FOLLOW-UP STUDY OF BOYS
A thesis submitted in conformity with the requirements
for the degree of Doctor of Philosophy
Department of Human Development and Applied Psychology
Ontario Institute for Studies in Education
University of Toronto
© Copyright by Devita
Devita Singh, Doctor of Philosophy
Department of Human Development and Applied Psychology
University of Toronto
This study provided information on the long term psychosexual and psychiatric outcomes of 139 boys with gender identity disorder (GID). Standardized assessment data in childhood (mean age, 7.49 years; range, 3–12 years) and at follow-up (mean age, 20.58 years; range, 13–39 years) were used to evaluate gender identity and sexual orientation outcome. At follow-up, 17 participants (12.2%) were judged to have persistent gender dysphoria. Regarding sexual orientation, 82 (63.6%) participants were classified as bisexual/ homosexual in fantasy and 51 (47.2%) participants were classified as bisexual/homosexual in behavior. The remaining participants were classified as either heterosexual or asexual. With gender identity and sexual orientation combined, the most common long-term outcome was desistence of GID with a bisexual/homosexual sexual orientation followed by desistence of GID with a heterosexual sexual orientation. The rates of persistent gender dysphoria and bisexual/homosexual sexual orientation were substantially higher than the base rates in the general male population. Childhood assessment data were used to identify within-group predictors of variation in gender identity and sexual orientation outcome. Social class and severity of cross-gender behavior in childhood were significant predictors of gender identity outcome. Severity of childhood cross-gender behavior was a significant predictor of sexual orientation at follow-up. Regarding psychiatric functioning, the heterosexual desisters reported significantly less behavioral and psychiatric difficulties compared to the bisexual/homosexual persisters and, to a lesser extent, the bisexual/homosexual desisters. Clinical and theoretical implications of these follow-up data are discussed.
There are a number of individuals who have contributed instrumentally in the various stages of the dissertation project and to whom I would like to express my most sincere gratitude.
First, to my supervisor, Dr. Ken Zucker, there are no words to express what your continued mentorship has meant to me. Thank you for graciously accepting me as a student seven years ago. Your impact in my life extends far beyond research and clinical training–you also helped me to develop as a person and I feel privileged to have worked with you. You have equally inspired, challenged, and, certainly, frustrated me. Looking back, however, it has been the most insightful and rewarding journey these past years. Your immense wisdom and insightfulness never ceased to amaze and inspire me, perhaps rivaled only by your astute attention to detail and editing skills. Thank you for allowing me to complete this fascinating project with you and for the endless hours discussing the meaning of it all. Perhaps most importantly, thank you for holding such high expectations of me. Your relentless ability to “push” me (and prevent me from “whining”) caused me endless grief, but alas helped to create this wonderful piece of work and fostered my growth in numerous ways.
I sincerely thank my thesis committee members for their invaluable contributions and insightful feedback that not only shaped this project in interesting ways, but also enhanced my learning. Dr. Michele Peterson-Badali, you have been instrumental in my Ph.D. research, first as the second reader for my M.A. thesis and now as a committee member for my dissertation. Your insight, feedback, and encouragement have been undoubtedly helpful, especially as I weathered seemingly insurmountable obstacles. Dr. Katreena Scott, I thank you for not only your intellectual commitment to this endeavor, but your ability to help me always remember to see the forest for the trees. I greatly appreciate your selflessness in always making time for me and your incredible capacity to problem solve. Dr. Susan Bradley, I am most grateful for your clinical insights. The commitment, empathy, and skills you bring to clinical work with children and their families have inspired me and provided me with skills which have enabled me to grow as a clinician. Dr. Lana Stermac, thank you for coming on board and offering your insights. Finally, to my external examiner, Dr. Michael Bailey, it was indeed my pleasure to have your insights and reflections on this project. Since I first read your book, I recognized your passion for understanding the lives of gender dysphoric children and was immensely pleased that you agreed to take on this role.
I express sincere gratitude and appreciation to the study participants, for without you this project could not come have come to fruition. Thank you for the hours spent sharing your stories–your candor made data collection enjoyable and added much depth and complexity to this project. Thank you to members of the Gender Identity Service who have been important supports and to the research assistants who tolerated my need for perfection as you entered the large volume of data needed for this project. I would also like to thank Dr. Ray Blanchard and Dr. James Cantor for your statistical support and clinical insights.
Finally, I think my family and friends. This arduous journey would not have been possible without your warmth, support, and patience with the inevitable stress associated with this project. To old and dear friends–Tomoko and Heidy–we have shared countless memories throughout the years and your support helped me complete this journey. Hamed, there are no words! You made so much possible and I am forever grateful for our friendship and for your kindness when I needed it most. Brian, thank you for being an amazing friend and the most patient person I have ever met. The many, many hours you spent listening to my venting and offering technical support made this journey substantially easier. Navin, your unconditional support has not gone unnoticed. To my parents, words cannot express the gratitude I have for unique ways in which you have both supported and encouraged me. Dad, thank you for always understanding and challenging me, and mom, thank you for inspiring me to be better than I think I am capable of. Finally, but by no means least, I sincerely thank my sister, Wanita. You get me in ways no one else can and made the most difficult phases of this project bearable.
Table of Contents
Gender identity is usually a central aspect of a person’s sense of self and, once developed, appears to be less malleable as development progresses (e.g., Egan & Perry, 2001; Ruble, Martin, & Berenbaum, 2006). The development of one’s gender identity and, by extension, gender role, is more than a cognitive milestone as it impacts on virtually all aspects of human functioning. In childhood, significant sex differences are seen in such behavioral domains as peer, toy, and activity preferences (e.g., Fagot, Leinbach, & Hagan, 1986; Ruble et al., 2006; Zucker, 2005b). In adolescence and adulthood, significant sex differences are seen in psychosocial domains such as interpersonal relational styles (e.g., Maccoby, 1998) and career choice (e.g., Lippa, 1998, 2005). One can imagine the profound implications on the person whose gender identity development departs from typical pathways and which results in much distress for the individual, a phenomenon that is recognized clinically as Gender Identity Disorder (GID).
Green and Money’s (1960) seminal article on boys with “incongruous gender role” was perhaps the first attempt in the literature to label, describe, and characterize the phenotype of young boys who exhibited a pattern of cross-gender behavior. Fourteen years later, Green’s (1974) seminal book, “Sexual Identity Conflict in Children and Adults,” provided a comprehensive description of children who were “discontent with the gender role expected of them.” Since the publication of these early works and, certainly, with the introduction of GID to the psychiatric nomenclature in the third edition of the Diagnostic and
Statistical Manual of Mental Disorders (DSM; American Psychiatric Association, 1980), tremendous progress has been made in understanding several aspects of this disorder. The phenomenology of GID is now fairly well documented (e.g., Cohen-Kettenis & Pfäfflin, 2003; Zucker, 2000, 2005a; Zucker & Bradley, 1995) and psychometrically robust assessment measures and procedures have been developed to allow for thorough diagnostic evaluation (for reviews, see Zucker, 1992, 2005b; Zucker & Bradley, 1995).
There are, however, some important gaps in the literature on children with GID, two of which are addressed in the current study. Although the natural history or outcome of boys with GID has received some empirical attention, the findings have not been consistent. Studies have generally found that not all boys with GID persist in having GID in adulthood and, in fact, the majority desist and have a homosexual sexual orientation. However, the rates of persistence of GID found across various studies have been variable, ranging from 2.3% to 30% (Green, 1987; Wallien & Cohen-Kettenis, 2008), but are considerably higher than the estimated prevalence of GID in the general population (Zucker & Lawrence, 2009). The reasons for this variability are a matter of conjecture as little is known about the factors that influence GID persistence into adolescence and adulthood. Further complicating the picture is the finding that some children with GID do grow up to have a heterosexual sexual orientation (e.g., Wallien & Cohen-Kettenis, 2008). Thus, given the variation observed in the long-term gender identity and sexual orientation outcome of boys with GID, it is important to examine childhood predictors of outcome in a sample of boys with atypical gender development. Second, very little is known about the long-term psychiatric outcome of boys with GID as the follow-up studies to date have primarily examined gender identity and sexual orientation outcomes.
The present study aimed to fill these gaps in the literature on boys with GID. First, the study examined the gender identity and sexual orientation outcome of boys with GID. Second, the study examined psychiatric outcome at follow-up. Third, using the extensive assessment data collected during childhood, in conjunction with the follow-up data, the study attempted to identify within-group childhood characteristics that were predictive of variations in gender identity and sexual orientation outcome in adolescence and adulthood.
This chapter will begin with a review of relevant psychosexual terminology. Information about the phenomenology of GID in children, adolescents and adulthood is summarized, including associated psychopathology in children, adolescents, and adults with GID. Current controversies in the field, particularly with regard to diagnosis and treatment, are summarized. This is followed by the results of studies that have examined gender identity and sexual orientation outcome in boys with GID. The literature on the relationship between childhood sex-typed behavior and sexual orientation in adulthood is reviewed. The remainder of the chapter includes a conceptual framework for the study within the field of developmental psychopathology. The chapter concludes with the rationale and goals for the present study. Given that the present study was a follow-up of boys with GID, the literature summarized in this chapter is primarily on that of boys with GID.
1.2.1 Sex and Gender
The terms sex and gender have been used interchangeably in the literature (Muehlenhard & Peterson, 2011). In this thesis, sex refers to whether a person is biologically male or female. Some of the common attributes that distinguish a person as male or female include the sex chromosomes, gonads, and internal and external genitalia (Vilain, 2000). Gender refers to psychological and behavioral characteristics associated with males and females (Kessler & McKenna, 1978; Ruble et al., 2006).
1.2.2 Gender Role
The term gender role, originally coined by Money (1955), refers to those behaviors, attitudes, and personality attributes that are consistent with cultural definitions and expectations of masculinity and femininity (Diamond, 2002; Zucker & Bradley, 1995). During childhood, gender role is commonly operationalized according to certain observable behaviors, referred to as sex-typed behaviors, including peer preference, interest in rough-and-tumble play, dress-up play, toy preference, and so on (Ruble et al., 2006). These gender role/sex-typed behaviors are often construed as indirect markers of a child’s gender identity as they are, on average, sex dimorphic (Zucker, 2005b). For example, boys tend to be more active than girls and engage more in rough-and-tumble play (Maccoby, 1998). Boys, on average, prefer to play with toy vehicles and weapons while girls, on average, prefer to play with dolls and toy household items (Berenbaum & Snyder, 1995). The diagnostic criteria for GID are defined, in part, by a profound and pervasive non-conformity to sex-typed behaviors (American Psychiatric Association, 2000). In children, quantitative measurement of sex-typed behavior is obtained through parent-report questionnaires as well as direct observation (Zucker, 2006a; Zucker & Bradley, 1995). In adolescents and adults with GID, descriptions of these childhood behaviors are obtained through retrospective self-report.
1.2.3 Gender Identity
Gender identity refers to a person’s basic sense of self as a male or female, that is, the inner experience of belonging to one gender (Fagot & Leinbach, 1985; Stoller, 1964). Most individuals develop a gender identity that is congruent with their biological sex, such that most biological females have a “female” gender identity and most biological males have a “male” gender identity. From a cognitive-developmental standpoint, gender identity refers to a child’s ability to not only accurately discriminate males from females, but to also correctly identify his or her own gender status as a boy or a girl. Within this framework, the development of gender identity is a cognitive milestone and is thought to represent the first stage in achieving gender constancy, that is, the understanding that being male or female is a biological characteristic and cannot be changed by altering superficial attributes, such as hair style or clothing (Fagot & Leinbach, 1985). In typically developing children, gender identity is established by age 3, at which point children can correctly answer the question, “Are you a boy or are you a girl?” Gender constancy is, on average, established by age 7. The development of a gender identity carries affective significance, as evidenced by the intensity of children’s “emotional commitment to doing what boys and girls are supposed to do” (Fagot & Leinbach, 1985, p. 687). It is also evidenced by the pride with which young children announce their gender and the embarrassment experienced if they are mislabelled by others (Zucker, 2005c).
1.2.4 Gender Dysphoria
The term gender dysphoria refers to the subjective experience of dissatisfaction and discontent about one’s biological status as male or female (Fisk, 1973; Zucker, 2006a). The concept of gender dysphoria is fundamental to clinical work with children, adolescents, and adults with GID as it captures the distress that results from the incongruity between one’s biological/assigned sex and internal experience of gender identity (Cohen-Kettenis & Gooren, 1999; Cohen-Kettenis & Pfäfflin, 2003; Money, 1994). In the DSM-IV criteria for GID (Appendix A), Criterion A1 (“repeatedly stated desire to be, or insistence that he or she is, the other sex”) can be considered the most concrete and direct expression of gender dysphoria (Zucker, 2010a). There are, however, developmental influences on the way in which children, adolescents and adults express their gender dysphoria. Cognitive development, language capacity, and social desirability are among some of the factors that may influence an individual’s expression of gender dysphoria.
Young children may actually state that they are members of the opposite sex (Cohen-Kettenis & Pfäfflin, 2003; for case examples, see Zucker, 1994; Zucker & Green, 1992). Although this misclassification could be related to the child’s age and cognitive development, it may also reflect the severity of the gender dysphoria if the child does truly believe he/she is of the opposite gender. However, most children with GID do not generally misclassify their sex and they know that they are male or female. Thus, when asked, “Are you a boy or a girl,” they answer correctly (Zucker et al., 1993); however, they will voice the desire to be of the opposite sex and find little that is positive about their own sex (Zucker & Green, 1992). Some children may express the wish to be of the opposite sex (i.e., implying that they know which sex they belong to) and simultaneously insist that they are of the opposite sex (e.g., Zucker, 2000, 2006c). These children may have confusion over gender constancy and may be uncertain whether changing aspects of one’s behavior (e.g., hair, clothing) will also change one’s gender. In fact, young children with GID demonstrate more cognitive confusion about gender compared to controls and appear to have a “developmental lag” in their gender constancy acquisition (Zucker et al., 1999).
There is some evidence that overt statements to be of the opposite sex tend to diminish with age. Older children with GID may not verbalize the wish to be of the opposite sex, perhaps for social desirability reasons (Bates, Skilbeck, Smith, & Benter, 1974). For example, they may have received feedback directly or indirectly from parents and peers regarding the appropriateness of their cross-gender wish (Bradley, 1999; Zucker & Bradley, 2004). During clinical evaluation, it is not uncommon for older children who are struggling with their gender identity to not endorse the desire to be of the opposite sex, but later in therapy reveal their cross-gender wishes once they have developed feelings of security in the therapeutic relationship. Indeed, during the preparation phase for the DSM-IV, this clinical observation served as the rationale for collapsing of the verbalized wish to be of the other sex with the other behavioral indicators of cross-gender identification (Bradley et al., 1991). In the DSM-III, the desire to be or insistence that one is of the opposite sex was a required criterion for the diagnosis. Zucker and Bradley (1995) conducted a re-analysis of data from Green’s (1987) follow-up study of effeminate boys and found some research data to support this clinical observation. Of the 60 boys seen in childhood for effeminate behavior, 47 were 3-9 years old and the remaining 13 were 10-12 years old. Of the 47 younger boys, 43 (91.5%) were reported by their mothers to occasionally or frequently state the wish to be a girl, compared to 9 (69.2%) of the boys in the older age group.
That overt statements to be of the opposite sex tend to diminish with age may not be a cross-national observation in children with GID. Wallien et al. (2009) compared a sample of children with GID seen at a specialized gender clinic in Toronto to a sample of children with GID seen at a gender clinic in the Netherlands on the Gender Identity Interview, which is a self-report measure of cognitive and affective gender identity confusion. An age effect was found for the Toronto patients such that the youngest children in the sample (5 years of age and younger) reported more cross-gender feelings than did the older children (6-12 years). In contrast, there was no significant association between age and scores on the Gender Identity Interivew for the Dutch patients. Thus, in the Dutch sample, older patients were as likely to report cross-gender feelings as younger patients.
