fantasy. In less formal language
it means that when we become sexually aroused we feel lustful but also fearful
and often angry, and we process mixtures of these feelings with our characteristic
defenses. As part of our mental processing activities, we create dramatic narratives
which we (later) return to in the form of memories. As we become more sexually
excited, the fantasy, with its different components, is experienced more and more
vividly. These fantasies provide the stimuli for masturbatory activities and also
endow specific, actually occurring sexual situations with meaning. During intense
interpersonal sexual activity, the fantasy constructs of people appear to come to
life and are shared. Both aspects of the actual activitythe enactment in external
reality of subjective wishes and narratives and the sharing of such experience with
its evoked mutual identificationsare experienced as thrilling.
Sexual fantasy is rooted in bodily experienceones own body and that of
others. The first relationship in which ones body is more or less completely explored by another is with the mother or mothering person. Hence, psychoanalysts
have noted that representations of the motherchild relationship and of the mothers
body are symbolically included in sexual fantasy experienced during adulthood
by both sexes (Chasseguet-Smirgel, 1986).
Kernberg put it as follows (Kernberg, 1995, p. 26):
The fantasied early polymorphous perverse relations to the parental objects are condensed
with the admiring and invasive relation to the lovers body parts. Erotic desire is rooted
in the pleasure of unconsciously enacting polymorphous perverse fantasies and activities,
including symbolic activation of the earliest object relations of the infant with mother and
of the small child with both parents. All this is expressed in the perverse components of

P1: FLF
Archives of Sexual Behavior [asb]

584

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Friedman and Downey


intercourse and sexual playin fellatio, cunnilingus and anal penetration and in exhibitionistic, voyeuristic and sadistic sexual play. Here the links between the early relationship
with mother of both genders and the enjoyment of the interpenetration of body surfaces,
protuberances and cavities is central.

Sexual passion brings the self to the boundary in the experience of pleasure
associated with emotional intensity. At the outer boundary of the self, there is
always a feeling of danger of dissolution, and then a feeling of relief, relaxation,
and of being soothed when the intensity passes, and the self inhabits its customary,
familiar surroundings. When sexual passion is shared with another person, the loss
of self-boundary is associated not only with what is usually termed communication in conventional usage, but rather with the sense that the inchoate matter of
ones interior and the others interior are mixed. This type of experience, whether
actually occurring and encoded in memory, or only wished for but nonetheless
psychologically represented in some form, endows sexual fantasy with the power
of mysterious enchantment.
The complex layers of meaning represented by the adult sexual experience
are an important reason that childhood and adult sexuality, although similar in
certain ways, are quite different in others. Children, even older ones, although
capable of sexual activity do not experience their inner worlds as adults do. Their
ideas about sex and love, virtue and vice, and rage and fear become modified with
life experience and cognitive and social growth. Artists have intuitively grasped
this. For example, Romeo and Juliet, Shakespeares play about adolescent lovers
tells quite a different story than that of Anthony and Cleopatra, his drama about
full-grown adults.
CONCLUSIONS: PSYCHOANALYSIS AND SEX RESEARCH
It is difficult to capture the subjective qualities of sexual fantasy in laboratory research settings or via questionnaire studies. On the other hand, research
using psychoanalytic techniques is time-consuming and extremely difficult to implement. Moreover, many aspects of sexual fantasy and sexual arousal were unnoticed by psychoanalytic psychology and could never have been discovered via
psychoanalytic exploration. It was only because of the physiological investigations of Masters and Johnson that psychoanalysis revised its erroneous models of
female development and psychosexual functioning (Masters and Johnson, 1966).
The pioneering work of Money and collaborators on gender identity led to similarly
far-reaching revisions of psychoanalytic models of the mind (Money and Ehrhardt,
1972; Stoller, 1968). Psychoanalytic exploration alone could never have described
the crucial distinction in women between genital and subjective arousal. Many
questions about the somatic, psychological, and social determinants and consequences of sexual fantasy during development remain to be illuminated. These
provide fertile ground for collaborative efforts between behavioral scientists and
psychoanalysts.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227175

September 11, 2000

Psychoanalysis and Sexual Fantasies

14:2

Style file version Nov. 19th, 1999

585

REFERENCES
Abel, G. A. G., Osborn, C., Anthony, D., and Gardos, P. (1992). Current treatments of paraphiliacs.
Annu. Rev. Sex Res. II: 255291.
American Psychiatric Association (1957). Diagnostic and Statistical Manual of Mental Disorders,
1st ed., American Psychiatric Press, Washington, DC.
American Psychiatric Association (1968). Diagnostic and Statistical Manual of Mental Disorders,
2nd ed., American Psychiatric Press, Washington, DC.
American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders,
4th ed., American Psychiatric Press, Washington, DC.
Benedek, T. (1973). Psychoanalytic Investigations: Selected Papers. Quandrangel: The New York
Times Book Co., New York, NY.
Benedek, T., and Rubenstein, B. B. (1942). The Sexual Cycle in Women: Psychosomatic Medicine:
Medicine Monographs III, Nos. 1 and 2. National Research Council, Washington, DC.
Bradford, S. M. W. (1995). In Oldham, J. M., and Bams, R. (Eds.), Annual Review of Psychiatry,
Vol. 14: Pharmacological Treatment of the Paraphilias, American Psychiatric Press, Washington,
DC, pp. 755779.
Buss, D. M., and Schmitt, D. R. (1993). Sexual strategies theory: An evolutionary perspective in human
mating. Psychol. Rev. 100: 204232.
Chasseguet-Smirgel, J. (1986). Sexuality and Mind: The Role of the Father and the Mother in the
Psyche, New York University Press, New York, NY.
Collaer, M. L., and Hines, M. (1995). Human behavioral sex differences: A role for gonadal hormones
during early development? Psychol. Bull. 118: 55107.
DeFries, Z. (1979). A comparison of political and apolitical lesbians. J. Am. Acad. Psychoanal. 17(1):
5766.
Downey, J. I., and Friedman, R. C. (1998). Female homosexuality: Classical psychoanalytic theory
reconsidered. J. Am. Psychoanal Assn. 46(2): 471506.
Erickson, M. T. (1993). Rethinking Oedipus: An evolutionary perspective of incest avoidance. Am. J.
Psychiat. 150: 411416.
Everaerd, W., and Laan, E. (1994). Cognitive aspects of sexual functionoing and dysfunctioning. Sex.
Marit. Ther. 9: 225230.
Fine, G. A. (1987). With the Boys: Little League Baseball and Preadolescent Culture, University of
Chicago Press, Chicago, IL.
Freud, S. (1905/1953). Three Essays on the Theory of Sexuality. In Strachey, J. (Ed. and trans.), Standard
Edition, (hereafter S.E.), Vol. 7, pp. 125243. Reprinted by Hogarth, London, England.
Freud, S. (1989). Some psychical consequences of the anatomical distinction between the sexes. In
Gay, P. (Ed.), The Freud Reader, 1925. W.W. Norton and Co., Inc., USA, pp. 670678.
Freud, S. (1933/1964). Femininity. In Strachey, J. (Ed.), SE, Vol. 22, pp. 112135. Reprinted by Hogarth,
London, England.
Freud, S. (19401949). An Outline of Psychoanalysis, W.W. Norton and Co., Inc., USA.
Freund, K., Langevin, R., Chamberlayne, R., Deosoran, A., and Zajac, Y. (1974). The phobic theory
of male homosexuality. Arch. Gen. Psychiat. 31: 495499.
Friedman, R. C. (1988). Male Homosexuality: A Contemporary Psychoanalytic Perspective, Yale
University Press, New Haven, CT; London, England.
Friedman, R. C., and Downey, J. (1993a). Neurobiology and sexual orientation: Current relationships.
J. Neuropsychiat. & Clin. Neurosci. 5: 131153.
Friedman, R. C., and Downey, J. (1993b). Psychoanalysis, psychobiology and homosexuality. J. Am.
Psychoanal. Assn. 41(5): 11591198.
Friedman, R. C., and Downey, J. I. (1995). Biology and the Oedipus complex. Psychoanal. Q. LXIV,
234264.
Friedman, R. C., and Downey, J. I. (1998). Psychoanalysis and the model of homosexuality as psychopathology: An historical overview. Am. J. Psychoanal. 58(3): 249270.
Galenson, E., and Roiphe, E. (1981). Infantile Origins of Sexual Activity, International Universities
Press, New York, NY.
Gilligan, C. (1982). In a Different Voice, Harvard University Press, Cambridge, MA.
Gorski, R. A. (1991). Sexual differentiation of the endocrine brain and its control. In Brain
Endocrinology, 2nd ed., Raven Press, New York, NY, pp. 71103.

P1: FLF
Archives of Sexual Behavior [asb]

586

PL113-227175

September 11, 2000

14:2

Style file version Nov. 19th, 1999

Friedman and Downey

Greenberg, D. F. (1988). The Construction of Homosexuality, University of Chicago Press, Chicago,


IL.
Hatfield, E., and Rapson, R. L. (1993). Historical and cross-cultural perspectives on passionate love
and sexual desire. Annu. Rev. Sex Res. IV: 6799.
Hedricks, C. A. (1994). Sexual behavior across the menstrual cycle: A biopsychosocial approach. Annu.
Rev. Sex Res. V: 122173.
Herdt, G., and McClintock, M. (2000). The Magical Age of 10. Arch. Sex. Behav. 29: 587606.
Horney, K. (1924). On the genesis of the castration complex in women. Int. J. Psychoanal. 5: 5065.
Horney, K. (1926). The flight from womanhood: The masculinity-complex in women as viewed by
men and women. Int. J. Psychoanal. 7: 324339.
Isay, R. (1989). Being Homosexual, Farrar Straus Giroux, New York, NY.
Kernberg, O. (1995). Love Relations: Normality and Pathology, Yale University Press, New Haven,
CT.
Kirkpatrick, M. (1984). Female homosexuality. Panel Report, Annual Meeting of American
Psychoanalytic Association, San Diego, CA, 4 May.
Kirkpatrick, M. (1989). Women in love in the 1980s. J. Am. Acad. Psychoanal. 17(4): 535542.
Kohlberg, L. (1981). The Philosophy of Moral Development: Moral States and the Idea of Justice.
Essays on Moral Development, Harper & Row, New York, NY.
Laan, E., and Everaerd, W. (1995). Determinants of female sexual arousal: Psychophysiological theory
and data. Annu. Rev. Sex Res. VI: 3277.37.
Lewes, K. (1988). The Psychoanalytic Theory of Male Homosexuality, Simon & Schuster, New York,
NY.
Maccoby, E. (1998). The Two Sexes: Growing Up Apart, Coming Together, Harvard University Press,
Cambridge, MA.
Maccoby, E. E., and Jacklin, C. N. (1974). The Psychology of Sex Differences, Stanford University
Press, Palo Alto, CA.
Masters, W. H., and Johnson, V. E. (1966). Human Sexual Response, Little, Brown, Boston, MA.
Money, J. (1988). Gay, Straight and In-Between, Oxford University Press, Oxford, England.
Money, J., and Ehrhardt, A. (1972). Man and Woman: Boy and Girl, John Hopkins University Press,
Baltimore, MD.
Olds, D. D. (1994). Connectionism and psychoanalysis. JAPA 42(2): 581613.
Paredes, R. G., and Baum, M. J. (1997). Role of the medial preoptic area/anterior hypothalamus in the
control of masculine sexual behavior. Annu. Rev. Sex Res. VIII: 68102.
Person, E. (1995). By Force of Fantasy, Basic Books, New York, NY.
Schrut, A. (1993). Is castration anxiety and the Oedipus complex a viable concept? Presented at the
meeting of The American Academy of Psychoanlaysis, San Francisco, CA, 22 May.
Severino, S. K., and Moline, M. L. (1989). Premenstrual Syndrome: A Clinicians Guide. The Guilford
Press, New York, NY, London, England.
Socarides, C. W. (1978). Homosexuality, J. Aronson, New York, NY.
Stoller, R. (1975a). Perversion: The Erotic Form of Hatred, Karnac Books Ltd, London, England.
Stoller, R. (1975b). Pornography and perversion. In Perversion: The Erotic Form of Hatred, Karnac
Books Ltd., London, England. pp. 6392.
Stoller, R. (1979). Sexual Excitement: Dynamics of Erotic Life, Pantheon Books, New York, NY.
Stoller, R. J. (1968). Sex and Gender, Science House, New York, NY.
Sulloway, F. J. (1979). Freud: Biologist of the Mind, Basic Books, New York, NY.
Tyson, P. (1982). A developmental line of gender identity, gender role and choice of love object. J Am.
Psychoanal Assn. 40: 5984.
Tyson, P. (1994). Theories of female psychology. JAPA 42(2): 447469.
Yates, A. (Ed.) (1993). Sexual and Gender Identity Disorders: Child Adolesc. Psychiatr. Clin. North
Am. 2(3).
Zucker, K., and Bradley, S. J. B. (1995). Gender Identity Disorder and Psychosexual Problems in
Children and Adolescents, Guilford Publications Inc, New York, NY.

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 29, No. 6, 2000

The Magical Age of 10


Gilbert Herdt, Ph.D.,1,3 and Martha McClintock, Ph.D.2

Developmental processes of puberty and their cultural contexts in understanding the emergence of sexual subjectivity, especially sexual attraction, prior to
gonadarche are critically examined. In particular, we consider the hypothesis
that sexual attraction follows the onset of adrenal puberty, termed adrenarche, precipitating the development of stable and memorable attraction toward
others approximately by the age of 10. In a prior study, the authors suggested
that adrenarche is a significant source of this developmental change in sexuality
(McClintock, M., and Herdt, G., 1996). The inferential evidence from New Guinea
is compared with recent studies from the United States, including clinical findings
on precocious puberty. We conclude with the question of whether the age of 10
is a human universal in the development of attraction and sexuality.
KEY WORDS: puberty; sexual attraction; adrenarche; gonadarche; culture.

INTRODUCTION
This paper critically examines developmental processes of puberty and their
cultural contexts in understanding the emergence of sexual subjectivity, especially
sexual attraction, prior to gonadarche. In particular, we consider the hypothesis
that sexual attraction follows the onset of adrenal puberty, termed adrenarche,
precipitating the development of attraction toward others approximately by the
age of 10. Inferential evidence from New Guinea cultures is compared with recent
studies from the United States to reconsider how societies deal with the expression
of sexual development before the age of 10. Finally, we conclude by asking whether
the age of 10 is a human universal in the development of attraction, in cross-cultural
and historical perspective.
1 Program in Human Sexuality Studies, San Francisco State University, San Francisco, California 94132.
2 Committee

on Human Development, University of Chicago, Chicago, Ilinois 60637.

3 To whom correspondence should be addressed at Program in Human Sexuality Studies, San Francisco

State University, 1600 Holloway Avenue, San Francisco, California 94132.


