simplebooklet thumbnail

of 0
~ Special Report ~
Sexuality and Gender
Findings from the Biological,
Psychological, and Social Sciences
Lawrence S. Mayer, M.B., M.S., Ph.D.
Paul R. McHugh, M.D.
Number 50 ~ Fall 2016 ~ $7.00.......
Editor’s Note: Questions related to sexuality and gender bear on
some of the most intimate and personal aspects of human life. In
recent years they have also vexed American politics. We offer this
report written by Dr. Lawrence S. Mayer, an epidemiologist
trained in psychiatry, and Dr. Paul R. McHugh, arguably the most
important American psychiatrist of the last half-century in
the hope of improving public understanding of these questions.
Examining research from the biological, psychological, and social
sciences, this report shows that some of the most frequently
heard claims about sexuality and gender are not supported by
scientific evidence. The report has a special focus on the higher
rates of mental health problems among LGBT populations, and
it questions the scientific basis of trends in the treatment of chil-
dren who do not identify with their biological sex. More effort is
called for to provide these people with the understanding, care,
and support they need to lead healthy, flourishing lives.
Preface 4
Lawrence S. Mayer
Executive Summary 7
Sexuality and Gender
Findings from the Biological, Psychological, and Social Sciences
Lawrence S. Mayer, M.B., M.S., Ph.D. and Paul R. McHugh, M.D.
Introduction 10
Part 1: Sexual Orientation 13
Abstract 13
Problems with Defining Key Concepts 15
The Context of Sexual Desire 19
Sexual Orientation 21
Challenging the “Born that Way” Hypothesis 25
Studies of Twins 26
Number 50 ~ Fall 2016
Molecular Genetics 32
The Limited Role of Genetics 33
The Influence of Hormones 34
Sexual Orientation and the Brain 39
Misreading the Research 41
Sexual Abuse Victimization 42
Distribution of Sexual Desires and Changes Over Time 50
Conclusion 57
Part 2: Sexuality, Mental Health Outcomes, and Social Stress 59
Abstract 59
Some Preliminaries 60
Sexuality and Mental Health 60
Sexuality and Suicide 66
Sexuality and Intimate Partner Violence 70
Transgender Health Outcomes 73
Explanations for the Poor Health Outcomes: The Social Stress Model 75
Discrimination and prejudice events 77
Stigma 79
Concealment 81
Testing the model 82
Conclusion 85
Part 3: Gender Identity 86
Abstract 86
Key Concepts and Their Origins 87
Gender Dysphoria 93
Gender and Physiology 98
Transgender Identity in Children 105
Therapeutic Interventions in Children 106
Therapeutic Interventions in Adults 108
Conclusion 114
Notes 117
Lawrence S. Mayer, M.B., M.S., Ph.D. is a scholar in residence in the Department of
Psychiatry at the Johns Hopkins University School of Medicine and a professor of statis-
tics and biostatistics at Arizona State University. Paul R. McHugh, M.D. is a professor
of psychiatry and behavioral sciences at the Johns Hopkins University School of Medicine
and was for twenty-five years the psychiatrist-in-chief at the Johns Hopkins Hospital. He
is the author or coauthor of several books, including, most recently, Try to Remember:
Psychiatry’s Clash over Meaning, Memory, and Mind (Dana Press, 2008).
The New Atlantis (1627) was the title Francis Bacon selected for his fable of a society living
with the benefits and challenges of advanced science and technology. Bacon, a founder and cham-
pion of modern science, sought not only to highlight the potential of technology to improve human
life, but also to foresee some of the social, moral, and political difficulties that confront a society
shaped by the great scientific enterprise. His book offers no obvious answers; perhaps it seduces
more than it warns. But the tale also hints at some of the dilemmas that arise with the ability to
remake and reconfigure the natural world: governing science, so that it might flourish freely with-
out destroying or dehumanizing us, and understanding the effect of technology on human life,
human aspiration, and the human good. To a great extent, we live in the world Bacon imagined,
and now we must find a way to live well with both its burdens and its blessings. This very chal-
lenge, which now confronts our own society most forcefully, is the focus of this journal.
Editorial Office:
The New Atlantis
1730 M Street N.W., Suite 910
Washington, D.C. 20036
Telephone: (202) 682-1200
Fax: (202) 408-0632
Subscription Office:
Postmaster and subscribers, please send
subscription orders and address changes to:
The New Atlantis Subscription Services,
P.O. Box 3000, Denville, N.J. 07834-3000,
or call toll-free at (866) 440-6916.
Rate: $24/year (4 Issues). Please add $10 for
delivery outside the United States.
Advertising Information:
Those interested in placing advertisements
should contact Samuel Matlack, Managing
Manuscripts and proposals should be directed
to Samuel Matlack by e-mail
( or
by post to our editorial office.
The New Atlantis (
ISSN 1543-1215) is
published quarterly in the Spring,
Summer, Fall, and Winter by the Center
for the Study of Technology and Society in
partnership with the Ethics and Public Policy
Center in Washington, D.C. It is printed by
Global Printing and distributed by Ingram
Periodicals, Inc.
Adam Keiper
anaging Editor
Samuel Matlack
associatE Editors
Brendan P. Foht
M. Anthony Mills
assistant Editor
Michael W. Begun
sEnior Editors
Caitrin Nicol Keiper
Yuval Levin
Christine Rosen
Ari N. Schulman
Eric Cohen
contributing Editors
James C. Capretta
Matthew B. Crawford
Alan Jacobs
Peter Augustine Lawler
Wilfred M. McClay
Gilbert Meilaender
Charles T. Rubin
Diana Schaub
Roger Scruton
Stephen L. Talbott
Raymond Tallis
Algis Valiunas
Adam J. White
Robert Zubrin
4 ~ The New Atlantis
Copyright 2016. All rights reserved. See for more information.
his report was written for the general public and for mental
health professionals in order to draw attention to and offer
some scientific insight about the mental health issues faced by
LGBT populations.
It arose from a request from Paul R. McHugh, M.D., the former chief of
psychiatry at Johns Hopkins Hospital and one of the leading psychiatrists
in the world. Dr. McHugh requested that I review a monograph he and
colleagues had drafted on subjects related to sexual orientation and iden-
tity; my original assignment was to guarantee the accuracy of statistical
inferences and to review additional sources. In the months that followed, I
closely read over five hundred scientific articles on these topics and perused
hundreds more. I was alarmed to learn that the LGBT community bears a
disproportionate rate of mental health problems compared to the popula-
tion as a whole.
As my interest grew, I explored research across a variety of scientific
fields, including epidemiology, genetics, endocrinology, psychiatry, neuro-
science, embryology, and pediatrics. I also reviewed many of the academic
empirical studies done in the social sciences including psychology, sociol-
ogy, political science, economics, and gender studies.
I agreed to take over as lead author, rewriting, reorganizing, and
expanding the text. I support every sentence in this report, without res-
ervation and without prejudice regarding any political or philosophical
debates. This report is about science and medicine, nothing more and
nothing less.
Readers wondering about this report’s synthesis of research from so
many different fields may wish to know a little about its lead author. I am
a full-time academic involved in all aspects of teaching, research, and pro-
fessional service. I am a biostatistician and epidemiologist who focuses on
the design, analysis, and interpretation of experimental and observational
data in public health and medicine, particularly when the data are complex
in terms of underlying scientific issues. I am a research physician, having
trained in medicine and psychiatry in the U.K. and received the British
equivalent (M.B.) to the American M.D. I have never practiced medicine
(including psychiatry) in the United States or abroad. I have testified in
dozens of federal and state legal proceedings and regulatory hearings, in
Fall 2016 ~ 5
Copyright 2016. All rights reserved. See for more information.
most cases reviewing scientific literature to clarify the issues under exami-
nation. I strongly support equality and oppose discrimination for the LGBT
community, and I have testified on their behalf as a statistical expert.
I have been a full-time tenured professor for over four decades. I have
held professorial appointments at eight universities, including Princeton,
the University of Pennsylvania, Stanford, Arizona State University, Johns
Hopkins University Bloomberg School of Public Health and School of
Medicine, Ohio State, Virginia Tech, and the University of Michigan.
I have also held research faculty appointments at several other institu-
tions, including the Mayo Clinic.
My full-time and part-time appointments have been in twenty-three
disciplines, including statistics, biostatistics, epidemiology, public health,
social methodology, psychiatry, mathematics, sociology, political science,
economics, and biomedical informatics. But my research interests have
varied far less than my academic appointments: the focus of my career has
been to learn how statistics and models are employed across disciplines,
with the goal of improving the use of models and data analytics in assess-
ing issues of interest in the policy, regulatory, or legal realms.
I have been published in many top-tier peer-reviewed journals (includ-
ing The Annals of Statistics, Biometrics, and American Journal of Political
Science) and have reviewed hundreds of manuscripts submitted for publica-
tion to many of the major medical, statistical, and epidemiological journals
(including The New England Journal of Medicine, Journal of the American
Statistical Association, and American Journal of Public Health).
I am currently a scholar in residence in the Department of Psychiatry
at Johns Hopkins School of Medicine and a professor of statistics and bio-
statistics at Arizona State University. Up until July 1, 2016, I also held
part-time faculty appointments at the Johns Hopkins Bloomberg School
of Public Health and School of Medicine, and at the Mayo Clinic.
n undertaking as ambitious as this report would not be possible
without the counsel and advice of many gifted scholars and editors.
I am grateful for the generous help of Laura E. Harrington, M.D., M.S.,
a psychiatrist with extensive training in internal medicine and neuroim-
munology, whose clinical practice focuses on women in life transition,
including affirmative treatment and therapy for the LGBT community.
She contributed to the entire report, particularly lending her expertise
to the sections on endocrinology and brain research. I am indebted also
to Bentley J. Hanish, B.S., a young geneticist who expects to graduate
medical school in 2021 with an M.D./Ph.D. in psychiatric epidemiology.
6 ~ The New Atlantis
Lawrence S. Mayer
He contributed to the entire report, particularly to those sections that
concern genetics.
I gratefully acknowledge the support of Johns Hopkins University
Bloomberg School of Public Health and School of Medicine, Arizona State
University, and the Mayo Clinic.
In the course of writing this report, I consulted a number of indi-
viduals who asked that I not thank them by name. Some feared an angry
response from the more militant elements of the LGBT community;
others feared an angry response from the more strident elements of
religiously conservative communities. Most bothersome, however, is
that some feared reprisals from their own universities for engaging such
controversial topics, regardless of the report’s content a sad statement
about academic freedom.
dedicate my work on this report, first, to the LGBT community, which
bears a disproportionate rate of mental health problems compared to
the population as a whole. We must find ways to relieve their suffering.
I dedicate it also to scholars doing impartial research on topics of pub-
lic controversy. May they never lose their way in political hurricanes.
And above all, I dedicate it to children struggling with their sexu-
ality and gender. Children are a special case when addressing gender
issues. In the course of their development, many children explore the
idea of being of the opposite sex. Some children may have improved
psychological well-being if they are encouraged and supported in their
cross-gender identification, particularly if the identification is strong
and persistent over time. But nearly all children ultimately identify with
their biological sex. The notion that a two-year-old, having expressed
thoughts or behaviors identified with the opposite sex, can be labeled
for life as transgender has absolutely no support in science. Indeed, it is
iniquitous to believe that all children who have gender-atypical thoughts
or behavior at some point in their development, particularly before
puberty, should be encouraged to become transgender.
As citizens, scholars, and clinicians concerned with the problems fac-
ing LGBT people, we should not be dogmatically committed to any par-
ticular views about the nature of sexuality or gender identity; rather, we
should be guided first and foremost by the needs of struggling patients,
and we should seek with open minds for ways to help them lead mean-
ingful, dignified lives.
awrence S. Mayer, M.B., M.S., Ph.D.
Copyright 2016. All rights reserved. See for more information.
Fall 2016 ~ 7
Copyright 2016. All rights reserved. See for more information.
This report presents a careful summary and an up-to-date explanation of
research from the biological, psychological, and social sciences related
to sexual orientation and gender identity. It is offered in the hope that
such an exposition can contribute to our capacity as physicians, scientists,
and citizens to address health issues faced by LGBT populations within
our society.
Some key findings:
Part One: Sexual Orientation
The understanding of sexual orientation as an innate, biologi-
cally fixed property of human beings the idea that people are
“born that way” is not supported by scientific evidence.
While there is evidence that biological factors such as genes
and hormones are associated with sexual behaviors and attrac-
tions, there are no compelling causal biological explanations
for human sexual orientation. While minor differences in the
brain structures and brain activity between homosexual and
heterosexual individuals have been identified by researchers,
such neurobiological findings do not demonstrate whether these
differences are innate or are the result of environmental and
psychological factors.
Longitudinal studies of adolescents suggest that sexual ori-
entation may be quite fluid over the life course for some people,
with one study estimating that as many as 80% of male adoles-
cents who report same-sex attractions no longer do so as adults
(although the extent to which this figure reflects actual changes
in same-sex attractions and not just artifacts of the survey pro-
cess has been contested by some researchers).