For developmental reasons, there are limitations on children’s ability to think abstractly about gender and to evaluate the meaning of their gender dysphoria. It is not uncommon that when asked to list reasons for wanting to be of the opposite sex, young children with GID will provide concrete advantages. As an example, one 8-year old boy with GID asserted that it would be better to be a girl because they have better bands, such as the Spice Girls (Zucker & Bradley, 2004). The extent to which a child engages in sex-typed behaviors typical of the opposite sex and their rejection of activities and clothing typical of their own sex can be construed as surface indicators of gender dysphoria. Some children may experience discomfort with their sexual anatomy, which serves as another indicator of their felt gender dysphoria.
Adolescents and adults with GID are generally more straightforward than children in their expression about their unhappiness with their biological sex. In adolescents and adults with GID, verbalization of an intense discomfort with both primary and secondary sex characteristics and the desire for medical treatment (e.g., hormonal treatment, sex reassignment surgery) to address this discomfort is one of the most salient ways in which gender dysphoria is expressed (Cohen-Kettenis & Pfäfflin, 2010; Zucker, 2010a). However, the experience of gender dysphoria does not necessarily imply a desire for sex-reassignment surgery as some adolescents are only interested in hormonal treatment for their gender dysphoria (Cohen-Kettenis & Pfäfflin, 2010). Compared to children, adolescents and adults with GID, on average, have the cognitive and language capacity to think abstractly about their gender identity and gender subjectivity (i.e., beyond surface behaviors) and dialogue about the meaning of their dysphoria and its genesis. They are also more capable of discussing their anatomic dysphoria, which is often at the core of their distress (Bower, 2001). In such discussions, it is not uncommon for these individuals to express feeling trapped or having been born in the wrong body (e.g., Shaffer, 2005).
1.2.5 Sexual Orientation
Sexual orientation refers to a person’s erotic responsiveness to sexual stimuli and is typically measured along the dimension of the sex of the person to whom one is sexually attracted, that is, whether one is attracted to a member of the opposite sex (heterosexual sexual orientation), the same sex (homosexual sexual orientation), or both sexes (bisexual sexual orientation) (LeVay, 1993; Zucker, 2006a; Zucker & Bradley, 1995). Individuals who do not experience sexual attraction are referred to as asexual. Sexual orientation is often assessed with regard to at least two parameters: sexual orientation in fantasy and sexual orientation in behavior (Diamond, 1993; Green, 1987; Sell, 1997). The former refers to erotic fantasies experienced during sexually stimulating events, such as masturbation or while watching erotic pictures or movies, and the latter refers to actual sexual behavior, such as kissing and intercourse.
In contemporary sexology, the assessment of sexual orientation may include psychophysiological techniques to measure sexual arousal (Chivers, Rieger, Latty, & Bailey, 2004), semi-structured interviews (Kinsey, Pomeroy, & Martin, 1948), and self-report questionnaires (e.g., Zucker et al., 1996). From the foregoing definitions, that gender identity and sexual orientation are distinct constructs is obvious, yet, unfortunately, these terms are often conflated (for discussion, see Drescher, 2010b). As discussed later, synonymous use of these terms has implications for how one conceptualizes therapeutic approaches to help children with gender dysphoria feel more comfortable about their biological sex.
1.2.6 Sexual Identity
Sexual identity refers to an individual’s experience and conception of their sexual attraction (Diamond, 2000). Thus, it is the individual’s recognition, definition/labelling, and acceptance of themselves as heterosexual, bisexual, or homosexual (Diamond, 2002; Savin-Williams & Diamond, 2000). It is important to uncouple the construct of sexual orientation from the construct of sexual identity as they are not always synonymous. For example, a person may be predominantly sexually aroused by homosexual stimuli but may not necessarily regard or accept himself as “homosexual” (e.g., Bailey, 2009; LeVay, 2011; Ross, 1983).
1.2.7 Transgender and Transsexualism
The word transgender is an informal (i.e., non-diagnostic) term broadly used to subsume expressions of gender variance or gender nonconformity regardless of whether criteria for GID are met. Typically, individuals who are considered transgendered exhibit significant cross-gendered behaviors or identity. Some adolescents and adults use the term as a self-label of their gender identity (e.g., “I am transgendered” or “I am a trans person”) (Lawrence & Zucker, 2012). The term does not imply a particular sexual orientation (Drescher, 2010b).
Transsexualism, used sometimes synonymously with GID in adolescents and adults (e.g., Cohen-Kettenis & Pfäfflin, 2003; Simon, Zsolt, Fogd, & Czobor, 2011), is not an official diagnostic category in the DSM-IV (APA, 2000), although it is a diagnosis in the ICD-10 classification system that is given to adolescents and adults (World Health Organization, 1993). The term male-to-female transsexual (MtF) refers to biological males who identify as and desire to live (or are actually living) as females, but does not imply degree of transition to the female gender role (e.g., presenting socially, taking cross-sex hormones, received some type of surgical intervention) (Cohen-Kettenis & Gooren, 1999).
1.3 Phenomenology of Gender Identity Disorder
1.3.1 GID in Children
Although GID was only first introduced to the psychiatric nomenclature in the third edition of the DSM (American Psychiatric Association, 1980), its historical background extends over 150 years ago with case descriptions of individuals who experienced conflict over what is now referred to as their gender identity (see Zucker & Bradley, 1995). The incipient DSM-III diagnoses, Gender Identity Disorder of Childhood and Transsexualism, have since been modified into one overarching diagnosis, Gender Identity Disorder, with distinct criteria sets for children versus adolescents and adults, which reflect developmental variations in clinical presentation. In the present revised fourth edition of the DSM, the diagnosis (Appendix A) requires the presence of two components: (1) evidence of a strong and persistent cross-gender identification, which is generally manifested as the desire to be, or insistence that one is, the other sex and/or through the adoption of cross-sex behaviors, and (2) evidence of persistent discomfort with one’s biological sex and/or a sense of inappropriateness in the gender role of that sex, which, in males, is manifested through such behaviors as aversion towards rough-and-tumble play (American Psychiatric Association, 2000). The onset of cross-gender behaviors generally occurs during the preschool period, and signs of GID may be visible as early as two years of age (Cohen-Kettenis & Pfäfflin, 2003). Typically, the behavioral signs of GID precede overt statements about feeling like or wanting to be of the opposite sex (Green 1976, 1987). Parents of boys with GID often report that, from the moment their sons could talk, they insisted on wearing their mothers’ clothes and shoes and were predominantly interested in girls’ toys (Cohen-Kettenis & Gooren, 1999).
The phenomenology of GID in boys has been well-described elsewhere (e.g., Zucker & Bradley, 1995; Zucker & Cohen-Kettenis, 2008; Cohen-Kettenis & Pfäfflin, 2003). Boys with GID experience a strong psychological identification with the other sex, as evidenced by an array of sex-typed behaviors more characteristic of females and a rejection of sex-typed behaviors characteristic of boys (Green, 1976; Zucker & Bradley, 1995). Zucker (2002, 2008a) identified eight categories of sex-typed behavior relevant to the clinical picture of boys with GID: (1) identity statements, (2) dress-up play/cross dressing, (3) toy play, (4) roles in fantasy play, (5) peer relations, (6) motoric and speech characteristics, (7) involvement in rough-and-tumble play, and (8) statements about sexual anatomy. Boys with GID are usually interested in playing with girls’ toys, such as Barbie dolls, and are more intrigued by girls’ games and activities (e.g., skipping rope) than boys’ activities (e.g., hockey). Although they may be equally interested in play as same-aged peers, they tend to dislike and refrain from rough-and-tumble play and gravitate more towards female peers than male peers.
Some boys with GID do not typically object to wearing stereotypically masculine clothing (e.g., pants) in social settings, such as school, but will engage in cross-dressing when the setting is amenable (Zucker, 2010a) while others will insist on wearing female clothing in public and may demonstrate oppositional behavior if allowed to cross-dress only in private (e.g., Ehrensaft, 2011). In a recent approach to treatment of children with GID, some parents and clinicians allow and even encourage gender transition in childhood. Boys with GID treated within this approach demonstrate strong resistance to wearing male-typical clothing. Instead, they insist on dressing socially and privately in female-typical clothing and are allowed to do so as part of their social gender transition (for case examples, see English, 2011; Rosin, 2008; Spiegel, 2008).
In pretend and dress-up play, boys with GID often take on the female role (e.g., a princess or mother). As discussed earlier, some boys with GID express distress about being a boy and having a male body and some will verbalize the wish to be a girl (for clinical examples, see Zucker 2006c, Zucker et al., 2012b). Some boys with GID also experience “anatomic dysphoria,” which is a dislike of one’s genitals. They may verbalize this dislike, attempt to hide their genitals or pretend to have female genitalia (Cohen-Kettenis & Pfäfflin, 2003; Zucker & Bradley, 1995). Indeed, mothers of boys with GID rated their sons as experiencing higher dissatisfaction with their sexual anatomy compared to mothers of clinical and community control boys (Johnson et al., 2004; Lambert, 2009).
On a terminological note, in referring to children with marked cross-gender behavior, some authors avoid formal nosology (i.e., GID) and, instead, use alternative terms such as “gender nonconforming” or “gender variant” on the premise that these terms are less stigmatizing than GID. The issue in using non-standardized terminology is that the populations to which the term refers is less well defined. “Gender variant,” for example, may refer broadly to children who display varying degrees of cross-gender behavior, some of whom may meet diagnostic criteria for GID but others may not. Further complicating matters, it is not always clear in the literature whether “gender variant” was used as an alternative to GID or as a general term to represent all children with marked cross-gender behavior. Some authors, however, use these alternative terms when referring to children in non-clinical samples (e.g., Rieger, Linsenmeier, Bailey, & Gygax, 2008). In these cases, a gender nonconforming boy is one who is relatively feminine or less masculine compared with other boys and a gender conforming boy is one who is relatively unfeminine compared to other boys. In this thesis, GID is used to refer to children who meet criteria for GID and gender atypical/gender nonconforming is used when referring more broadly to children with marked cross-gender behavior whose GID status is unknown.
1.3.2 GID in Adolescents and Adults
A core characteristic of adolescents and adults with GID is psychological identification with the opposite sex (American Psychiatric Association, 2000). This generally manifests in the verbalization of an intense desire to be a member of the opposite sex. Some adolescents and adults with GID attempt to adopt the social role or “pass” as a member of the opposite sex through alteration of surface level physical attributes such as hair or clothing style. Another core characteristic of adolescents and adults with GID is discomfort with their sexual anatomy (anatomic dysphoria), though this is not experienced by all individuals with GID (American Psychiatric Association, 2000; Bradley & Zucker, 1997). Anatomic dysphoria may manifest as an interest in taking contra-sex hormones and, in some cases, receiving sex reassignment surgery to alter their physical appearance (Cohen-Kettenis, Delemarre-van de Waal, & Gooren, 2008; Cohen-Kettenis & Pfäfflin, 2003; de Vries, Steensma, Doreleijers, & Cohen-Kettenis, 2011a; Smith, van Goozen, & Cohen-Kettenis, 2001; Zucker, Bradley, Owen-Anderson, Kibblewhite, & Cantor, 2008; Zucker, 2006a). Treatment for adolescents with GID typically involves biomedical interventions that facilitate the transition from one gender to another. It is also recommended and, at times, required that the adolescent also engage in psychotherapy, though with a different treatment philosophy compared to psychotherapy for children with GID (Zucker et al., 2011; Zucker, Wood, Singh, & Bradley, 2012a). In general, this approach to treating adolescents and adults with GID is uncontroversial, though there may be cross-clinic/clinician variations in timing of treatment (e.g., minimum age for cross-sex hormones).
1.4 Prevalence of Gender Identity Disorder
More than 25 years ago, Meyer-Bahlburg (1985) characterized GID as a rare phenomenon. While there have been no formal epidemiological studies on the prevalence of GID in children, adolescents, and adults, other lines of evidence suggest that Meyer-Bahlburg’s observation still holds true. Information about the prevalence of cross-gender behavior in children has come from studies using the Child Behavior Checklist (CBCL) (Achenbach & Edelbrock, 1983), a parent-report measure of emotional and behavior problems. The CBCL has two items that measure cross-gender identification, “Behaves like opposite sex” and “Wishes to be of opposite sex,” which can be summed (range, 0-4) to provide a composite of gender identity. In a Dutch study of 7526 7-year-old twin pairs from the general population, 4.7% of children had a summed score (across the two items) of 1 or higher (van Beijsterveldt, Hudziak, & Boomsma, 2006). More recently, Steensma, van der Ende, Verhulst, and Cohen-Kettenis (2012) reported on 879 Dutch children (406 boys, 473 girls) from the general population followed prospectively for 24 years. The mean age in childhood was 7.5 years (range, 4-11 years). Fifty one (5.8%) of the 879 children were classified as gender variant (i.e., summed score on gender identity items was 1 or higher), which is similar to the percentage found by van Beijsterveldt et al.
Since the 1960s, a number of studies have reported estimated prevalence rates for GID in adults (for a review, see Zucker & Lawrence, 2009). Rates have varied, in part, depending on the inclusion criteria (e.g., including individuals who have had, at least, hormonal treatment but have not necessarily had any surgical interventions vs. only including individuals who have had sex reassignment surgery). For example, De Cuypere et al. (2007) estimated that 1 in 12,900 biological adult males in Belgium have GID, while Weitze and Osburg (1996) estimated a prevalence rate of 1 in 42,000 in Germany. The estimated prevalence rate in most other studies have fallen within this range (i.e., 1/12,900-1/42,000). Based on these estimated rates, it seems reasonable to presume that the prevalence of GID is low. In Steensma et al.’s (2012) prospective study, only 1 (0.1%) of the 879 participants, a biological male, had undergone gender reassignment (cross-sex hormonal treatment and surgery) when followed up in adulthood.
1.5 Treatment of Children with Gender Identity Disorder
At present, there are three general approaches that guide the clinical management of children with GID, each of which rests on its own conceptualization of gender identity development and GID. It is beyond the scope of this thesis to review in detail treatment approaches and the debates surrounding them (for detailed reviews, discussions, and clinical examples see, for example, Dreger, 2009; Stein, 2012; Zucker, 2001a, 2006c, 2007, 2008b; Zucker & Bradley, 1995; Zucker et al., 2012b).
1.5.1 The Therapeutic Model
In one approach, the goals of treatment are: (1) to circumvent the consistently observed sequelae of GID (e.g., ostracism by peers, depression), (2) to help children feel more comfortable with their biological sex, thereby reducing/alleviating gender dysphoria, (3) to increase the likelihood of desistance of GID in adolescence and adulthood, and (4) to alleviate co-occurring socioemotional problems in the child or difficulties within the family dynamic that may play a role in the child’s gender confusion (e.g., Meyer-Bahlburg, 2002; Zucker & Bradley, 1995; Zucker et al., 2012b). Dreger (2009) labeled this approach the “therapeutic model,” in contrast to the “accommodating” model described below. However, depending on the clinician’s theoretical perspective on the etiology of GID, the specific interventions used may vary.