587
C 2000 Plenum Publishing Corporation
0004-0002/00/1200-0587$18.00/0

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

588

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

Herdt and McClintock

A major stream of research now posits two sequential but distinct forms of
pubertal processes: adrenal puberty and gonadal puberty. Adrenal puberty is the
process that occurs in middle childhood, between 6 and 10 years of age. It is hypothesized to be a critical source of developmental subjectivity, including feelings
of attraction and sexual awareness, becoming stable and memorable around the
age of 10. By contrast, gonadal puberty begins later, normatively around the ages
of 11 and 12 for girls and boys, respectively, and continues into the late teens. It is
coincident with adolescent maturation and morphological developmental changes,
which are commonly referred to as secondary sex traits, as well as fertility. The
evidence reviewed later, though anecdotal and incomplete, nevertheless suggests
that the creation of sexual subjectivity begins well before the onset of adolescence.
It is intrinsically driven by hormonal forces but, nevertheless, is informed by cultural meanings and social roles. Furthermore, it is sufficiently marked by the age
of 10 or so that even in cultures where age variation in developmental transitions
is common, a variety of cultures in ancient and modern times have sensed the age
of 10 to be of critical importance in defining this age as the sexual juncture between
childhood and adulthood. Thus, the strong inference is that in society sexual
attraction emerges as a significant developmental subjectivity during adrenarche,
but before gonadarche and adolescence.
Although a variety of researchers, since the time of Freud (1905), have suggested that sexuality emerges in childhood, the question whether this development is anchored in the unconscious, whether it involves subjectivities like those
of the adult, as well as the intrinsic mechanisms of this developmental transformation, remains obscure (Bem, 2000; Green, 1987; Stoller, 1968). Moreover, many
scholars working within this paradigm continue to ignore the evidence from recent biological developmental, historical, and cross-cultural studies. For example,
a recent textbook, by Kimmel (2000, p. 37), a leading sociologist of gender, asserts:
Sex differentiation faces its most critical events . . . [at] puberty, when the bodies of boys
and girls are transformed by a flood of sex hormones that cause the development of facial
hair for boys, and the development of all secondary sex characteristics.

Nonetheless, the literature on sexuality generally attests to a greater range and


diversity of sexual development than textbooks and the popular culture allow,
especially in cross-cultural child development and what we in the west refer to as
adolescence (Elder, 1975; Herdt and Leavitt, 1998).
Recent developmental, neurological, linguistic, and hormonal studies generally suggest that in western countries, a critical developmental change occurs
by the age of 8 or so, establishing the context for major cognitive and behavioral transformations in human development, including preparation for a more
adult-like understanding of sexuality (Gelman et al., 1986; Money, 1997; Piaget,
1971). Likewise, instead of viewing late childhood as a time of latency in sexual development, as did Freud, we conceptualize middle childhood as a time in
which general psychophysiological arousal, including erotic feelings and events

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

The Magical Age of 10

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

589

previously unknown or unrecognized, produces increasingly memorable and stable


sexual attraction (see Friedman and Downey, 2000).4
The question of the chronological age of first sexual arousal and attraction
have long intrigued researchers in sex and human development. In this century,
Freud (1905), Ehrhardt and Meyer-Bahlburg (1981), Gagnon (1971), Kinsey and
colleagues (1948, p. 299ff.), Maccoby (1979), Mead (1927), Money (1997), and
Stoller (1968) are among the distinguished thinkers who have conceptualized the
age and maturational conditions of the first experience of sexual attraction toward
others. It has long been recognized that gonadal morphological changes in western
populations vary widely and are distinct for males and females, and that chronological and maturational ages differ. It is also widely believed that historical change and
modernization impact strongly upon maturation, as noted later. Today, in general,
the mean age of onset of gonadal puberty is 11.5 years in boys, but may begin
as early as 9 years or as late as 15 years and still be considered within normal
limits (Money and Lewis 1990, p. 241). Likewise, the relationship between the
achievement of gonadal puberty and sexual maturation is problematic, as Kinsey
and colleagues (1948) noted long ago: first ejaculation in males, for example, is inadequate for understanding sexual subjectivity. Adrenarche, we surmisewith
its accompanying rise in sex steroidshas not even been considered as a factor in
the development of sexuality, much less sexual subjectivity.
Certainly gender differences inflect the emergence of sexuality and sexual
subjectivity, and male and female experiences must not be lumped together when
it comes to the development of sexual attraction. As is well known, gender differences in a range of behavioral domains tend to increase during middle childhood
(Moller et al., 1992); for example, Maccoby and Jacklin (1974) demonstrated the
preference of nursery school children for their same-gendered peers. By the age
of 4.5 years, children spent three times as much social time with the same gender;
by the age of 6.5 years the increase is at the huge ratio of 11 to 1. When gender
differences of this magnitude are enhanced or even exaggerated, and where gender
segregation exists throughout life, such as among the Sambia of Papua New Guinea
(Herdt, 1981), the combination of intrinsic and extrinsic (social) influences may especially affect the development of sexual subjectivity and the conscious formation
of objects of attraction before adolescence.
A word about definitions and semantics. The notion of attraction is culturally and emotionally loaded and although very imprecise, nevertheless, the
construct is well established in the literature (Gagnon, 1990; Herdt and Boxer,
1993; Laumann et al., 1994; Kinsey et al., 1948). By sexual attraction is meant
the subjective state, within the adult person, of feelings of desire or fantasies about
another person, known or imagined, that may or may not lead to sexual intimacy
4 This

article deals only with the normative developmental experience, which we are attempting to
reconceptualize; we are aware of the importance of precocious and delayed puberty for the phenomena
under review, but along with others, we chose to separate these from the discussion of normative sexual
attraction development (see Money and Lewis, 1990, p. 245ff).

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

590

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

Herdt and McClintock

with another person. The definition marks the adult experience as mature and separates the subjective state from the actual sexual behavior that might result from
it. This concept should not be confused with sexual orientation, which is a more
global, and often assumed to be a core or fixed trait (Byne and Parsons, 1993;
Gorman, 1994; Meyer-Bahlberg, 1997; Stoller and Herdt, 1985; Storms, 1981;
Townsend and Wasserman, 1997).
Attraction indicates psychophysiological arousal, but this is not a necessary
precursor of sexual arousal, at least in all cases. People may at times be aroused
but deny their feelings or have their culture deny them; the relationships between
experience and self-conscious recognition of sexual attraction are especially problematic in sex-negative or repressive societies.
It is important to mark the difference between childhood and adult subjectivities as well. For a child attraction is not the same as for an adult; a childs
diffuse, more emergent properties of liking, friendship, and emotional closeness or
intimacy have meanings different from those of an adult that is sexually aroused.
Moreover, the content of such experiences obviously varies by culture, as does
the ability of the person to express their feelings, particularly in public. Such
differences are especially critical to keep in mind when sexuality is approved
in childhood play and carried into the teen years, as Mead (1927, 1935, 1961)
repeatedly suggested in her ethnographies. The childhood experienceand the
subsequent adult retrospection of thisis also distinct and should not be confused (Plummer, 1995). Indeed, sexual attraction is manifest throughout middle
childhood and is not yet linked with sexual desire or fantasy. It is around the age
of 10 that it becomes a robust, memorable experience (McClintock and Herdt,
1996). Both subjectivities are nevertheless dependent upon cultural meanings in
interpreting the preadolescent experiences as sexual, a point long recognized
in script theory (reviewed in Gagnon, 1990). In some societies, such meanings
are restricted to reproductive genital sexuality, whereas in others, recreational and
pleasurable sexuality are encouraged outside of and beyond marriage, such as
among the Trobriand Islanders, the !Kung people of Southern Africa, and others
(Herdt, 1997; Vance, 1991).
Adolescence, as a time of increasing demands upon the persons social and
moral responsibility in many societies, often compresses sexual behavior, because
parents, families, and communities may regulate intimacy and sexual relations,
especially when relationships may lead to reproductive unions (Herdt and Leavitt,
1998). Coincidentally, of course, this change in moral attitudes occurs as the childs
body changes morphologically into an adults. Individuals may experience desire
and attraction toward others without the cultural means either to recognize or to
express their feelings in public. However, eventually, attraction typically becomes
genitally arousing, and such arousal must be reconciled with customary arrangements for sexuality and marriage recognized in the local culture (Mead, 1961).
Sexual maturity in many societies is, finally, dependent upon the achievement of
gonadal puberty and menarche in males and females, respectively, though neither

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

The Magical Age of 10

591

of these achievements should be reduced to the presence or absence of orgasm or


menstrual flow (see the discussion of preadolescent orgasm in Kinsey et al., 1948,
pp. 175180).5
ADRENAL PUBERTY AND ATTRACTION
The inference that sexual attractionas denoted by intimate romantic and/or
erotic attraction to anotherbecomes stable and memorable by the age of 10, was
previously formulated by McClintock and Herdt (1996). In a study of self-identified
gay and lesbian adolescents in Chicago, Herdt, and Boxer (1993) had observed that
the mean age of first awareness of attraction was between 9.5 and 10 years for boys
and girls, respectively. Subsequently, McClintock and Herdt reviewed the evidence
to discover similar patterns in other studies. It was argued that American children
typically become aware of a new way of sexual thinking about their bodies, body
imagery, gender roles, and emotional and intimate relations with their playmates
between the fourth and fifth grades (e.g., a normative chronological age range of
911, for most fourth and fifth graders). Studies on sex-typing in play relationships,
and popularity among children, also support the importance of these sexual and
gender changes around fourth grade (Moller et al., 1992).
The concept of puberty is still widely seen in the minds of many researchers
and the public as adolescence or as the morphologic changes (incorrectly labeled
secondary sex changes; see Money and Ehrhardt, 1972) of adolescence. In prior
models of developmental sexual psychology, gonadarche was typically seen as a
kind of internal force that changed attraction into sexual action, culminating in the
developmental sequelae of adult sexuality, especially reproduction. As was previously argued (McClintock and Herdt, 1996), however, biopsychosocial puberty
should be expanded to encompass two distinct, sequential processes of puberty,
adrenal puberty followed by gonadal puberty, each of which is independent but
temporally processed by different mechanisms of development (McClintock et al.,
1998; and see Hopper, 1975; Korth-Schutz, 1989). Rather than viewing sexual attraction as following gonadarche, this model conceptualizes the subjectivity of
attraction as a longer sequence of developments that begins endocrinologically
at age 6 on average (McClintock et al., 1998, p. 1).
Traditionally, in the general population, the gonads were seen as the cause
of gonadal puberty, anticipating reproduction, now around the age of 10.75 years,
for females (McClintock et al., 1998, p. 1). Money and Lewis (1990, pp. 241,
242) find that for males, the visible evidence of gonadal puberty is in the enlargement of the testes, scrotum, and penis, with a mean age onset of these changes
at 11.5 years. Ejaculation and nocturnal emissions of viable sperm complete the
5 As

Simon and Gagnon (1973, p. 34) once wisely stated: An important source of guilt in children
comes from the imputation to them by adults of sexual appetites or abilities that they may not have,
but they learn, however imperfectly, to pretend they have.

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

592

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

Herdt and McClintock

process. However, these changes may begin as early as 9, and as late as 15, and
still be considered within the normal range. The duration of the process is typically 24.5 years until completion, which culminates in attainment of adult height
and pubic hair distribution. For females, the visible and invisible changes begin
between the ages of 9 and 13, with growth accompanied by pubic hair and breast
growth. The average age of the growth spurt is just after 12, with the average
first ovulation and menstruation occurring between 13 and 13.5, with a range of
11.515.5 years. The female process may last only 1.5 years or as long as 6 years.
Between the ages of 5 and 15, the childs developing hormones change and
influence both physical and psychological (interpersonal and intrapsychic) development. Both genders begin neonatal development with adult levels of testosterone
and estrogen. However, the sex hormone levels begin to fall and remain low until
the maturation of the adrenal glands, i.e., ages 68 (reviewed in McClintock and
Herdt, 1996; McClintock et al., 1998). Increased adrenal activity and hormone production begins to increase exponentially until it reaches the low adult range around
the age of 10, and then plateaus in both boys and girls. Androgens released during
preadolescence continue to rise from adrenal gland secretion until ages 12 and 13,
when the maturation of the gonads continue to augment androgen production.
Children between the ages of 6 and 8 begin to experience increasing adrenal
function, in both males and females. The adrenal glands (specifically the adrenal
cortex) secrete low levels of androgens (typically identified as male sex hormones),
primarily of dehydroepiandrosterone (DHEA). The specific androgen released by
the adrenal glands is in the same metabolic pathways as testosterone and estrogen.
There is no sex difference in the rate or onset, though the conversion rate may vary
between the genders, until the onset of gonadarche. The levels of these hormones
begin to steadily climb upwards until adult levels of DHEA are reached by the age
of 12 for girls and the age of 13 for boys, respectively. By the age of 10, they reach
the low end of the adult range. Although these levels are low as compared to normal adult levels, they are many (1020) times what typical young children exhibit.
Moreover, although the hormone levels required for an organizational (long-term,
permanent) effect are unknown, there are brain changes at this age, which are
indicative of neural proliferation and sculpting; there is an overproduction of neurons followed by a selective loss, presumably of nonfunctional connections, which
sculpts the neural networks of the cortex (Blumenthal et al., 1999). The levels
experienced between ages 6 and 10 are within activational (short-term, temporary)
range. Thus, it is highly probable that the levels of hormones secreted during these
age ranges have a significant influence on preadolescent brains.
Something prior to puberty is transforming the childs body and psyche in
the direction of sexual arousal. What might the precursor of this development be?
Adrenarche is the best candidate for conceptualizing the development of attraction in this hypothetical model. In addition to the factors already outlined, four
additional developmental influences stimulate the development of attraction. First,

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

PL113-227176

September 16, 2000

10:6

The Magical Age of 10

Style file version Nov. 19th, 1999

593

the adrenals release hormones that have already been identified as being relevant
in adults for sexual attraction. Second, there does not appear to be a significant
age difference in adrenal pubertal development for a girl and a boy. Third, the
same hormones continue to rise in concentration during gonadarche. If the gonads
constitute the structure for biological priming in sexual attraction development,
then it intuitively follows that the same hormones at earlier ages have a similar
effect. Finally, DHEA is the primary sex hormone released by the adrenals. It is
only two metabolic steps away from testosterone, but another three steps away
from estradiolthe major adult sex hormones. These hormonal changes, it is postulated, stimulate awareness of the body and sensations when interacting with
others, heightening perceptions of sexual and/or romantic attraction and their cues
before gonadal puberty (McClintock and Herdt, 1996). Eventually these developmental subjectivities are strong enough to become memorable in the childs
experience.