Compared to heterosexuals, non-heterosexuals are about two
to three times as likely to have experienced childhood sexual
Executive Summary
8 ~ The New Atlantis
Executive Summary
Copyright 2016. All rights reserved. See for more information.
Part Two: Sexuality, Mental Health Outcomes, and Social Stress
Compared to the general population, non-heterosexual sub-
populations are at an elevated risk for a variety of adverse health
and mental health outcomes.
Members of the non-heterosexual population are estimated
to have about 1.5 times higher risk of experiencing anxiety dis-
orders than members of the heterosexual population, as well as
roughly double the risk of depression, 1.5 times the risk of sub-
stance abuse, and nearly 2.5 times the risk of suicide.
Members of the transgender population are also at higher risk
of a variety of mental health problems compared to members of
the non-transgender population. Especially alarmingly, the rate
of lifetime suicide attempts across all ages of transgender indi-
viduals is estimated at 41%, compared to under 5% in the overall
U.S. population.
There is evidence, albeit limited, that social stressors such as
discrimination and stigma contribute to the elevated risk of poor
mental health outcomes for non-heterosexual and transgender
populations. More high-quality longitudinal studies are neces-
sary for the “social stress model” to be a useful tool for under-
standing public health concerns.
Part Three: Gender Identity
The hypothesis that gender identity is an innate, fixed prop-
erty of human beings that is independent of biological sex that
a person might be “a man trapped in a woman’s body” or “a
woman trapped in a man’s body” is not supported by scientific
According to a recent estimate, about 0.6% of U.S. adults iden-
tify as a gender that does not correspond to their biological sex.
Studies comparing the brain structures of transgender and
non-transgender individuals have demonstrated weak correla-
tions between brain structure and cross-gender identification.
These correlations do not provide any evidence for a neurobio-
logical basis for cross-gender identification.
Fall 2016 ~ 9
Executive Summary
Copyright 2016. All rights reserved. See for more information.
Compared to the general population, adults who have under-
gone sex-reassignment surgery continue to have a higher risk
of experiencing poor mental health outcomes. One study found
that, compared to controls, sex-reassigned individuals were
about 5 times more likely to attempt suicide and about 19 times
more likely to die by suicide.
Children are a special case when addressing transgender issues.
Only a minority of children who experience cross-gender identi-
fication will continue to do so into adolescence or adulthood.
There is little scientific evidence for the therapeutic value of
interventions that delay puberty or modify the secondary sex
characteristics of adolescents, although some children may have
improved psychological well-being if they are encouraged and
supported in their cross-gender identification. There is no evi-
dence that all children who express gender-atypical thoughts or
behavior should be encouraged to become transgender.
10 ~ The New Atlantis
Copyright 2016. All rights reserved. See for more information.
Few topics are as complex and controversial as human sexual orienta-
tion and gender identity. These matters touch upon our most intimate
thoughts and feelings, and help to define us as both individuals and social
beings. Discussions of the ethical questions raised by sexual orientation
and gender identity can become heated and personal, and the associated
policy issues sometimes provoke intense controversies. The disputants,
journalists, and lawmakers in these debates often invoke the authority of
science, and in our news and social media and our broader popular culture
we hear claims about what “science says” on these matters.
This report offers a careful summary and an up-to-date explana-
tion of many of the most rigorous findings produced by the biologi-
cal, psychological, and social sciences related to sexual orientation
and gender identity. We examine a vast body of scientific literature from
several disciplines. We try to acknowledge the limitations of the research
and to avoid premature conclusions that would result in over-interpreta-
tion of scientific findings. Since the relevant literature is rife with incon-
sistent and ambiguous definitions, we not only examine the empirical
evidence but also delve into underlying conceptual problems. This report
does not, however, discuss matters of morality or policy; our focus is on the
scientific evidence what it shows and what it does not show.
We begin in Part One by critically examining whether concepts such
as heterosexuality, homosexuality, and bisexuality represent distinct,
fixed, and biologically determined properties of human beings. As part of
this discussion, we look at the popular “born that way” hypothesis, which
Sexuality and Gender
Findings from the Biological,
Psychological, and Social Sciences
Lawrence S. Mayer, M.B., M.S., Ph.D. and Paul R. McHugh, M.D.
Fall 2016 ~ 11
Copyright 2016. All rights reserved. See for more information.
posits that human sexual orientation is biologically innate; we examine
the evidence for this claim across several subspecialties of the biologi-
cal sciences. We explore the developmental origins of sexual attractions,
the degree to which such attractions may change over time, and the
complexities inherent in the incorporation of these attractions into one’s
sexual identity. Drawing on evidence from twin studies and other types
of research, we explore genetic, environmental, and hormonal factors.
We also explore some of the scientific evidence relating brain science to
sexual orientation.
In Part Two we examine research on health outcomes as they relate
to sexual orientation and gender identity. There is a consistently observed
higher risk of poor physical and mental health outcomes for lesbian, gay,
bisexual, and transgender subpopulations compared to the general popu-
lation. These outcomes include depression, anxiety, substance abuse, and
most alarmingly, suicide. For example, among the transgender subpopula-
tion in the United States, the rate of attempted suicide is estimated to be
as high as 41%, ten times higher than in the general population. As phy-
sicians, academics, and scientists, we believe all of the subsequent discus-
sions in this report must be cast in the light of this public health issue.
We also examine some ideas proposed to explain these differential
health outcomes, including the “social stress model.” This hypothesis
which holds that stressors like stigma and prejudice account for much of
the additional suffering observed in these subpopulations does not seem
to offer a complete explanation for the disparities in the outcomes.
Much as Part One investigates the conjecture that sexual orientation
is fixed with a causal biological basis, a portion of Part Three examines
similar issues with respect to gender identity. Biological sex (the binary
categories of male and female) is a fixed aspect of human nature, even
though some individuals affected by disorders of sex development may
exhibit ambiguous sex characteristics. By contrast, gender identity is a
social and psychological concept that is not well defined, and there is little
scientific evidence that it is an innate, fixed biological property.
Part Three also examines sex-reassignment procedures and the evi-
dence for their effectiveness at alleviating the poor mental health outcomes
experienced by many people who identify as transgender. Compared to
the general population, postoperative transgender individuals continue to
be at high risk of poor mental health outcomes.
An area of particular concern involves medical interventions for
gender-nonconforming youth. They are increasingly receiving therapies
that affirm their felt genders, and even hormone treatments or surgical
12 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
modifications at young ages. But the majority of children who identify as
a gender that does not conform to their biological sex will no longer do
so by the time they reach adulthood. We are disturbed and alarmed by the
severity and irreversibility of some interventions being publicly discussed
and employed for children.
Sexual orientation and gender identity resist explanation by simple
theories. There is a large gap between the certainty with which beliefs
are held about these matters and what a sober assessment of the science
reveals. In the face of this complexity and uncertainty, we need to be hum-
ble about what we know and do not know. We readily acknowledge that
this report is neither an exhaustive analysis of the subjects it addresses
nor the last word on them. Science is by no means the only avenue for
understanding these astoundingly complex, multifaceted topics; there are
other sources of wisdom and knowledge including art, religion, philoso-
phy, and lived human experience. And much of our scientific knowledge
in this area remains unsettled. However, we offer this overview of the
scientific literature in the hope that it can provide a shared framework for
intelligent, enlightened discourse in political, professional, and scientific
exchanges and may add to our capacity as concerned citizens to alleviate
suffering and promote human health and flourishing.
Fall 2016 ~ 13
Copyright 2016. All rights reserved. See for more information.
While some people are under the impression that sexual orientation is an innate,
fixed, and biological trait of human beings
—that, whether heterosexual, homosexual,
or bisexual, we are “born that way”
there is insufficient scientific evidence
to support that claim. In fact, the concept of sexual orientation itself is highly
ambiguous; it can refer to a set of behaviors, to feelings of attraction, or to a sense of
identity. Epidemiological studies show a rather modest association between genetic
factors and sexual attractions or behaviors, but do not provide significant evidence
pointing to particular genes. There is also evidence for other hypothesized biologi-
cal causes of homosexual behaviors, attractions, or identity
such as the influence
of hormones on prenatal development but that evidence, too, is limited. Studies
of the brains of homosexuals and heterosexuals have found some differences, but
have not demonstrated that these differences are inborn rather than the result of
environmental factors that influenced both psychological and neurobiological traits.
One environmental factor that appears to be correlated with non-heterosexuality is
childhood sexual abuse victimization, which may also contribute to the higher rates
of poor mental health outcomes among non-heterosexual subpopulations, compared
to the general population. Overall, the evidence suggests some measure of fluidity
in patterns of sexual attraction and behavior
contrary to the “born that way”
notion that oversimplifies the vast complexity of human sexuality.
The popular discussion of sexual orientation is characterized by two
conflicting ideas about why some individuals are lesbian, gay, or bisexual.
While some claim that sexual orientation is a choice, others say that sexu-
al orientation is a fixed feature of one’s nature, that one is “born that way.”
We hope to show here that, though sexual orientation is not a choice,
neither is there scientific evidence for the view that sexual orientation is
a fixed and innate biological property.
A prominent recent example of a person describing sexual orientation
as a choice is Cynthia Nixon, a star of the popular television series Sex and
the City, who in a January 2012 New York Times interview explained, “For
me it’s a choice, and you don’t get to define my gayness for me,” and com-
mented that she was “very annoyed” about the issue of whether or not gay
people are born that way. “Why can’t it be a choice? Why is that any less
Similarly, Brandon Ambrosino wrote in The New Republic in
Sexual Orientation
Part One
14 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
2014 that “It’s time for the LGBT community to stop fearing the word
‘choice,’ and to reclaim the dignity of sexual autonomy.”
By contrast, proponents of the “born that way” hypothesis expressed
for instance in Lady Gaga’s 2011 song “Born This Way posit that there
is a causal biological basis for sexual orientation and often try to bolster
their claims with scientific findings. Citing three scientific studies
an article from Science magazine,
Mark Joseph Stern, writing for Slate in
2014, claims that “homosexuality, at least in men, is clearly, undoubtedly,
inarguably an inborn trait.”
However, as neuroscientist Simon LeVay,
whose work in 1991 showed brain differences in homosexual men com-
pared to heterosexual men, explained some years after his study, “It’s
important to stress what I didn’t find. I did not prove that homosexuality
is genetic, or find a genetic cause for being gay. I didn’t show that gay men
are ‘born that way,’ the most common mistake people make in interpreting
my work. Nor did I locate a gay center in the brain.”
Many recent books contain popular treatments of science that make
claims about the innateness of sexual orientation. These books often
exaggerate or at least oversimplify complex scientific findings. For
example, in a 2005 book, psychologist and science writer Leonard Sax
responds to a worried mother’s question as to whether her teenage son will
outgrow his homosexual attractions: “Biologically, the difference between
a gay man and a straight man is something like the difference between a
left-handed person and a right-handed person. Being left-handed isn’t just
a phase. A left-handed person won’t someday magically turn into a right-
handed person.. .. Some children are destined at birth to be left-handed,
and some boys are destined at birth to grow up to be gay.”
As we argue in this part of the report, however, there is little scientific
evidence to support the claim that sexual attraction is simply fixed by
innate and deterministic factors such as genes. Popular understandings
of scientific findings often presume deterministic causality when the find-
ings do not warrant that presumption.
Another important limitation for research and for interpretation of
scientific studies on this topic is that some central concepts —including
“sexual orientation itself are often ambiguous, making reliable mea-
surements difficult both within individual studies and when comparing
results across studies. So before turning to the scientific evidence concern-
ing the development of sexual orientation and sexual desire, we will exam-
ine at some length several of the most troublesome conceptual ambiguities
in the study of human sexuality in order to arrive at a fuller picture of the
relevant concepts.
Fall 2016 ~ 15
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
Problems with Defining Key Concepts
A 2014 New York Times Magazine piece titled “The Scientific Quest to
Prove Bisexuality Exists”
provides an illustration of the themes explored
in this Part sexual desire, attraction, orientation, and identity and of
the difficulties with defining and studying these concepts. Specifically, the
article shows how a scientific approach to studying human sexuality can
conflict with culturally prevalent views of sexual orientation, or with the
self-understanding that many people have of their own sexual desires and
identities. Such conflicts raise important questions about whether sexual
orientation and related concepts are as coherent and well-defined as is
often assumed by researchers and the public alike.
The author of the article, Benoit Denizet-Lewis, an openly gay
man, describes the work of scientists and others trying to demonstrate
the existence of a stable bisexual orientation. He visited researchers
at Cornell University and participated in tests used to measure sexual
arousal, tests that include observing the way pupils dilate in response to
sexually explicit imagery. To his surprise, he found that, according to this
scientific measure, he was aroused when watching pornographic films of
women masturbating:
Might I actually be bisexual? Have I been so wedded to my gay
identity one I adopted in college and announced with great fanfare to
family and friends that I haven’t allowed myself to experience another
part of myself ? In some ways, even asking those questions is anathema
to many gays and lesbians. That kind of publicly shared uncertainty is
catnip to the Christian Right and to the scientifically dubious, psycho-
logically damaging ex-gay movement it helped spawn. As out gay men
and lesbians, after all, we’re supposed to be sure we’re supposed to
be “born this way.”