Some clinicians view cross-gender behaviors as a result of inappropriate learning and attempt to extinguish them using principles of behavior therapy (e.g., Reker & Lovaas, 1974). Zucker et al. (2012b) proposed a multifactorial theory in which cross-gender identification is influenced by several factors, including biological, psychosocial, psychological, and psychodynamic variables. Within this framework, a biopsychosocial model of treatment is used to address the underlying factors that contribute to the child’s cross-gender identification (e.g., socioemotional problems within the child, family dynamics). In addition to therapy with the child, intervention may also include parent and/or family counselling. Some clinicians use a strictly psychodynamic formulation in which GID is viewed as a defense against distress and anxiety. Thus, psychodynamically informed therapy is used to address the underlying factors that perpetuate this defensive response (Coates & Wolfe, 1997). Regardless of the etiological framework, a common thread among these clinicians is the assumption that it is possible to modify a child’s gender identity (e.g., Meyer-Bahlburg, 2002; Zucker, 2008b). In a variation of the therapeutic approach, clinicians in the Netherlands place the emphasis of treatment on concomitant emotional/behavioral problems in the child as well as family dynamics rather than on direct attempts to modify gender identity (Cohen-Kettenis & Pfäfflin, 2003; de Vries & Cohen-Kettenis, 2012). The rationale for this approach is that, if the concomitant problems have contributed to causing or maintaining the gender dysphoria, then the dysphoria will likely disappear by addressing these problems. de Vries and Cohen-Kettenis have recently referred to the approach used in the Netherlands as the “Dutch approach.”
Both the Dutch approach as well as that espoused by Zucker et al. (2012b) utilizes a developmental perspective to treatment. When gender dysphoria persists from childhood into adolescence, it is less likely alleviated by psychological intervention and more likely to be treated by hormonal and surgical interventions (e.g., Cohen-Kettenis & van Goozen, 1997; de Vries et al, 2011a; Zucker, 2006). Thus, the therapeutic approach for adolescents is one that supports transitioning on the grounds that it will lead to better psychosocial adjustment (Zucker et al., 2011, 2012b).
The therapeutic model has faced intense criticism because some clinicians have claimed that, in addition to treating gender dysphoria, they were also preventing homosexuality, which they viewed as disordered (e.g., Rekers, Bentler, Rosen, & Lovaas, 1977). Some critics of the therapeutic model, and of Zucker’s approach in particular, view it as “homophobic” and similar to reparative therapy that has been used in attempts to change an individual’s sexual orientation (e.g., Pickstone-Taylor, 2003). Most contemporary clinicians emphasize that the goal of treatment is to resolve conflicts associated with the GID, regardless of the child’s eventual sexual orientation (Cohen-Kettenis, 2001; Zucker & Cohen-Kettenis, 2008). Moreover, Bradley and Zucker (2003) have explicitly stated that they do not endorse the prevention of homosexuality as a therapeutic goal. However, some parents of children with GID who request treatment do so, in part, because they hope to prevent homosexuality in their child (Zucker, 2008c). The therapeutic approach has also been criticized on the grounds that it does not appreciate the distinction between children with GID (i.e., children with gender dysphoria) and children who show gender-variant behaviors but without concomitant gender dysphoria (e.g., Stein, 2012). Discussed later, this criticism is a reflection of a broader conceptual and diagnostic debate in the field regarding the conflation of GID proper with presumably innocuous cross-gender behavior.
1.5.2 Accommodation Model
A second approach to treatment of GID has been referred to as the “wait and see” or “accommodation model” (Dreger, 2009; Hill, Rozanski, Carfagnini, &Willoughby, 2005). Within this framework, there is no direct attempt to help the child feel more comfortable about their biological sex or to modify their cross-gender behaviors. Rather, parents are encouraged to support the child’s cross-gender behaviors in order to reduce feelings of stigmatization in the child and to promote the child’s overall adjustment (Ehrensaft, 2012; Menvielle, 2012; Menvielle & Hill, 2011; Menvielle & Tuerk, 2002;). Ehrensaft (2011), in her case description of a 6-year-old biological male, explained that, essentially, the family and therapist tolerate a state of “not knowing” until the child “unfolds an authentic gender identity and expression,” which may or may not be aligned with their biological sex. If a child’s “authentic gender self” is not aligned with their biological sex, early social gender transition is then supported (e.g., Ehrensaft, 2012). The accommodation treatment approach is viewed as supportive and accepting of children’s authentic gender role expression on the premise that it does not steer children down a particular gender path (e.g., Bocking & Ehrbar, 2005; Hill et al., 2005). It is arguable, however, that by allowing cross-gender behavior, one is, in fact, steering children down a cross-gendered path. More than two decades ago, Green (1987) speculated that boys whose parents do not attempt to discourage cross-sex behavior might be more likely to become transsexuals as adults. Within this treatment approach, there appears to be an assumption that gender identity can change as indicated by the recognition that some children who socially transition at an early age may want to reverse the gender role transition later on (Ehrensaft, 2012; Menvielle, 2012).
1.5.3 Early Transition Approach
A third, more recent, approach takes an extreme stance on childhood cross-gender behavior and has likely been fuelled by changing ideas about what constitutes appropriate expression of gender (Drescher, 2010a). In this model, pre-pubescent children with GID, sometimes as young as 5 years of age, are allowed and encouraged to socially transition from one gender to another (e.g., Vanderburgh, 2009; see also Brown, 2006; English, 2011; Rosin, 2008; Spiegel, 2008). There is no attempt to decrease cross-gender behavior and identification. A social transition may involve, for example, a biological male using a female name and registering at school as a female (e.g., Saeger, 2006). This approach is partly rooted in the assumption that the onset of cross-gender behavior is an indication of innate (cross) gender identity rather than as a sign of gender confusion or a GID. Further, it is argued that an early transition (i.e., before puberty) may circumvent associated mental health issues seen in individuals with GID (Vanderburgh, 2009). The role of the therapist is to help families navigate aspects of the transition process, such as advocacy within the social setting and educating families about the medical aspects of transitioning.
There are some serious concerns about this approach. The most striking implication of an approach that facilitates early transitioning is that it may steer some children down a transgendered path who might have otherwise not desired to transition as they progress in development. Proponents of the early transitioning model have not addressed how this approach fits conceptually or clinically with the finding that the majority of children with GID show a desistence in adolescence (e.g, Drummond et al., 2008; Green, 1987; Wallien and Cohen-Kettenis, 2008). This is an important issue because an approach that encourages transitioning in childhood assumes that these children would persist in their GID into adolescence, which is not supported by the follow-up studies of children with GID.
There have been no quantitative follow-up studies on children who socially transition in childhood, probably, in part, because this approach is still relatively recent. However, one qualitative study conducted in The Netherlands suggests that socially transitioning children is not without its drawbacks (Steensma, Biemond, Boer, & Cohen-Kettenis, 2011). In this study, 25 adolescents who had met criteria for GID in childhood were interviewed regarding stability/instability of their gender identity from childhood into adolescence, among other things. The results of two adolescent (biological) females are of relevance to this discussion. In childhood, these females were seen and treated as boys by other children and they dressed in male-typical clothing all the time. It is unclear, however, the extent to which these females were socially transitioned (e.g., name and pronoun use). In adolescence, both girls experienced a desistence of their gender dysphoria and wanted to live in the female gender role. Both girls found it a struggle to attempt living in the female role after having lived to some extent in the male gender role. One girl commented, “At high school, I wanted to make a new start. I did not want people to know that I had looked like a boy and had wanted to be a boy in childhood.” While it is arguable that an approach that supports social transition in childhood may be beneficial to children who will turn out to be persisters, it is not the advisable approach for children who will desist. The challenge, however, is the difficulty in predicting the gender identity outcome of very young children with GID (Steensma & Cohen-Kettenis, 2011).
To date, there is no consensus on the best treatment approach for children with GID. This state of affairs has been maintained by the paucity of empirical data on treatment and also, in part, by theoretical disagreements among clinicians about gender identity development and its malleability in childhood. As a point of agreement, proponents of both the therapeutic and accommodation model agree that, if it is apparent that an adolescent is committed to transitioning, the recommended treatment approach is to provide cross-sex hormonal therapy, to be followed by surgery, if desired, in adulthood. Unfortunately, the debate about therapeutics for children is far from over largely because of scant research attention in this area. There have been no rigorous treatment outcome studies on children with GID and, certainly, no randomized controlled treatment trials that have compared the effects of these therapeutic approaches on gender identity outcome (Bradley & Zucker, 2003; de Vries & Cohen-Kettenis, 2012; Zucker, 2001a). In addition, there have been no studies that compared any of the different treatment approaches for GID to a condition of no treatment. Beyond resolving debate, there is an even more important reason to evaluate treatment approaches. As noted previously, most children with GID seem to desist in their gender dysphoria by adolescence. It remains unknown whether the aforementioned treatment approaches are associated with different long term outcomes (e.g., persistence vs. desistence of GID, general psychiatric functioning, psychosocial adjustment).
GID is arguably one of the most contentious diagnoses in the DSM. A detailed review of the controversies surrounding the diagnosis is beyond the scope of this chapter, but can be found elsewhere (e.g., Bockting, 2009; Bradley & Zucker, 1998; Bryant, 2006; Drescher 2010a, 2010b; Hill et al., 2005; Meyer-Bahlburg, 2010; Wilson, Griffin, & Wren, 2002; Zucker & Bradley, 1995; Zucker, Drummond, Bradley, & Peterson-Badali, 2009). Essentially, one group of critics argue for a reform of the diagnosis while a second group question the legitimacy of GID as a diagnostic category.
1.6.1 Diagnostic Reform
One major criticism of the GID diagnosis is that it fails to differentiate between children who have both cross-gender identity (gender dysphoria) and pervasive cross-gender behaviors from those who show signs of pervasive cross-gender behavior but without the co-occurring unhappiness about their biological sex (i.e., without co-occurring gender dysphoria) (Bockting, 1997). In the current form of the GID diagnosis (Appendix A), the Point A criterion is met if a child has at least 4 of 5 markers of persistent cross-gender identification: the desire to be, or insistence that one is, of the other sex (Criterion A1) and marked/pervasive cross-gender role behaviors, such as peer and clothing preference (Criteria A2-A5). Critics have argued that Criterion A1 (which is viewed as capturing gender dysphoria) should not be condensed with criteria pertaining to cross-gender behaviors; otherwise, a child may receive a diagnosis of GID through demonstration of cross-gender behaviors but in the absence of gender dysphoria (e.g., Bartlett, Vasey, & Bukowski, 2000; Bockting & Ehrbar, 2005; Hill et al., 2005; Richardson, 1996, 1999; Wilson et al., 2002). The concern is that a diagnosis and subsequent treatment may be harmful to the child (e.g., Langer & Martin, 2004). Presumably, these critics are arguing that the absence of verbal statements of cross-gender identification or wish is an indicator that the child is not gender dysphoric, regardless of the degree of cross-gender behavior. However, as discussed earlier, a child may experience unhappiness with their biological sex but not verbalize it. It is conceptually possible for children to meet diagnostic criteria for GID if they endorse items A2-A5 and also express unhappiness about their sexual anatomy (i.e., anatomic dysphoria, Criterion B), but yet do not make explicit statements of wanting to be of the opposite sex. These children may actually be struggling with their gender identity and, without a diagnosis, the way in which the cross-gendered behaviors are managed may not be in the best interest of the child (Zucker, 2010a). From a clinical standpoint, however, it is not common for a child to express anatomic dysphoria but not verbalize cross-gender identification.
It has been recommended that the diagnostic criteria be revised such that a distinction is made between children who have both cross-gender identification (manifested as explicit statements of wanting to be of the opposite sex) and cross-gender behaviors from children who only demonstrate cross-gender behaviors (e.g., Bartlett et al., 2000). Zucker (2005c) suggested that one solution to this debate is a modification of the DSM-IV diagnostic criteria such that it would be necessary for the child to systematically verbalize the wish to be of the opposite sex in order for the Point A criterion to be met. In a re-analysis of a parent-report measure of cross-gender identification, Zucker (2010a) found that children who frequently stated the desire to be of the other gender also showed more pervasive cross-gender behaviors. In part because of this finding, the DSM-5 Workgroup on GID has recommended that the persistent desire to be or insistence that one is of the opposite gender should be a necessary criterion for the diagnosis of GID. It is hoped that this change would result in a tightening of the diagnostic criteria and may better separate children with GID from those displaying marked variance in their gender role behaviors but without the desire to be of the other gender. The proposed revision to the DSM-IV diagnostic criteria for GID in children is summarized in Appendix B. The Workgroup on GID proposed retention of the diagnosis in DSM-5 with a name change (“Gender Dysphoria in Children”). In addition to statements of cross-gender identification (Criterion A1), children need to have at least 5 of 7 other manifestations of incongruence between expressed and assigned gender. In the proposed diagnostic criteria, rejection of sex-typical toys, games and activities, and anatomic dysphoria are part of Point A criteria. Point B criteria pertain to distress or impairment.
1.6.2 Is GID a Mental Disorder?
That GID is not a mental disorder and should, therefore, be removed from the DSM has been argued from at least four perspectives: (1) the GID diagnosis pathologizes normal variation in gender role expression, (2) children with GID are not impaired or inherently distressed, (3) the diagnosis was introduced to the DSM as a veiled attempt to repathologize homosexuality, and (4) GID is a childhood manifestation of homosexuality.
It has been argued that the cross-gender behaviors observed in children with GID are no more than normal, though sometimes extreme, variation in gender role behavior. The GID diagnosis, therefore, pathologizes children who exhibit harmless gender non-conformity and who are simply expressing their interests and inherent tendencies (e.g., Langer & Martin, 2004; Pickstone-Taylor, 2003). Proponents of the GID diagnosis argue that this line of thinking represents biological essentialism and is a simplistic view of a complex phenomenon that is influenced by biological, psychological, and interpersonal processes (Bradley & Zucker, 2003; Zucker, 2006b; Zucker et al., 2012b). Thus, while the critics who argue for a reform of the diagnosis recommend that the criteria should be more stringent to better distinguish gender dysphoric from non-gender dysphoric children with cross-gender behaviors, these critics argue that all children who receive the GID diagnosis are actually displaying nonpathological gender nonconformity.
The GID diagnosis has also been criticized on the grounds that it does not meet the criteria for a mental disorder because children with GID do not show evidence of inherent distress or impairment in functioning (Point D criterion), and, if they do experience distress or socioemotional difficulties, it is simply a reaction to social intolerance of their cross-gender behaviors (e.g., Bartlett et al., 2000; Menvielle, 1998; Wilson et al., 2002). Some argue that the disorder is more a reflection of a gender oppressive society rather than signaling a disorder within the individual (Ault & Brzuzy, 2009). Supporters of the diagnosis have provided compelling reasons to retain GID in the diagnostic nomenclature (Bradley & Zucker, 1998, 2003). For instance, clinic-referred children with GID sometimes verbalize significant unhappiness over their status as males or females and often state the desire to change themselves into the opposite gender (for clinical examples, see Zucker et al., 2012b; Zucker & Bradley, 1995). It is also argued that, even in the absence of explicit statements to be of the opposite sex, pervasive enactments of cross-gender fantasies, such as through role-play and dress-up, is a behavioral manifestation of underlying unhappiness with one’s biological sex (Zucker, 2006b).
On a more political level, it has been argued that the inclusion of GID into the DSM was a veiled political maneuver to repathologize homosexuality, which was simultaneously removed from the DSM at the time that GID was introduced (e.g., Ault & Brzuzy, 2009; Sedgwick, 1991). Zucker and Spitzer (2005) noted that the DSM-III included a diagnostic category of ego-dystonic homosexuality; thus, there was no need for a backdoor diagnosis to replace homosexuality. These authors also brought attention to the fact that several clinicians and scientists who recommended the inclusion of GID in the DSM had argued in favor of delisting homosexuality.