THE CASE OF NEW GUINEA


Anthropology has stressed social, rather than the biological elements of
puberty, since the time of the French scholar Van Gennep (1960), and it is
commonly accepted today that adolescence is not a universal category. Moreover, for a variety of reasons, the social and biological dimensions were typically
lumped together with a notion of gonadarche or puberty at 13 (reviewed in Herdt
and Leavitt, 1998), as one clearly sees in the early work of Mead (1927). Thus,
distinctions between adolescence, puberty, and sexuality are controversial
and require detailed study of the ethnographic data from a particular culture area
to make sense of what is local and universalor something in between.
The island of New Guinea is home to more than 700 cultures and within the
Melanesian area, 2000 languages. The area has also been known, since the classical
studies of Malinowski (1929) and Mead (1935), for the immense range of variation
in sexual behavior found across the life course (Herdt, 1984). As Herdt has worked
among the Sambia of New Guinea since 1974 and contextualized sexuality, we have
chosen to highlight this culture area in comparing findings from the United States
on the emergence of sexual attraction before gonadarche. We shall emphasize the
precolonial situation of these societies prior to western contact and globalization.
What is remarkable about the anthropological record of this area is the attention to the age of 10 or so in many reports of ritual initiation for over more than half
a century. These cultures in the precolonial period were characterized by unwritten
languages, and the lack of attention to individual ages, typically used for recording birth and developmental events in the west. What stands out is the dramatic
attention showered upon the tenth year, especially in male development, but also
in female development. Ritual separation, especially from the natal households,

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

594

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

Herdt and McClintock

gender segregation, ordeals and taboos, as well as the introduction into sexual life,
either homoerotic or heteroerotic, characterized the late childhood years. The folk
psychology of these peoples strongly connected the need for ritual initiation with
an inferred internal development in the attractions, desires, and motivations of the
child, especially boys. The societal response seems widely to have involved the
need for change in the persons social identity, living arrangements, and sexual
regulation before gonadarche. When it is remembered that gonadarche was much
later than in contemporary western cultures, the extreme attention to changing the
childs psychosexual status by the age of 10 is even more impressive.
Contemporary study of these issues began in 1954, when the Australian anthropologist K. E. Readthen the leading authority on New Guineapublished
the first ethnological survey of the New Guinea Highlands. Read had served in
New Guinea during WW II and later returned to conduct the first long-term field
study. He also initiated first contact with certain indigenous Highlands peoples,
such as the Bena Bena, described as follows:
Men and women did not sleep under the same roof. A man had a house for each of his
wives, and he kept many of his personal possessions in them, but since constant association
with women was thought to be weakening, he regularly slept in the club house with his
male contemporaries and seniors and all boys over the age of about ten. (Read, 1954, p. 13)

Read claimed that this generalization applied to all Highland groups.


Reports from other anthropologists since then have confirmed the insight of
his original discovery. In these precolonial societies, the age of 10 seems to be a
baseline for removal from the childhood category, followed by ritual advancement into a new social and sexual category. The following are but a few of the
relevant cases known from the literature. The earliest report comes from British anthropologist Deacon, who described Malekula Island in the New Hebrides, specifically that boys are initiated at the age of 10 (Deacon, 1934, p. 41). American
anthropologist Langness (1967, p. 164), writing on the same Bena Bena people
observed by Read years earlier, independently confirmed that boys are taken to the
mens house around the age of 10 or 12. Further afield, the Dutch anthropologist
Van Baal (1966, p. 52, 1984, p. 133), who described the Marind-anim peoples
occupying the entire Southwest coast of New Guinea, noted that boys stayed with
their mothers until the age of 5 or 6, and then went to live nearer their fathers;
however, they were formally admitted to the mens house when they had reached
the approximate age of 10. Initiation was the introduction to sexual life, as the
boys were inseminated by older males (Van Baal, 1984, p. 133). Again, among the
Kaluli people in Northwest Papua New Guinea, the ethnographer states: Homosexual intercourse for boys also took place in everyday life . . . whenever a boy
reached the age of about ten or eleven (Schieffelin, 1976, p. 152).
It is remarkable that in spite of the diverse nationalities, generations, and
cultural theories of these different anthropologists, all of them have drawn attention
to the age of 10. Whatever projection might be involved in guessing the age of

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

The Magical Age of 10

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

595

the child and his or her age-cohort, a good deal of objective understanding of the
relevant age differences in these local communities was obviously available to
these fieldworkers. In fact, our own work among the Sambia suggests that the age
of 10 is probably a regional critical period for sexual development.
But what is so critical about the age of 10 in New Guinea? As recently
suggested elsewhere for the Sambia people, the accumulating evidence on sexual
development supports the hypothesis that at least for males,6 adults recognized
the age of 9 or 10 in the child as commencing the emergence of sexual attraction
to others (Herdt, 2000). That is, parents and community leaders inferredeither
directly from their own observations of children, or indirectly from retrospections
of their own psychosexual developmentthat sexual attraction and desire were
budding in late childhood. Following initiation, Sambia elders and fathers taught
boys about the need to physically separate from their mothers in order to grow into
strong warriors. This gender segregation was strongly sanctioned and prevented
any possibility of sexual interaction between boys and girls. Concomitantly, older
males introduced boys to ritual insemination, typically occurring between younger
(ages 714) and older males (ages 15 and above) before marriage (Herdt, 1981),
a pattern found in 1020% of Melanesian societies (Herdt, 1984).
Homoerotic initiation rites of these kinds are common throughout the area
(Herdt, 1984). However, boys and girls are elsewhere introduced to heterosexual
relations by the age of 10 or so, suggesting that the process applies both to males
and females, as well as to homoerotic and heteroerotic customs (Knauft, 1993). The
reason for this intense focus on sexuality by the age of 10 may have had to do with
how these cultures regarded sexual and gender development before gonadarche as
a social problem that required the dramatic solution of rites de passage.
Sambia male initiation rites required treatment of the boy before the age
of 10, otherwise it was thought that he would weaken and die. The Sambia believe
that a boy must be initiated before he is too old or too big in order for the
rite of passage to have its necessary and desired effect (Stoller and Herdt, 1982).
By contrast, girls were not initiated until their betrothed husbands had attained the
stage of being late adolescent warriors, which was about the age of 1012 for girls.
Further, female initiation waited upon menarche in girls, which was in the late teens
throughout Highlands New Guinea. Initiation of the boy moved him away from
women and mother directly and led into the mens house, where he was inseminated, paving the way for a 1015-year period of being exclusively homoerotic/
homosocial, during the preadolescent and adolescent development, until his late
teens or early twenties, depending upon the exact age at which a particular boy
would have married and fathered a child (Herdt, 1981, 1987). Sambia pinpoint
this transition point at between the ages of 7 and 10 years, for an age-cohort of
6 We

would not want to claim that all traditional societies in Melanesia had initiation practices moving
boys into the mens house by the age of 10, because examples to the contrary can certainly be found
(Herdt, 1991).

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

596

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

Herdt and McClintock

boys from a group of villages, with individual boys ages averaging approximately
8.5 years.
We do not know at what age adrenal puberty begins in these populations,
but the onset of the independent process of gonadal puberty was slower than in
contemporary western society. Child mortality was also high and in some areas
created a definite challenge to the survival of the society (Herdt and Leavitt, 1998),
resulting in prolonged breast-feeding of children, in some cases, upto the age of 4
or 5. Long-term field studies have attested to the resilience and robustness of
surviving children in New Guinea, however, as well as to the seemingly slow rate
of growth (Mead, 1956). We should note that the onset of gonadal puberty was very
slow, perhaps slower than any comparable area of the world, in these precolonial
societies of Papua New Guinea. These societies were protein deficient, parasitically
challenged, and they lacked prenatal care or modern medicine in dealing with such
diseases as yaws and malaria. Infant mortality was very high and reached more than
50% in some villages in some years (Herdt, 1987). Adrenal puberty may have been
at the same or slightly older ages for Sambia children. Generally, gonadal puberty
remained late in precolonial times and well into the 1970s, with boys achieving
gonardarche between the ages of 13 and 14 years, and girls achieving menarche by
the age of 18 among the Bundi people, and as late as 19.2 years among the Sambia
and their neighbors (Malcolm, 1968; Worthman, 1999; see Danker-Hopfe, 1986,
for European comparisons).
Because of warfare, an extreme imperative was placed on the need to achieve
gonadarche and adult masculinity in boys. Sambia parents felt very strongly that
the rituals would only have their desired effect before the age of 10. In their
minds, proper sexual development and reproduction were placed at a risk without
insemination by older warriors. However, in other cultures, heterosexual relations
were viewed as logical outcome of parallel ritual processes (Knauft, 1993).
In sum, Sambia adults imagined that the boys awareness of his body and
sexuality took a pivotal turn in late childhood, well before gonardarche. Sambia
initiation, commencing after the age of 7, directly followed or was concurrent with
the onset of adrenarche, we would hypothesize, in males. Furthermore, although
the culture did not recognize the boys as being physically mature, their folk psychology recognized the age of 10 as the critical period for sexual subjectivity.
Given that the boys begin their sexual role as fellators on average by the age of
8.5 years, it is not surprising that by the age of 10 or so they would be regarded
as having developed sexual interest. Moreover, by the age of 1112, the period
of the second-stage initiation that advances the boys to the next level of the male
hierarchy, the boys have become aggressive fellators who actively pursue semen
to masculinize their bodies (Herdt and Stoller, 1990, p. 103). The ethnographic
evidence suggests that Sambia boys, by the age of 910, had begun to experience
awareness of their attractions toward others, including, in some cases, sexual attraction toward other boys or girls (Herdt, 2000). The culture feared the expression

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

The Magical Age of 10

597

of these attractions, if left unchecked. By initiating boys by the age of 10, the men
reduced the undesired effects of sexual attractions toward others, especially incest,
premarital sex, moral and social challenges to male control, including the use of
ritualized boy-inseminating to regulate male/male sexuality before marriage. Far
from being rare or exotic, these cases in New Guinea may actually highlight a
more general process in preliterate societies that has previously been ignored.

CONTEMPORARY STUDIESUNITED STATES


Accumulating studies from the United States over the past decade suggest
that the development of sexual attraction may commence in middle childhood and
achieve individual subjective recognition sometime around the age of 10 (Herdt
and Boxer, 1993; Pattatuci and Hamer, 1995; Hamer et al., 1993, as previously
reported in McClintock and Herdt, 1996, Fig. 1). As these studies have shown, first
same-sex attraction for males and females typically occurs at the mean age of 9.6
for boys (Herdt and Boxer, 1993), and between the ages of 10 and 10.5 for girls
(Hamer et al., 1993; Pattutuci and Hamer, 1995). It is significant to note that within
the range of these samples, males and females, heterosexuals and homosexuals, all
experienced sexual attraction at or near the age of 10, with male sexual subjectivity
a bit ahead of female sexual subjectivity.
In a prior study, it was speculated that adrenal puberty may be the source
of this change in samples of American men and women (McClintock and Herdt,
1996). In two separate studies, conducted by different investigators from diverse
academic fields in different parts of the country, the age of 10 was shown to be the
developmental marker for first memorable attraction toward others, regardless of
the gender of the object. Furthermore, these studies have a mean age of 37 years
(Hamer et al., 1993) in one case, and a mean age of 17.9 years in the other (Herdt
and Boxer, 1993). This differenceapproximately one generationis critical, for
it suggests that first attraction is independent of social age or generational cohort
differences. It also hints that cultural change (such as the mass media attention to
intervening events of the AIDS epidemic) have not directly affected the age of onset
of attraction in the youngest cohort. Furthermore, if the reports of first attraction
were biased by retrospection to a significant degree, one would expect that the
difference in social experience and the persons proximity to the developmental
marker would have resulted in different reported outcomes. Since the reported age
is the same for both generations, a deeper biopsychosocial structure of influence
is hypothesized to be at work in both cohorts.
The study of Herdt and Boxer (1993) reported first attraction by an average age
of 9.6 for boys and 10.1 for girls. Typically, the age of first homoerotic fantasy was
11.2 for males and 11.9 for females, with sexual conduct with the same gender
delayed by 2 years for males, and about 4 years (to 15.2 years on average) for

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

598

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

Herdt and McClintock

females (Herdt and Boxer, 1993, chapter 5). It was not until completing the study
that the investigators went to the larger research literature for comparison, and
discovered that the same age had previously been reported by Saghir and Robins
(1973, p. 232) from their study a generation before. They assert:
The natural history for the development of homosexual responses, overt behavior and finally
recognition of identification of other homosexuals with their groups could be illustrated by
the following chronological account: As early as the age of 10 I would feel attracted to
my teachers and would want to be with them and do things for them. Shortly later, I started
having crushes on my classmates. I would think about them and desire to be with them.

In a different study, focused on risk factors for suicide in gay and bisexual
youth, Remafedi et al. (1991, Table 3) found that the mean age for first homosexual
attraction in their sample of 137 males ranged from 9.27 to 10.66 years. The work
of Savin-Williams (1998) shows a similar age of onset of sexual attraction in
younger gays and lesbians (see also Savin-Williams and Diamond, 2000). An
important early study by DAugelli (1991, p. 141) of 77 college males also found
an awareness of attraction toward other males on average by the age of 10.8 years.
Given the difficulties of remembering and reporting common to survey studies,
and given the strong cultural bias of the folk psychology to mark sexuality after
gonadarche, it is remarkable that the age of attraction clusters around the age of 10
in all these studies.
What role does sexual orientation play in the emergence of attraction before
gonadarche? Clearly, more research is needed to answer this question with authority; however, one should be cautious in concluding that the development of
sexual attraction differs significantly by sexual orientation. First, the development
of sexual attraction in heterosexuals and homosexuals, (defined by fantasy and
desire and their vicissitudes) does not seem to be dependent upon the biological
or social concomitants of gonadarche. Second, differences in sexual precocity for
heterosexuals versus homosexuals do not seem to be significant predeterminants of
adolescent sexual outcomes (Bailey and Zucker, 1995; Money and Lewis, 1990).7
Indeed, a recent study has found the reverse: Bailey and Oberschneider (1997,
pp. 438439) in an intriguing retrospective survey of 136 professional dancers
found that straight males experience their first heterosexual feelings at an average
age of 8.9 years, whereas the gay males report their first attractions at the age of
10.4 years, a significant difference.
What role do gender differences play in the development of attraction? The
slightly younger age for males compared to females may be influenced by the social
and historical conditions of gender role performance, as in so many other areas of
gender development. Summarizing the Kinsey data on male sexual development,
7 Survey studies have found differences in adolescent attraction levels between homosexuals and hetero-

sexuals. Remafedi et al. (1992, pp. 716717) report that the average experience of first sexual attraction
is at a younger age for heterosexuals than for homosexuals in a large Minnesota school survey. On
average, the straights reported an average age of 15 for first heterosexual attractions, whereas for gays
it was 15.6 years for their first homosexual attraction.