Despite the apparently scientific (though admittedly limited) evidence
of his bisexual-typical patterns of arousal, Denizet-Lewis rejected the
idea that he was actually bisexual, because “It doesn’t feel true as a sexual
orientation, nor does it feel right as my identity.”
Denizet-Lewis’s concerns here illustrate a number of the quandaries
raised by the scientific study of human sexuality. The objective measures
the researchers used seemed to be at odds with the more intuitive, subjec-
tive understanding of what it is to be sexually aroused; our own under-
standing of what we are sexually aroused by is tied up with the entirety of
our lived experience of sexuality. Furthermore, Denizet-Lewis’s insistence
16 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
that he is gay, not bisexual, and his concern that uncertainty about his
identity could have social and political implications, points to the fact that
sexual orientation and identity are understood not only in scientific and
personal terms, but in social, moral, and political terms as well.
But how do categories of sexual orientation with labels such as
“bisexual” or “gay” or “straight” help scientists study the complex phe-
nomenon of human sexuality? When we examine the concept of sexual
orientation, it becomes apparent, as this part will show, that it is too vague
and poorly defined to be very useful in science, and that in its place we
need more clearly defined concepts. We strive in this report to use clear
terms; when discussing scientific studies that rely on the concept of “sex-
ual orientation,” we try as much as possible to specify how the scientists
defined the term, or related terms.
One of the central difficulties in examining and researching sexual
orientation is that the underlying concepts of “sexual desire,” “sexual
attraction,” and “sexual arousal” can be ambiguous, and it is even less
clear what it means that a person identifies as having a sexual orientation
grounded in some pattern of desires, attractions, or states of arousal.
The word “desire” all by itself might be used to cover an aspect of
volition more naturally expressed by “want”: I want to go out for din-
ner, or to take a road trip with my friends next summer, or to finish this
project. When “desire” is used in this sense, the objects of desire are fairly
determinate goals
some may be perfectly achievable, such as moving to
a new city or finding a new job; others may be more ambitious and out of
reach, like the dream of becoming a world-famous movie star. Often, how-
ever, the language of desire is meant to include things that are less clear:
indefinite longings for a life that is, in some unspecified sense, different or
better; an inchoate sense of something being missing or lacking in one-
self or one’s world; or, in psychoanalytic literature, unconscious dynamic
forces that shape one’s cognitive, emotional, and social behaviors, but that
are separate from one’s ordinary, conscious sense of self.
This more full-blooded notion of desire is, itself, ambiguous. It might
refer to a hoped-for state of affairs like finding a sense of meaning, fulfill-
ment, and satisfaction with one’s life, a desire that, while not completely
clear in its implications, is presumably not entirely out of reach, although
such longings may also be forms of fantasizing about a radically altered or
perhaps even unattainable state of affairs. If I want to take a road trip with
my friends, the steps are clear: call up my friends, pick a date, map out a
route, and so on. However, if I have an inchoate longing for change, a hope
for sustainable intimacy, love, and belonging, or an unconscious conflict
Fall 2016 ~ 17
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
that is disrupting my ability to move forward in the life I have tried to
build for myself, I face a different sort of challenge. There is not necessar-
ily a set of well-defined or conscious goals, much less established ways of
achieving them. This is not to say that the satisfaction of these longings is
impossible, but doing so often involves not only choosing concrete actions
to achieve particular goals but the more complex shaping of one’s own life
through acting in and making sense of the world and one’s place in it.
So the first thing to note when considering both popular discussions
and scientific studies of sexuality is that the use of the term “desire” could
refer to distinct aspects of human life and experience.
Just as the meanings that might be intended by the term “desire are
many, so also is each of these meanings varied, making clear delineations
a challenge. For example, a commonsense understanding might suggest
that the term “sexual desire” means wanting to engage in specific sexual
acts with particular individuals (or categories of individuals). Psychiatrist
Steven Levine articulated this common view in his definition of sexual
desire as “the sum of the forces that incline us toward and away from sexual
But it is not obvious how one might study this “sum in a rig-
orous way. Nor is it obvious why all the diverse factors that can potentially
influence sexual behavior, such as material poverty in the case of prosti-
tution, for instance alcohol consumption, and intimate affection, should
all be grouped together as aspects of sexual desire. As Levine himself
points out, “In anyone’s hands, sexual desire can be a slippery concept.
Consider a few of the ways that the term “sexual desire” has been
employed in scientific contexts designating one or more of the follow-
ing distinct phenomena:
1. States of physical arousal that may or may not be linked to a
specific physical activity and may or may not be objects of con-
scious awareness.
2. Conscious erotic interest in response to finding others attrac-
tive (in perception, memory, or fantasy), which may or may not
involve any of the bodily processes associated with measurable
states of physical arousal.
3. Strong interest in finding a companion or establishing a
durable relationship.
4. The romantic aspirations and feelings associated with infatu-
ation or falling in love with a specific individual.
18 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
5. Inclination towards attachment to specific individuals.
6. The general motivation to seek intimacy with a member of
some specific group.
7. An aesthetic measure that latches onto perceived beauty in
In a given social science study, the concepts mentioned above will
often each have its own particular operational definition for the purposes
of research. But they cannot all mean the same thing. Strong interest in
finding a companion, for example, is clearly distinguishable from physical
arousal. Looking at this list of experiential and psychological phenomena,
one can easily envision what confusions might arise from using the term
“sexual desire” without sufficient care.
The philosopher Alexander Pruss provides a helpful summary of
some of the difficulties involved in characterizing the related concept of
sexual attraction:
What does it mean to be “sexually attracted” to someone? Does it mean
to have a tendency to be aroused in their presence? But surely it is pos-
sible to find someone sexually attractive without being aroused. Does
it mean to form the belief that someone is sexually attractive to one?
Surely not, since a belief about who is sexually attractive to one might
be wrong for instance, one might confuse admiration of form with
sexual attraction. Does it mean to have a noninstrumental desire for a
sexual or romantic relationship with the person? Probably not: we can
imagine a person who has no sexual attraction to anybody, but who has
a noninstrumental desire for a romantic relationship because of a belief,
based on the testimony of others, that romantic relationships have
noninstrumental value. These and similar questions suggest that there
is a cluster of related concepts under the head of “sexual attraction,”
and any precise definition is likely to be an undesirable shoehorning.
But if the concept of sexual attraction is a cluster of concepts, neither
are there simply univocal concepts of heterosexuality, homosexuality,
and bisexuality.
The ambiguity of the term “sexual desire” (and similar terms) should
give us pause to consider the diverse aspects of human experience that
are often associated with it. The problem is neither irresolvable nor
unique to this subject matter. Other social science concepts aggression
and addiction, for example may likewise be difficult to define and to
Fall 2016 ~ 19
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
operationalize and for this reason admit of various usages.
the ambiguity presents a significant challenge for both research design
and interpretation, requiring that we take care in attending to the mean-
ings, contexts, and findings specific to each study. It is also important to
bracket any subjective associations with or uses of these terms that do not
conform to well-defined scientific classifications and techniques.
It would be a mistake, at any rate, to ignore the varied uses of this and
related terms or to try to reduce the many and distinct experiences to
which they might refer to a single concept or experience. As we shall see,
doing so could in some cases adversely affect the evaluation and treatment
of patients.
The Context of Sexual Desire
We can further clarify the complex phenomenon of sexual desire if we
examine what relationship it has to other aspects of our lives. To do so,
we borrow some conceptual tools from a philosophical tradition known
as phenomenology, which conceives of human experience as deriving its
meaning from the whole context in which it appears.
The testimony of experience suggests that one’s experience of sexual
desire and sexual attraction is not voluntary, at least not in any immedi-
ate way. The whole set of inclinations that we generally associate with the
experience of sexual desire whether the impulse to engage in particular
acts or to enjoy certain relationships does not appear to be the sole prod-
uct of any deliberate choice. Our sexual appetites (like other natural appe-
tites) are experienced as given, even if their expression is shaped in subtle
ways by many factors, which might very well include volition. Indeed, far
from appearing as a product of our will, sexual desire however we define
it is often experienced as a powerful force, akin to hunger, that many
struggle (especially in adolescence) to bring under direction and control.
Furthermore, sexual desire can impact one’s attention involuntarily or
color one’s day-to-day perceptions, experiences, and encounters. What
seems to be to some extent in our control is how we choose to live with
this appetite, how we integrate it into the rest of our lives.
But the question remains: What is sexual desire? What is this part
of our lives that we consider to be given, prior even to our capacity to
“Operationalizing” refers to the way social scientists make a variable measurable. Homosexuality
may be operationalized as the answers that survey respondents give to questions about their sexual
orientation. Or it could be operationalized as answers to questions about their desires, attractions,
and behavior. Operationalizing variables in ways that will reliably measure the trait or behavior
being studied is a difficult but important part of any social science research.
20 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
deliberate and make rational choices about it? We know that some sort
of sexual appetite is present in non-human animals, as is evident in the
mammalian estrous cycle; in most mammalian species sexual arousal and
receptivity are linked to the phase of the ovulation cycle during which the
female is reproductively receptive.
One of the relatively unique features
of Homo sapiens, shared with only a few other primates, is that sexual
desire is not exclusively linked to the woman’s ovulatory cycle.
biologists have argued that this means that sexual desire in humans has
evolved to facilitate the formation of sustaining relationships between
parents, in addition to the more basic biological purpose of reproduc-
tion. Whatever the explanation for the origins and biological functions
of human sexuality, the lived experience of sexual desires is laden with
significance that goes beyond the biological purposes that sexual desires
and behaviors serve. This significance is not just a subjective add-on to
the more basic physiological and functional realities, but something that
pervades our lived experience of sexuality.
As philosophers who study the structure of conscious experience have
observed, our way of experiencing the world is shaped by our “embodi-
ment, bodily skills, cultural context, language and other social practic-
Long before most of us experience anything like what we typically
associate with sexual desire, we are already enmeshed in a cultural and
social context involving other persons, feelings, emotions, opportunities,
deprivations, and so on. Perhaps sexuality, like other human phenomena
that gradually become part of our psychological constitution, has roots in
these early meaning-making experiences. If meaning-making is integral
to human experience in general, it is likely to play a key role in sexual
experience in particular. And given that volition is operative in these
other aspects of our lives, it stands to reason that volition will be operative
in our experience of sexuality too, if only as one of many other factors.
This is not to suggest that sexuality including sexual desire, attrac-
tion, and identity is the result of any deliberate, rational decision cal-
culus. Even if volition plays an important role in sexuality, volition itself
is quite complex: many, perhaps most, of our volitional choices do not
seem to come in the form of discrete, conscious, or deliberate decisions;
“volitional” does not necessarily mean “deliberate.” The life of a desiring,
volitional agent involves many tacit patterns of behavior owing to habits,
past experiences, memories, and subtle ways of adopting and abandoning
different stances on one’s life.
If something like this way of understanding the life of a desiring, voli-
tional agent is true, then we do not deliberately “choose” the objects of our
Fall 2016 ~ 21
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
sexual desires any more than we choose the objects of our other desires.
It might be more accurate to say that we gradually guide and give our-
selves over to them over the course of our growth and development. This
process of forming and reforming ourselves as human beings is similar to
what Abraham Maslow calls self-actualization.
Why should sexuality
be an exception to this process? In the picture we are offering, internal
factors, such as our genetic make-up, and external environmental factors,
such as past experiences, are only ingredients, however important, in the
complex human experience of sexual desire.
Sexual Orientation
Just as the concept of “sexual desire” is complex and difficult to define,
there are currently no agreed-upon definitions of “sexual orientation,”
“homosexuality,” or “heterosexuality” for purposes of empirical research.
Should homosexuality, for example, be characterized by reference to
desires to engage in particular acts with individuals of the same sex, or
to a patterned history of having engaged in such acts, or to particular
features of one’s private wishes or fantasies, or to a consistent impulse
to seek intimacy with members of the same sex, or to a social identity
imposed by oneself or others, or to something else entirely?
As early as 1896, in a book on homosexuality, the French thinker Marc-
André Raffalovich argued that there were more than ten different types of
affective inclination or behavior captured by the term “homosexuality” (or
what he called “unisexuality”).
Raffalovich knew his subject matter up
close: he chronicled the trial, imprisonment, and resulting social disgrace
of the writer Oscar Wilde, who had been prosecuted for “gross indecency”
with other men. Raffalovich himself maintained a prolonged and intimate
relationship with John Gray, a man of letters thought to be the inspiration
for Wilde’s classic The Picture of Dorian Gray.