The strong association between GID in childhood and homosexuality in adulthood has also added to the controversy surrounding the diagnosis. Follow-up studies of boys with GID have found that the most common outcome in adulthood is desistance of GID with a homosexual sexual orientation (e.g., Green, 1987). Some have interpreted this finding to mean that the cross-gender behaviors observed in children with GID is simply an early manifestation of later homosexuality (e.g., Minter, 1999) and, therefore, should not be pathologized or treated (e.g., Corbett, 1998; Isay, 1997). However, cross-gender behaviors in childhood are not isomorphic with a later homosexual sexual orientation. Some boys with GID grow up to have a later heterosexual sexual orientation (e.g., Green, 1987; Wallien & Cohen-Kettenis, 2008). A second response to this particular criticism is that what constitutes a mental disorder is its operational definition (Green, 2011). Thus, if a child meets diagnostic criteria for a disorder, it is irrelevant to the assignment of a current diagnosis whether the child will meet the diagnosis in the future.
As discussed above, GID is a controversial diagnosis. The diagnosis itself has received much criticism and there remains a significant lack of consensus in the field regarding clinical management of the disorder. Discussions and debates on best treatment practices have raised the issue of the long-term outcomes for boys with GID, particularly in regard to gender identity and sexual orientation (e.g., Zucker, 2008b), a topic now addressed in the present literature review.
1.7 Psychosexual Outcome of Boys with GID
One approach to understanding the developmental trajectory of boys with GID is to retrospectively assess the childhood experiences of adult male-to-female transsexuals. These studies have found that adolescents and adults with GID, particularly those with a co-occurring homosexual sexual orientation (in relation to their birth sex), invariably recall a pattern of childhood cross-gender behavior that corresponds to the DSM criteria for GID (Green, 1974; Smith, van Goozen, Kuiper, & Cohen-Kettenis, 2005; Zucker et al., 2006). However, given the potential problems with retrospective research (for an overview, see Hardt & Rutter, 2004), most notably that the recollections may not be accurate, the ideal methodology to understand the long-term outcome of boys with GID is to identify a group of such children and follow them prospectively. Since the 1960s, a number of such studies have been conducted. Of these, Green’s (1987) study and Wallien and Cohen-Kettenis’ (2008) study constitute the two most comprehensive long-term follow-up of boys with GID. In addition, the results of 6 other follow-up studies which utilized much smaller sample sizes are also summarized. For clarity, the results on gender identity outcome are presented first followed by the results on sexual orientation outcome. Across all studies, sexual orientation is classified in relation to birth sex.
1.7.1 Gender Identity Outcome of Children with GID
Zucker and Bradley (1995) summarized data from six published follow-up studies of
boys who displayed marked cross-gender behavior (Bakwin, 1968; Davenport, 1986; Kosky, 1987; Lebovitz, 1972; Money & Russo, 1979; Zuger, 1978). The results of these studies are presented as a group due to the small sample size of each study. Across these six studies, a total of 55 boys were seen at follow-up (range, 16-36 years). Of these, 5 (9.1%) were classified as transsexual at follow-up (i.e., they showed persistent gender dysphoria). All 5 persisters had a homosexual sexual orientation.
One of the earliest prospective follow-up studies to utilize a reasonably large sample size was conducted by Green (1987). Green’s sample consisted of 66 behaviorally “feminine” boys and 56 control boys who were unselected for their gender identity. Both groups of boys were initially assessed at a mean age of 7 years (range, 4-12 years) and were recruited through various forms of advertisement. Although Green did not utilize a formal DSM diagnosis, from his clinical descriptions it appears that most of the behaviorally feminine boys would have met criteria for GID. Most of the feminine boys stated their wish to be girls or to grow up to be women, avoided male-typical activities (e.g., rough-and-tumble play, sports), preferred female roles in pretend play, and showed a preference for girls’ clothes, toys, and peers (Green, 1974, 1976). Forty-four feminine boys and 35 control boys were available for follow-up assessment in adolescence and adulthood (M age, 18.9 years; range, 14-24). Only a minority of the feminine boys (n = 12) received formal therapy between the childhood assessment and the follow-up interview. At follow-up, only one (2.3%) of the 44 behaviorally feminine boys continued to experience gender dysphoria and desired sex reassignment surgery. None of the control boys reported any gender dysphoria.
More recently, Wallien and Cohen-Kettenis (2008) conducted the largest follow-up study to date on boys and girls with GID (77 children; 59 boys, 18 girls). The childhood data were collected as part of the standard assessment of children seen in their specialized gender identity clinic in The Netherlands. At follow-up, 54 participants (40 boys, 14 girls) were successfully traced and completed the follow-up assessment. The remaining 23 participants (19 boys, 4 girls) could not be traced. However, Wallien and Cohen-Kettenis assumed that these untraced participants were desisters on the premise that had they been persisters they would have likely had contact with the clinic and, therefore, included them in the calculation of a persistence rate. Of the 77 children followed prospectively, Wallien and Cohen-Kettenis reported that 21 (12 boys, 9 girls) were still gender dysphoric at follow-up, which yielded a persistence rate of 27% for the total sample of boys and girls with GID. However, when calculated based only on those participants who were actively involved in the follow-up assessment (i.e., excluding the 23 participants who could not be traced at follow-up), the persistence rate was 38.8%.
For comparative purposes, I provide additional details for the boys in Wallien and Cohen-Kettenis’ study. The mean age for all 59 boys was 8.3 years (range, 5-12 years) at the initial assessment and 19.4 years (range, 16-28 years) at follow-up. Unfortunately, separate demographic data were not available on the subgroup of boys who were successfully contacted at follow-up. When calculated using only the 40 boys who were successfully traced at follow-up, the rate of persistent gender dysphoria was 30% (12 persisters, 28 desisters). If, however, one used Wallien and Cohen-Kettenis’ more liberal calculation of persistence by also including the 19 boys who could not be traced and classifying them as desisters, the persistence rate for boys was 20.3%. All of the persisters met complete diagnostic criteria for GID at follow-up, were treated with puberty suppressing hormonal treatment before the age of 16, and with cross-sex hormones after the age of 16 years.
The rates of GID persistence found by Wallien and Cohen-Kettenis (2008), regardless of whether one uses the liberal or conservative calculation, were considerably higher than that reported by Green (1987) (in fact, more than 8 times as high) and approximately 2-3 times as high as the rate found by Zucker and Bradley (1995) in their review of six studies.
Wallien and Cohen-Kettenis’ sample size was large enough to allow for group comparisons to identify potential differences in childhood between the boys who persisted in their gender dysphoria from those who desisted. There was a significant difference between the persisters and desisters on the diagnosis received in childhood. All of the boys who showed persistent gender dysphoria had met the full criteria for GID in childhood. However, of the 28 (traced) boys who desisted, 17 met full criteria for GID while the remaining 11 were subthreshold. Of the 19 boys who could not be traced at follow-up, 15 met full criteria for GID and 4 were subthreshold. The persisters were also more extreme in their childhood cross-gender behaviors and identification on two dimensional measures, both of which were also used in the present study: (1) the Gender Identity Interview for Children (Wallien et al., 2009; Zucker et al., 1993), a child-report measure, and (2) the Gender Identity Questionnaire for Children (Johnson et al., 2004), a parent-report measure. These data indicated that boys with more extreme cross-gender identification in childhood were more likely to be persisters than desisters.
18.104.22.168 Methodological Issues
Two methodological issues regarding the follow-up studies need to be mentioned. First, with the exception of Green’s (1987) study, none of the follow-up studies included a clinical control group (i.e., referred for reasons other than gender identity concerns) or community control group. Therefore, interpretation of follow-up data relied on evidence from the literature on adults with GID (Zucker, 1985). Second, during the follow-up assessments, particularly those in which data were obtained through clinical interview (e.g., Green, 1987), participants were asked to discuss an extremely sensitive and personal aspect of their lives–their psychosexual development. Due to the potential effects of social desirability on participants’ responses, it is possible that the rates of persistent gender dysphoria (and also homosexual outcomes) in these studies were an underestimate. It is unlikely that participants would have reported a homosexual or GID outcome if, in fact, they were not homosexual or experiencing gender dysphoria; thus, an overestimation was unlikely to have occurred. Unfortunately, social desirability was not measured by Green or Wallien and Cohen-Kettenis; thus, the effects of this phenomenon on their respective results are unknown.
22.214.171.124 Process of GID Desistence
There have been no quantitative follow-up studies that have systematically examined the developmental process through which GID desists (e.g., how and at what age). Some authors suggest that desistence typically occurs sometime around puberty or early adolescence (de Vries & Cohen-Kettenis, 2012; Wallien & Cohen-Kettenis, 2008). However, one should be skeptical in viewing puberty as the transformative period in the lives of children with GID with regard to their gender identity. Some gender-referred children show changes in their gender identity before puberty and, in fact, desist in their dysphoria during childhood (for case examples, see Zucker, 2006c; Zucker & Bradley, 1995).
The results of a recent qualitative study by Steensma et al. (2011) suggested that desistence may occur between the ages of 10-13 years and is likely influenced by psychosocial and psychosexual factors. In this study, 25 adolescents (13 boys, 12 girls; M age, 15.88 years; range, 14-18) who had been seen in a specialized gender identity clinic in The Netherlands in childhood and diagnosed with GID were contacted in adolescence for a follow-up interview. During the follow-up interview, participants were asked to discuss their childhood gender role behaviors, stability of gender identity, sexual orientation, and physical development. Thus, the data on childhood gender-role behavior and change in gender identity between the initial assessment and follow-up were based on retrospective recall. The primary goal of the study was to obtain qualitative data on the developmental course of GID, including factors that contributed to GID persistence and desistence. Of the 13 biological males in this study, 7 had applied for sex reassignment surgery and can, therefore, be considered persisters. The remaining 6 were desisters. One should not, however, extract a persistence rate from these data because this was not the goal of this study. Second, persistence and desistence were not clearly operationalized. Third, as this was a qualitative study, there was no quantitative comparison between the participants and non-participants and differences between the groups could have affected gender identity outcome. Thus, given that the participants might have represented a biased sample, caution should be exercised in drawing conclusions about the obtained persistence rate.
Several findings from this study warrant mention. First, both persisters and desisters recalled that around the age of 6 or 7 years they started to identify with and expressed a wish to be of the opposite sex, though they had exhibited cross-gender behavior prior to this. The underlying “motives” for childhood cross-gender identification appeared to be different for the persisters and desisters. The persisters recalled having felt as though they were the opposite sex. In contrast, the desisters recalled having wished they were of the opposite sex but did not feel they were the opposite sex. Second, the period from age 10 to 13 years of age appeared to be significant in the developmental trajectory for both persisters and desisters. Both groups of adolescents reported that three factors affected their feelings of gender dysphoria and either lessened or intensified it: their social environment (e.g., peer relations), anticipation about and experience of puberty, and emerging sexual attraction. This study highlighted the need for studies to prospectively follow children with GID beyond adolescence and into adulthood while periodically monitoring stability and change in their psychosexual development, particularly during the transition from childhood to adolescence. For example, one of the males in the desister group continued to experience some feelings of gender dysphoria but was not interested in living as woman. As the authors noted, a longer follow-up period may help to clarify whether this participant was experiencing protracted gender dysphoria that would eventually diminish or whether some form of mild gender dysphoria would remain a stable aspect of his psychological functioning.
The results of Steensma et al.’s study can be considered preliminary given the small sample size and retrospective recall of gender experiences from assessment to follow-up. This study also represents the first attempt in the literature to account for the process through which GID persists or desists. Future studies utilizing a prospective systematic design (i.e., regular follow-up intervals following the childhood assessment) with quantitative measurements of relevant variables are needed. Further, there are other factors beyond those examined by Steensma et al. that may potentially contribute to the developmental course of GID (e.g., psychiatric functioning, psychotherapy) and that would require empirical investigation.
1.7.2 Gender Identity Outcome of Adolescents with GID
Clinical observation and empirical evidence suggest that persistence of gender dysphoria, including the desire for sex change, is higher among patients assessed for the first time during adolescence and then followed up than among patients first assessed in childhood and then followed prospectively (Zucker & Bradley, 1995).
Cohen-Kettenis and van Goozen (1997) reported follow-up data on 33 adolescents with GID seen at a specialized gender identity clinic in The Netherlands. Of the 33 adolescents (M age at initial assessment, 17.5 years; range, 15-20), 22 (66.6%) went on to receive cross-sex hormonal therapy and some form of sex reassignment surgery. Of the remaining 11 adolescents, 8 were not diagnosed with “transsexualism” (presumably, they did not meet criteria for GID) and were, therefore, not recommended for sex reassignment surgery. The remaining 3 participants were diagnosed with transsexualism but, for a variety of reasons, including severe comorbid psychopathology, were not recommended for any cross-sex hormonal treatment. The “true” persistence rate of these adolescents could be higher than 66.6% because there were no data on whether any of the 11 participants who did not initially receive cross-sex hormonal or surgical interventions did so at a later date. This persistence rate of 66.6% is much higher than the persistence rates found in follow-up studies of boys who are first referred in childhood.
Zucker et al. (2011) reported on 109 adolescents who were assessed in the Gender Identity Service at the Centre for Addiction and Mental Health (CAMH) in Toronto, Canada. Of the total sample, 66 (60.6%) were recommended for puberty blocking hormonal therapy and 43 (39.4%) were not. The percentage of adolescents referred for hormonal therapy was similar to the 66.6% reported by Cohen-Kettenis and van Goozen (1997). It should be noted that the Zucker et al. study was geared towards examining which factors (e.g., demographic, psychosexual) influenced the clinical decision to recommend or not recommend puberty blocking hormones. Blockers were more likely to be recommended for those adolescents who were more extreme in their current gender dysphoria and childhood cross-gender behavior. Thus, while follow-up data were available for some participants, this study was not a systematic follow-up study per se. It would be informative to know whether the participants who did not receive blockers persisted or desisted in their gender dysphoria.
1.7.3 Sexual Orientation Outcome 
In their review of six follow-up studies of boys with GID (Bakwin, 1968; Davenport, 1986; Kosky, 1987; Lebovitz, 1972; Money & Russo, 1979; Zuger, 1978), Zucker and Bradley (1995) also summarized data on sexual orientation outcome. Of the 55 boys reported on in these studies, 13 were classified as “Uncertain” with regard to sexual orientation, in part, because they were not sexually active and data regarding sexual orientation in fantasy were not definitive. Of the remaining 42 cases, 26 (61.9%) were classified as homosexual (this includes the 5 individuals who were persisters with a homosexual sexual orientation) and 16 (38.1%) were classified as heterosexual. In these six follow-up studies, it is unclear if sexual orientation was classified according to fantasy or behavior.
In Green’s (1987) study, sexual orientation in fantasy and behavior was assessed using a semi-structured interview and rated using Kinsey’s 7-point scale where 0 = exclusively heterosexual and 6 = exclusively homosexual (Kinsey et al., 1948). Of the 44 feminine boys assessed at follow-up, 33 (75%) were classified as bisexual/homosexual in fantasy (Kinsey ratings of 2-6) and 11 (25%) were classified heterosexual in fantasy (Kinsey ratings of 0-1). Of note, the persister in Green’s study reported a homosexual sexual orientation, in both fantasy and behavior. Unlike the feminine boys, all of the clinical control boys were classified as heterosexual in fantasy. At follow-up, only 30 of the feminine boys reported having had sexual experience. Of these, 24 (80%) were classified as bisexual/homosexual in their sexual behavior and the remaining 6 (20%) were classified as heterosexual in behavior. In the control group, 25 of the 35 boys reported sexual experiences at follow-up. Of these, 1 (4.0%) was classified as non-heterosexual (he reported bisexual sexual experiences) and the others were classified as heterosexual. Therefore, depending on whether sexual orientation was classified according to fantasy or behavior, 75-80% of the 44 feminine boys were classified as bisexual/homosexual in their sexual orientation compared to only 0-4% of the control boys.