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

The Magical Age of 10

599

Gagnon (1971, p. 239) once noted: For males early adolescence is commonly
characterized by the onset of early sexual activity which is conducted in the context
of secrecy experienced in tension with the public masculine striving associated with
homosexuality. In contrast, among females overt sexual activity is infrequent; they,
like males, live in a world dominated by their own gender, but it is a more public
world designed to promote future heterosociality. Hence, about 20% of boys, but
only 10% of girls in Kinseys sample, had experienced orgasm to masturbation by
the age of 12. A generation later, Baldwin and Baldwin (1997, p. 193) state, Boys
report noticing the pubertal changes in sexual excitability 2 or 3 years earlier (and
much more often per week) than girls do, giving boys yet another several years
head start over girls in learning about the physical pleasures of sex. In a study by
Knoth et al. (1988, p. 79, and Table 1), The modal age of first arousal for boys
across all our samples was between 11 and 12. The modal age for the first arousal
for girls was 23 years later than for boys. Perhaps even more important for the
present discussion of adrenal puberty, Knoth et al. (1988, p. 79, and Table 1) found
that 40% of males reported having their first sexual arousal by the age of 8, whereas
the aggregate of 60% reported theirs by the age of 10.
Cultural change is occurring all the time, and we should be surprised if its
effects were not felt in the area of the development of sexuality. However, a comparison of the Kinsey recall data with those of the recent National Health and Social
Life Survey suggest only modest age cohort differences related to the emergence
of sexual attraction in young people (Laumann et al., 1994; Michaels, 1996). It is
tempting to posit that whatever is changing in the context of desires and objects
of attraction, the deep structure of the development of attraction as a subjective
process is only crudely associated, if at all, with cultural change.
EARLY MATURATION AND SEXUAL ATTRACTION
It has long been theorized that the age of menarche varies with social and
historical conditions, and indeed, that modernity has brought about increasingly
earlier ages of maturation (Danker-Hopfe, 1986; Herdt and Leavitt, 1998; Khan
et al., 1996). Thus, as modernity advances, with increments in diet, health care,
education, and maternal care, gonadal puberty has a tendency to emerge earlier in
development. Although a variety of studies have focused on this point, a startling
new clinical study indicates that precocious gonadal puberty is increasingly apparent in the United States and this observation bears upon the conceptualization
of sexual attraction.
In a report based upon a study of 17,077 girls seen by a cross-section of 225
clinicians in the United States, Herman-Giddens and colleagues (1997) reported a
substantially younger age of gonadal pubertal traits. At the age of 8, the clinicians
found that 48.3% of African-American girls and 14.7% of white girls had begun
pubertal development. Breast and pubic hair development in proportions increasing

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

600

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

Herdt and McClintock

toward adult norms were registered for these girls, with nearly 17% of the AfricanAmerican revealing axillary hair. The authors note that for each characteristic,
African-American girls were more advanced than white girls at the same age
(Herman-Giddens et al., 1997, p. 507). By the age of 12, they report that 62%
of the African-Americans and 35% of the white girls had begun menses. With
respect to breast developmentin terms of Tanner stagesthe mean age of onset
for breast development was 8.87 years for African-American girls and 9.96 years
for white girls. These developmental characteristics are markedly younger than
suggested by standard pediatric textbooks (Herman-Giddens et al., 1997, p. 509)
on the subject. The authors conclude that More appropriate standards for defining
precocious and delayed puberty may need to be developed, taking into account
racial differences (Herman-Giddens et al., 1997, p. 511). The implications for
the development of sexual attraction and sexual behavior before adolescence are
obvious but bear scrutiny (McClintock et al., 1998).
The authors conclusion seems to be directly related to what typically would
be called gonadal puberty, though, in fact, the early ages observed by the clinicians fall between adrenal and gonadal puberty. The standard thinking in pediatric
practice previously did not differentiate between these two processes, and the
statement just made seems to conflate them. The question arises: If gonadarche is
demonstrably related to social and historical changes in maturation, might some
of these changes also result from adrenal puberty? At the present we must rely
again upon inferences drawn from studies of the social conditions of responses to
early sexual attraction and sexual behavior prior to the onset of adolescence.
Cultures may, as has long been known, either support or inhibit the expression of sexual expressions before adolescence (Ford and Beach, 1951; Herdt, 1997;
Mead, 1961). Sexually approving culturesthose in which sexual play in childhood is tolerated or even encouraged, differ strikingly from sexually disapproving
cultureswherein family and community may frown upon or even punish sexual
exploration or curiosity prior to the age at which the sexual culture thinks this is
natural and normal (Carrier, 1980; Herdt, 1997). It might be predicted that early
physical maturation, and/or sexual behavior, would be more readily approved in
such sex positive cultures. In contemporary Norway, for example, Langfeldt
(1981) describes sexual relations among normative 812-year-old males as a regular part of boys subcultures. He asserts that it is common for these male peer
groups to sexually experiment in secret (Langfeldt, 1981, pp. 6768). Langfeldt
(1990) estimates that about 10% of all Norwegian children between the ages of 4
and 10 masturbate to orgasm. It is interesting to note that clinical reports reveal
the age of 10 as typical for the beginning of sexual feelings. If early maturation
and sexuality are positively regarded in Norway, we might then predict that the
expression of early sexual attraction and the expression of sexual feelings will
develop relatively unimpeded.
Conversely, Musaph (1990) has reviewed the statistical evidence for first intercourse in western countries and finds the average age onset to be in the teens. Both

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

PL113-227176

September 16, 2000

The Magical Age of 10

10:6

Style file version Nov. 19th, 1999

601

parental control and religious affiliation strongly influenced the age of first sexual
intercourse. It may be that sexual intercourse, especially in sexually disapproving cultures, may serve to inhibit knowledge of early sexual attraction. Ironically,
however, the same conditions of sexual ignorance, when accompanied by a reign
of silence or taboo on sex talk (Fine, 1988) in these sexual cultures, may allow
for the emergence of pristine attractions and sexual fantasies, before adolescence.
Diaz (1998, p. 108) has noted how Latino culture in the United States creates
conditions of this kind in the early initiation of sexual intercourse and bisexual
behavior among males.
To illustrate this effect, recall the earlier cited study of the development of
same-sex feelings among gay and lesbian self-identified youth in Chicago (Herdt
and Boxer, 1993). Many of the 202 youth (age range of 1420) in the Chicago
study said that in growing up they always felt different, and their parents often
described how their sons and daughters were perceived to be different than the
others. Their parents, in turn, described their gay or lesbian childs development
as being more creative, academic, or artistic (for boys), or more athletic,
introspective, or competitive (for girls), compared to the peers and sibs of
these children; however, such stereotypes also reflect our societys attitudes about
normative gender development in childhood (Herdt and Koff, 2000). There is no
reason to believe that gonadal puberty arrived earlier than normal in this population.
However, by the age of nine and half on the average, the Chicago boys and girls
had experienced their first erotic attraction to the same-sex. In short, they were
aware and in some cases excited or aroused by another person, typically a peer or
friend, suggesting that these children had already recognized sexual attraction in
themselves and were on the path to sexual maturity.
Thus, sexual subjectivity (probably including sexual orientation) had achieved
an adult-like state well before gonadarche. Granted, this does not mean that the
sexuality of these boys and girls was complete or finished, nor that subsequent
psychosexual transformations would ensue. Instead, it seems plausible to infer
that whatever form their sexual and social careers take, sexual attraction following
adrenarche creates the conditions for sexual subjectivity and behavior in children
well before the society expects this to happen. In this respect, the precolonial New
Guinea societies may have anticipated outcomes ahead of those in contemporary
American society.

CONCLUSION
This article has examined the hypothesis that sexual attraction emerges after
the advent of adrenal puberty, typically precipitating the development of stable
and memorable sexual attraction by the age of 10 across cultures. Two pubertal
processesadrenarche and later gonadarcheare suggested as doing the work of
maturation, including the development of phenotypic gonadarche, with important

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

602

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

Herdt and McClintock

implications for the emergence of sexual awareness and behavior. This argument
effectively expands the period of puberty to encompass a wider span of human
development after the age of 6. It also suggests that the emergence of attraction, as
found in the New Guinea culturesboth in homoerotic and heteroerotic forms, and
in the contemporary United States among males and females in both homosexuals
and heterosexualsmay constitute a good candidate for being a human universal
of sexuality.
Middle childhood should no longer be viewed as a period of hormonal quiescence. Nor should we believe that for all children, there is an absence of sexual
subjectivity before gonadarche. Rather, the accumulating evidence suggests that
there is more sexual subjectivity occurring during childhood than previously believed, especially from the age of 6 onward, with the onset of adrenarche. The
key in the United States is that between the fourth and fifth grades, the childs
sexual attractions have already begun to stabilize or consolidate, becoming robust
and memorable, suggesting the results of an earlier developmental process. The
stability of the attraction is manifest by its memorability, accessible even in late
adulthood. When thinking of how sexual risk-taking is regarded in development,
and is sensitive to the context of relationships, it is critical to reconsider the early
onset of sexual attraction before adolescence and its implications for social policy
(Ehrhardt, 1996).
Although cross-cultural differences in the meanings of sexual arousal and
attraction are impressive, the evidence for a deeper structure of adrenal hormonal
development that influences the sequence and timing of sexual attraction before
adolescence is profound. This is not to say that cultures may of course thwart the
emergence of developmental subjectivities of sexual attraction in late childhood,
through the use of beliefs, taboos, rituals, and social gender roles. Are the internal
processes associated with adrenal puberty robust enough to overcome these social
barriers in the development of individual development of the body and fantasy
before gonadal puberty? We do not know the answer to this question; however, as
Freud (1905) speculated long ago, cultures may exercise an enormous constraint
upon the emergence of sexuality and hence, the subjective memory of, as well as
the expression of, sexual aim and object attractions. When a culture completely
denies or forgets the earlier experience of childhood upon adult development,
we have what Benedict (1938) once referred to as cultural discontinuity. It is
tempting to argue that if attraction typically develops during adrenarche but is
ignored or repressed by adults retrospection about sexual development, particularly before it becomes stabilized around the age of 10, the contemporary United
States may be a good example of a society in which discontinuity in sexuality is a
common developmental experience, and may affect the memory of earliest sexual
attraction (Herdt, 1990). Because male and female, as well as homosexual and heterosexual experiences of attraction were found before the age of 10, the internal
representation of sexual attraction is robust and memorable enough to overcome
these societal constraints (McClintock and Herdt, 1996).

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

PL113-227176

September 16, 2000

The Magical Age of 10

10:6

Style file version Nov. 19th, 1999

603

We should not ignore the context of political power in the social regulation of
childhood and adolescent sexuality. In precolonial New Guinea, it may well have
been the case that adrenal puberty led to sexual attraction in ways that directly
or indirectly challenged male power and gender hierarchy. Clearly, the implementation of strict avoidance taboos and gender segregation constitute powerful
indicators of adult male authority and the attempt to control adolescent sexual
attraction and behavior. The need for strict identification with the same-gender
parent, and political solidarity in times of warfare, may have produced a general
structural effort to exaggerate gender differences and assert sexual control. These
points lead to a generalization about the New Guinea societies: When a society
worries over the effects of early gender development, and the expression of sexual
attraction before adulthood, its folk psychology and institutions will implement
controls on the childs sexuality well before gonadarche. It is remarkable that our
own postindustrial society continues to exert similar powerful controls over childhood sexuality in the face of enormous change and access to sexual knowledge
and the media. Sexuality in the western liberal democracies, it would seem, is still
a challenge to forces of social regulation and authority.
That western and nonwestern societies have focused upon the age of 10 as a
memory marker for development is thus no coincidence, but neither should it be
regarded as a great mystery. The age of 10 is not magicalonly a convenience
marker in the cultural reasoning of societies about powerful hormonal processes.

ACKNOWLEDGMENTS
We would like to thank Todd Rawls, and Niels F. Teunis, for their helpful
comments on this paper.

REFERENCES
Bailey, J. M., and Oberschneider, M. (1997). Sexual orientation and professional dance. Arch. Sex.
Behav. 26: 433444.
Bailey, J. M., and Zucker, K. J. (1995). Childhood sex-typed behavior and sexual orientation: A conceptual analysis and quantitative review. Dev. Psychal. 31: 4355.
Baldwin, J. D., and Baldwin, J. I. (1997). Gender differences in sexual interest. Arch. Sex. Behav. 26:
179210.
Bell, A. P., Weinberg, M. S., and Hammersmith, S. (1981). Sexual Preference, Indiana University Press,
Bloomington.
Bem, D. (1996). Exotic becomes erotic: A developmental theory of sexual orientation. Psychol. Rev.
103: 320335.
Benedict, R. (1938). Continuities and discontinuities in cultural conditioning. Psychiatry 1: 161167.
Blumenthal, J., Giedd, J. N., Jeffries, N. O., Castellanos, F. X., Liu, H., Tijdenbos, A., Paris, T.,
Evans, A. C., and Rapaport, J. L. (1999). Brain development during childhood and adolescence:
A longitudinal MRI study. Natl. Neurol. 2(10): 861863.
Byne, W., and Parsons, B. (1993). Human sexual orientation: The biological theories reappraised. Arch.
Gen. Psychiat. 50: 228239.