We might also consider
the vast psychoanalytic literature from the early twentieth century on
the topic of sexual desire, in which the experiences of individual subjects
and their clinical cases are catalogued in great detail. These historical
examples bring into relief the complexity that researchers still face today
when attempting to arrive at clean categorizations of the richly varied
affective and behavioral phenomena associated with sexual desire, in both
same-sex and opposite-sex attractions.
We may contrast such inherent complexity with a different phenom-
enon that can be delineated unambiguously, such as pregnancy. With very
few exceptions, a woman is or is not pregnant, which makes classification
22 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
of research subjects for the purposes of study relatively easy: compare
pregnant women with other, non-pregnant women. But how can research-
ers compare, say, “gay” men to “straight” men in a single study, or across
a range of studies, without mutually exclusive and exhaustive definitions
of the terms “gay” and “straight”?
To increase precision, some researchers categorize concepts associ-
ated with human sexuality along a continuum or scale according to varia-
tions in pervasiveness, prominence, or intensity. Some scales focus on both
intensity and the objects of sexual desire. Among the most familiar and
widely used is the Kinsey scale, developed in the 1940s to classify sexual
desires and orientations using purportedly measurable criteria. People are
asked to choose one of the following options:
0 - Exclusively heterosexual
1 - Predominantly heterosexual, only incidentally homosexual
2 - Predominantly heterosexual, but more than incidentally homosexual
3 - Equally heterosexual and homosexual
4 - Predominantly homosexual, but more than incidentally heterosexual
5 - Predominantly homosexual, only incidentally heterosexual
6 - Exclusively homosexual
But there are considerable limitations to this approach. In prin-
ciple, measurements of this sort are valuable for social science research.
They can be used, for example, in empirical tests such as the classic
“t-test,” which helps researchers measure statistically meaningful dif-
ferences between data sets. Many measurements in social science, how-
ever, are “ordinal,” meaning that variables are rank-ordered along a
single, one-dimensional continuum but are not intrinsically significant
beyond that. In the case of the Kinsey scale, this situation is even worse,
because it measures the self-identification of individuals, while leaving
unclear whether the values they report all refer to the same aspect of
sexuality different people may understand the terms “heterosexual”
and homosexual” to refer to feelings of attraction, or to arousal, or to
fantasies, or to behavior, or to any combination of these. The ambigu-
ity of the terms severely limits the use of the Kinsey scale as an ordinal
measurement that gives a rank order to variables along a single, one-
dimensional continuum. So it is not clear that this scale helps research-
ers to make even rudimentary classifications among the relevant groups
using qualitative criteria, much less to rank-order variables or conduct
controlled experiments.
Fall 2016 ~ 23
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
Perhaps, given the inherent complexity of the subject matter, attempts
to devise “objective” scales of this sort are misguided. In a critique of such
approaches to social science, philosopher and neuropsychologist Daniel
Robinson points out that “statements that lend themselves to different
interpretation do not become ‘objective’ merely by putting a numeral in
front of them.”
It may be that self-reported identifications with cultural-
ly fraught and inherently complex labels simply cannot provide an objec-
tive basis for quantitative measurements in individuals or across groups.
Another obstacle for research in this area may be the popular, but not
well-supported, belief that romantic desires are sublimations of sexual
desires. This idea, traceable to Freud’s theory of unconscious drives, has
been challenged by research on “attachment theory,” developed by John
Bowlby in the 1950s.
Very roughly, attachment theory holds that later
affective experiences that are often grouped under the general rubric
“romantic” are explained in part by early childhood attachment behaviors
(associated with maternal figures or caregivers) not by unconscious,
sexual drives. Romantic desires, following this line of thought, might not
be as strongly correlated with sexual desires as is commonly thought. All
of this is to suggest that simple delineations of the concepts relating to
human sexuality cannot be taken at face value and that ongoing empirical
research sometimes changes or complicates the meanings of the concepts.
If we look at recent research, we find that scientists often use at least
one of three categories when attempting to classify people as “homo-
sexual” or “heterosexual”: sexual behavior; sexual fantasies (or related
emotional or affective experiences); and self-identification (as “gay,” “les-
bian,” “bisexual,” “asexual,” and so forth).
Some add a fourth: inclusion
in a community defined by sexual orientation. Consider, for example, the
American Psychological Association’s definition of sexual orientation in a
2008 document designed to educate the public:
Sexual orientation refers to an enduring pattern of emotional, romantic
and/or sexual attractions to men, women or both sexes. Sexual orienta-
tion also refers to a person’s sense of identity based on those attractions,
related behaviors, and membership in a community of others who share
those attractions. Research over several decades has demonstrated that
sexual orientation ranges along a continuum, from exclusive attraction
to the other sex to exclusive attraction to the same sex.
One difficulty with grouping these categories together under the same
general rubric of “sexual orientation” is that research suggests they often
24 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
do not coincide in real life. Sociologist Edward O. Laumann and col-
leagues summarize this point clearly in a 1994 book:
While there is a core group (about 2.4 percent of the total men and
about 1.3 percent of the total women) in our survey who define themselves
as homosexual or bisexual, have same-gender partners, and express
homosexual desires, there are also sizable groups who do not consider
themselves to be either homosexual or bisexual but have had adult
homosexual experiences or express some degree of desire.. . . [T]his
preliminary analysis provides unambiguous evidence that no single
number can be used to provide an accurate and valid characterization
of the incidence and prevalence of homosexuality in the population at
large. In sum, homosexuality is fundamentally a multidimensional phe-
nomenon that has manifold meanings and interpretations, depending
on context and purpose.
[Emphases added.]
More recently, in a 2002 study, psychologists Lisa M. Diamond and Ritch
C. Savin-Williams make a similar point:
The more carefully researchers map these constellations differen-
tiating, for example, between gender identity and sexual identity, desire
and behavior, sexual versus affectionate feelings, early-appearing versus
late-appearing attractions and fantasies, or social identifications and
sexual profiles
the more complicated the picture becomes because few
individuals report uniform inter-correlations among these domains.
[Emphases added.]
Some researchers acknowledge the difficulties with grouping these
various components under a single rubric. For example, researchers John
C. Gonsiorek and James D. Weinrich write in a 1991 book: “It can be
safely assumed that there is no necessary relationship between a person’s
sexual behavior and self-identity unless both are individually assessed.”
Likewise, in a 1999 review of research on the development of sexual orien-
tation in women, social psychologist Letitia Anne Peplau argues: “There
is ample documentation that same-sex attractions and behaviors are not
inevitably or inherently linked to one’s identity.”
In sum, the complexities surrounding the concept of “sexual orienta-
tion” present considerable challenges for empirical research on the sub-
ject. While the general public may be under the impression that there are
widely accepted scientific definitions of terms such as “sexual orientation,”
in fact, there are not. Diamond’s assessment of the situation in 2003 is still
true today, that “there is currently no scientific or popular consensus on
Fall 2016 ~ 25
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
the exact constellation of experiences that definitively ‘qualify’ an indi-
vidual as lesbian, gay, or bisexual.”
It is owing to such complexities that some researchers, for instance
Laumann, proceed by characterizing sexual orientation as a “multidi-
mensional phenomenon.But one might just as well wonder whether, in
trying to shoehorn this “multidimensional phenomenon” into a single
category, we are not reifying a concept that corresponds to something
far too plastic and diffuse in reality to be of much value in scientific
research. While labels such as “heterosexual” and “homosexual” are
often taken to designate stable psychological or even biological traits,
perhaps they do not. It may be that individuals’ affective, sexual, and
behavioral experiences do not conform well to such categorical labels
because these labels do not, in fact, refer to natural (psychological or
biological) kinds. At the very least, we should recognize that we do not
yet possess a clear and well-established framework for research on these
topics. Rather than attempting to research sexual desire, attraction,
identity, and behavior under the general rubric of sexual orientation,
we might do better to examine empirically each domain separately and
in its own specificity.
To that end, this part of our report considers research on sexual desire
and sexual attraction, focusing on the empirical findings related to etiol-
ogy and development, and highlighting the underlying complexities. We
will continue to employ ambiguous terms like “sexual orientation” where
they are used by the authors we discuss, but we will try to be attentive to
the context of their use and the ambiguities attaching to them.
Challenging the “Born that Way” Hypothesis
Keeping in mind these reflections on the problems of definitions, we turn
to the question of how sexual desires originate and develop. Consider the
different patterns of attraction between individuals who report experi-
encing predominant sexual or romantic attraction toward members of
the same sex and those who report experiencing predominant sexual or
romantic attraction toward members of the opposite sex. What are the
causes of these two patterns of attraction? Are such attractions or pref-
erences innate traits, perhaps determined by our genes or prenatal hor-
mones; are they acquired by experiential, environmental, or volitional fac-
tors; or do they develop out of some combination of both kinds of causes?
What role, if any, does human agency play in the genesis of patterns of
attraction? What role, if any, do cultural or social influences play?
26 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
Research suggests that while genetic or innate factors may influence
the emergence of same-sex attractions, these biological factors cannot
provide a complete explanation, and environmental and experiential fac-
tors may also play an important role.
The most commonly accepted view in popular discourse we men-
tioned above the “born that way” notion that homosexuality and het-
erosexuality are biologically innate or the product of very early develop-
mental factors has led many non-specialists to think that homosexuality
or heterosexuality is in any given person unchangeable and determined
entirely apart from choices, behaviors, life experiences, and social contexts.
However, as the following discussion of the relevant scientific literature
shows, this is not a view that is well-supported by research.
Studies of Twins
One powerful research design for assessing whether biological or psy-
chological traits have a genetic basis is the study of identical twins. If the
probability is high that both members in a pair of identical twins, who
share the same genome, exhibit a trait when one of them does this is
known as the concordance rate then one can infer that genetic factors
are likely to be involved in the trait. If, however, the concordance rate for
identical twins is no higher than the concordance rate of the same trait
in fraternal twins, who share (on average) only half their genes, this indi-
cates that the shared environment may be a more important factor than
shared genes.
One of the pioneers of behavioral genetics and one of the first
researchers to use twins to study the effect of genes on traits, including
sexual orientation, was psychiatrist Franz Josef Kallmann. In a landmark
paper published in 1952, he reported that for all the pairs of identical
twins he studied, if one of the twins was gay then both were gay, yield-
ing an astonishing 100% concordance rate for homosexuality in identi-
cal twins.
Were this result replicated and the study designed better, it
would have given early support to the “born that way” hypothesis. But
the study was heavily criticized. For example, philosopher and law profes-
sor Edward Stein notes that Kallmann did not present any evidence that
the twins in his study were in fact genetically identical, and his sample
was drawn from psychiatric patients, prisoners, and others through what
Kallmann described as “direct contacts with the clandestine homosexual
world,” leading Stein to argue that Kallmann’s sample “in no way con-
stituted a reasonable cross-section of the homosexual population.”
Fall 2016 ~ 27
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
(Samples such as Kallmann’s are known as convenience samples, which
involve selecting subjects from populations that are conveniently acces-
sible to the researcher.)
Nevertheless, well-designed twin studies examining the genetics of
homosexuality indicate that genetic factors likely play some role in deter-
mining sexual orientation. For example, in 2000, psychologist J. Michael
Bailey and colleagues conducted a major study of sexual orientation using
twins in the Australian National Health and Medical Research Council
Twin Registry, a large probability sample, which was therefore more
likely to be representative of the general population than Kallmann’s.
The study employed the Kinsey scale to operationalize sexual orientation
and estimated concordance rates for being homosexual of 20% for men
and 24% for women in identical (maternal, monozygotic) twins, compared
to 0% for men and 10% for women in non-identical (fraternal, dizygotic)
The difference in the estimated concordance rates was statisti-
cally significant for men but not for women. On the basis of these findings,
the researchers estimated that the heritability of homosexuality for men
was 0.45 with a wide 95% confidence interval of 0.00 0.71; for women,
it was 0.08 with a similarly wide confidence interval of 0.00 0.67. These
estimates suggest that for males 45% of the differences between certain
sexual orientations (homosexual versus heterosexuals as measured by the
Kinsey scale) could be attributed to differences in genes.
The large confidence intervals in the study by Bailey and colleagues
mean that we must be careful in assessing the substantive significance of
these findings. The authors interpret their findings to suggest that “any
major gene for strictly defined homosexuality has either low penetrance
or low frequency,”
but their data did show (marginal) statistical signifi-
cance. While the concordance estimates seem somewhat high in the mod-
els used, the confidence intervals are so wide that it is difficult to judge
the reliability, including the replicability, of these estimates.
It is worth clarifying here what “heritability” means in these studies,
since the technical meaning in population genetics is narrower and more
precise than the everyday meaning of the word. Heritability is a measure
of how much variation in a particular trait within a population can be
attributed to variation in genes in that population. It is not, however, a
measure of how much a trait is genetically determined.