In Wallien and Cohen-Kettenis’ (2008) study, sexual orientation was assessed using the 9-item Sexual Orientation Questionnaire (SOQ) (http://links.lww/com/A569). Items on the SOQ pertained to four domains of sexual orientation: sexual behavior (4 items), sexual fantasy (2 items), sexual attraction (2 items), and sexual identity (1 item). Response options for questions pertaining to sexual behavior and fantasy reflected Kinsey’s 7-point scale. Although this study included biological males and females, only data for males are presented here. Of the 40 males who were contacted for follow-up, data on sexual orientation in fantasy were only available for 21. Of these, 17 (81%) were classified as bisexual/homosexual and 4 (19%) as heterosexual. Data on sexual orientation in behavior were available for 19 males (13 desisters, 6 persisters). Of these, 15 (79%) were classified as bisexual/homosexual and 4 (21%) were classified as heterosexual. When sexual orientation was examined according to gender identity outcome,  10 (77%) and 13 (81%) of the desisters were classified as bisexual/homosexual in behavior and fantasy, respectively. The remaining desisters were classified as heterosexual, (3) 23% in behavior and (3) 19% in fantasy. For the 6 persisters on which requisite data were available, 5 (83%) were classified as bisexual/homosexual in both fantasy and behavior and 1 (17%) was classified as heterosexual in fantasy and behavior.
In the studies by Green (1987) and Wallien and Cohen-Kettenis (2008), most boys with GID later developed a bisexual/homosexual sexual orientation, with rates that ranged from 75-81%. These rates are substantially higher than the currently accepted base rate of a homosexual sexual orientation in males of 3.1% (Laumann, Gagnon, Michael, & Michaels, 1994).
That some males with GID develop a heterosexual sexual orientation in adulthood is not fully understood. It has been suggested that perhaps the degree of cross-gender identification may affect sexual orientation outcome (Zucker, 1985). On the latter point, Wallien and Cohen-Kettenis (2008) found a significant difference between their sexual orientation groups on a parent-report measure of childhood sex-typed behavior (the Gender Identity Questionnaire for Children). However, when these analyses were redone for the desisters only, the difference was no longer significant and suggested that the extreme cross-gender scores of the persisters was responsible for the overall group difference on this measure.
Green (1987) also examined the relationship between degree of cross-gender behavior in childhood and sexual orientation at follow-up. For all participants, Green (1987) computed a childhood composite “extent of femininity” score on the basis of six behaviors: cross-dressing, rough-and-tumble play, wish to be a girl, desire to be like father, attention to mother’s fashion, and female-type doll-play. Across the entire sample, there was a significant correlation between “extent of femininity” and sexual orientation; however, within the “feminine” group only, this association was not significant for sexual orientation in fantasy or behavior. Green (1987) concluded that the lack of “range” (i.e., variability) in “extent of femininity” and sexual orientation contributed to the insignificant finding.
Green also examined whether there were specific childhood features that distinguished which “feminine” boys developed a homosexual sexual orientation from those who developed a heterosexual sexual orientation. Of 14 variables of childhood sex-typed behavior, three variables were related to sexual orientation at follow-up: female role-play, doll-play, and female peers. With age at childhood assessment controlled for, preference for female peers and doll play were significantly correlated with sexual orientation in fantasy, but not behavior. Female role play was, however, significantly correlated with both fantasy and behavior (Green, Roberts, William, Goodman, & Mixon, 1987).
One criticism of these prospective studies is that they followed up children who showed extreme forms of cross-sex behavior and identification, most of whom were clinically referred. As such, the generalizability of these studies is limited to similar groups of children. In the context of this limitation, Steensma et al. (2012) conducted a 24-year prospective study of 879 (406 boys, 473 girls) non-clinically referred children, unselected for their gender identity, who were part of a population-based study (for a description of the study, see section on Prevalence of GID). At follow-up, sexual orientation was assessed by asking four questions, each of which pertained to a different domain of sexual orientation: sexual attraction (“To whom do you feel attracted?”), sexual fantasy (“About whom do you fantasize sexually?”), sexual behavior (“With whom do you have sexual contact?”), and sexual identity (“How do you identify yourself?”). For sexual attraction, fantasy, and behavior, participants’ responses were coded using Kinsey’s 7-point scale. At follow-up, 11 (2.7%) of the 406 boys were bisexual/homosexual in fantasy and 10 (2.5%) were bisexual/homosexual in behavior. Steensma et al. also examined sexual orientation outcome separately for the gender-variant (10 males, 41 females) and non-gender variant (396 males, 432 females) groups. Of the 10 gender variant boys, 2 (20%) were bisexual/homosexual in fantasy and behavior and the remaining 8 (80%) were heterosexual in fantasy and behavior. Of the 396 non-gender variant boys, 9 (2.3%) and 8 (2.1%) were bisexual/homosexual in fantasy and behavior, respectively. I used Fisher’s exact test to determine if these represented significant differences in sexual orientation across the two subgroups of boys (i.e., gender variant versus non gender-variant). For both fantasy and behavior, there was a significant difference in the number of boys classified as bisexual/homosexual versus heterosexual (both ps < .05). Depending on whether one looks at fantasy or behavior, the prevalence of bisexuality/ homosexuality in the gender variant boys was 8.7-9.5 times higher than the prevalence rate of bisexuality/homosexuality in the non-gender variant boys. Thus, when a sample of boys unselected for their gender identity were followed up into adulthood, those who were at the cross-gendered end of the spectrum were significantly more likely to develop a bisexual/ homosexual sexual orientation.
The rates of bisexuality/homosexuality obtained by Steensma et al. for the entire sample of males (2.5-2.7%) were substantially lower than those obtained by Green (1987) and Wallien and Cohen-Kettenis (2008) in their follow-up of boys with gender dysphoria. This difference is not surprising and may be attributable to sample differences; however, an interpretative caution is in order. Steensma et al. measured gender variance in childhood using 2 items on the CBCL; they did not include specific measures of gender identity, gender role, or gender dysphoria. Thus, the actual extent of cross-gender behaviors in their sample of males is unclear. On the other hand, the males in Green’s and Wallien and Cohen-Kettenis’ studies were extreme in their cross-gender behaviors on measures of gender identity and gender role. Given that the boys in Steensma et al.’s study were taken from the general population and were unselected for their gender identity it is likely, and expected, they would have significantly less cross-gender behaviors compared to the boys seen by Green and by Wallien and Cohen-Kettenis. Indeed, the rate of bisexuality/homosexuality obtained by Steensma et al. is similar to the base rate of a homosexual sexual orientation in males of 3.1% (Laumann, Gagnon, Michael, & Michaels, 1994). At the same time, and consistent with Green’s (1987) study, which included a control group, the gender variant boys in Steensma et al.’s study were more likely to develop a bisexual/ homosexual sexual orientation compared to the non-gender variant boys. Similar to the developmental course of GID, these results suggest that there may also be a dosage effect on sexual orientation outcome–the more gender variance in childhood, the higher the likelihood of homosexuality.
1.7.4 Gender Identity and Sexual Orientation Outcomes in Boys with GID: Summary
The prospective follow-up studies of boys with GID (or pervasive cross-gender behavior in the case of studies conducted prior to 1980) suggests that there are four outcomes: (1) persistence of GID, with a co-occurring bisexual or homosexual sexual orientation, (2) persistence of GID, with a co-occurring heterosexual sexual orientation, (3) desistance of GID, with a co-occurring homosexual sexual orientation, and (4) desistance of GID, with a co-occurring heterosexual sexual orientation. Of these, a desistence of GID with a co-occurring homosexual sexual orientation appears to be the most common (e.g., Green, 1987; Wallien & Cohen-Ketternis, 2008).
1.8 Retrospective Studies of Homosexual Men: Summary of Key Findings
As the current study utilized a longitudinal follow-up design, the results of retrospective studies are only briefly summarized to highlight the key findings and to inform the extent to which prospective and retrospective results converge. In the 1960s, 1970s, and 1980s, studies were conducted which identified significant differences between heterosexual and homosexual adults on their recollections of childhood sex-typed behaviors (e.g., Whitam, 1977). However, some authors were skeptical about these findings. Carrier (1986), for example, felt these results were obtained using skewed samples and, therefore, could not be generalized to all homosexual men and women. Following the publication of a meta-analysis by Bailey and Zucker (1995), the strong relationship between childhood sex-typed behavior and sexual orientation was more or less confirmed.
Bailey and Zucker reviewed 41 studies, 32 of which compared the recalled childhood (≤ 12 years of age) sex-typed behaviors of heterosexual and homosexual men. Bailey and Zucker found that, on average, homosexual men recalled substantially more cross-gendered behavior during childhood than heterosexual men. The effect sizes were large (Cohen’s d = 1.31) (Cohen, 1988) and “were among the largest effect sizes ever reported in the realm of sex-dimorphic behaviors.” The heterosexual and homosexual adults differed on various domains of childhood sex-typed behavior, including rough-and-tumble play, toy and activity preferences, role playing, cross-dressing, and preferred sex of peers. Several studies conducted after the Bailey and Zucker (1995) meta-analysis yielded similar results (e.g., Bailey & Oberschneider, 1997; Bogaert, 2003; Cohen, 2002; Zucker et al., 2006; for a review, see Zucker, 2008c).
Similar results have also been found in retrospective studies of non-Western cultures, including Samoa (Bartlett & Vasey, 2006), Turkey, and Thailand (Cardoso, 2009). As one example, Bartlett and Vasey conducted a retrospective study of childhood sex-typed behavior in the Samoan fa’afafine, Samoan men, and Samoan women. Fa’afafine are biological males who manifest gender atypical behavior. In Samoa, the fa’afafine are essentially a third gender group and, for the most part, self-identify as such. Some adult fa’afafine present socially as females through clothing choice, hair, voice, and mannerisms, while others adopt only some aspects of female gender roles. Most fa’afafine are androphilic, meaning they are sexually attracted to biological males. The fa’afafine recalled engaging in significantly more female-typical (e.g., playing with girls’ toys, putting on make-up) and less-male typical behaviors (e.g., playing with boys, playing rough sports) in childhood compared to men. Further, their recollections of childhood sex-typed behaviors did not differ significantly from those of Samoan women. Some fa’afafine also recalled an aversion towards male-typical activities.
Despite the consistency of retrospective studies, some authors continue to challenge that a relationship exists between childhood sex-typed behavior and adult sexual orientation, partly on the premise that retrospective recall of childhood behavior may be distorted (e.g., Gottschalk, 2003). A number of studies have utilized methodology that minimizes the chance of a recall bias and the results speak against the retrospective distortion hypothesis. In one study, researchers examined childhood gender nonconformity in homosexual and heterosexual adults by examining videos from their childhood (1-15 years of age) as a visual recording will not be susceptible to memory bias (Rieger, Linsenmeier, Gygax, & Bailey, 2008). The adults who self-labeled as homosexual were judged (by raters masked to their sexual identity) to be significantly more gender nonconforming as children than the adults who self-labeled as heterosexual. In another line of research that utilized heterosexual and homosexual participants, there was a significant correlation between the retrospective recall of childhood sex-typed behavior by the participants and ratings of the participants by his or her mother (Bailey, Miller, & Willerman, 1993; Bailey, Nothnagel, & Wolfe, 1995; Bailey, Willerman, & Parks, 1991).
In sum, the prospective data on sexual orientation outcome of boys with GID converge decently with retrospective studies of homosexual adults. Owing to these two lines of evidence, it is now generally accepted that childhood gender role behavior is not only strongly related to sexual orientation in adulthood in clinical and non-clinical samples but that it can also predict sexual orientation outcome (for a review, see Bailey & Zucker, 1995; Zucker, 2008c). There are, however, a few caveats to keep in mind. Some boys with GID followed prospectively into adolescence and adulthood have a heterosexual sexual orientation (e.g., Cohen-Kettenis & Wallien, 2008; Green, 1987). Studies that systematically examined childhood correlates of sexual orientation differentiation are needed to help us better understand this variability in outcome. As well, a proportion of homosexual men do not recall a childhood history of cross-gender behavior (Bailey & Zucker, 1995). Of the homosexual men who recall a cross-gendered childhood, it is unclear how many of these individuals would have met the full diagnostic criteria for GID as children.
1.9 Childhood Sex-Typed Behavior and Sexual Orientation: Explaining the Linkage
Both biological and psychosocial explanations have been offered to explain the linkage between childhood sex-typed behavior and sexual orientation (for reviews, see Green, 2008; LeVay, 2011; Zucker, 2008c).
1.9.1 Biological Explanation: Influence of Genes
One biological explanation is that childhood cross-gendered behaviors and adult sexual
orientation are strongly linked because they are both components of psychosexual differentiation that develop under the influence of genes. A number of studies have found that homosexuality is highly familial–gay men have more gay siblings than do straight men (e.g., Bailey et al., 1999; Bailey & Bell, 1993; Bailey, Dunne, & Martin, 2000; Bailey, Willerman, & Parks, 1991; Hershberger, 1997; Pillard, Poumadere & Carretta, 1982; Pillard & Weinrich, 1986; Schwartz, Kim, Kolundzija, Rieger, & Sanders, 2010). These studies have found that brothers of homosexual men are homosexual 7% to 22% of the time, with a median rate of approximately 10% (for reviews, see Bailey & Pillard, 1995; Mustanski, Chivers, & Bailey, 2002) which is well above the estimated base rate of homosexuality in the general population of biological males (Laumann et al., 1994). Differences across studies in recruitment process, assessment of sexual orientation, source of information (e.g., the gay men themselves reporting about their siblings vs. interviewing the siblings of gay men) likely account for the variation in these findings. That homosexuality tends to cluster in families could be attributed to genetic or shared environmental factors. If genes do contribute to sexual orientation, one would also expect a clustering of homosexuality among family members without shared environment (e.g., nonsibling relatives). Indeed, increased rates of homosexuality or bisexuality have been found among uncles and male cousins of gay men (Bailey, Bobrow, Wolfe, & Mikach, 1995).
Twin studies have supported a genetic interpretation of the familiality findings described above. One of the earliest twin studies of male homosexuality reported a 100% concordance rate for 37 monozygotic (MZ) twin pairs compared with a 15% rate for 26 dizygotic (DZ) pairs (Kallmann, 1952). This study has since been strongly criticized because subjects were primarily recruited from psychiatric institutions and because the methodology through which zygosity was established was not clearly delineated (Mustanksi, Chivers, & Bailey, 2002). The results of subsequent twin studies suggest that the concordance rates for homosexuality among MZ twins are much lower than 100% and appear to be closer to 50%. In a study of 115 male twin pairs, Bailey and Pillard (1991) found a concordance rate for homosexuality of 52% among MZ twins compared to 22% among DZ twins. Whitam, Diamond, and Martin (1993) found higher concordance rates: 65% for MZ twins and 29% for DZ twins. One criticism of these studies is that they relied on advertisement; thus, there may have been a self-referral bias such that gay individuals with a gay twin might have been more likely to respond to the study advertisement compared to gay individuals without a gay twin and this could have resulted in an inflation of concordance rates (Bailey, Dunne, & Martin, 2000; LeVay, 2011). More recent studies have examined concordance rates for homosexuality among pairs of twin from large registries that were created without reference to the twins’ sexual orientation and these have reported lower concordance rates compared to self-selected samples. In a study of Australian twins, for example, Bailey, Dunne, and Martin (2000) found a concordance rate of 20% among male MZ twins.
The findings from behavior genetics studies also support a genetic basis for homosexuality. In a reanalysis of the Australian twin data (Bailey, Dunne, & Martin, 2000), Kirk, Bailey, Dunne, and Martin (2000) estimated the heritability of homosexuality in males to be approximately 30%. In other words, genetic influences accounted for 30% of the variation of sexual orientation in men. In a more recent population based study of Swedish twins, the heritability of homosexuality in males ranged from 34%-39% (Längström, Rahman, Carlström, and Lichtenstein, 2010).