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

604

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

Herdt and McClintock

Carrier, J. (1980). Homosexuality in cross-cultural perspective. In Marmor, J. (Ed.), Homosexual Behavior: A Modern Reappraisal, Basic Books, New York, pp. 100122.
Danker-Hopfe, H. (1986). Menarchal ages in Europe. Yearbook Phys. Anthropol. 29: 81112.
Deacon, A. B. (1934). Malekula: A Vanishing People in the New Hebrides, George Routledge, London.
Diaz, R. M. (1998). Latino Gay Men and HIV, Routledge, New York.
DAugelli, A. (1991). Gay men in college: Identity processes and adaptations. J. Col. Stud. Dev. 32:
140146.
Ehrhardt, A. A. (1996). Our view of adolescent sexuality: Risk behavior without developmental context.
Am. J. Pub. Health 86(11): 15231525.
Ehrhardt, A., and Meyer-Bahlburg, H. F. L. (1981). Effects of prenatal sex hormones on gender-related
behavior. Science 211: 13121318.
Elder, G. (1975). Adolescence in the Life Cycle: Psychological Change and Social Context, Halsted
Press, New York.
Fine, M. (1988). The Missing Discourse of Desire, Harvard Ed. Rev. Vol. 2, pp. 2953.
Ford, C. S., and Beach, F. A. (1951). Patterns of Sexual Behavior, Harper and Brothers, New York.
Freud, S. (1905/1953). Three Essays on the Theory of Sexuality. Standard Edition of the Complete
Psychological Works of Sigmund Freud, Vol. 8, Hogarth, London.
Gagnon, J. (1971). The creation of the sexual in adolescence. In Kagan, J., and Coles, R. (Eds.), Twelve
to Sixteen: Early Adolescence, W. W. Norton, New York, pp. 231257.
Gagnon, J. (1990). The explicit and implicit use of the scripting perspective in sex research. Annu. Rev.
Sex Res. l: 144.
Gelman, S., Collman, P., and Maccoby, E. E. (1986). Inferring properties from categories versus
inferring categories from properties: The case of gender. Ch. Dev. 57.
Gorman, M. R. (1994). Male homosexual desire: Neurological investigations and scientific bias. Perspect. Biol. Med. 38: 6181.
Green, R. (1987). The Sissy Boy Syndrome and the Development of Homosexuality, Yale University
Press, New Haven.
Hamer, D. H., Hu, S., Magnuson, V. L., Hu, N., and Pattutucci, A. M. L. (1993). Linkage between and
markers on the X chromosome and male sexual orientation. Science 261: 321327.
Herdt, G. (1981). Guardians of the Flutes, McGraw-Hill, New York.
Herdt, G. (1984). Ritualized homosexual behavior in the male cults of Melanesia, 18621983: An
introduction. In Herdt, G. (Ed.), Ritualized Homosexuality in Melanesia, University of California
Press, Berkeley, pp. 181.
Herdt, G. (1987). The Sambia: Ritual and Gender in New Guinea, Holt, Rinehart, and Winston, New
York.
Herdt, G. (1990). Developmental continuity as a dimension of sexual orientation across cultures. In
McWhirter, D., Reinisch, J., and Sanders, S. (Eds.), Homosexuality and Heterosexuality: The
Kinsey Scale and Current Research, Oxford University Press, New York, pp. 208238.
Herdt, G. (1991). Representations of homosexuality in traditional societies: An essay on cultural
ontology and historical comparison, Part I. J. Hist. Sex. 1: 481504.
Herdt, G. (1997). Same Sex, Different Cultures: Perspectives on Gay and Lesbian Lives, Westview
Press, New York.
Herdt, G. (1999). Sambia Sexual Culture: Essays from the Field, University of Chicago Press, Chicago.
Herdt, G. (2000). Why the Sambia initiate boys before age 10. In Bancroft, J. (Ed.), Sexual Theory,
Indiana University Press, Bloomington.
Herdt, G., and Boxer, A. (1993). Children of Horizons, Beacon Press, Boston.
Herdt, G., and Koff, B. (2000). Something to Tell You: The Road Families Travel When a Child is Gay,
Columbia University Press, New York.
Herdt, G., and Leavitt, S. C. (1998). Adolescence in Pacific Island Societies, University of Pittsburgh
Press, Pittsburgh.
Herdt, G., and Stoller, R. J. (1990). Intimate Communications: Erotics and the Study of Culture,
Columbia University Press, New York.
Herman-Giddens, M. E., Slora, E. J., Wasserman, R. C., Bourdony, C. J., Bhapkar, M. R., Koch, G. G.,
and Hasemeier, C. M. (1997). Secondary sexual characteristics and menses in young girls seen
in office practice: A study from the pediatric research in office settings network. Pediatrics 99:
505512.

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

The Magical Age of 10

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

605

Hopper, B. R., and Yen, S. S. (1975). Circulating concentrations of dehydroepiandrosterone and dehydroepiandrosterone sulphate during puberty. J. Clin. Endocrinol. Met. 40(3): 458461.
Khan, A. D. (1996). Early childhood determinants of age at Menarche in rural Guatemala. Am. J.
Human. Biol. 8: 717723.
Kimmel, M. S. (2000). The Gendered Society, Oxford University Press, New York.
Kinsey, A. C., Pomeroy, W. B., and Martin, C. E. (1948). Sexual Behavior in the Human Male,
W. Saunders, Philadelphia.
Knauft, B. (1993). South Coast New Guinea Cultures, Cambridge University Press, New York.
Knoth, R., Boyd, K., and Singer, B. (1988). Empirical tests of sexual selection theory: Predictions of
sex difference in onset, intensity, and time course of sexual arousal. J. Sex Res. 24: 7389.
Korth-Schutz, S. (1989). Precocious adrenarche. In F. M. G. (Ed.), Pediatric Adolescent Endocrinology,
Karger, New York, pp. 226235.
Langfeldt, T. (1981). Childhood masturbation: Individual and social organization. In Constantine, L. L.,
and Martinson, F. M. (Eds.), Children and Sex, Little, Brown and Co., Boston, pp. 6374.
Langfeldt, T. (1990). Early childhood and juvenile sexuality, development and problems. Handbook of
Sexology, Vol. 7, Elsevier Science, Amsterdam, pp. 179200.
Langness, L. L. (1967). Sexual antagonism in the New Guinea Highlands: A Bena Bena example.
Oceania, 37(3): 161177.
Laumann, E. O., Gagnon, J. H., Michael, R. T., and Michaels, S. (1994). The Social Organization of
Sexuality: Sexual Practices in the United States, University of Chicago Press, Chicago.
Malinowski, B. (1929). The Sexual Life of Savages in North-Western Melanesia, Harcourt, Brace and
World, New York.
McClintock, M., and Herdt, G. (1996). Rethinking puberty: The development of sexual attraction. Curr.
Direc. Psychol. Sci. 5: 178183.
McClintock, M., Herdt, G., and Rosenfield, R. (1998). Preadolescent determinants of sexuality. Pediat.
Update 19(9), 110.
Maccoby, E. E. (1988). Gender as a social category. Dev. Psychol. 24: 755765.
Maccoby, E. E., and Jacklin, C. (1974). The Psychology of Sex Differences, Stanford University Press,
Stanford.
Malcolm, L. A. (1968). Determination of the growth curve of the Kukukuku people of New Guinea
from dental eruption in children and adult height. Archaeol. Phys. Anthropol. Oceania 4: 7278.
Mead, M. (1927). Coming of Age in Samoa, Norton, New York.
Mead, M. (1935). Sex and Temperament in Three Primitive Societies, Dutton, New York.
Mead, M. (1956). New Lives for Old. Cultural Transformation: Manus 19281953, William Morrow,
New York.
Mead, M. (1961). Cultural determinants of sexual behavior. In Young, W. C. (Ed.), Sex and Internal
Secretions, Williams & Williams, Baltimore, MD, pp. 14331479.
Meyer-Bahlburg, H. (1984). Psychoendocrine research on sexual orientation. Current status and future
options. Prog. Brain Res. 61: 375398.
Meyer-Bahlburg, H. (1997). The role of prenatal estrogens in sexual orientation. In Ellis, L., and
Ebertz, L. (Eds.), Sexual Orientation: Toward Biological Understanding, Ct. Praeger, Westport,
pp. 4151.
Michaels, S. (1996). The prevalence of homosexuality in the United States. In Cabaj, R. P., and Stein,
T. S. (Eds.), Textbook of Homosexuality and Mental Health, pp. 4365.
Moller, L. C., Hymel, S., and Rubin, K. H. (1992). Sex typing in play and popularity in middle
childhood. Sex Roles 26: 331353.
Money, J. (1987). Sin, sickness, or society? Am. Psychol. 42: 384399.
Money, J. (1997). Principles of Developmental Sexology, Continuum, New York.
Money, J., and Ehrhardt, A. (1972). Man and Woman, Boy and Girl, Johns Hopkins University Press,
Baltimore, MD.
Money, J., and Lewis, V. G. (1990). Puberty: Precocious, delayed and incongruous. In Perry, M. E.
(Ed.), Handbook of Sexology, 7: Childhood and Adolescence, pp. 236262.
Musaph, H. (1990). First sexual intercourse. In Perry, M. E. (Ed.), Childhood and Adolescent Sexology,
Elsevier, Amsterdam, pp. 287296.
Pattatuci, A., and Hamer, D. (1995). Developmental and familiarity of sexual orientation in females.
Behav. Genet. 25: 407420.

P1: FLF/FGL

P2: FLF

Archives of Sexual Behavior [asb]

606

PL113-227176

September 16, 2000

10:6

Style file version Nov. 19th, 1999

Herdt and McClintock

Piaget, J. (1971). Structuralism (Trans. C. Maschler), Harper Torchbooks, New York.


Plummer, K. (1995). Telling Sexual Stories, Routledge, New York.
Read, K. E. (1954). Cultures of the central highlands, New Guinea. Southwestern J. Anthropol. 10:
143.
Remafedi, G., Farrow, J. A., and Deisher, R. W. (1991). Risk factors for attempted suicide in gay and
bisexual youth. Pediatrics 87: 869875.
Remafedi, G., Resnick, M., Blum, R., and Harris, L. (1992). Demography of sexual orientation in
adolescents. Pediatrics 89: 714721.
Saghir, and Robins. (1973). Male and Female Homosexuality, Williams and Wilkins, Baltimore.
Schieffelin, E. L. (1976). The Sorrow of the Lonely and the Burning of the Dancers, St. Martins, New
York.
Simon, W., and Gagnon, J. H. (1973). Psychosexual development. In Gagnon, J. H., and Simon, W.
(Eds.), The Sexual Scene, Transaction Books, New Brunswick, NJ, pp. 2947.
Stoller, R. J. (1968). Sex and Gender, Science House, New York.
Stoller, R. J., and Herdt, G. (1982). The development of masculinity: A cross-cultural contribution.
J. Am. Psychoanal. Assoc. 30: 2959.
Stoller, R. J., and Herdt, G. (1985). Theories of origins of male homosexuality: A cross-cultural look.
Arch. Gen. Psychiat. 42: 399404.
Storms, M. (1981). A theory of erotic orientation development. Psychol. Rev. 88: 340353.
Tanner, J. M. (1973). Trend toward earlier menarche in London, Oslo, Copenhagen, The Netherlands
and Hungary. Nature 243: 9596.
Townsend, J. M., and Wasserman, T. (1997). The perception of sexual attractiveness: Sex differences
in variability. Arch. Sex. Behav. 26.
Van Baal, J. (1966). Dema, Martinus Nijhoff, The Hague.
Van Baal, J. (1984). The dialectics of sex in Marind-anim Culture. In Herdt, G. (Ed.), Ritualized
Homosexuality in Melanesia, University of California Press, Berkeley, pp. 128166.
Van Gennep, A. (1960). The Rites of Passage, In Vizedom, M. K., and Caffee, G. L. (Eds.), University
of Chicago Press, Chicago.
Vance, C. S. (1991). Anthropology rediscovers sexuality: A theoretical comment. Soc. Sci. Med. 33:
875884.
Worthman, C. (1999). Faster, farther, and higher: Biology and the discourses on human sexuality. In
Suggs, D. N., and Miracle, A. (Eds.), Culture, Biology and Sexuality, University of Georgia Press,
Athens, GA, pp. 6475.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

10:10

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 29, No. 6, 2000

Sexual Identity Trajectories Among Sexual-Minority


Youths: Gender Comparisons
Ritch C. Savin-Williams, Ph.D.1,3 and Lisa M. Diamond, Ph.D.2

The present investigation explored gender differences in sexual identity developmentfirst same-sex attractions, self-labeling, same-sex sexual contact, and
disclosureamong 164 sexual-minority young adults. Based on interviews, results indicated the value of assessing gender differences in the context, timing,
spacing, and sequencing of sexual identity milestones. Adolescent males had an
earlier onset of all milestones except disclosure. The context for sexual identity
milestones were likely to be emotionally oriented for young women and sexually
oriented for young men. The gap from first same-sex attractions (89 years of
age) to first disclosure (around 18 years) averaged 10 years for both sexes. Young
women followed label-first developmental trajectories; men were more likely to
pursue sex before identifying themselves as gay. In terms of achieving sexual identity milestones, gender mattered, but it was not everything.
KEY WORDS: sexual-minority youth; sex differences; sexual identity; coming out; developmental
milestones.

INTRODUCTION
Theoretical models describing the advent of a same-sex erotic identity were
first proposed by clinical and developmental psychologists over two decades ago.
These coming out or sexual identity models remain the most prevalent conceptualization of sexual-minority development, characterizing the process by which
individuals recognize, define, and accept their status as lesbian, gay, or bisexual. Although diverse in conceptual underpinnings, sexual identity models are
1 Department

of Human Development, Cornell University, Ithaca, New York 14853; RCS15@cornell.


edu.
2 Department of Psychology, University of Utah, Salt Lake City, Utah 84112.
3 To whom correspondence should be addressed.

607
C 2000 Plenum Publishing Corporation
0004-0002/00/1200-0607$18.00/0

P1: FLF
Archives of Sexual Behavior [asb]

608

PL113-227177

September 16, 2000

10:10

Style file version Nov. 19th, 1999

Savin-Williams and Diamond

nearly universal in their linear stage sequences (Cohen and Savin-Williams, 1996;
McConnell, 1994). For example, most models mark the onset of sexual identity
development as the individuals first awareness of same-sex attractions, presumed
to occur in late childhood or early adolescence. This is followed some years later
by a period of testing and exploration, during which youths seek information about
gay, lesbian, and bisexual lifestyles and communities and/or engage in experimentation with same-sex sexual contact. Succeeding stages of identity development
entail adopting a sexual-minority label, disclosing this sexual identity to others, becoming involved in a same-sex romantic relationship, and celebrating ones sexual
identity within a larger social context (e.g., the political arena).
Although this linear progression is intuitively appealing, extant research suggests that it is far from universal. Rather, considerable diversity exists among
sexual-minority youth of different backgrounds, cohorts, and ethnicities regarding the paths taken to sexual-minority identification, particularly regarding the
relative ordering of first same-sex sexual contact and self-labeling among young
gay and bisexual males (Dube, 2000; Dube and Savin-Williams, 1999; SavinWilliams, 1998). However, the most notable deviations from the standard model
have been documented among lesbians and bisexual women (Diamond, 1998,
2000; Golden, 1996; Rust, 1995). This is perhaps to be expected, given that most
coming-out models were originally derived from exclusively male samples. Not
only do women typically initiate sexual identity development at later ages than
men, but they also appear to do so for different reasons, often referencing emotional rather than sexual feelings for women. In some cases, the order in which
they complete stages of sexual identity development is reversed (Diamond, 1998).
In critiquing such models, numerous researchers (Eliason, 1996a,b; Morris, 1997;
Peplau et al., 1998) have called for a multidimensional approach to sexual identity
development that seriously considers the unique ways in which females and males
experience their erotic desires. The research presented here accepts this challenge
by investigating diversity in the timing, spacing, and context of traditional sexual identity milestones both within and across male and female sexual-minority
youth.