Traits that are almost entirely genetically determined can have very
low heritability values, while traits that have almost no genetic basis can
be found to be highly heritable. For instance, the number of fingers human
beings have is almost completely genetically determined. But there is little
28 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
variation in the number of fingers humans have, and most of the variation
we do see is due to non-genetic factors such as accidents, which would
lead to low heritability estimates for the trait. Conversely, cultural traits
can sometimes be found to be highly heritable. For instance, whether a
given individual in mid-twentieth century America wore earrings would
have been found to be highly heritable, because it was highly associated
with being male or female, which is in turn associated with possessing XX
or XY sex chromosomes, making variability in earring-wearing behavior
highly associated with genetic differences, despite the fact that wearing
earrings is a cultural rather than biological phenomenon. Today, herita-
bility estimates for earring-wearing behavior would be lower than they
were in mid-twentieth century America, not because of any changes in
the American gene pool, but because of the increased acceptance of men
wearing earrings.
So, a heritability estimate of 0.45 does not mean that 45% of sexual-
ity is determined by genes. Rather, it means that 45% of the variation
between individuals in the population studied can be attributed in some
way to genetic factors, as opposed to environmental factors.
In 2010, psychiatric epidemiologist Niklas Långström and colleagues
conducted a large, sophisticated twin study of sexual orientation, analyz-
ing data from 3,826 identical and fraternal same-sex twin pairs (2,320
identical and 1,506 fraternal pairs).
The researchers operational-
ized homosexuality in terms of lifetime same-sex sexual partners. The
sample’s concordance rates were somewhat lower than those found in
the study by Bailey and colleagues. For having had at least one same-sex
partner, the concordance for men was 18% in identical twins and 11% in
fraternal twins; for women, 22% and 17%, respectively. For total number
of sexual partners, concordance rates for men were 5% in identical twins
and 0% in fraternal twins; for women, 11% and 7%, respectively.
For men, these rates suggest an estimated heritability rate of 0.39 for
having had at least one lifetime same-sex partner (with a 95% confidence
interval of 0.00 0.59), and 0.34 for total number of same-sex partners
(with a 95% confidence interval of 0.00 0.53). Environmental factors
experienced by one twin but not the other explained 61% and 66% of the
variance, respectively, while environmental factors shared by the twins
failed to explain any of the variance. For women, the heritability rate for
having had at least one lifetime same-sex partner was 0.19 (95% confi-
dence interval of 0.00 0.49); for total number of same-sex partners, it
was 0.18 (95% confidence interval of 0.11 0.45). Unique environmental
factors accounted for 64% and 66% of the variance, respectively, while
Fall 2016 ~ 29
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
shared environmental factors accounted for 17% and 16%, respectively.
These values indicate that, while the genetic component of homosexual
behavior is far from negligible, non-shared environmental factors play
a critical, perhaps preponderant, role. The authors conclude that sexual
orientation arises from both heritable and environmental influences
unique to the individual, stating that “the present results support the
notion that the individual-specific environment does indeed influence
sexual preference.”
Another large and nationally representative study of twins published
by sociologists Peter S. Bearman and Hannah Brückner in 2002 used data
from the National Longitudinal Study of Adolescent to Adult Health
(commonly abbreviated as “Add Health”) of adolescents in grades 7 12.
They attempted to estimate the relative influence of social factors, genetic
factors, and prenatal hormonal factors on the development of same-sex
attractions. Overall, 8.7% of the 18,841 adolescents in their study reported
same-sex attractions, 3.1% reported a same-sex romantic relationship,
and 1.5% reported same-sex sexual behavior. The authors first analyzed
the “social influence hypothesis,” according to which opposite-sex twins
receive less gendered socialization from their families than same-sex twins
or opposite-sex siblings, and found that this hypothesis was well-supported
in the case of males. While female opposite-sex twins in the study were
the least likely of all the groups to report same-sex attractions (5.3%),
male opposite-sex twins were the likeliest to report same-sex attractions
(16.8%) more than twice as likely as males with a full, non-twin sister
(16.8% vs. 7.3%). The authors concluded there was “substantial indirect
evidence in support of a socialization model at the individual level.”
The authors also examined the “intrauterine hormone transfer hypoth-
esis,” according to which prenatal hormone transfers between opposite-
sex twin fetuses influences the sexual orientation of the twins. (Note that
this is different from the more general hypothesis that prenatal hormones
influence the development of sexual orientation.) In the study, the propor-
tion of male opposite-sex twins reporting same-sex attraction was about
twice as high for those without older brothers (18.7%) as for those with
older brothers (8.8%). The authors argued that this finding was strong
evidence against the hormone-transfer hypothesis, since the presence of
older brothers should not decrease the likelihood of same-sex attraction
if that attraction has a basis in prenatal hormonal transfers. However,
that conclusion seems premature: the observations are consistent with the
possibility of both hormonal factors and the presence of an older brother
having an effect (especially if the latter influences the former). This study
30 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
also found no correlation between experiencing same-sex attraction and
having multiple older brothers, which had been reported in some earlier
Finally, Bearman and Brückner did not find evidence of significant
genetic influence on sexual attraction. Significant influence would require
that identical twins have significantly higher concordance rates for same-
sex attraction than fraternal twins or non-twin siblings. But in the study,
the rates were statistically similar: identical twins were 6.7% concordant,
dizygotic pairs 7.2% concordant, and full siblings 5.5% concordant. The
authors concluded that “it is more likely that any genetic influence, if
present, can only be expressed in specific and circumscribed social struc-
Based on their data, they suggested the one observed social
structure that might enable this genetic expression is the more limited
“gender socialization associated with firstborn OS [opposite-sex] twin
Thus, they inferred that their results “support the hypothesis
that less gendered socialization in early childhood and preadolescence
shapes subsequent same-sex romantic preferences.”
While the findings
here are suggestive, further research is needed to confirm this hypothesis.
The authors also argued that the higher concordance rates for same-sex
attraction reported in previous studies may be unreliable due to method-
ological problems such as non-representative samples and small sample
sizes. (It should be noted, however, that these remarks were published
prior to the study by Långström and colleagues discussed above, which
uses a study design that does not appear to have these limitations.)
To reconcile the somewhat mixed data on heritability, we could hypoth-
esize that attraction to the same sex may have a stronger heritable compo-
nent as people age that is, when researchers attempt to measure sexual
orientation later in life (as in the 2010 study by Långström and colleagues)
than when measured earlier in life. Heritability estimates can change
depending on the age at which a trait is measured because changes in the
environmental factors that might influence variation in the trait may vary
for individuals at different ages, and because genetically influenced traits
may become more fixed at a later stage in an individual’s development
(height, for instance, becomes fixed in early adulthood). This hypothesis is
also suggested by findings, discussed below, that same-sex attraction may
be more fluid in adolescence than in later stages of adulthood.
In contrast to the studies just summarized, psychiatrist Kenneth S.
Kendler and colleagues conducted a large twin study using a probabil-
ity sample of 794 twin pairs and 1,380 non-twin siblings.
Based on
concordance rates for sexual orientation (defined in this study as self-iden-
Fall 2016 ~ 31
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
tification based on attraction), the authors state that their results “suggest
that genetic factors may provide an important influence on sexual orienta-
The study does not, however, appear to be sufficiently powerful to
draw strong conclusions about the degree of genetic influence on sexual-
ity: only 19 of 324 identical twin pairs had any non-heterosexual member,
with 6 of the 19 pairs concordant; 15 of 240 same-sex fraternal twin pairs
had any non-heterosexual member, with 2 of the 15 pairs concordant.
Because only 8 twin pairs were concordant for non-heterosexuality, the
study’s ability to draw substantively significant comparisons between
identical and fraternal twins (or between twins and non-twin siblings) is
Overall, these studies suggest that (depending on how homosexual-
ity is defined) in anywhere from 6% to 32% of cases, both members of an
identical twin pair would be homosexual if at least one member is. Since
some twin studies found higher concordance rates in identical twins than
in fraternal twins or non-twin siblings, there may be genetic influences on
sexual desire and behavioral preferences. One needs to bear in mind that
identical twins typically have even more similar environments early
attachment experiences, peer relationships, and the like than fraternal
twins or non-twin siblings. Because of their similar appearances and tem-
peraments, for example, identical twins may be more likely than fraternal
twins or other siblings to be treated similarly. So some of the higher con-
cordance rates may be attributable to environmental factors rather than
genetic factors. In any case, if genes do play a role in predisposing people
toward certain sexual desires or behaviors, these studies make clear that
genetic influences cannot be the whole story.
Summarizing the studies of twins, we can say that there is no reliable
scientific evidence that sexual orientation is determined by a person’s
genes. But there is evidence that genes play a role in influencing sexual
orientation. So the question Are gay people born that way?” requires
clarification. There is virtually no evidence that anyone, gay or straight,
is “born that way” if that means their sexual orientation was genetically
determined. But there is some evidence from the twin studies that certain
genetic profiles probably increase the likelihood the person later identifies
as gay or engages in same-sex sexual behavior.
Future twin studies on the heritability of sexual orientation should
include analyses of larger samples or meta-analyses or other systematic
reviews to overcome the limited sample size and statistical power of some
of the existing studies, and analyses of heritability rates across different
dimensions of sexuality (such as attraction, behavior, and identity) to
32 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
overcome the imprecisions of the ambiguous concept of sexual orienta-
tion and the limits of studies that look at only one of these dimensions of
Molecular Genetics
In examining the question whether, and perhaps to what extent, there
may be genetic contributions to homosexuality, we have so far looked at
studies that employ methods of classical genetics to estimate the herita-
bility of a trait like sexual orientation but that do not identify particular
genes that may be associated with the trait.
But genetics can also be
studied using what are often called molecular methods that provide esti-
mates of which particular genetic variations are associated with traits,
whether physical or behavioral.
One early attempt to identify a more specific genetic basis for homo-
sexuality was a 1993 study by geneticist Dean Hamer and colleagues of
40 pairs of homosexual brothers.
By examining the family history of
homosexuality for these individuals, they identified a possible linkage
between homosexuality in males and genetic markers on the Xq28 region
of the X chromosome. Attempts to replicate this influential study’s results
have had mixed results: George Rice and colleagues attempted and failed
to replicate Hamer’s findings,
though in 2015 Alan R. Sanders and col-
leagues were able to replicate Hamer’s original findings using a larger
population size of 409 male twin pairs of homosexual brothers, and to find
additional genetic linkage sites.
(Since the effect was small, however, the
genetic marker would not be a good predictor of sexual orientation.)
Genetic linkage studies like the ones discussed above are able to
identify particular regions of chromosomes that may be associated with a
trait by looking at patterns of inheritance. Today, one of the chief meth-
ods for inferring which genetic variants are associated with a trait is the
genome-wide association study, which uses DNA sequencing technologies
to identify particular differences in DNA that may be associated with a
trait. Scientists examine millions of genetic variants in large numbers of
individuals who have a particular trait, as well as individuals who do not
have the trait, and compare the frequency of genetic variants among those
who do and do not have the trait. Specific genetic variants that occur more
frequently among those who have than those who do not have the trait are
inferred to have some association with that trait. Genome-wide associa-
tion studies have become popular in recent years, yet few such scientific
studies have found significant associations of genetic variants with sexual
Fall 2016 ~ 33
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
orientation. The largest attempt to identify genetic variants associated
with homosexuality, a study of over 23,000 individuals from the 23andMe
database presented at the American Society of Human Genetics annual
meeting in 2012, found no linkages reaching genome-wide significance for
same-sex sexual identity for males or females.
So, again, the evidence for a genetic basis for homosexuality is inconsis-
tent and inconclusive, which suggests that, though genetic factors explain
some of the variation in sexual orientation, the genetic contribution to this
trait is not likely to be strong and even less likely to be decisive.
As is often true of human behavioral tendencies, there may be genetic
contributions to the tendency toward homosexual inclinations or behav-
iors. Phenotypic expression of genes is usually influenced by environmen-
tal factors different environments may lead to different phenotypes even
for the same genes. So even if there are genetic factors that contribute to
homosexuality, an individual’s sexual attractions or preferences may also
be influenced by a number of environmental factors, such as social stress-
ors, including emotional, physical, or sexual abuse. Looking to develop-
mental, environmental, experiential, social, or volitional factors will be
necessary to arrive at a fuller picture of how sexual interests, attractions,
and desires develop.
The Limited Role of Genetics
Lay readers might note at this point that even at the purely biological
level of genetics, the shopworn “nature vs. nurture” debates regarding
human psychology have been abandoned by scientists, who recognize that
no credible hypothesis can be offered for any particular traits that would
be determined either purely by genetics or the environment. The grow-
ing field of epigenetics, for example, demonstrates that even for relatively
simple traits, gene expression itself can be influenced by innumerable
other external factors that can shape the functioning of genes.
This is
even more relevant when it comes to the relationship between genes and
complex traits like sexual attraction, drives, and behaviors.