The strong relationship between cross-gender behavior in childhood and adult homosexuality combined with twin studies demonstrating a genetic basis for homosexuality raises at least two questions: (1) Is there a genetic basis for cross-gender behavior?; and (2) Is there a common genetic basis for both cross-gender behavior and sexual orientation? Bailey, Dunne, and Martin (2000) also measured childhood gender nonconformity of the twins in their study and found that heritability accounted for 50% of variance in recalled childhood gender nonconformity among men. Monozygotic twins who are both gay were more similar in their cross-gender behavior (correlation of gender nonconformity = .54) compared to dizygotic twins who are both gay (correlation of gender nonconformity = .14). Similar results were obtained by Alonko et al. (2010) in a study of Finnish twins taken from a national register: monozygotic (male) twins had higher correlations on both gender atypical behavior and sexual orientation (r = .56 and .50, respectively) compared to dizygotic twins (r = .27 and .25, respectively). These finding suggested that childhood gender nonconformity, like homosexuality, is heritable. In fact, a study of Dutch children taken from a national twin register, and who were therefore unselected for gender identity or sexual orientation, estimated that 70% of the variance in cross-gender behavior was accounted for by genetic factors (van Beijsterveldt, Huzdiak, & Boomsma, 2006). Alonko et al. (2010) found a large genetic correlation (r = .73) between gender atypical behavior and sexual orientation for the male twin pairs, which suggests that a shared set of genes is partially responsible for both childhood gender nonconformity and adult homosexuality.
1.9.2 Biological Explanation: Role of Prenatal Hormones
Another prominent biological explanation for the linkage between childhood sex-typed behavior and sexual orientation is that they are both influenced by common biological processes involving prenatal hormones. Research on biological females with congenital adrenal hyperplasia (CAH) provides evidence for the influence of prenatal hormones on sex-typed behaviors. In CAH, a defect in the enzyme involved in cortisol production results in the adrenal glands producing higher than normal levels of androgens. As the condition is recognized at birth and corrected, the period of abnormal androgen exposure is generally limited to the prenatal period. Data from several groups of researchers consistently show that girls with CAH show a male-typical pattern of toy and activity preferences from childhood through adulthood (for reviews, see Hines, 2002, 2010, 2011). For example, they are, on average, more active and aggressive than girls without CAH (e.g., Pasterski et al., 2007) and have toy preferences similar to those of boy (e.g, Berenbaum & Hines, 1992; Berenbaum & Snyder, 1995; Pasterski et al., 2005, 2011; Servin, Nordenström, Larsson, & Bohlin, 2003). Girls with CAH also appear to have an enhanced preference for boys as playmates (e.g., Pasterski et al., 2011; Servin et al., 2003) and engage in more rough-and-tumble play compared to unaffected girls (e.g., Pasterski et al., 2011).The extent of male-typed interests in childhood in females with CAH appears to be correlated with the degree of prenatal androgen exposure (Nordenström, Servin, Bohlin, Larsson, & Wedell, 2002).
In adulthood, women with CAH have a higher rate of bisexuality/homosexuality compared to unaffected women (Hines, Brook & Conway, 2004; Meyer-Bahlburg, Dolezal, Baker, Ehrhardt, & New, 2006; Meyer-Bahlburg, Dolezal, Baker, & New, 2006; Zucker et al., 1996). Women with CAH also report unhappiness with their female gender role and gender identity (e.g., Hines et al., 2004). In a review of the literature on women with CAH (total n = 250), Dessens, Slijper, & Drop (2005) estimated that approximately 5% experienced gender dysphoria and 1.6% pursued a gender transition (see also for a review, Hines, 2010). These percentages are significantly higher than the estimated prevalence rate of FtM transsexualism in the general population of biological females (1:30,400) (Bakker, van Kestern, Gooren, & Bezemer, 1993). Based on their review, women with CAH were approximately 500 times more likely than women in the general population to experience severe gender dysphoria and transition to the male gender role. It has been argued that the excessive androgen exposure in women with CAH is the linkage factor that explains the behavioral masculinity in childhood and gender dysphoria and bisexuality/homosexuality in adulthood (Zucker, 2008c).
Most boys and adults with GID have normally developed genitalia (Green, 1976). Thus, if prenatal androgen levels are implicated in the etiology of GID, the effects are such that genital development is not affected in any obvious way. Animal studies have shown that there are different sensitive periods for prenatal androgen effects on behavior and genital anatomy. For example, in female rhesus macaques, Goy, Bercovitch, and McBrair (1988) were able to induce behavioral masculinization without accompanying genital ambiguity (i.e., the genitals were not masculinized) by altering the timing of prenatal androgen exposure. These results suggest that, depending on the timing of prenatal exposure to androgens, there may be subsequent effects on behavior, anatomy or both.
1.9.3 Psychosocial Explanations
Psychosocial theories have also been offered in explanation for the association between childhood sex-typed behavior and adult sexual orientation. Green (1987) theorized that, in childhood, pre-homosexual boys identify with their mothers and lack a close relationship with their fathers and other boys, which results in “male affect starvation.” In adolescence, homoerotic contact is used in some compensatory manner to achieve closeness with other males. It is unclear how this formulation would account for males who do not have a close relationship with their father or other adult males but grow up to be heterosexual. In Bem’s (1996, 2008) “exotic becomes erotic” theory of sexual orientation, biological factors influences a child’s temperament which, in turn, influences a child’s preference for same-sex versus opposite-sex peers. Children feel similar to the peers with whom they socialize and different from those with whom they do not socialize. This feeling of being different results in physiological arousal, which is then transformed to sexual arousal. For example, effeminate boys who prefer female peers initially feel different from boys and experience them as “exotic.” Later in development, these exotic feelings become erotic feelings. It is unclear how Bem’s theory would account for homosexual adults who were stereotypically masculine in childhood.
1.10 Associated Psychopathology in GID
One goal of the present study is to identify whether there are within-group childhood
characteristics, including childhood behavior problems, that were predictive of long-term outcomes of boys with GID. Thus, the remainder of this chapter will focus on associated psychopathology in GID and will conclude with a conceptual framework for and goals of the present study.
1.10.1 Children with GID
126.96.36.199 Behavior Problems in Children with GID
Studies on associated psychopathology in children with GID have taken two general approaches. One approach is to examine the presence of general psychopathology or behavior problems as would be reported on such measures as the CBCL. A second approach has been to examine the presence of other DSM diagnoses in children with GID.
Information on associated psychopathology or general behavior problems in children with GID has been systematically obtained from parent-report data on the CBCL. Using maternal-report data, Zucker and Bradley (1995) compared 161 gender-referred boys and 90 siblings on five indices of disturbance: the number of elevated narrow-band scales (T > 70), number and sum of items rated 1 or 2, and T scores for the Internalizing and Externalizing broad-band scales. On all five indices, gender-referred boys had significantly higher levels of behavioral disturbance compared to their siblings. Further, the boys with GID had significantly higher Internalizing T scores than Externalizing T scores; however, there was no significant difference between these scores for their siblings. Zucker and Bradley also compared 46 gender-referred boys pair-matched to 46 clinical control boys on the five CBCL indices described above: there was no significant difference between the groups on any of the indices. Cohen-Kettenis et al. (2003) found similar rates of behavior problems in a cross-national, cross-clinic comparison of 358 gender-referred children assessed at the CAMH in Toronto and 130 gender-referred children assessed in The Netherlands. In this study, boys with GID also showed more internalizing than externalizing problems on the CBCL.
CBCL studies of gender-referred children have found two other noteworthy patterns. First, there appears to be a relationship between age and degree of behavior problems, with older children having more behavior problems than younger children (Cohen-Kettenis et al., 2003; Zucker & Bradley, 1995). For example, in the Cohen-Kettenis et al. study, 26.1% of 3- to 5-year-old children in the Toronto sample had a CBCL sum score that was in the clinical range (>90th percentile) compared to 62.1% of 6- to 12-year-old children. The corresponding percentages for the Dutch sample were 43.8% and 61.7%, respectively.
Second, children with GID appear to have more peer relationship difficulties compared to their siblings (Zucker, Bradley, & Sanikhani, 1997). Further, gender-referred boys seem to have poorer peer relationships and experience more negative social consequences for their cross-gender behavior compared to gender-referred girls (Cohen-Kettenis et al., 2003; Wallien, Veenstra, Kreukels, & Cohen-Kettenis, 2010). Cohen-Kettenis et al. also found that poor peer relations in boys with GID was a significant predictor of CBCL behavior problems and accounted for 32% of the variance. Thus, poor peer relations may mediate the relationship between cross-gender behavior and psychopathology.
These findings are not surprising. Even among children unselected for their gender identity, there is evidence that gender nonconforming behavior is associated with overall adjustment problems (Carver et al., 2003; Rieger & Savin-Williams, 2012; Yunger et al., 2004). Studies of children unselected for their gender identity have also demonstrated that children tend to react negatively to gender-atypical peers and may resist developing friendships with these children (Fagot, 1977; Martin, 1989) and effeminate boys, in particular, receive more negative feedback from peers than masculine girls about their cross-gender behavior (Blakemore, 2003; Fagot, 1977). In an observational study of playmate preferences, non-referred boys were more likely to choose another non-referred boy as a favourite playmate than to choose a boy with GID (Fridell, 2001). On the other hand, a recent study showed that gender referred children are accepted by peers of the opposite sex and have better relations with opposite-sex than same-sex peers (Wallien et al., 2010). Thus, while gender-referred boys may be accepted by female peers, there is likely still a significant amount of ostracism experienced from same-sex peers. Zucker (2005c) conjectured that peer ostracism may also contribute to the above noted relationship between age and behavior problems in children with GID as older children with marked cross-gender behavior may experience relatively more social ostracism compared to younger children who engage in similar behaviors (see Zucker, Wilson-Smith, Kurita, & Stern, 1995).
In recent years, a new line of clinical research has suggested that some children with GID may have comorbid Pervasive Developmental Disorder (PDD), particularly Asperger’s Disorder (e.g., de Vries, Noens, Cohen-Kettenis, van Berckelaer-Onnes, & Doreleijers, 2010; Landén & Rasmussen, 1997; Mukkades, 2002; Zucker et al., 2012b). One explanation for a possible linkage between these two relatively rare disorders is the intense focus/obsessional interest in specific activities (e.g., Klin, Danovitch, Merz, & Volkmar, 2007). Postema et al. (2011) examined two items on the CBCL which measure obsessions (Item 9: “Can’t get his/her mind over certain thoughts; obsession”) and compulsions (Item 66: “Repeats certain acts over and over; compulsions”) in 528 gender referred children (435 boys, 93 girls) and 414 siblings (239 boys, 175 girls). Items are scored based on the past 6 months on a 0-2 scale where 0 = not true, 1 = somewhat or sometimes true, and 2 = very true or often true. For Item 9, 61.5% of gender-referred boys received a score of 1 or 2, compared to 27.3% of their male siblings, and 49% of referred boys in the CBCL standardization sample (Achenbach, 1991). For Item 66, 26.2% of gender referred boys received a score of 1 or 2, compared to 10.5% of their male siblings, and 26% of referred boys in the CBCL standardization sample. For both Items 9 and 66, the comparison between gender-referred boys and their siblings was significant. These preliminary results support the idea that boys with GID (and, generally, children with GID) show an apparent elevation in obsessional interests, which may be one reason for the observed comorbidity of GID and PDD. For the gender-referred boys, gender-related themes constituted more than half (54.6%) of the examples provided by their mothers.
In summary, children with GID display significantly more behavior problems compared to their siblings and non-referred children but typically display as many behavior problems as other clinical populations, with older children with GID having more behavior problems than younger children. Boys with GID typically present with more internalizing than externalizing difficulties, such as separation anxiety. Furthermore, gender-referred children, boys in particular, have poor peer relations and this was postulated to mediate the relationship between cross-gender behavior and behavior problems. The role of peer relations in the lives of children with GID is emphasized by the fact that, for some clinicians, a primary goal of treatment for gender-referred children is to prevent social ostracism and improve peer relations (Zucker & Bradley, 1995).
188.8.131.52 Comorbidity in Children with GID
Over 25 years ago, Coates and Person (1985) found that, in 25 boys (M age, 7.4 years) who met criteria for GID, 15 (60%) also met criteria for separation anxiety disorder (SAD). Coates and Person did not provide details on how they assessed for the presence of SAD, beyond a general statement that DSM-III criteria were used nor did they provide data on inter-rater reliability. A decade later, Zucker, Bradley, and Lowry Sullivan (1996b) examined the relationship between GID and SAD in 115 boys (M age, 6.3 years), of whom 73 (63.5%) met the full diagnostic criteria for GID and the remaining 42 (36.5%) were subthreshold for the diagnosis. To assess the presence of SAD, Zucker et al. developed a 21-item parent-report interview based on the DSM-III diagnostic criteria. The items assessed nine content domains which reflected the DSM-III criteria for SAD. Questions were phrased such that they could be answered by mothers as “Yes,” “Sometimes,” or “No.” Zucker et al. used conservative and liberal criteria to assign a diagnosis of SAD that differed in terms of the frequency of the symptoms reported. A conservative diagnosis was given if the mother answered “Yes” to questions in three of the nine content domains. A liberal diagnosis was given if the mother answered “Yes” or “Sometimes” in three of the nine content domains. Using the conservative criterion for SAD, there was no significant association between the presence of GID and SAD. A significant association existed when the liberal criterion was used. Of the 73 boys who meet full criteria for GID, 47 (64.4%) also met the criteria for SAD. In contrast, only 16 (38.1%) of the 42 boys who were subthreshold for the diagnosis of GID also met criteria for SAD.
A limitation of Coates and Person’s (1985) and Zucker et al.’s (1996) study is that neither included a group of clinical control participants referred for reasons other than GID. Nonetheless, the rates of SAD obtained in these studies are higher than the prevalence rates of SAD in children (for a review, see Masi, Mucci, & Milliepedi, 2005) and suggests that SAD may be overrepresented in boys with GID.
A recent line of research has led some authors to conclude that elevated levels of separation anxiety in childhood may be a component of the more general pattern of childhood femininity seen in some homosexual men (for a review, see VanderLaan, Gothreau, Bartlett, & Vasey, 2011a). Using a retrospective design, VanderLaan, Gothreau, Bartlett, and Vasey (2011b) found that homosexual men recalled, on average, significantly higher levels of separation anxiety compared to heterosexual men but similar to those of women. For the homosexual men but not heterosexual men, increases in recalled childhood femininity were associated with increases in separation anxiety. There is also cross-cultural support for this finding. Vasey, VanderLaan, Gothreau, and Bartlett (2011) found that Samoan fa’afafine recalled significantly more childhood separation anxiety compared to Samoan women and (heterosexual) men. These results raise the question of whether boys with GID who develop a homosexual sexual orientation would have experienced more separation anxiety in childhood compared to those who develop a heterosexual sexual orientation. None of the prospective studies on boys with GID have specifically examined this.
Researchers in The Netherlands systematically assessed the prevalence of DSM diagnoses in children with GID using a structured parent-report interview (Wallien, Swaab, & Cohen-Kettenis, 2007). Wallien et al. compared 120 children referred to a specialized gender identity clinic and diagnosed with GID (86 boys, 34 girls; age range, 4-11 years) to 47 children (37 boys, 10 girls) referred to an ADHD clinic and who served as a clinical control group. Of the 120 children with GID, 62 (51.6%) met criteria for a DSM diagnosis other than GID. There was no significant difference between the GID and ADHD children on number and type of comorbid diagnoses. Of the 86 GID boys, 48 (55.8%) met criteria for a diagnosis other than GID. Of these, more than half (n = 26, 56.1%) met criteria for an anxiety disorder (mostly specific phobia), 22 (25.5%) met criteria for a disruptive disorder (e.g., oppositional defiant disorder), and 4 (4.7%) met criteria for a mood disorder. In contrast to the studies by Coates and Person (1985) and Zucker et al. (1996b), only 5 (5.8%) of boys with GID met criteria for SAD.