RECENT EMPIRICAL RESEARCH ON SEXUAL


IDENTITY MILESTONES
The recent increase in the availability of diverse populations of sexual-minority
youths for research purposes has spawned several investigations that document various developmental aspects of growing up as a lesbian, gay, or bisexual person. As
a matter of recent protocol, researchers generally test for gender differences in the
timing of sexual identity milestones, with or without a conceptualization of how
or why the sexes might differ on these domains. Furthermore, these analyses often

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

Gender Differences in Sexual Identity Development

10:10

Style file version Nov. 19th, 1999

609

overlook factors such as the sequence of milestones, the duration between them,
and the context in which they are experienced.
Examined first are four recent studies of child and adolescent sexual identity
development that contain samples large enough for gender comparisons. The data
are summarized in Table I. All four studies recruited participants from communitybased support groups for youths; one (Rosario et al., 1996) supplemented its
sample with youths from college-based organizations. Youths of color were well
represented, with whites accounting for 2278% of the samples.
Herdt and Boxer (1993) interviewed youths from the Chicago Horizons support group. Although sexual identity labels were not reported for the sample, age of
first label (16 years) was the oldest of the four studies. Reported gender differences
included an earlier onset of same-sex behavior among males, over 2 years before
female youths; the greater likelihood of girls having opposite-sex experiences
before same-sex experiences; and the larger proportion of boys who had exclusive same-sex experiences. Perhaps most significantly, Herdt and Boxer proposed
gender variation in developmental sequencinga term unfortunately limited to
whether opposite-sex or same-sex behavior occurred first.
With a sample of youths from 14 community centers, DAugelli and
Hershberger (1993) reported that female adolescents became aware at a later age
than did males that they were attracted to same-sex membersthe only one of the
four studies to report this finding. The investigators uniquely assessed the length
of time between milestones. Males had a significantly longer period of time between first awareness and self-labeling (5 vs. 4 years) and between first awareness
and first disclosure (7 vs. 5 years). No gender differences emerged between selflabeling and first disclosure (just under 2 years) or between first awareness and first
same-sex encounter (between 4 and 5 years). Nearly 90% of youths had engaged
in same-sex activity, but males had many more such encounters.
Rosario et al. (1996) recruited youths from community-based and college
organizations in New York City. It is noteworthy that female youths composed
one half of the sample. They were significantly older than male youths when they
first considered (13.9 versus 12.5 years) or were certain of their sexual identity.
Similar to DAugelli and Hershberger (1993) but distinct from Herdt and Boxer
(1993) and DAugelli (1998), no gender differences were reported for age of first
same-sex sexual behavior.
In a follow-up anonymous questionnaire study, DAugelli (1998) solicited
a large sample of youths from 37 community-based youth groups. Significantly
more female than male youths identified themselves as bisexual during initial selflabeling and presently. Similar to his previous findings, no gender differences were
reported in the gap between first awareness and first same-sex sexual behavior;
male youths reporting a greater number of same-sex partners was also confirmed.
However, inconsistent with his previous findings, the sexes differed in age of first
self-labeling, age of first same-sex encounter, and the length of time between

16.0

16.8
89

16.0

16.8
92

14.6

16.0
86

74
26
11.1e
14.6

Female

14.6
14.7

95

65
31
10.8
13.3

Male

83
17
9.8
14.3
15.0

16.4
94

15.9e
15.3

88

Male

67
32
10.5
14.0

Female

15.7e

16.6
86

64
36
10.4
15.9e

Female

DAugelli (1998)d

610

b N = 194; percent females 27%; mean age 18.9.


c N = 156; percent females 49%; mean age 18.3.
d N = 260; percent females 44%; mean age 18.9.
e Significant gender differences.

14.8

16.7
91

75
25
9.8
14.9

Male

Rosario et al. (1996)c

September 16, 2000

= 202; percent females 27%; mean age 18.3.

10.1
15.2e

9.6
13.1

Female

DAugelli and Hershberger (1993)b

PL113-227177

aN

Sex label (%)


Gay/lesbian
Bisexual
Age of first awareness
Age of first same-sex
activity
Age of first label
Gay/lesbian
Bisexual
Age of first disclosure
Percent with same-sex
activity (%)

Male

Herdt and Boxer (1993)a

Archives of Sexual Behavior [asb]

Table I. Summary of Four Studies Documenting Gender Differences in Child and Adolescent Developmental Milestones

P1: FLF
10:10
Style file version Nov. 19th, 1999

Savin-Williams and Diamond

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

10:10

Gender Differences in Sexual Identity Development

Style file version Nov. 19th, 1999

611

self-labeling and first disclosure; compared to his earlier findings, the sexes no
longer differed in the time gap between first awareness and self-labeling and between first awareness and first disclosure. Over 80% of both male and female
youths reported that an awareness of same-sex attractions preceded same-sex sexual behavior. For both sexes, early age of awareness was significantly associated
with earlier self-labeling; for males, also with earlier onset of same-sex behavior.
It is difficult to discern generalized patterns from these studies. Investigations
agreed regarding the percent of youths who have had same-sex sexual behavior,
age of first awareness (during the 10th year), and age of first disclosure (during the
16th year). However, less agreement was evident for ages of first same-sex activity
and first self-labeling, both within and across the sexes. None of the studies found
gender differences regarding age of disclosure, but conflicting results emerged regarding ages of first awareness, same-sex activity, and self-labeling. Most importantly, the sequence of these events was not uniform from study to study. Although
both male and female sequence data from the study of Herdt and Boxer (1993)
appeared to agree with the proposed sexual identity models, the other three studies
did not support them. For example, most youths in DAugelli and Hershbergers
study (DAugelli and Hershberger, 1993) self-labeled before engaging in same-sex
sexual contact. Rosario et al.s lesbian but not bisexual female youths conformed
to sexual identity models (Rosario et al., 1996). Overall, the sexual identity model
sequence of stages fits male data slightly better than it does female data.
These conflicting findings constitute important evidence for the existence
of diverse trajectories of sexual-minority development among female and male
youths. Yet in order to interpret this diversity, and the extent to which it is attributable to gender, research must move beyond the conventional approach of
comparing mean ages at which female and male sexual minorities reach various
milestones. For example, as demonstrated by DAugelli (1998), there is value in
investigating the time gap between various milestones. The experience of sexual
identity development might be quite different for those who initiate and complete
this process within a single year versus those who remain within its grip for a
decade. The sequencing of events is also critically understudied. Herdt and Boxer
(1993) have examined whether female and male youths pursue same-sex sexual
contact before or after their first other-sex sexual contact, but clearly other possible sequences might be investigated. Most importantly, research should aim for
coordinated assessment of the timing, context, spacing, and sequencing of sexual
identity milestones, and these investigations should be grounded in current understanding of gender differences in sexual and social behavior in order to produce
meaningfully interpretable findings.
WHY SHOULD GENDER MATTER?
Some 20 years ago, Kirkpatrick and Morgan (1980) raised the question
of whether female homosexuality is a mirror image of male homosexuality or a

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

612

10:10

Style file version Nov. 19th, 1999

Savin-Williams and Diamond

specifically female phenomenon. That is, are female and male sexual minorities
more alike on the basis of sexual orientation than they are different on the basis
of gender? The balance of research over the past two decades supports the overwhelming importance of gender. For example, Saghir and Robins (1980) found
that in matters of sexual behavior, number of partners, and age of onset, gender
differences were greater than sexual orientation differences. Similar findings were
reported by Bailey et al. (1994), who examined gender and sexual orientation
differences in preferences for uncommitted sex; interest in visual sexual stimuli;
perceived importance of emotional and sexual fidelity; and perceived value of
partner characteristics such as age, physical attractiveness, and social status. In
all domains, gender effects prevailed. Independent of sexual orientation, women
showed less interest than men did in uncommitted sex and visual sexual stimuli,
ascribed less importance to a partners age and attractiveness, and attributed more
importance to emotional fidelity. Bailey and colleagues concluded that gender
had a considerably greater impact on these sexual and romantic domains than did
sexual orientation.
It is on the basis of findings such as these that Peplau et al. (1998, p. 393)
asserted, A sensible research strategy will be to develop separate causal analyses
of womens and mens sexual orientation, each grounded in generalizations that are
accurate descriptions of the phenomena associated with sexual orientation for that
gender. On the basis of extant research, we believe that the most important differences between males and females concern the subjective experience and perceived
importance of explicit sexual desire and activity. Thus, attempts to systematically
probe for multiple trajectories of female and male sexual-minority development
should focus on the precise role that sexual feelings and behaviors play in this
process. The current research adopts this focus.

CURRENT STUDY
We investigate four milestones that occupy a significant role in sexual identity
development: first same-sex attractions, first same-sex sexual contact, first selflabeling as nonheterosexual, and first disclosure of a nonheterosexual identity to
others. We broaden past investigations that report mean ages of these events in three
ways: by examining the contexts of these transitional events, the duration of time
between events, and variation in the ordering of two key transitionsfirst same-sex
contact and first self-labeling. Our aim is to probe for multiple trajectories of sexual
identity development and to determine the extent to which certain trajectories are
gender-specific.
In light of well-documented gender differences regarding sexual and social
behavior, explicit sexual desire and activity should play a more substantive role in
the sexual questioning of sexual-minority males than of females. Accordingly, we

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

Gender Differences in Sexual Identity Development

10:10

Style file version Nov. 19th, 1999

613

make the following predictions:


1. The context for first memories of same-sex attractions and self-labeling
will be sexually oriented for boys and emotionally oriented for girls.
2. Female sexual-minority youths will be more likely than would males to
pursue their first same-sex sexual contact within a romantic relationship.
3. Gender differences in age of first same-sex sexual contact will be mediated
by the context in which this contact occurs; specifically, youths who have
their first same-sex sexual contact with a romantic partner will have this
contact at a later age, independent of gender.
4. Male sexual-minority youths will be more likely to pursue same-sex sexual
contact at least 1 year before self-labeling, whereas females will be more
likely to pursue same-sex sexual contact at least 1 year after self-labeling.
In addition to conducting conventional tests of gender differences in milestone
ages and time spans between events, we examine whether such differences are
significant after controlling for variation in the relative sequencing of first samesex contact and first self-labeling (outlined in Hypothesis 4). Given the lack of data
of this nature, we do not make predictions regarding the results of these analyses.
Finally, additional exploratory analyses are also planned to address associations
between the context of first same-sex attractions and first self-labeling and the ages
at which these milestones occur.

METHOD
Participants
Participating youths included 78 women and 86 men between the ages of 17
and 25 years. The young women were engaged in a research project on the developmental trajectories of sexual-minority young adults; the young men composed
Sample Two in an earlier study on sexual-minority male youths (Savin-Williams,
1998). All youths met the inclusion criterion of claiming some degree of physical
or romantic interest in same-sex others and were diverse in social class, religious
affiliation, and size of hometown community, but less so in educational level and
ethnic/racial identification.
Youths were recruited through announcements in college classes on gender
and sexuality; flyers to campus social and political organizations; advertisements in
a community newsletter and public places (bar, bookstore, cafe) for sexual minorities; postings on Internet list-serves for sexual-minority students on several college
campuses; and referrals from other participants. This multiple recruitment strategy
was undertaken to draw participants along the spectrum of same-sex attractions.
For example, campus political organizations tend to include very out individuals

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

10:10

614

Style file version Nov. 19th, 1999

Savin-Williams and Diamond

who openly identify their sexuality; college Women Studies courses often draw
students who are just beginning to acknowledge their same-sex attractions.
The research project was advertised as an interview focusing on growing up
in the 1990s with physical or romantic attractions for the same sex. To volunteer,
youths contacted the principal investigator (RSW) either in person or by telephone
or electronic mail. Special efforts were made to include individuals who were not
comfortable openly identifying themselves as lesbian, gay, or bisexual by assuring
them that such identifications were not necessary for participation and that the
interviews were confidential and would not be audio- or video-taped.
Due to the nature of the recruitment strategy, response rate cannot be calculated because it is unknown how many potential participants who met selection
criteria did not volunteer for the study. However, 9% of women and 4% of men
who originally contacted the investigator either did not return efforts to contact
them or did not show up for the face-to-face or phone interviews. Due to human
subject considerations, no attempt was made to discover reasons for refusal to
participate.
Procedures and Measures
Participants were interviewed in-person or by telephone. A male research
assistant interviewed one-third of the male participants; the principal investigator
interviewed all other youths. Interviews were conducted in the principal investigators office or a place chosen by youths that afforded privacy and confidentiality. The nature and aims of the research project were explained, questions were
answered, and consent was secured in accordance with human subjects stipulations. Over 90% of interviews with male participants lasted from 45 to 90 min,
with a median of 60 min. The same format and content were followed for interviews with young women (median = 45 min), but several domains were omitted
(community activities, current harassment, attitudes toward AIDS). To increase
diversity of the female sample, the announcement was posted on several southern and eastern college campus list-serves. Forty-one percent of women participants took advantage of this opportunity, completing the interview by telephone
(median time = 45 min). There were no demographic differences between women
recruited in this manner and the other female participants, but women interviewed
by phone were more likely to identify as lesbian rather than bisexual or unlabeled,
2 (2, N = 71) = 6.67, p < 0.04.
Initial questions ascertained participants age, ethnicity, hometown community size, and family social class (based on occupational status of parents). The
remainder of the interview focused on milestones of sexual identity development,
from earliest memories of same-sex attractions to current feelings about ones sexual identity (for review see Savin-Williams, 1998). For purposes of this project,

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

10:10

Gender Differences in Sexual Identity Development

Style file version Nov. 19th, 1999

615

only data concerning the transitions of first same-sex attractions, first same-sex
sexual contact, first self-labeling, and first disclosure were analyzed.
First Same-Sex Attractions
Participants were asked, Describe your first memories of being attracted
to girls/boys. How old were you and what specifically do you remember? You
need not have interpreted the attractions as sexual in nature at that time. How far
back can you recall such an experience? The specific context of first same-sex
attractions and the age at which they occurred were recorded.
First Same-Sex Sexual Activity
Sex was defined as an act in which there was genital contact on the part
of either you, your partner, or both. If participants acknowledged that this had
occurred with someone of the same sex, they were asked to describe the encounter,
including their age, participants age, and the nature of their relationship.
Labeling Sexual Identity
Participants were asked, When did you first realize that you were not heterosexual? What memories do you have of how you came to the conclusion that
you were not heterosexual? The specific context of first self-labeling and the age
at which this event took place were recorded.
First Disclosure
Participants reported the age at which they first disclosed their nonheterosexual identity to another person and described their relationship to this individual.