These gene-environment relationships are complex and multidimen-
sional. Non-genetic developmental factors and environmental experiences
may be sculpted, in part, by genetic factors working in subtle ways. For
example, social geneticists have documented the indirect role of genes
in peer-aligned behaviors, such that an individual’s physical appearance
could influence whether a particular social group will include or exclude
that individual.
34 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
Contemporary geneticists know that genes can influence a person’s
range of interests and motivations, therefore indirectly affecting behavior.
While genes may in this way incline a person to certain behaviors, com-
pelling behavior directly, independently of a wide range of other factors,
seems less plausible. They may influence behavior in more subtle ways,
depending on external environmental stimuli (for instance, peer pressure,
suggestion, and behavioral rewards) in conjunction with psychological
factors and physical makeup. Dean Hamer, whose work on the possible
role of genetics in homosexuality was examined above, explained some
of the limitations of behavioral genetics in a 2002 article in Science: “The
real culprit [of lack of progress in behavioral genetics] is the assumption
that the rich complexity of human thought and emotion can be reduced to
a simple, linear relation between individual genes and behaviors. .. . This
oversimplified model, which underlies most current research in behavior
genetics, ignores the critical importance of the brain, the environment,
and gene expression networks.”
The genetic influences affecting any complex human behavior
whether sexual behaviors, or interpersonal interactions depend in part
on individuals’ life experiences as they mature. Genes constitute only
one of the many key influences on behavior in addition to environmental
influences, personal choices, and interpersonal experiences. The weight
of evidence to date strongly suggests that the contribution of genetic fac-
tors is modest. We can say with confidence that genes are not the sole,
essential cause of sexual orientation; there is evidence that genes play a
modest role in contributing to the development of sexual attractions and
behaviors but little evidence to support a simplistic “born that way” nar-
rative concerning the nature of sexual orientation.
The Influence of Hormones
Another area of research relevant to the hypothesis that people are born
with dispositions toward different sexual orientations involves prenatal
hormonal influences on physical development and subsequent male- or
female-typical behaviors in early childhood. For ethical and practical
reasons, the experimental work in this field is carried out in non-human
mammals, which limits how this research can be generalized to human
cases. However, children who are born with disorders of sexual develop-
ment (DSD) serve as a population in which to examine the influence of
genetic and hormonal abnormalities on the subsequent development of
non-typical sexual identity and sexual orientation.
Fall 2016 ~ 35
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
Hormones responsible for sexual differentiation are generally thought
to exert on the developing fetus either organizational
effects which pro-
duce permanent changes in the wiring and sensitivity of the brain, and thus
are considered largely irreversible or activating effects, which occur later
in an individual’s life (at puberty, and into adulthood).
hormones may prime the fetal systems (including the brain) structurally,
and set the stage for sensitivity to hormones presenting at puberty and
beyond, when the hormone will then “activate” systems which were “orga-
nized” prenatally.
Periods of peak response to the hormonal environment are thought
to occur during gestation. For example, testosterone is thought to influ-
ence the male fetus maximally between weeks 8 and 24, and then again at
birth, until about three months of age.
Estrogens are provided through-
out gestation by the placenta and the mother’s blood system.
in animals reveal there may even be multiple periods of sensitivity for a
variety of hormones, that the presence of one hormone may influence the
action of another hormone, and the sensitivity of the receptors for these
hormones can influence their actions.
Sexual differentiation, alone, is a
highly complex system.
Specific hormones of interest in this area of research are testosterone,
dihydrotestosterone (a metabolite of testosterone, and more potent than
testosterone), estradiol (which can be metabolized into testosterone),
progesterone, and cortisol. The generally accepted pathways of normal
hormonal influence of development in utero are as follows. The typical
pattern of sex differentiation in human fetuses begins with the differen-
tiation of the sex organs into testes or ovaries, a process that is largely
genetically controlled. Once these organs have differentiated, they produce
specific hormones that determine development of external genitalia. This
window of time in gestation is when hormones exert their phenotypic and
neurological effects. Testosterone secreted by the testes contributes to the
development of male external genitalia and affects neurological develop-
ment in males;
it is the absence of testosterone in females which allows
for the female pattern of external genitalia to develop.
Imbalances of
testosterone or estrogen, as well as their presence or absence at specific
critical periods of gestation, may cause disorders of sexual development.
(Genetic or environmental effects can also lead to disorders of sexual
Stress may also play some role in influencing the way hormones shape
gonadal development, neurodevelopment, and subsequent sex-typical
behaviors in early childhood.
Cortisol is the main hormone associated
36 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
with stress responses. It may originate from the mother, if she experiences
severe stressors during her pregnancy, or from the fetus under stress.
Elevated levels of cortisol may also occur from genetic defects.
of the most extensively studied disorders of sexual development is con-
genital adrenal hyperplasia (CAH), which in females can result in genital
Over 90% of cases of CAH result from a mutation in a gene
that codes for an enzyme that helps synthesize cortisol.
This results in
an overproduction of cortisol precursors, some of which are converted
into androgens (hormones associated with male sex development).
a result, girls are born with some degree of virilization of their genitalia,
depending on the severity of the genetic defect.
For severe cases of geni-
tal virilization, surgical intervention is sometimes performed to normalize
the genitalia. Hormone therapies are also often administered to mitigate
the effects of excess androgen production.
Females with CAH, who as
fetuses were exposed to above-average levels of androgens, are less likely
to be exclusively heterosexual than females without CAH, and females
with more severe forms of CAH are more likely to be non-heterosexual
than females with milder forms of the condition.
Likewise, there are disorders of sexual development in genetic males
affected by androgen insensitivity. In males with androgen insensitivity
syndrome, the testes produce testosterone normally, but the receptors
to testosterone are not functional.
The genitalia, at birth, appear to
be female, and the child is usually raised as a female. The individual’s
endogenous testosterone is broken down into estrogen, such that the
individual begins to develop female secondary sex characteristics.
does not become apparent that there is a problem until puberty, when the
individual does not start menses appropriately.
These patients generally
prefer to continue life as females, and their sexual orientation does not dif-
fer from females having an XX genotype.
Studies have suggested that
they are just as likely if not more likely to be exclusively interested in male
partners than XX females.
There are other disorders of sexual development affecting some genet-
ic males (i.e., with an XY genotype) in whom androgen deficiencies are a
direct result of the lack of enzymes either to synthesize dihydrotestoster-
one from testosterone or to produce testosterone from its precursor hor-
Individuals with these deficiencies are born with varied degrees
of ambiguous genitalia, and are sometimes raised as girls. During puberty,
however, these individuals often experience physical virilization, and must
then decide whether to live as men or women. Peggy T. Cohen-Kettenis,
a professor of gender development and psychopathology, found that 39 to
Fall 2016 ~ 37
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
64% of individuals with these deficiencies who are raised as girls change
to live as men in adolescence and early adulthood, and she also reported
that “the degree of external genital masculinization at birth does not seem
to be related to gender role changes in a systematic way.”
The twin studies reviewed earlier may shed light on the role of
maternal hormonal influences, since both identical and fraternal twins are
exposed to similar maternal hormonal influences in utero. The relatively
weak concordance rates in the twin studies suggest that prenatal hor-
mones, like genetic factors, do not play a strongly determinative role in
sexual orientation. Other attempts at finding significant hormonal influ-
ences on sexual development have likewise been mixed, and the salience
of the findings is not yet clear. Since direct studies of prenatal hormonal
influences on sexual development are methodologically difficult, some
studies have tried to develop models whereby differences in prenatal hor-
monal exposure can be inferred indirectly by measuring subtle morpho-
logical changes or by examining hormonal disorders that are present later
during development.
For example, one rough proxy of prenatal testosterone levels used by
researchers is the ratio between the length of the second finger (index
finger) and the fourth finger (ring finger), which is commonly called the
“2D:4D ratio.” Some evidence suggests that the ratio may be influenced
by prenatal exposure to testosterone, such that in males higher levels of
exposure to testosterone cause shorter index fingers relative to the ring
finger (or having a low 2D:4D ratio), and vice versa.
According to one
hypothesis, homosexual men may have a higher 2D:4D ratio (closer to the
ratio found in females than in heterosexual males), while another hypoth-
esis suggests the opposite, that homosexual men may be hypermasculin-
ized by prenatal testosterone, resulting in a lower ratio than in hetero-
sexual men. For women, the hypothesis for homosexuality that they have
been hypermasculinized (lower ratio, higher testosterone) has also been
proposed. Several studies comparing this trait in homosexually versus
heterosexually identified men and women have shown mixed results.
A study published in Nature in 2000 found that in a sample of 720
California adults, the right-hand 2D:4D ratio of homosexual women was
significantly more masculine (that is, the ratio was smaller) than that of
heterosexual women and did not differ significantly from that of hetero-
sexual men.
This study also found no significant difference in mean
2D:4D ratio between heterosexual and homosexual men. Another study
that year, which used a relatively small sample of homosexual and het-
erosexual men from the United Kingdom, reported a lower 2D:4D (that
38 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
is, more masculine) ratio in homosexual men.
A 2003 study using a
London-based sample also found that homosexual men had a lower 2D:4D
ratio than heterosexuals,
while two other studies with samples from
California and Texas showed higher 2D:4D ratios for homosexual men.
A 2003 twin study compared seven female monozygotic twin pairs
discordant for homosexuality (one twin was lesbian) and five female
monozygotic twin pairs concordant for homosexuality (both twins were
In the twin pairs discordant for sexual orientation, the indi-
viduals identifying as homosexual had significantly lower 2D:4D ratios
than their twins, whereas the concordant twins showed no difference.
The authors interpreted this result as suggesting that “low 2D:4D ratio
is a result of differences in prenatal environment.”
Finally, a 2005 study
of 2D:4D ratios in an Austrian sample of 95 homosexual and 79 hetero-
sexual men found that the 2D:4D ratios of heterosexual men were not
significantly different from those of homosexual men.
After reviewing
the several studies on this trait, the authors conclude that “more data are
essential before we can be sure whether there is a 2D:4D effect for sexual
orientation in men when ethnic variation is controlled for.”
Much research has examined the effects of prenatal hormones on
behavior and brain structure. Again, these results come primarily from
studies of non-human primates, but the study of disorders of sexual
development has provided helpful insights into the effects of hormones on
sexual development in humans. Since hormonal influences typically occur
during time-sensitive periods of development, when their effects manifest
physically, it is reasonable to assume that organizational effects of these
early, time-linked hormonal patterns are likely to direct aspects of neural
development. Neuroanatomical connectivity and neurochemical sensitivi-
ties may be among such influences.
In 1983, Günter Dörner and colleagues performed a study investi-
gating whether there is any relationship between maternal stress during
pregnancy and later sexual identity of their children, interviewing two
hundred men about stressful events that may have occurred to their moth-
ers during their prenatal lives.
Many of these events occurred as a con-
sequence of World War II. Of men who reported that their mothers had
experienced moderately to severely stressful events during pregnancy,
65% were homosexual, 25% were bisexual, and 10% were heterosexual.
(Sexual orientation was assessed using the Kinsey scale.) However, more
recent studies have shown much smaller or no significant correlations.
In a 2002 prospective study on the relationship between sexual orienta-
tion and prenatal stress during the second and third trimesters, Hines
Fall 2016 ~ 39
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
and colleagues found that stress reported by mothers during pregnancy
showed “only a small relationship” to male-typical behaviors in their
daughters at the age of 42 months, “and no relationship at all” to female-
typical behaviors in their sons.
In summary, some forms of prenatal hormone exposure, particularly
CAH in females, are associated with differences in sexual orientation,
while other factors are often important in determining the physical and
psychological effects of those exposures. Hormonal conditions that con-
tribute to disorders of sex development may contribute to the develop-
ment of non-heterosexual orientations in some individuals, but this does
not demonstrate that such factors explain the development of sexual
attractions, desires, and behaviors in the majority of cases.
Sexual Orientation and the Brain
There have been several studies examining neurobiological differences
between individuals who identify as heterosexual and those who iden-
tify as homosexual. This work began with neuroscientist Simon LeVay’s
1991 study that reported biological differences in the brains of gay men
as compared to straight men specifically, a difference in volume in a
particular cell group of the interstitial nuclei of the anterior hypothala-
mus (INAH3).
Later work by psychiatrist William Byne and colleagues
showed more nuanced findings: “In agreement with two prior studies. ..
we found INAH3 to be sexually dimorphic, occupying a significantly
greater volume in males than females. In addition, we determined that the
sex difference in volume was attributable to a sex difference in neuronal
number and not in neuronal size or density.”
The authors noted that,
Although there was a trend for INAH3 to occupy a smaller volume in
homosexual men than in heterosexual men, there was no difference in the
number of neurons within the nucleus based on sexual orientation.” They
speculated that “postnatal experience” may account for the differences in
volume in this region between homosexual and heterosexual men, though
this would require further research to confirm.
They also noted that
the functional significance of sexual dimorphism in INAH3 is unknown.
The authors conclude: “Based on the results of the present study as well
as those of LeVay (1991), sexual orientation cannot be reliably predicted
on the basis of INAH3 volume alone.”