The above described associated psychopathology in children with GID can be understood in several ways. First, as noted earlier, it may be the result of social ostracism (Zucker, 2005c). Second, it has been shown that behavior problems in gender-referred children may be significantly associated with measures of parental psychopathology and this may reflect generic familial vulnerability to psychopathology (Zucker & Bradley, 1995; Zucker et al., 2003). Third, associated psychopathology in children with GID may be implicated in the etiology of GID and, therefore, may contribute to its genesis (Coates & Person, 1985; Zucker et al., 2012b). Regardless of the mechanisms underlying the relationship between childhood cross-gender behavior and general behavior problems in gender-referred children, it raises the question of whether the associated psychopathology in childhood is related to long-term psychosexual and psychiatric outcomes, a question the present study will attempt to address.
1.10.2 Adolescents and Adults with GID
CBCL data suggest that adolescents with GID have significantly higher levels of behavioral problems (e.g., Internalizing T score) and poorer peer relations compared to children with GID (Zucker, Owen, Bradley, & Ameeriar, 2002). As discussed earlier, this “age effect” could represent a proxy for social ostracism (i.e., youth with gender identity issues encounter more problems in their social relationships as they age). Similar to that observed in children with GID, gender-referred adolescents are as likely to have as many behavior problems and similar peer relations as other clinically referred adolescents (i.e., referred for reasons other than gender) and to have considerably more behavioral problems and poorer peer relations than nonreferred adolescents (Zucker et al., 2012b; for a review, see Zucker, 2006c).
The presence of comorbid psychiatric diagnoses in adolescents with GID is an extremely important clinical issue as it can influence decisions regarding treatment recommendations (Cohen-Kettenis, Delemarre-van de Waal, & Gooren, 2008; Zucker et al., 2011). Only one study has systematically examined comorbidity of DSM diagnoses in adolescents with GID. Using a structured parent-report interview, de Vries, Doreleijers, Steensma, and Cohen-Kettenis (2011b) examined the presence of DSM diagnoses (other than GID) in 105 gender referred adolescents (53 males, 52 females; M age at assessment, 14.6 years). Of the 105 adolescents, 32.4% met criteria for at least one DSM diagnosis other than GID, which is lower than the obtained prevalence rate of comorbid diagnoses of 52% seen in children with GID (see Wallien et al., 2008). These results are, therefore, in contrast to studies showing more problem behaviors in adolescents compared to children. For the entire sample, the most common disorders were social phobia (9.5%) and major depression (8.6%). The presence of comorbid diagnoses did not vary according to GID diagnosis (i.e., full criteria versus sub-threshold). Of the 53 boys in the study, 39.6% met criteria for at least one concurrent disorder. Social phobia, major depression, and oppositional defiant disorder were the most common diagnoses. Further, the biological males in this study were at increased odds of having a mood or anxiety disorder compared to the natal females. de Vries et al. suggested that perhaps natal males with GID show patterns of psychopathology most typically associated with their desired gender (i.e., females). The most prominent limitation of this study was that parents, not the adolescent themselves, were the informants. This might have resulted in an underestimate of some diagnoses, such as alcohol or substance abuse and, by extension, an underestimate of the prevalence of comorbid diagnoses in adolescents with GID.
Comorbidity with other Axis I psychiatric disorders have been examined in adults with GID. The rates obtained have varied across studies. Cole, O’Boyle, Emory, and Meyer (1997) relied on historical information to assess comorbidity in 318 male-to-female transsexuals seen in a gender identity clinic. Only 6% of the patients reported a history of treatment for an Axis I disorder other than substance use or GID. The most common comorbid diagnosis was major depression. Other studies have used unstructured and semi-structured interviews to assess for other psychiatric conditions and these have typically founds higher rates of comorbidity. In some studies, data were reported for male-to-females and females-to-males combined. Bodlund and Armelius (1994) reported that 44% of a mixed group of male-to-female and female-to-male had another Axis I diagnosis other than GID. Haraldsen and Dahl (2000) reported similar results: in a mixed group of male-to-female and female-to-male transsexuals that included both pre-operative (i.e., applied for sex reassignment surgery) and post-operative patients (i.e., completed sex reassignment surgery), 33% met criteria for another Axis I diagnosis. De Cuypere, Jannes, and Rubens (1995) reported a lower rate of comorbidity. Only 23% of the male-to-female patients in their study met criteria for another Axis I disorder. In a recent study of adult patients with GID in Japan, Hoshiai et al. (2010) reported that 18% of the male-to-female patients had another Axis I diagnosis. Across these studies, the most common comorbid diagnoses were a mood or anxiety disorder. Hoshiai et al. reported that 76% if the male-to-female patients reported a lifetime history of suicidal ideation. De Cuypere, Jannes, and Rubens (1995) and Verschoor and Poortinga (1988) reported high rates of lifetime suicide attempt in their samples of male-to-female patients, 54% and 19% respectively.
Prevalence estimates of comorbid substance use in male-to-female adults with GID have varied. None of the male-to-female patients in Hoshiai et al.’s (2010) study had comorbid substance-related disorders. In contrast, Cole et al. (1997), Hepp, Kraemer, Schnyder, Miller & Delsignore (2005), and De Cuypere et al. (1995) reported fairly high lifetime prevalence figures of substance abuse in the male-to-female patients: 29%, 50%, and 50%, respectively.
De Cuypere et al. (1995) reported personality disorders in 70% of the male-to-female patients in their sample. Most other studies, however, have found lower rates of comorbid Axis II disorders. For example, Hepp et al. (2005), Bodlund and Armelius (1994), and Haraldsen and Dahl (2000) reported prevalence rates of 42%, 33%, and 20% respectively. Of note, these latter studies reported on mixed groups of male-to-female and female-to-male patients. With the exception of De Cuypere et al.’s study, these rates are similar to those seen in other clinical populations but higher than in nonclinical populations (e.g., Newton-Howes et al., 2010; Zimmerman, Rothschild, & Chelminski, 2005).
1.10.3 Suicidality and Victimization in Transgendered Populations
Even among youth unselected for their gender identity, the presence of gender atypical behavior appears to be associated with poorer well being (Rieger & Savin-Williams, 2012) and places these youth at higher risk for victimization compared to gender-typical youth. Thus, transgender youth are particularly vulnerable to victimization because of their atypical gender role presentation (for a discussion, see Stieglitz, 2010). Nuttbrock et al. (2010) found that 50.1% of male-to-female transgendered adults (from a sample of 517) experienced physical abuse while 78.1% experienced verbal abuse that was related to their gender identity/gender role. In an earlier study, Grossman, D’Augelli, and Salter (2006) found that 87% of male-to-female transgendered youth had experienced some type of victimization. These percentages appear to be higher than the estimated victimization rates for youth in North America of 50%-62% (e.g., Kilpatrick, Saunders, & Smith, 2002; Romano, Bell, & Billette, 2011; Saunders, 2003).
A number of population-based studies have found that transgendered youth/adults may be at an increased risk for suicide attempts compared to youth in the general population (for a review, see Haas et al., 2011). It has been estimated that 3.13%-4.6% of adults in the United States have made at least one suicide attempt (Kessler, Borges, & Walters, 1999; Weissman et al., 1999). In staggering contrast, studies of community samples of youth who self-label as transgendered have found that approximately 30-40% have made at least one suicide attempt (Clements-Nolle, Marx, & Katz, 2006; Grossman & D’Augelli, 2007; Maugen & Shipherd, 2010).
Transgendered individuals may be at even greater risk for suicide ideation and attempts than homosexual males (Mathy, 2003) and this may be directly related to discrimination and victimization pertaining to their gender atypical behaviors (Clements-Nolle et al., 2006; Nuttbrock et al. 2010). Given that the majority of boys with GID later develop a homosexual sexual orientation with desistence of GID, it has been suggested that they may share some of the same risk factors that have been identified in gay and lesbian adolescents, including stigmatization, rejection by the peer group, discrimination, and mental health difficulties (see e.g., D’Augelli, 2002; Lombardi, Wilchins, Priesing, & Malfouf, 2001; Mathy, 2003; Meyer, 2003; Morrow, 2004; Sandfort, Bakker, Schellevis, & Vanwesenbeeck, 2006; Savin-Williams & Ream, 2003). A large body of evidence, including well conducted epidemiological studies indicates that individuals from sexual minority populations are at an increased risk for various mental health difficulties, most notably depression, suicide, anxiety, and substance use, compared to their heterosexual counterparts (Cochran, 2001; Cochran & Mays, 2000; Cochran, Sullivan, & Mays, 2003; D’Augelli, 2002; Faulkner & Cranston, 1998; Ferguson, Horwood, & Beautrais, 1999; Gilman et al. 2001; King et al., 2008; Meyer, 2003; Remafedi, French, Story, Resnick, & Blum, 1998), with prevalence rates of suicide attempts ranging from 23% to 42% (D’Augelli & Hershberger, 1993; D’Augelli, Hershberger, & Pilkington, 1998, 2001; Safren & Heimberg, 1999; for a review, see McDaniel, Purcell, & D’Augelli, 2001). Furthermore, studies have found that the increased risk for suicide attempts among sexual minority youth is maintained even after controlling for substance abuse and other psychiatric comorbidity, such as depression (e.g., Herrell et al., 1999). That an association exists between homosexuality and mental health risk is now generally accepted (see meta-analysis by King et al., 2008).
1.11 A Conceptual Framework for the Present Study
The extent to which childhood behavior is predictive of behavior in adulthood has been a central question in the field of developmental psychology. Historically, researchers took a retrospective approach to examining the relationship between adult and child psychopathology. The major methodological limitation with retrospective studies, however, pertains to possible distortions in memory in recalling childhood symptoms or behavior (e.g., Verhulst, 1995). This limitation, combined with the increased recognition that some childhood problems continued into adulthood while others did not, argued for a prospective approach (Achenbach, 1997) and resulted in a wave of prospective studies which examined the developmental course of psychopathology in children.
Prospective studies of children with a particular disorder have found that some children continue to have the same disorder in adolescence and adulthood. Other children, however, show a desistence of the disorder from childhood to adolescence and, therefore, demonstrate discontinuity of the disorder over time (Maughan & Rutter, 2008). Thus, children with the same diagnosis in childhood can have multiple outcomes in adolescence and adulthood, which is commonly referred to as multifinality (Cicchetti & Rogosch, 1996). A longitudinal approach is necessary if one is to understand both continuities and discontinuities in psychopathology over time (Verhulst & Koot, 1991).
These developmental psychopathology concepts are particularly relevant to any prospective follow-up studies of boys with GID. The extant literature on boys with GID indicates that most grow up to feel comfortable with their biological sex and show a desistence of GID. Of the boys who show desistence of GID, most develop a homosexual sexual orientation and a minority develops a heterosexual sexual orientation. Thus, it appears that only a minority of boys with GID show persistence of the disorder into adolescence and adulthood. Therefore, within a group of boys with GID, discontinuity of the disorder from childhood to adolescence appears to be the most common developmental trajectory; however, there are multiple long-term outcomes for boys with GID. To understand this finding, it is necessary to conduct within-group analyses (McNeil & Kaij, 1979) to identify childhood factors that can predict outcome in adolescence and adulthood (e.g., with regard to persistent GID vs. desistent GID). Further, the percentage of children with GID who show persistent gender dysphoria into adolescence and adulthood is higher than the estimated prevalence rate of GID in the general population. Children with GID, then, can be conceptualized as “high-risk” for GID in adulthood when compared to children without GID.
The present study can be conceptualized as a within-group comparison of a group of children “at risk” for GID to develop an understanding of the childhood factors that can predict the various observed outcomes. Four decades ago, Green (1970) recognized that the developmental picture leading to “transsexualism” in adulthood can only be clearly understood by conducting longitudinal research. On understanding the development of adult transsexualism, Green wrote, “One way is to study a group of young children who would appear to be of high risk with respect to later manifestation of anomalous sexual and gender behavior, and to study their families. Most adults who request sex reassignment report difficulties in adopting appropriate gender-typic behavior during childhood (p. 271).”
Several studies using prospective designs have been conducted to examine predictors of outcome in children with other psychiatric disorders, including schizophrenia (Werry & McCleenan, 1991), bipolar disorder (Geller, Fox, & Clark, 1993; Werry, McClellan, & Chard, 1991), attention deficit-hyperactivity disorder (Dalsgaard, Mortensen, Frydenberg, & Thomsen, 2002), conduct disorder (see Robins, 1966), and pervasive developmental disorders (Szatmari, Bryson, Boyle, Streiner, & Duku, 2003). Notably, in adults with antisocial disorder, a similar disjunction between retrospective and prospective studies as observed in GID is also found: severely antisocial adults were antisocial children; however, only about half of antisocial/ conduct disordered children grow up to become antisocial adults (for a review, see Maughan & Rutter, 2008). Longitudinal studies have also been conducted on children identified as “at-risk” for developing schizophrenia later in life, with the goal of identifying predictors of within-group outcome (i.e., development of schizophrenia versus no schizophrenia later in life) (Erlenmeyer-Kimling et al., 1997; Neale & Weintraub, 1975;). These studies have suggested attempts to predict outcome should include demographic, psychosocial, and psychological factors.
1.11.1 Rationale for the Present Study
There are several reasons why a systematic examination of the long-term outcome of boys with GID is urgently needed, with both theoretical and clinical implications. The follow-up studies on boys with GID have focused predominantly on examining outcome vis-à-vis gender identity and sexual orientation. At present, very little is known about the long-term psychiatric functioning among boys with GID.
Although a large majority of boys with GID no longer have the disorder in adulthood, there is variation in the rates of persistence obtained in follow-up studies. Additional follow-up studies are needed to better clarify what proportion of boys may persist in having the disorder in adolescence and adulthood. Further, there are very limited data on childhood characteristics that may differentiate persisters from desisters. For example, it appears that children with more severe cross-gender identification are more likely to show persistence of GID compared to children who are less severe (Wallien & Cohen-Kettenis, 2008). However, other childhood factors may also contribute to the developmental course of GID. None of the follow-up studies to date have systematically conducted an evaluation of childhood demographic, psychosocial, and psychological variables and their role in the persistence and desistence of GID. Green (1970) stated, “The developmental picture leading to adult transsexualism remains smudgy” (p. 271). Four decades later, the picture is still “smudgy.”
The identification of predictors of GID outcome also carries clinical implications. In children with GID who appear to be on a trajectory for GID in adolescence and adulthood, intervention efforts can attempt to steer them away from a transsexual outcome onto one in which comfort with one’s biological sex is the outcome. The rationale for such an intervention is the observed complexity of embarking on a life course of cross-sex hormonal and surgical treatment combined with the pervasive social discrimination and victimization that many people with gender dysphoria encounter (Zucker, Drummond, Bradley, & Peterson-Badali, 2009). Green (2008), more directly, expressed that living as a homosexual adult is generally easier than living as a transsexual. Some clinicians have argued for early recognition of children who may persist in their gender dysphoria, but from a different standpoint. If a clinician is certain that a child with GID is committed to a pathway leading to transsexualism (i.e., persisting GID), interventions with gonadotropin releasing hormone (GnRH) blockers (commonly referred to as “puberty blockers”) could be used to delay the onset of puberty (Wallien & Cohen-Kettenis, 2008).
It is also clinically relevant to examine childhood factors that may predict outcome with regard to psychological functioning in adolescence and adulthood. If it were possible to identify children who may be at risk for developing psychiatric problems in adulthood, treatment can also focus on reducing the risk for psychopathology. In the field of developmental psychopathology, in general, the identification of predictors of change in psychopathology has been regarded theoretically important for advancing our knowledge of the development of psychopathology and as clinically relevant for informing treatment guidelines (Mathijssen, Koot, & Verhulst, 1999).