RESULTS
General Characteristics of the Sample
An alpha of 0.05 was used for all significance tests. Table II presents sample characteristics separate by sex. Young women were significantly more likely
than young men to claim bisexual-based identities, 2 (4, N = 164) = 40.8, p <
0.0001. An analysis of variance detected no sex or sexual identity differences
regarding age of respondent.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

10:10

616

Style file version Nov. 19th, 1999

Savin-Williams and Diamond


Table II. Sample Characteristics Separate by Sex

Age (years)
Mean (SD)
Sexual identification [N (%)]
Lesbian/gay
Bisexual
Unlabeled
Bi-Lesbian/gay
Questioning
Ethnic/racial identification [N (%)]
White
African American
Asian/Pacific Islanders
Latina/o
Mixed race & other
Social class [N (%)]
Upper middle
Middle
Lower middle
Community size [N (%)]
Farm/rural
Small town
Small city/suburbs
Urban

Males

Females

21.6 (2.2)

20.8 (1.7)

71 (83)
6 (7)
4 (5)
1 (1)
4 (5)

27 (35)
26 (33)
7 (9)
8 (10)
10 (13)

62 (72)
5 (6)
6 (7)
10 (12)
3 (3)

61 (78)
2 (3)
10 (13)
2 (3)
3 (4)

39 (45)
27 (31)
20 (23)

30 (38)
31 (40)
17 (22)

13 (15)
17 (20)
31 (36)
25 (29)

8 (10)
7 (9)
48 (62)
15 (19)

A significant association was found between sex and community size, with
males more likely to have come from suburban environments, 2 (3, N = 164) =
11.2, p = 0.01. There were no other differences between male and female respondents, between sexual identity groups, and in 3-way interactions with sex and
sexual identity.

Context of First Attractions, Labeling, and Sexual Contact


Respondents were categorized into two groups according to the context in
which they experienced their first same-sex attractions (Table III). The sexual group
included individuals who described this experience as involving explicit same-sex
sexual thoughts or sexual activity. The emotional group included individuals who
described their first same-sex attractions as involving only emotional feelings for
the same sex. Not included were individuals who described this experience in
terms that were neither explicitly sexual nor emotional, such as fascinated with a
woman/man or admired the beauty of a woman/man (denoted ambiguous in
Table III). A chi-square test established that males were significantly more likely
than females to list a sexual context and females were more likely to list an emotional context, supporting Hypothesis 1, 2 (1, N = 130) = 23, 6, p < 0.00001.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

10:10

Style file version Nov. 19th, 1999

Gender Differences in Sexual Identity Development

617

Table III. Contexts of First Same-Sex Attractions, Same-Sex Sexual Contact,


and Self-Labeling

First same-sex attractions


Sexual
Emotional
Ambiguous
First same-sex sexual contact
Stranger
Relative (usually a cousin)
Friend
Romantic partner
First self-labeling as nonheterosexual
Sexual
Emotional
Facilitative environment

Males [N (%)]

Females [N (%)]

59 (69)
6 (7)
21 (24)

34 (44)
31 (40)
13 (17)

16 (20)
9 (12)
50 (64)
4 (5)

0 (0)
3 (5)
20 (33)
38 (62)

52 (63)
12 (15)
18 (22)

31 (40)
22 (28)
25 (32)

Respondents were categorized into three groups according to the context in


which they first labeled themselves nonheterosexual (Table III). In addition to the
sexual and emotional groups just described, the facilitative environment group
included individuals who described this experience as involving exposure to the
concept of same-sex sexuality through television, books, or other sexual-minority
individuals. A chi-square test detected a significant association between sex and
distribution into these categories, 2 (2, N = 160) = 9.3, p = 0.01. Examination
of standardized cell deviates again confirmed Hypothesis 1: Males were significantly more likely than females to list a sexual context and females were more
likely to list an emotional context.
To determine whether the context of a youths first same-sex attractions was
related to the age at which he or she experienced those attractions or pursued
same-sex contact, a multivariate analysis of variance (MANOVA) was carried out
with sex and attraction context as independent variables (excluding those in the
ambiguous category) and age of first same-sex attractions and first same-sex
sexual contact as dependent variables. This analysis did not include individuals
who had not yet experienced same-sex sexual contact (8 males and 17 females).
There was a significant effect of sex, Wilks lambda = 0.93 (2, 110), p = 0.014.
Bonferroni-corrected univariate tests demonstrated that males had earlier same-sex
sexual attractions than females had, controlling for attraction context, F(1, 111) =
7.0, MSE = 12.7, p < 0.02.
Predicting age of sexual identity labeling, a MANOVA revealed no significant
effects for sex of the individual, F(1, 154), MSE = 11.2, p > 0.05, or context of
labeling, F(2, 154), MSE = 18.8, p > 0.05. However, the interaction effect was
significant, F(2, 154), MSE = 24.2, p = 0.024. Males who first labeled within
a sexual context were significantly younger (15.6 years) than all other groups
(17.6 years).

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

10:10

618

Style file version Nov. 19th, 1999

Savin-Williams and Diamond

Table III also presents the proportion of male and female youth who had their
first same-sex sexual contact with relatives (usually cousins), strangers, friends,
or romantic partners. A chi-square test detected a significant association between
sex and distribution into these categories, 2 (3, N = 139) = 57.5, p < 0.00001.
Examination of standardized cell deviates revealed that males were significantly
more likely than females to have their first same-sex sexual contact with a stranger
and females were more likely than males to have their first same-sex sexual contact
with a romantic partner, consistent with Hypothesis 2.
To investigate associations between the age of first same-sex sexual contact
and the context in which that contact occurred, two Bonferroni-corrected t-tests
were performed. Females who experienced their first same-sex contact with someone other than a romantic partner experienced this contact at significantly earlier ages, (Mrelationship = 18.5, Mno relationship = 13.8) t = 3.5, p = 0.002. These
women were then compared to male youths who experienced their first same-sex
contact with someone other than a romantic partner (only four had their first samesex contact within a romantic relationship). There was no difference between
these groups in age of first same-sex sexual contact (Mfemale = 13.8, Mmale =
13.8), t = 0.5, p = ns, demonstrating that the context in which first same-sex
sexual contact occurs is significantly associated with its timing, independent
of gender.
Age and Time Spans of Sexual Identity Milestones
The mean ages at which respondents reported their first same-sex attractions,
first same-sex sexual contact, first self-labeling, and first disclosure of nonheterosexuality are presented in Table IV. A MANOVA with sex as the independent
Table IV. Gender and Trajectory Differences in Sexual Identity Milestonesa

First same-sex attractions


First same-sex sexual contact
First self-labeling
First disclosure
Gap (in years) between first attractions
and first sexual contact
Gap (in years) between first sexual
contact and self-labeling
Gap (in years) between self-labeling
and disclosure
Gap (in years) between first attractions
and self-labeling/disclosure
(whichever came last)
a Values are represented as
p < 0.10; p < 0.05.

Mean (SD).

Males

Females

Sex-first

Label-first

7.7 (3.0)
14.1 (4.7)
16.4 (2.9)
17.9 (2.4)
7.0 (5.2)

9.0 (4.1)
16.4 (4.4)
17.6 (2.1)
17.9 (1.9)
8.7 (4.7)

7.5 (2.8)
11.5 (4.0)
17.0 (2.8)
17.9 (2.6)
4.0 (4.3)

8.9 (4.0)
18.4 (2.0)
17.0 (2.5)
17.9 (1.9)
10.1 (4.4)

2.4 (4.7)

0.9 (3.9)

5.4 (4.1)

2.0 (1.9)

1.5 (2.4)

0.3 (1.1)

1.0 (2.0)

0.9 (1.9)

10.4 (3.7)

9.3 (4.3)

10.6 (3.6)

9.1 (4.3)

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

10:10

Gender Differences in Sexual Identity Development

Style file version Nov. 19th, 1999

619

variable and the aforementioned milestone ages as dependent variables detected a


significant effect of sex, Wilks lambda = 0.83 (4, 134), p = 0.0002. This analysis
excluded individuals who had not engaged in same-sex sexual contact. Bonferronicorrected univariate tests established that females experienced their first samesex contact at significantly later ages than did males, F(1, 137) = 17.1, MSE =
20.5, corrected p = 0.012, and self-labeled at significantly later ages, F(1, 162) =
10.2, MSE = 6.6, corrected p < 0.008. There was a trend-level gender difference in age of first same-sex attractions, F(1, 162) = 6.1, MSE = 12.5, corrected
p = 0.06. There was no gender difference in age of first disclosure, F(1, 162) =
0.01, MSE = 4.8, p = ns.
Table IV also presents means and standard deviations for time spans (in
years) between milestone events, stratified by gender. To test the significance of
gender differences regarding these variables, a MANOVA was carried out with
gender as the independent variable and the following dependent variables: Years
difference between first same-sex attractions and first same-sex sexual contact,
first same-sex sexual contact and self-labeling, self-labeling and disclosure, and
first same-sex attractions and labeling/disclosure (whichever was most recent).
For this analysis, current age was used as the substitute for age of first same-sex
contact to calculate spacing scores among individuals who had no same-sex contact. The overall analysis detected a significant effect of gender, Wilks lambda
= 0.82 (4, 158), p = 0.000002. Bonferroni-corrected univariate tests established
that females reported a smaller gap than males between labeling and disclosure,
F(1, 161) = 21.4, MSE = 3.3, corrected p < 0.00001. The sexes did not differ
in the gap between first same-sex attractions and first same-sex sexual contact,
F(1, 161) = 2.5, MSE = 27.6, uncorrected p = ns, between first same-sex contact and first self-labeling, F(1, 161) = 3.2, MSE = 22.1, uncorrected p = ns,
or between first same-sex attractions and eventual labeling/disclosure (whichever
came last), F(1, 161) = 3.2, MSE = 16.4, corrected p = ns.
SEXUAL IDENTITY TRAJECTORIES
Respondents were divided into two groups according to the relative sequencing of same-sex sexual contact and self-labeling in their developmental trajectory. The sex-first group (20% of females; 51% of males) engaged in same-sex
sexual contact prior to labeling their nonheterosexual identity or disclosing it to
others; label-first (80% of females; 49% of males), the reverse (this group included individuals who had not engaged in same-sex sexual contact). Females
were significantly more likely to follow the label-first than the sex-first trajectory,
2 (1, N = 164) = 21.1, p < 0.00001. There were no associations between sexual identity trajectory and any background characteristic presented in Table II; nor
were there any associations between sexual identity trajectory and current sexual
identity.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

620

10:10

Style file version Nov. 19th, 1999

Savin-Williams and Diamond

The next series of analyses addressed the relative importance of sex of respondent versus identity trajectory in explaining variation in the timing of identity milestones. First, a MANOVA was carried out with sex of respondent and identity trajectory as independent variables and age of first same-sex attractions, same-sex sexual
contact, self-labeling, and disclosure as dependent variables. This overall test did
not include individuals who had not engaged in same-sex sexual contact. There
were significant independent effects of both sex of respondent, Wilks lambda =
0.86 (4, 133), p = 0.0004, and trajectory, Wilks lambda = 0.39 (4, 133), p <
0.00001, but no interaction effect. Bonferroni-corrected univariate tests established
that the only significant sex difference after controlling for trajectory was the earlier age of first labeling among males, F(1, 161) = 11.3, MSE = 6.6, p < 0.004.
The aforementioned gender difference in age of first same-sex sexual contact was
not significant after controlling for trajectory. The only trajectory effect, independent of gender, was the earlier age of first same-sex sexual contact among those
in the sex-first group, F(1, 136) = 146, MSE = 10, corrected p < 0.00001.
Analyses next addressed the relative importance of sex of respondent versus
identity trajectory in explaining the spacing between identity milestones. First,
a MANOVA was carried out with sex of respondent and identity trajectory as
independent variables and the following dependent variables: Years difference between first same-sex attractions and first same-sex sexual contact, self-labeling and
disclosure, and first same-sex attractions and labeling/disclosure (whichever was
most recent). The gap between first sexual contact and first labeling was not tested
because the sex-first and label-first groups are defined on the basis of this gap.
Current age was used as the substitute for age of first same-sex contact to calculate spacing scores among individuals with no same-sex contact. This analysis
detected a significant effect of trajectory, controlling for sex of respondent, Wilks
lambda = 0.39 (3, 158), p < 0.00001, and a significant effect of sex of respondent,
controlling for trajectory, Wilks lambda = 0.85 (3, 158), p = 0.00002. There was
no interaction effect. Bonferroni-corrected univariate tests established that controlling for trajectory, the only significant gender difference was the shorter female gap
between first labeling and disclosure, F(1, 160) = 23.11, MSE = 3.3, corrected
p = 0.0001. Individuals in the sex-first trajectory had a shorter gap between first
same-sex attractions and first same-sex contact, F(1, 160) = 74, MSE = 18.9,
corrected p = 0.00001. These findings are presented in Table IV, and graphically
depicted in Fig. 1, which displays means and standard deviations for the ages at
which female and male respondents in each identity trajectory experienced first
same-sex attractions, same-sex sexual contact, self-labeling, and disclosure.
DISCUSSION
Results from this study demonstrate the value of assessing the context, timing,
spacing, and sequencing of sexual identity milestones when investigating gender

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

10:10

Style file version Nov. 19th, 1999

Gender Differences in Sexual Identity Development

621

Fig. 1. Means and standard deviations for milestones of sexual


identity development, stratified by gender and trajectory.

differences in sexual-minority development. When this full range of variables is


taken into account, a pattern emerges that is consistent with extant social psychological research on gender differences in sexual and social behavior. Specifically,
the most robust differences between sexual-minority males and females concern
the relative role of explicit sexual feelings and behaviors in the process of sexual
identity development. Thus, sexual identity models derived from the experiences
of sexual-minority males are likely to overemphasize the importance of explicitly
sexual feelings and same-sex sexual contact. By documenting variation both within
and across males and females regarding the relative sequencing of same-sex sexual
contact and self-labeling, the current research suggests important new directions
for research on pathways of sexual identity development.