In 2002, psychologist Mitchell S.
Lasco and colleagues published a study examining a different part of the
brain the anterior commissure and found that there were no signifi-
cant differences in that area based either on sex or sexual orientation.
40 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
Other studies have since been conducted to ascertain structural or
functional differences between the brains of heterosexual and homosexual
individuals (using a variety of criteria to define these categories). Findings
from several of these studies are summarized in a 2008 commentary pub-
lished in the Proceedings of the National Academy of Sciences.
Research of
this kind, however, does not seem to reveal much of relevance regarding the
etiology or biological origins of sexual orientation. Due to inherent limi-
tations, this research literature is fairly unremarkable. For example, in one
study functional MRI was used to measure activity changes in the brain
when pictures of men and women were shown to subjects, finding that
viewing a female face produced stronger activity in the thalamus and orbi-
tofrontal cortex of heterosexual men and homosexual women, whereas in
homosexual men and heterosexual women these structures reacted more
strongly to the face of a man.
That the brains of heterosexual women
and homosexual men reacted distinctively to the faces of men, whereas the
brains of heterosexual men and homosexual women reacted distinctively
to the faces of women, is a finding that seems rather trivial with respect
to understanding the etiology of homosexual attractions. In a similar vein,
one study reported different responses to pheromones between homosex-
ual and heterosexual men,
and a follow-up study showed a similar find-
ing in homosexual compared to heterosexual women.
Another study
showed differences in cerebral asymmetry and functional connectivity
between homosexual and heterosexual subjects.
While findings of this kind may suggest avenues for future investiga-
tion, they do not move us much closer to an understanding of the biologi-
cal or environmental determinants of sexual attractions, interests, prefer-
ences, or behaviors. We will say more about this below. For now, we will
briefly illustrate a few of the inherent limitations in this area of research
with the following hypothetical example. Suppose we were to study the
brains of yoga teachers and compare them to the brains of bodybuilders.
If we search long enough, we will eventually find statistically significant
differences in some area of brain morphology or brain function between
these two groups. But this would not imply that such differences deter-
mined the different life trajectories of the yoga teacher and the body-
builder. The brain differences could have been the result, rather than the
cause, of distinctive patterns of behavior or interests.
Consider another
example. Suppose that gay men tend to have less body fat than straight
men (as indicated by lower average scores on body mass indices). Even
though body mass is, in part, determined by genetics, we could not claim
based on this finding that there is some innate, genetic cause of both body
Fall 2016 ~ 41
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
mass and homosexuality at work. It could be the case, for instance, that
being gay is associated with a diet that lowers body mass. These examples
illustrate one of the common problems encountered in the popular inter-
pretation of such research: the suggestion that the neurobiological pattern
determines a particular behavioral expression.
With this overview of studies on biological factors that might influ-
ence sexual attraction, preferences, or desires, we can understand the
rather strong conclusion by social psychologist Letitia Anne Peplau
and colleagues in a 1999 review article: “To recap, more than 50 years
of research has failed to demonstrate that biological factors are a major
influence in the development of women’s sexual orientation.. .. Contrary
to popular belief, scientists have not convincingly demonstrated that biol-
ogy determines women’s sexual orientation.”
In light of the studies we
have summarized here, this statement could also be made for research on
male sexual orientation, however this concept is defined.
Misreading the Research
There are some significant built-in limitations to what the kind of empiri-
cal research summarized in the preceding sections can show. Ignoring
these limitations is one of the main reasons the research is routinely
misinterpreted in the public sphere. It may be tempting to assume, as we
just saw with the example of brain structure, that if a particular biological
profile is associated with some behavioral or psychological trait, then that
biological profile causes that trait. This reasoning relies on a fallacy, and
in this section we explain why, using concepts from the field of epidemiol-
ogy. While some of these issues are rather technical in detail, we will try
to explain them in a general way that is accessible to the non-specialist
Suppose for the sake of illustration that one or more differences in
a biological trait are found between homosexual and heterosexual men.
That difference could be a discrete measure (call this D) such as presence
of a genetic marker, or it could be a continuous measure (call this C) such
as the average volume of a particular part of the brain.
Showing that a risk factor significantly increases the chances of a
particular health outcome or a behavior might give us a clue to develop-
ment of that health outcome or that behavior, but it does not provide
evidence of causation. Indeed, it may not provide evidence of anything
but the weakest of correlations. The inference is sometimes made that if
it can be shown that gay men and straight men differ significantly in the
42 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
probability that D is present (whether a gene, a hormonal factor, or some-
thing else), no matter how low that probability, then this finding suggests
that being gay has a biological basis. But this inference is unwarranted.
Doubling (or even tripling or quadrupling) the probability of a relatively
rare trait can have little value in terms of predicting who will or will not
identify as gay.
The same would be true for any continuous variable (C). Showing a
significant difference at the mean or average for a given trait (such as the
volume of a particular brain region) between men who identify as het-
erosexual and men who identify as homosexual does not suffice to show
that this average difference contributes to the probability of identifying as
heterosexual or homosexual. In addition to the reasons explained above, a
significant difference at the means of two distributions can be consistent
with a great deal of overlap between the distributions. That is, there may
be virtually no separation in terms of distinguishing between some indi-
vidual members of each group, and thus the measure would not provide
much predictability for sexual orientation or preference.
Some of these issues could, in part, be addressed by additional meth-
odological approaches, such as the use of a training sample or cross-
validation procedures. A training sample is a small sample used to develop
a model (or hypothesis); this model is then tested on a larger independent
sample. This method avoids testing a hypothesis on the same data used
to develop the hypothesis. Cross-validation includes procedures used to
examine whether a statistically significant effect is really there or just due
to chance. If one wants to show the result did not occur by chance (and if
the sample is large), one can run the same tests on a random split of the
relevant sample. After finding a difference in the prevalence of trait D or C
between a gay sample and a straight sample, researchers could randomly
split the gay sample into two groups and then show that these two groups
do not differ regarding D or C. Suppose one finds five differences out of
100 comparing gay to straight men in the overall samples, then finds five
differences out of 100 when comparing the split gay samples. This would
cast additional doubt on the initial finding of a difference between the
means of gay and straight individuals.
Sexual Abuse Victimization
Whereas the preceding discussion considered the part that biological fac-
tors might play in the development of sexual orientation, this section will
summarize evidence that a particular environmental factor childhood
Fall 2016 ~ 43
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
sexual abuse is reported significantly more often among those who later
identify as homosexual. The results presented below raise the question
whether there is an association between sexual abuse, particularly in child-
hood, and later expressions of sexual attraction, behavior, or identity. If so,
might child abuse increase the probability of having a non-heterosexual
Correlations, at least, have been found, as we will summarize below.
But we should note first that they might be accounted for by one or more
of the following conjectures:
1. Abuse might contribute to the development of non-hetero-
sexual orientation.
2. Children with (signs of future) non-heterosexual tendencies
might attract abusers, placing them at elevated risk.
3. Certain factors might contribute to both childhood sexual
abuse and non-heterosexual tendencies (for instance, a dysfunc-
tional family or an alcoholic parent).
It should be kept in mind that these three hypotheses are not mutually
exclusive; all three, and perhaps others, might be operative. As we sum-
marize the studies on this issue, we will try to evaluate each of these
hypotheses in light of current scientific research.
Behavioral and community health professor Mark S. Friedman and
colleagues conducted a 2011 meta-analysis of 37 studies from the United
States and Canada examining sexual abuse, physical abuse, and peer vic-
timization in heterosexuals as compared to non-heterosexuals.
results showed that non-heterosexuals were on average 2.9 times more
likely to report having been abused as children (under 18 years of age).
In particular, non-heterosexual males were 4.9 times likelier and non-
heterosexual females, 1.5 times likelier than their heterosexual coun-
terparts to report sexual abuse. Non-heterosexual adolescents as a whole
were 1.3 times likelier to indicate physical abuse by parents than their
heterosexual peers, but gay and lesbian adolescents were only 0.9 times as
likely (bisexuals were 1.4 times as likely). As for peer victimization, non-
heterosexuals were 1.7 times likelier to report being injured or threatened
with a weapon or being attacked.
The authors note that although they hypothesized that the rates of
abuse would decrease as social acceptance of homosexuality rose, “dispari-
ties in prevalence rates of sexual abuse, parental physical abuse, and peer
44 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
victimization between sexual minority and sexual nonminority youths
did not change from the 1990s to the first decade of the 2000s.”
these authors cite authorities who claim that sexual abuse does not “cause
individuals to become gay, lesbian, or bisexual,”
their data do not give
evidence against the hypothesis that childhood sexual abuse might affect
sexual orientation. On the other hand, the causal path could be in the
opposite direction or bi-directional. The evidence does not refute or sup-
port this conjecture; the study’s design is not capable of shedding much
light on the question of directionality.
The authors invoke a widely-cited hypothesis to explain the higher
rates of sexual abuse among non-heterosexuals, the hypothesis that
“sexual minority individuals are. .. more likely to be targeted for sexual
abuse, as youths who are perceived to be gay, lesbian, or bisexual are more
likely to be bullied by their peers.”
The two conjectures that abuse
is a cause and that it is a result of non-heterosexual tendencies are
not mutually exclusive: abuse may be a causal factor in the development
of non-heterosexual attractions and desires, and at the same time non-
heterosexual attractions, desires, and behaviors may increase the risk of
being targeted for abuse.
Community health sciences professor Emily Faith Rothman and col-
leagues conducted a 2011 systematic review of the research investigat-
ing the prevalence of sexual assault against people who identify as gay,
lesbian, or bisexual in the United States.
They examined 75 studies
(25 of which used probability sampling) involving a total of 139,635 gay
or bisexual (GB) men and lesbian or bisexual (LB) women, which mea-
sured the prevalence of victimization due to lifetime sexual assault (LSA),
childhood sexual assault (CSA), adult sexual assault (ASA), intimate
partner sexual assault (IPSA), and hate-crime-related sexual assault (HC).
Although the study was limited by not having a heterosexual control
group, it showed alarmingly high rates of sexual assault, including child-
hood sexual assault, for this population, as summarized in Table 1.
Using a multi-state probability-based sample in a 2013 study, psy-
chologist Judith Anderson and colleagues compared differences in adverse
childhood experiences including dysfunctional households; physical,
sexual, or emotional abuse; and parental discord among self-identified
homosexual, heterosexual, and bisexual adults.
They found that bisex-
uals had significantly higher proportions than heterosexuals of all adverse
childhood experience factors, and that gays and lesbians had significantly
higher proportions than heterosexuals of all these measures except paren-
tal separation or divorce. Overall, gays and lesbians had nearly 1.7 times,
Fall 2016 ~ 45
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
and bisexuals 1.6 times, the heterosexual rate of adverse childhood experi-
ences. The data for abuse are summarized in Table 2.
While this study, like some others we have discussed, may be limited
by recall bias that is, inaccuracies introduced by errors of memory it
has the merit of having a control group of self-identified heterosexuals
to compare with self-identified gay/lesbian and bisexual cohorts. In their
discussion of findings, the authors critique the hypothesis that childhood
trauma has a causal relationship to homosexual preferences. Among their
reasons for skepticism, they note that the vast majority of individuals who
suffer childhood trauma do not become gay or bisexual, and that gender-
nonconforming behavior may help explain the elevated rates of abuse.
However, it is plausible from these and related results to hypothesize
Table 1. Sexual Assault among Gay/Bisexual Men
and Lesbian/Bisexual Women
GB Men (%) LB Women (%)
CSA: 4.1 59.2 (median 22.7) CSA: 14.9 76.0 (median 34.5)
ASA: 10.8 44.7 (median 14.7) ASA: 11.3 53.2 (median 23.2)
LSA: 11.8 54.0 (median 30.4) LSA: 15.6 85.0 (median 43.4)
IPSA: 9.5 57.0 (median 12.1) IPSA: 3.0 45.0 (median 13.3)
HC: 3.0 19.8 (median 14.0) HC: 1.0 12.3 (median 5.0)
Sexual Abuse (%)
GLs Bisexuals Heterosexuals
29.7 34.9 14.8
Emotional Abuse (%)
GLs Bisexuals Heterosexuals
47.9 48.4 29.6
Physical Abuse (%)
GLs Bisexuals Heterosexuals
29.3 30.3 16.7
Table 2. Adverse Childhood Experiences among
Gays/Lesbians, Bisexuals, and Heterosexuals
46 ~ The New Atlantis
Special Report: Sexuality and Gender
that adverse childhood experiences may be a significant but not a
determinative factor in developing homosexual preferences. Further
studies are needed to see whether either or both hypotheses have merit.
A 2010 study by professor of social and behavioral sciences Andrea
Roberts and colleagues examined sexual orientation and risk of post-
traumatic stress disorder (PTSD) using data from a national epidemiological
face-to-face survey of nearly 35,000 adults.