Contexts of Sexual Identity Transitions


As predicted in Hypotheses 1 and 2, the context for first same-sex attractions,
first same-sex sexual contact, and self-labeling were more likely to be emotionally
or relationship oriented for young women and sexually oriented for young men.
When reflecting on their early memories of same-sex attractions, female youths
typically recalled crushes on friends; intense best friendships; and emotional infatuations with camp counselors, coaches, and teachers. Young men, to the contrary,
more frequently recalled explicitly sexual memoriesfeeling aroused by the sight

P1: FLF
Archives of Sexual Behavior [asb]

622

PL113-227177

September 16, 2000

10:10

Style file version Nov. 19th, 1999

Savin-Williams and Diamond

of another boy in the locker room or experiencing a furtive sexual encounter with
a male friend or cousin.
Just as female sexual-minority youths often experienced their first same-sex
attractions for close friends, they frequently had their first same-sex sexual contact
within full-fledged romantic relationships. One-fifth of male youths had their first
same-sex contact with total strangers, a scenario reported by none of the female
youths. This gender difference in context is directly related to the gender difference
in the timing of first same-sex contact. Females who did not have their first same-sex
sexual contact within a romantic relationship experienced this contact significantly
earlier than those who did, at ages comparable to male youths. This demonstrates
the importance of assessing the context of various sexual identity milestones before
interpreting gender differences in their timing.
Age and Time Spans of Sexual Identity Milestones
Youths in the current sample on average reached sexual identity milestones
12 years later than youths sampled in previous studies, except age of first samesex attractions and (males) first same-sex sexual contact. Age of first same-sex
attractions was 12 years earlier than data reported by youths in the four studies
reviewed earlier, ages which are consistent with McClintock and Herdts assertion
(McClintock and Herdt, 1996) that sexual attractions emerge during adrenarche
for both sexes around the age of 10 years. The earlier ages in the current sample
are within the age span of adrenarche; or the earlier ages can be attributed either
to the characteristics of the population of youths sampled or to the research protocol. College students may be more reflective or have better memories of early
experiences and feelings than do support group youths. Research protocol could
also be a factor. In the current study, youths were asked to describe a concrete
memory of their earliest same-sex attractions, with the caveat that they need not
have interpreted their feelings as sexual at that time. They were also encouraged
to think as far back in their memory as possible. This protocol differs markedly
from surveys that ask participants to state their age of first awareness.
In terms of first same-sex sexual contact among the current sample of sexualminority youths, it is notable that in all previous studies, gay and bisexual male
adolescents and adults report first sex at around the same time, during their thirteenth or fourteenth year (Savin-Williams, 1998). Apparently, regardless of historical context, research protocol, or population sampled, males with same-sex
attractions begin their sexual careers shortly after pubertal onset. This is consistent with longitudinal research on heterosexual adolescents demonstrating that
pubertal increases in androgen levels are strongly associated with increases in sexual motivation and with sexual debut among males (Udry et al., 1985). Udry
(1990) found that the ability of androgen levels to predict sexual debut among
white males was independent of potential social influences on sexual behavior,

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

Gender Differences in Sexual Identity Development

10:10

Style file version Nov. 19th, 1999

623

none of which were significant. Sexual debut among white females, however, was
entirely unrelated to androgen levels and was exclusively predicted by social factors (Udry et al., 1986; Udry and Billy, 1987), in spite of the fact that increased
androgens produced the same increases in sexual motivation among females as
among males.
Our research converges with these findings to suggest that the basic pathways leading from sexual desire to sexual activity are not equivalent among males
and females, independent of sexual orientation. For males, desire for a certain
sexual activity appears to be sufficient motivation to pursue this activity, whether
it is same-sex sexual contact (among sexual-minority youths) or other-sex sexual contact (among heterosexual youths). For females, social context is critical.
Few sexual-minority or heterosexual females pursue sexual contact on the basis of
sexual motivation alone. Overall, these results are consistent with theoretical and
empirical research cited earlier (Bailey et al., 1994; Peplau et al., 1998), which
suggests that gender is a more powerful predictor of sexual behavior than is sexual
orientation.
The later ages of reaching sexual identity milestones in our sample of youths
is likely the result of our recruitment of sexual-minority populations who differ in
a number of ways from youths who enter research projects through participation in
urban youth support groups. For example, many had not publicly adopted a sexualminority identity, had consciously adopted unconventional sexual identities, and
were heterogeneous regarding social class and community size. In contrast, samples in previous studies have typically contained more youths of lower socioeconomic status. Other differences, not generally assessed, could also account for
variability within and across samples in the timing of sexual identity milestones,
including personality characteristics such as narcissism, comfort level with sexuality, sexual experience, history of peer and family rejection, and gender atypicality. These possibilities should be pursued in future research on sexual-minority
development.
In terms of time spans, the gap from first same-sex attractions to first disclosure averaged around 10 years for both sexes. This is several years longer than
that reported in other studies of sexual-minority youths, primarily because our
youths had an earlier age of first awareness and a later age of disclosure. As noted
earlier, this discrepancy could be the result of sample characteristics. The delay
in disclosing sexual identity until the 18th birthday, 2 years later than youths recruited from urban support groups, could be due to the ease with which youths in
our sample could disguise their sexual identity from others; it is our impression
that youths who attend urban support groups are more likely to be gender atypical
and thus conceivably were pushed out of the closet by others. Alternatively, the
disclosure lag among the current sample of youths could simply be a consequence
of their later age of self-knowledge, higher social class, or greater fears that their
careers/educational status would be threatened if they came out.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

10:10

624

Style file version Nov. 19th, 1999

Savin-Williams and Diamond

Finally, analyses of gender differences in time spans between sexual identity


milestones revealed that the only significant difference concerned the shorter gap
among females between self-labeling and disclosure, consistent with previous research (DAugelli and Hershberger, 1993; DAugelli, 1998). That is, once young
sexual-minority women knew that they are not heterosexual they immediately disclosed this information to someone, usually a best female friend. Some young
women labeled themselves after or during the act of disclosurewhile talking to
friends about their attractions to women they concluded that these attractions had
meaning for their sexual identity. Again, this reflects the importance of interpersonal relationships in the understanding and acknowledgment of sexuality among
sexual-minority women.
Sexual Identity Trajectories
Consistent with Hypothesis 4, young men were disproportionately likely
to pursue same-sex sexual contact well in advance of labeling themselves nonheterosexual. Young women were disproportionately likely to label themselves
nonheterosexual before pursuing same-sex sexual contact. These results strongly
suggest that no singular sexual identity model is capable of representing the
diverse trajectories of male and female sexual identity development. They further imply that differences among youths cannot be explained by gender alone.
Whatever gender differences exist in sequencing of sexual identity milestones,
they are neither absolute nor interpretable without reference to complex modeling
of behavior. As Fig. 1 indicates, female youths in the sex-first trajectory look like
male youths with regard to timing and spacing between their sexual identity milestones, and male youths in the label-first trajectory look remarkably like female
youths. After controlling for trajectory, the only gender differences that remained
were located solely within the label-first group.
How might these trajectories be characterized and of what significance are
they? The distinguishing characteristic of the sex-first trajectory is early same-sex
sexual contact, with the resultant consequence of a significantly smaller gap between first same-sex attractions and first same-sex contact. The factors that might
motivate men and women in this group to pursue early same-sex sexual contact
are as yet unknown, but might include chance opportunities for same-sex sexual contact, personality characteristics related to novelty-seeking and inhibition,
and parental supervision. These same factors might influence heterosexual sexual activity as well, among both heterosexual and sexual-minority youths. These
are important areas for future research, particularly for discerning variables that
mediate gender differences in early sexual behavior. For example, what are the
distinguishing characteristics of females in the sex-first trajectory? What factors
launched them onto a developmental trajectory so much more common among
males?

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

Gender Differences in Sexual Identity Development

10:10

Style file version Nov. 19th, 1999

625

The same might be asked of males in the label-first trajectory. The picture
presented by the predominantly female label-first group is one of consolidating
milestones during a relatively brief 2-year period beginning just prior to their
17th birthday. Individuals in the predominantly male sex-first group on the other
hand, tended to space these events over a longer time period. In fact, it was our
impression that many of the young women in the label-first group experienced
these transitions as if they were not three separate events but one interconnected
transformation, sometimes even beginning and ending within a single 24-h period.
Why might this occur? Again, little is known about factors that account for
such a pathway, and the ways in which these factors interact with gender to propel
more women than men into the label-first trajectory. Perhaps the most fruitful area
for future research concerns identification of individual differences that influence
youths placement into one of these trajectories. All too often, research on sexual
orientation has overemphasized differences between sexual minorities and heterosexuals and underemphasized differences within sexual-minority populations. So
far, gender and ethnicity have been the only within-group differences to receive
systematic study. The current research indicates that such efforts should be expanded if robust models of sexual identity development capable of describing and
explaining diversity within and across males and females are to be built.

Limitation of Findings
The potential significance of the findings is limited by several characteristics
of the methods used. First, the youth participants cannot be assumed to be representative of the general population of sexual-minority adolescents and young
adults. They were recruited primarily from university settings and were disproportionately white and highly educated. Previous research has shown that college
students differ from youths enlisted from urban and community support groups in
age at which they reached a number of developmental milestones (Savin-Williams,
1998). It should be noted, however, that most North American youths are college
students and thus the studys participants might be more representative of sexualminority youths than the noncollege youths who attend community support groups.
Second, only one methodology was usedinterviewsand the findings must be
considered in light of limitations inherent in this approach (e.g., interviewer bias,
reduction of anonymity, inhibition of shy participants, gender bias in autobiographical memory recall). Third, even though most data on developmental milestones
were collected within a few years of their occurrence, the possibility of retrospective memory bias on recalled developmental milestones cannot be ruled out.
Furthermore, youths memories might be affected by the highly emotional nature
of the issues being assessed. Every attempt was made to counter these biases by
rooting all memories in concrete events.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

626

10:10

Style file version Nov. 19th, 1999

Savin-Williams and Diamond

Conclusion
Prior studies of sexual identity development have been essential in furthering an understanding of the processes by which sexual-minority children and
adolescents come to an understanding of their sexuality. Yet it is now glaringly
apparent that research focusing on group averages of developmental milestones is
limited. Such research conveys little about the inherent diversity within populations of sexual-minority youths or how female and male adolescents experience
their sexuality in similar and dissimilar ways. The current study represents a first
step toward differentiating patterns in the timing, spacing, and sequencing of sexual identity milestones that might reveal critical factors shaping female and male
sexual identity development.
Results from the present study suggest two general conclusions. First, gender
matters. The development of young sexual-minority women cannot be extrapolated from findings of gay male youths. To make sense of these differences,
researchers should reference well-documented gender distinctions among heterosexuals. Sexual-minority youths are more similar to heterosexual peers of the same
sex than to sexual-minority peers of the other sex. Second, gender is not everything.
Just as some male youths follow female-typical trajectories (self-labeling far in
advance of same-sex sexual contact), some female youths follow male-typical
trajectories (the reverse pattern). The most productive way to proceed is to systematically explore factors influencing the onset, context, spacing, and sequencing
of developmental milestones within and across males and females. In addition to
ethnicity, class, and cohort, research should consider individual differences at the
level of personality, early experience, and family relations. To explore these possibilities, researchers must pursue recruitment strategies that include as diverse a
range of sexual-minority youths as possible.
REFERENCES
Bailey, J. M., Gaulin, S., Agyei, Y., and Gladue, B. A. (1994). Effects of gender and sexual orientation on evolutionarily relevant aspects of human mating psychology. J. Pers. Soc. Psych. 66:
10811093.
Cohen, K. M., and Savin-Williams, R. C. (1996). Developmental perspectives on coming out to self
and others. In Savin-Williams, R. C., and Cohen, K. M. (Eds.), The Lives of Lesbians, Gays, and
Bisexuals: Children to Adults, Harcourt Brace College Publishing, Fort Worth, TX, pp. 113151.
DAugelli, A. R. (1998). Victimization history and mental health among lesbian, gay, and bisexual
youths. Paper presented at the meetings of the Society for Research on Adolescence, San Diego,
CA.
DAugelli, A. R., and Hershberger, S. L. (1993). Lesbian, gay, and bisexual youth in community
settings: Personal challenges and mental health problems. Am. J. Comm. Psych. 21: 421448.
Diamond, L. M. (1998). Development of sexual orientation among adolescent and young adult women.
Dev. Psych. 34: 10851095.
Diamond, L. M. (2000). Sexual identity, attractions, and behavior among young sexual-minority women
over a two-year period. Dev. Psych. 36: 241250.

P1: FLF
Archives of Sexual Behavior [asb]

PL113-227177

September 16, 2000

Gender Differences in Sexual Identity Development

10:10

Style file version Nov. 19th, 1999

627

Dube, E. M. (2000). Sexual identity and intimacy development among two cohorts of sexual minority
men. J. Sex Res.
Dube, E. M., and Savin-Williams, R. C. (1999). Sexual identity development among ethnic sexualminority male youths. Dev. Psych. 35: 13891399.
Eliason, M. J. (1996a). An inclusive model of lesbian identity assumption. J. Gay, Lesbian, and Bisex.
Identity 1: 319.
Eliason, M. J. (1996b). Identity formation for lesbian, bisexual, and gay persons: Beyond a minoritizing view. J. Homosexual. 30: 3158.
Golden, C. (1996). Whats in a name? Sexual self-identification among women. In Savin-Williams,
R. C., and Cohen, K. M. (Eds.), The Lives of Lesbians, Gays, and Bisexuals: Children to Adults,
Harcourt Brace College Publishing, Fort Worth, TX, pp. 229249.
Herdt, G., and Boxer, A. M. (1993). Children of Horizons: How Gay and Lesbian Teens are Leading
a New Way Out of the Closet, Beacon Press, Boston.
Kirkpatrick, M., and Morgan, C. (1980). Psychodynamic psychotherapy of female homosexuality.
In Marmor, J. (Ed.), Homosexual Behavior: A Modern Reappraisal, Basic Books, New York,
pp. 357375.
McClintock, M. K., and Herdt, G. (1996). Rethinking puberty: The development of sexual attraction.
Curr. Direc. Psychol. Sci. 5: 178183.
McConnell, J. H. (1994). Lesbian and gay male identities as paradigms. In Archer, S. L. (Ed.), Interventions for Adolescent Identity Development, Sage, Thousand Oaks, pp. 103118.
Morris, J. F. (1997). Lesbian coming out as a multidimensional process. J. Homosexual. 33: 122.
Peplau, L. A., Garnets, L. D., Spalding, L. R., Conley, T. D., and Veniegas, R. C. (1998). A critique of
Bems Exotic Becomes Erotic theory of sexual orientation. Psychol. Rev., 105: 387394.
Rosario, M., Rotheram-Borus, M. J., and Reid, H. (1996). Gay-related stress and its correlates among
gay and bisexual male adolescents of predominantly Black and Hispanic background. J. Community Psych. 24: 136159.
Rust, P. C. (1995). Bisexuality and the Challenge to Lesbian Politics: Sex, Loyalty, and Revolution,
New York University Press, New York.
Saghir, M. T., and Robins, E. (1980). Clinical aspects of female homosexuality. In Marmor, J. (Ed.),
Homosexual Behavior: A Modern Reappraisal, Basic Books, New York, pp. 280295.
Savin-Williams, R. C. (1998). . . . And Then I Became Gay: Young Mens Stories, Routledge, New
York.
Udry, J. R. (1990). Hormonal and social determinants of adolescent sexual initiation. In Bancroft, J.,
and Reinisch, J. M. (Eds.), Adolescence and Puberty, Oxford University Press, New York,
pp. 7087.
Udry, J. R., and Billy, J. O. G. (1987). Initiation of coitus in early adolescence. Am. Sociol. Rev. 52:
841855.
Udry, J. R., Billy, J. O. G., Morris, N. M., Groff, T. R., and Raj, M. H. (1985). Serum androgenic
hormones motivate sexual behavior in adolescent boys. Fertility and Sterility, 43: 9094.
Udry, J. R., Talbert, L. M., and Morris, N. M. (1986). Biosocial foundations for adolescent female
sexuality. Demography, 23: 217230.