Individuals were placed into
several categories: heterosexual with no same-sex attraction or partners
(reference group); heterosexual with same-sex attraction but no same-sex
partners; heterosexual with same-sex partners; self-identified gay/lesbian;
and self-identified bisexual. Among those reporting exposure to traumatic
events, gay and lesbian individuals as well as bisexuals had about twice
the lifetime risk of PTSD compared to the heterosexual reference group.
Differences were found in rates of childhood maltreatment and interpersonal
violence: gays, lesbians, bisexuals, and heterosexuals with same-sex partners
reported experiencing worse traumas during childhood and adolescence
than the reference group. The findings are summarized in Table 3.
Similar patterns emerged in a 2012 study by psychologist Brendan
Zietsch and colleagues that primarily focused on the distinct question of
whether common causal factors could explain the association between sexual
orientation in this study defined as sexual preference and depression.
In a community sample of 9,884 adult twins, the authors found that non-het-
erosexuals had significantly elevated prevalence of lifetime depression (odds
ratio for males 2.8; odds ratio for females 2.7). As the authors point out, the
data raised questions about whether higher rates of depression for non-het-
erosexuals could be explained, in their entirety, by the social stress hypoth-
esis (the idea, discussed in depth in Part Two of this report, that social stress
Table 3. Childhood Exposure to Maltreatment
or Interpersonal Violence (before Age 18)
Women Men
49.2% of lesbians 31.5% of gays
51.2% of bisexuals Approximately 32% of bisexuals
40.9% of heterosexuals with same-sex
27.9% of heterosexuals with same-sex
21.2% of heterosexuals 19.8% of heterosexuals
Copyright 2016. All rights reserved. See for more information.
Fall 2016 ~ 47
Part One: Sexual Orientation
experienced by sexual minorities accounts for their elevated risks of poor
mental health outcomes). Heterosexuals with a non-heterosexual twin had
higher rates of depression (39%) than heterosexual twin pairs (31%), sug-
gesting that genetic, familial, or other factors may play a role.
The authors note that “in both males and females, significantly higher
rates of non-heterosexuality were found in participants who experienced
childhood sexual abuse and in those with a risky childhood family environ-
Indeed, 41% of non-heterosexual males and 42% of non-hetero-
sexual females reported childhood family dysfunction, compared to 24% and
30% of heterosexual males and females, respectively. And 12% of non-het-
erosexual males and 24% of non-heterosexual females reported sexual abuse
before the age of 14, compared with 4% and 11% of heterosexual males and
females, respectively. The authors are careful to emphasize that their find-
ings should not be interpreted as disproving the social stress hypothesis, but
suggest that there may be other factors at work. Their findings do, however,
suggest there could be common etiological factors for depression and non-
heterosexual preferences, as they found that genetic factors account for 60%
of the correlation between sexual orientation and depression.
In a 2001 study, psychologist Marie E. Tomeo and colleagues noted that
the previous literature had consistently found increased rates of reported
childhood molestation in the homosexual population, with somewhere
between 10% and 46% reporting that they had experienced childhood sexual
The authors found that 46% of homosexual men and 22% of homo-
sexual women reported that they had been molested by a person of the same
gender, as compared with 7% of heterosexual men and 1% of heterosexual
women. Moreover, 38% of homosexual women interviewed did not identify
as homosexual until after the abuse, while the authors report conflicting
figures 68% in one part of the paper and (by inference) 32% in another
for the number of homosexual men who did not identify as homosexual until
after the abuse. The sample for this study was relatively small, only 267
individuals; also, the “sexual contact” measure of abuse in the survey was
somewhat vague, and the subjects were recruited from participants in gay
pride events in California. But the authors state that “it is most unlikely that
all the present findings apply only to homosexual persons who go to homo-
sexual fairs and volunteer to participate in questionnaire research.
In 2010, psychologists Helen Wilson and Cathy S. Widom published a
prospective 30-year follow-up study one that looked at children who had
experienced abuse or neglect between 1961 and 1971, and then followed up
with those children after 30 years to ascertain whether physical abuse,
sexual abuse, or neglect in childhood increased the likelihood of same-sex
Copyright 2016. All rights reserved. See for more information.
48 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
sexual relationships later in life.
An original sample of 908 abused and/
or neglected children was matched with a non-maltreated control group
of 667 individuals (matched for age, sex, race or ethnicity, and approxi-
mate socioeconomic status). Homosexuality was operationalized as anyone
who had cohabited with a same-sex romantic partner or had a same-sex
sexual partner, which made up 8% of the sample. Among these 8%, most
individuals also reported having had opposite-sex partners, suggesting
high rates of bisexuality or fluidity in sexual attractions or behaviors. The
study found that those who reported histories of childhood sexual abuse
were 2.8 times more likely to report having had same-sex sexual relation-
ships, though the “relationship between childhood sexual abuse and same-
sex sexual orientation was significant only for men.”
This finding sug-
gested that boys who are sexually abused may be more likely to establish
both heterosexual and homosexual relationships.
The authors advised caution in interpreting this result, because the
sample size of sexually abused men was small, but the association remained
statistically significant when they controlled for total lifetime number of
sexual partners and for engaging in prostitution. The study was also
limited by a definition of sexual orientation that was not sensitive to how
participants identified themselves. It may have failed to capture people
with same-sex attractions but no same-sex romantic relationship history.
The study had two notable methodological strengths. The prospective
design is better suited for evaluating causal relationships than the typical
retrospective design. Also, the childhood abuse recorded was documented
when it occurred, thus mitigating recall bias.
Having examined the statistical association between childhood sexual
abuse and later homosexuality, we turn to the question of whether the
association suggests causation.
A 2013 analysis by health researcher Andrea Roberts and colleagues
attempted to provide an answer to this question.
The authors noted
that while studies show 1.6 to 4 times more reported childhood sexual and
physical abuse among gay and lesbian individuals than among heterosexu-
als, conventional statistical methods cannot demonstrate a strong enough
statistical relationship to support the argument of causation. They argued
that a sophisticated statistical method called “instrumental variables,”
imported from econometrics and economic analysis, could increase the
level of association.
(The method is somewhat similar to the method of
“propensity scores,” which is more sophisticated and more familiar to pub-
lic health researchers.) The authors applied the method of instrumental
variables to data collected from a nationally representative sample.
Fall 2016 ~ 49
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
They used three dichotomous measures of sexual orientation: any vs.
no same-sex attraction; any vs. no lifetime same-sex sexual partners; and
lesbian, gay, or bisexual vs. heterosexual self-identification. As in other
studies, the data showed associations between childhood sexual abuse or
maltreatment and all three dimensions of non-heterosexuality (attraction,
partners, identity), with associations between sexual abuse and sexual
identity being the strongest.
The authors’ instrumental variable models suggested that early sexual
abuse increased the predicted rate of same-sex attraction by 2.0 percent-
age points, same-sex partnering by 1.4 percentage points, and same-sex
identity by 0.7 percentage points. The authors estimated the rate of
homosexuality that might be attributable to sexual abuse “using effect
estimates from conventional models” and found that on conventional effect
estimates, “9% of same-sex attraction, 21% of any lifetime same-sex sexual
partnering, and 23% of homosexual or bisexual identity was due to child-
hood sexual abuse.”
We should note that these correlations are cross-
sectional: they compare groups of people to groups of people, rather than
model the course of individuals over time. (A study design with a time-
series analysis would give the strongest statistical support to the claim
of causality.) Additionally, these results have been strongly criticized on
methodological grounds for having made unjustified assumptions in the
instrumental variables regression; a commentary by Drew H. Bailey and J.
Michael Bailey claims, “Not only do Roberts et al.’s results fail to provide
support for the idea that childhood maltreatment causes adult homosexu-
ality, the pattern of differences between males and females is opposite what
should be expected based on better evidence.”
Roberts and colleagues conclude their study with several conjec-
tures to explain the epidemiological associations. They echo suggestions
made elsewhere that sexual abuse perpetrated by men might cause boys
to think they are gay or make girls averse to sexual contact with men.
They also conjecture that sexual abuse might leave victims feeling stig-
matized, which in turn might make them more likely to act in ways that
are socially stigmatized (as by engaging in same-sex sexual relationships).
The authors also point to the biological effects of maltreatment, citing
studies that show that “quality of parenting” can affect chemical and hor-
monal receptors in children, and hypothesizing that this might influence
sexuality “through epigenetic changes, particularly in the stria terminalis
and the medial amygdala, brain regions that regulate social behavior.”
They also mention the possibilities that emotional numbing caused by
maltreatment may drive victims to seek out risky behaviors associated
50 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 2016. All rights reserved. See for more information.
with same-sex sexuality, or that same-sex attractions and partnering may
result from “the drive for intimacy and sex to repair depressed, stressed,
or angry moods,” or from borderline personality disorder, which is a risk
factor in individuals who have been maltreated.
In short, while this study suggests that sexual abuse may sometimes
be a causal contributor to having a non-heterosexual orientation, more
research is needed to elucidate the biological or psychological mechanisms.
Without such research, the idea that sexual abuse may be a causal factor
in sexual orientation remains speculative.
Distribution of Sexual Desires and Changes Over Time
However sexual desires and interests develop, there is a related issue that
scientists debate: whether sexual desires and attractions tend to remain
fixed and unalterable across the lifespan of a person or are fluid and
subject to change over time but tend to become fixed after a certain age
or developmental period. Advocates of the “born that way” hypothesis, as
mentioned earlier, sometimes argue that a person is not only born with a
sexual orientation but that that orientation is immutable; it is fixed for life.
There is now considerable scientific evidence that sexual desires,
attractions, behaviors, and even identities can, and sometimes do, change
over time. For findings in this area we can turn to the most comprehensive
study of sexuality to date, the 1992 National Health and Social Life Survey
conducted by the National Opinion Research Center at the University of
Chicago (NORC).
Two important publications have appeared using data
from NORC’s comprehensive survey: The Social Organization of Sexuality:
Sexual Practices in the United States, a large tome of data intended for the
research community, and Sex in America: A Definitive Survey, a smaller
and more accessible book summarizing the findings for the general pub-
These books present data from a reliable probability sample of the
American population between ages 18 and 59.
According to data from the NORC survey, the estimated prevalence
of non-heterosexuality, depending on how it was operationalized, and on
whether the subjects were male or female, ranged between roughly 1%
and 9%.
The NORC studies added scientific respectability to sexual
surveys, and these findings have been largely replicated in the United
States and abroad. For example, the British National Survey of Sexual
Attitudes and Lifestyles (Natsal) is probably the most reliable source of
information on sexual behavior in that country a study conducted every
ten years since 1990.
Fall 2016 ~ 51
Part One: Sexual Orientation
Copyright 2016. All rights reserved. See for more information.
The NORC study also suggested ways in which sexual behaviors and
identities can vary significantly under different social and environmental
circumstances. The findings revealed, for example, a sizable difference in
rates of male homosexual behavior among individuals who spent their
adolescence in rural as compared to large metropolitan cities in America,
suggesting the influence of social and cultural environments. Whereas
only 1.2% of males who had spent their adolescence in a rural environ-
ment responded that they had had a male sexual partner in the year of the
survey, those who had spent adolescence living in metropolitan areas were
close to four times (4.4%) more likely to report that they had had such an
From these data one cannot infer differences between these
environments in the prevalence of sexual interests or attractions, but the
data do suggest differences in sexual behaviors. Also of note is that women
who attended college were nine times more likely to identify as lesbians
than women who did not.
Moreover, other population-based surveys suggest that sexual desire
may be fluid for a considerable number of individuals, especially among
adolescents as they mature through the early stages of adult development.
In this regard, opposite-sex attraction and identity seem to be more stable
than same-sex or bisexual attraction and identity. This is suggested by
data from the National Longitudinal Study of Adolescent to Adult Health
(the Add Health” study discussed earlier). This prospective longitudinal
study of a nationally representative sample of U.S. adolescents starting in
grades 7 12 began during the 1994 1995 school year, and followed the
cohort into young adulthood, with four follow-up interviews (referred
to as Waves I, II, III, IV in the literature).
The most recent was in
2007 2008, when the sample was aged 24 32.
Same-sex or both-sex romantic attractions were quite prevalent in the
study’s first wave, with rates of approximately 7% for the males and 5% for
the females.
However, 80% of the adolescent males who had reported
same-sex attractions at Wave I later identified themselves as exclusively
heterosexual as young adults at Wave IV.
Similarly, for adolescent
males who, at Wave I, reported romantic attraction to both sexes, over
80% of them reported no same-sex romantic attraction at Wave III.
The data for the females surveyed were similar but less striking: for ado-
lescent females who had both-sex attractions at Wave I, more than half
reported exclusive attraction to males at Wave III.
J. Richard Udry, the director of Add Health for Waves I, II, and III,
was among the first to point out the fluidity and instability of romantic
attraction between the first two waves. He reported that among boys who
52 ~ The New Atlantis
Special Report: Sexuality and Gender
Copyright 20<