The present study is also urgently needed given the intense controversy in the field regarding best treatment practice for children with GID. While some parents seek therapeutic support that helps their child feel more comfortable with a gender identity that matches the child’s birth sex, a more extreme group of parents, and some therapists, actively promote an early social transition in the child, sometimes as young as preschool (Zucker et al., 2009). Unfortunately, there are no comparative data on the effects of these treatment approaches on long-term psychosexual and psychological outcome of children with GID. Given that the follow-up studies conducted thus far have found that most children with GID show a desistence, treatment approaches, particularly those that espouse early transitioning, need to be evaluated. Although the present study does not include an evaluation of therapeutics, data on persistence, desistence, and the factors that can predict outcome may, in the interim, inform treatment approaches.
Beyond the contextual importance of this study, the strong methodological design warrants specific mention. This present study differs from previous follow-up studies in several ways. First, the number of boys in our sample was considerably larger than the sample sizes of its predecessors. Second, I have extensive assessment data on psychological functioning in childhood and at follow-up. Third, at follow-up, cross-informant data (via parent-report) on psychological functioning and gender role are available. Fourth, the present study includes a measure of social desirability, which is a particularly important construct to measure in studies that examine gender identity and sexual orientation.
1.11.2 Goals of the Present Study
The first goal of the study was to provide descriptive data on the long-term outcome of boys with GID with regard to four possible outcomes: (1) persistence of gender dysphoria with a bisexual/homosexual sexual orientation, (2) persistence of gender dysphoria with a heterosexual sexual orientation, (3) desistence of GID with a bisexual/homosexual sexual orientation, and (4) desistence of GID with a heterosexual sexual orientation.
The second goal of the study was to provide descriptive data on the long-term psychiatric outcomes of boys with GID in terms of DSM psychiatric diagnoses and general behavior problems, both for the overall group of boys and according to their psychosexual outcome.
The third goal of the study was to identify childhood factors which can predict outcome at follow-up with regard to gender identity and sexual orientation. This included a systematic examination of demographic, psychological, and psychosocial variables as potential predictors of outcome.
The fourth goal of the study was to provide preliminary data on adult victimization experiences of boys with GID and its relationship to their psychiatric functioning.
The participants were biological males who had been referred to and then assessed in the Gender Identity Service, Child, Youth, and Family Program at the Centre for Addiction and Mental Health (CAMH) in Toronto, Ontario during their childhood and were adolescents or adults at the time of follow-up. Data collection occurred over three decades, 1986-2011 (Table 1).
Participants entered the follow-up study through two methods of recruitment. Most participants were recruited through routine contact for research follow-up. From 1986-2011, there were two main waves of participant recruitment through research contact, from 1986-1993 and, more recently, from 2009-2011. In addition, during the period of data collection, some adolescents who had been assessed in the clinic during childhood contacted the service for clinical reasons (e.g., persistent gender dysphoria, emerging sexual identity, or other clinical issues). These participants were informed about the opportunity to participate in the follow-up study and subsequently completed the study protocol. The majority of the patient-initiated participants had contacted the clinic between the two main waves of research recruitment. Thus, from 1994-2008, the participants who entered the study were primarily those who had contacted the service for clinical reasons. As such, the final study sample consisted of participants who entered the study through standard research contact or following participant-initiated involvement with the clinic.
2.1.1 Routine Contact for Research
A chart review was conducted of all boys (range, 3-12 years) assessed in the Gender
Identity Clinic Service at the CAMH since 1975 to identify eligible participants. Between 1975
and 2009, the clinic evaluated 463 boys referred for gender identity issues. To participate in the follow-up study, patients from this sample had to be at least 16 years of age. Using this cut-off,
the chart review identified 294 eligible participants, of which contact was attempted in 132 cases. For the remaining 162 cases, contact was not attempted due to lack of study resources and time constraints.
Of the 132 cases where contact was attempted, 19 (14.3%) potential participants could not be reached/traced through previous addresses, registrars, and personal contacts (e.g., the patient and/or family had moved and a current telephone number, mailing address, or e-mail could not be identified). Initial telephone contact was first made with the parents because participants were minors at the time of the childhood assessment and may have had no recollection of their clinic attendance (see Appendix C for phone script). In general, the response from parents and participants was positive. Of the 113 patients where contact was successful, 79 (69.9%) agreed to complete the study protocol and came into the clinic for a face-to-face assessment. In 28 (24.8%) cases, some follow-up data on gender identity and sexual orientation were provided over the phone by the parents or the patients themselves or were obtained through a chart review. For a variety of reasons, these 28 participants did not complete the standard follow-up assessment (e.g., geographic limitations, too busy). Appendix D summarizes the information acquired during a telephone conversation with the parent of a follow-up participant and serves as an exemplar of the information provided by parents, and which was used as the basis for group classification at follow-up. In the remaining 6 (5.3%) cases, either the parents did not allow us to speak to their now grown up child or the potential participant himself declined to participate in the study. In total, 107 participants entered the study through routine research recruitment.
2.1.2 Participant-Initiated Clinical Contact
An additional 32 participants were recruited into the study after they had contacted the Gender Identity Service for clinical reasons. In 7 cases, either the participant or the parents contacted the clinic due to persistent gender dysphoria. In 6 cases, the clinic was contacted because either the participant or the parent was concerned about sexual orientation and in 1 case there was concern about gender dysphoria and sexual orientation. Lastly, in the remaining 18 cases, the participants or parents contacted the clinic for other heterogeneous clinical concerns, including depression, substance abuse, parent-child conflict, and conduct problems. Appendix E lists the reasons for contacting the clinic for each of the 32 participants.
Some of the participants who contacted the clinic were younger than 16 years of age. It is important to note that all participants were aware that completing the follow-up study was voluntary and a decision to decline participation in the study would not have affected their involvement with the clinic. Of these 32 participants, 31 completed the standard follow-up protocol and 1 participant provided some data on gender identity and sexual orientation but did not complete the formal study.
2.1.3 Participation Rate
In total, 145 participants were approached about participating in the follow-up study, either through routine contact (n = 113) or following their clinical involvement with the Gender Identity Service (n = 32). Six participants declined, which yielded a participation rate of 95.9%.
2.1.4 Demographic Characteristics of Participants
The demographic characteristics of the participants in childhood and at follow-up are shown in Table 2. The GID diagnosis in childhood was based on the DSM (3rd ed. [DSM-III]; 3rd ed., rev. [DSM-III-R]; or 4th ed. [DSM-IV]; American Psychiatric Association [APA], 1980, 1987, and 1994, respectively) criteria applicable at the time of assessment. A total of 88 (63.3%) boys met complete DSM criteria for GID in childhood. The remaining 51 (36.7%) boys were subthreshold for a DSM diagnosis, but all had some indicators of GID, and, based on the historical information provided during the assessment, some would have met the complete DSM criteria at some point in their lives prior to their assessment in childhood.
The mean age of assessment in childhood was 7.49 years (SD = 2.66; range, 3.33-12.99) and the mean age at follow-up was 20.58 (SD = 5.22; range, 13.07-39.15). The mean time interval between childhood assessment and follow-up was 12.88 years (SD = 6.07; range, 2.77-29.29).
The study was approved by the Institutional Review Boards at the CAMH and the University of Toronto. The majority of participants who completed the face-to-face assessment were evaluated on a single day. Three participants were seen twice. In these instances, the participants completed the self-report measures during their second visit as the complexity of their clinical presentation extended the duration of the assessment. Participants were provided a
stipend for their participation in the follow-up assessment and reimbursement for travel expenses.
All participants who completed the face-to-face assessment gave written informed consent prior to their involvement in the follow-up assessment (see Appendix F). Participants were explained their right to participate, their right to withdraw at any time during the study, and assured anonymity. Concerns about emotional distress (e.g., endorsement of suicidality in the psychiatric interview) were addressed; although no participants required acute immediate care, there were instances in which intervention or clinical care was required. The majority of the follow-up procedures were conducted by the author of this thesis (n = 71) under the supervision of Dr. Kenneth Zucker. However, for participants assessed prior to 2009, the study procedures were carried out by Dr. Kenneth Zucker (n = 68).
Participants who completed the face-to-face interview responded positively to the assessment process. With one exception, rapport with the participants appeared to be good. At the time of follow-up, two participants were currently being seen in our clinic for therapy.
2.3.1 Childhood Assessment
184.108.40.206 Cognitive Functioning
Based on the age of clients at the time of assessment, the appropriate version of the Wechsler Intelligence Scale for Children was administered (WISC-R, WISC-III, WISC-IV or the WPPSI-R). Full scale IQ scores were used to characterize level of cognitive functioning.
220.127.116.11 Sex-typed Behavior
A variety of methods and measures were used to assess sex-typed behavior in childhood (i.e., gender identity and gender role), which are summarized in Table 3. A total of five child informant and two parent informant measures were used to assess the participants’ sex-typed behavior in childhood: (1) Draw-a-Person test (Zucker, Finegan, Doering, & Bradley, 1983); (2) a free-play task (Zucker, Doering, Bradley, & Finegan, 1982); (3) the Playmate and Playstyle Preferences Structured Interview (Fridell, Owen-Anderson, Johnson, Bradley, & Zucker, 2006); (4) sex-typed responses on the Rorschach test (Zucker, Lozinksi, Bradley & Doering, 1993); (5) the Gender Identity Interview for Children (Wallien et al., 2009; Zucker et al., 1993); (6) The Gender Identity Questionnaire for Children (Johnson et al., 2004); and (7) a measure of activity
level/extraversion (Zucker & Bradley, 1995). Some of these measures have been designed to assess gender identity constructs or a child’s general discomfort with his or her gender status, as well as gender role behaviors (e.g., toy preferences, fantasy play). Other measures assess well-established normative gender differences for dimensions of temperament that included ratings of activity level and involvement in rough-and-tumble play (see review in Zucker, 2005b).
18.104.22.168 Behavior Problems
The Child Behavior Checklist (CBCL; Achenbach, 1991; Achenbach & Edelbrock, 1983) was completed by a parent or guardian during the child’s initial assessment. This measure is a standardized parent-report questionnaire for ratings of behavior problems in children 4 to 18 years of age. It consists of 118 behavior problems that are rated on a 3-point scale (0 = not true, 1 = sometimes true or somewhat true, 2 = very true or often true). The questionnaire identifies two “broad-band” factors, “Internalizing” and “Externalizing,” that assess dimensions of child psychopathology. Internalizing disorders are described as covert emotional disturbances (e.g., depression, social withdrawal, anxiety) and the Externalizing disorders are described as disorders of overt conduct (e.g., aggression, hyperactivity). Parent ratings on the CBCL yield T scores for Total, Internalizing, and Externalizing problems. Reports indicate extensive evidence for the reliability and validity of the CBCL as an index of behavioral psychopathology in clinical and nonclinical populations (Achenbach, 1985, 1991; Achenbach & Edelbrock, 1983; Achenbach, McConaughy, & Howell, 1987). The intra-class correlations for individual items were more than .90. Test-retest reliability over a one-week period was .89, and inter-parent agreement for total behavior problem scores was .66. Furthermore, the long-term stability estimates for total behavior problem scores over 3-, 6-, and 18-month periods were .74, .60, and .46, respectively (Achenbach & Edelbrock, 1983). The CBCL discriminates well between children referred to mental health agencies from demographically-matched non-referred children (Achenbach & Edelbrock, 1983).
22.214.171.124 Peer Relations
Zucker, Bradley, and Sanikhani (1997) constructed a three-item Peer Relations Scale derived from the CBCL using Items 25, (“Doesn’t get along with other kids”), 38 (“Gets teased a lot”), and 48, (“Not liked by other kids”). Based on maternal ratings on the CBCL, both boys and girls with GID had poorer peer relations than did their siblings. Cronbach’s alpha was .81 and the mother-father correlation for the scale was .66 (n = 312, p < .001). The Peer Relations Scale was subsequently used in comparative study of children and adolescents with GID (Zucker, Owen, Bradley, & Ameeriar, 2002) where it was the strongest predictor of behavioral psychopathology on the CBCL for both children and adolescents. Cohen-Kettenis et al. (2003) used the Peer Relations Scale in a cross-clinic study in which children with gender identity issues seen at the Gender Identity Service at the CAMH were compared to children with gender identity issues seen at the Gender Clinic housed within the Department of Child and Adolescent Psychiatry at the University Medical Center Utrecht (The Netherlands) on demographic characteristics, social competence, and behavior problems. Across both clinics, boys with gender identity issues had poorer peer relations than did girls.
2.3.2 Follow-up Assessment
Table 4 summarizes the follow-up assessment protocol. The gender identity and sexual orientation measures, along with self-report measures of behavior problems, are used as part of the current standardized clinical assessment of adolescents referred to the Gender Identity Service (Zucker, 2005b; Zucker et al., 2012b; Zucker & Bradley, 1995). The victimization and suicidality questionnaires were novel to this follow-up evaluation, as was the standardized interview schedule to assess the presence of DSM psychiatric disorders.
All interviews were conducted in a private office. The follow-up assessment protocol was administered in the following order to all participants: (a) cognitive testing, (b) psychiatric diagnostic interview, (c) semi-structured clinical interview, and (d) self-report questionnaires. The semi-structured clinical interview consisted of four parts: (a) the participant’s current functioning (e.g., family, school, and work), (b) recollections of and thoughts about the childhood assessment and childhood gender role/gender identity, (c) current gender identity, and (d) sexual orientation in fantasy and behavior. The semi-structured interview always preceded the self-report questionnaires and was audiotaped, except in one instance where the participant did not give consent. The self-report questionnaires provided an immediate cross-check of the interview and served additional quantitative purposes. The contents of the follow-up assessment are described below.
126.96.36.199 Cognitive Functioning
Four subtests from the age-appropriate version of the Wechsler Intelligence Scales were administered (Vocabulary, Comprehension, Block Design, and Object Assembly). The scaled scores from these subtests were used to provide an estimated IQ score for cognitive functioning (Sattler, 2001). Data collection occurred over a period of 24 years. During this time, the Wechsler Intelligence Scales underwent revisions and the most recent version at the time of assessment was used for cognitive testing. As a result, over the period of data collection, several versions of the Wechsler scales were used. Participants 17 years or older were administered the Wechsler Adult Intelligence Scale-Revised (n = 8; WAIS-R; Wechsler, 1981), the Wechsler Adult Intelligence Scale-Third Edition (n = 16; WAIS-III;
Wechsler, 1997) or the Wechsler Adult Intelligence Scale-Fourth Edition (n = 32; WAIS-IV; Wechsler, 2008). Participants under the age of 17 were administered either the Wechsler Intelligence Scale for Children-Revised (n = 25; WISC-R; Wechsler, 1974), the Wechsler Intelligence Scale for Children-Third Edition (n = 16; WISC-III; Wechsler, 1991), or the Wechsler Intelligence Scale for Children-Fourth Edition (n = 12; WISC-IV, Wechsler, 2003).
188.8.131.52 Behavioral Functioning
For participants younger than 19 years, the Youth Self-Report (YSR; Achenbach, 1991) was used to assess behavior and emotional problems. For participants 19 years and older, the Young Adult Self-Report (YASR; Achenbach, 1997) or the Adult Self-Report (ASR; Achenbach & Rescorla, 2003) was used. T scores for Internalizing, Externalizing, and Total Behavior Problems were derived for each participant. The YSR is a 103 item self-report questionnaire that assesses emotional and behavior problems for ages 11 to 18 years. The YASR, the predecessor of the Adult Self-Report, contains 119 items to measure psychopathology in individuals 18 to 30 years. The YASR was utilized in data collection until the ASR was published. The ASR contains 126 items to measure psychopathology for individuals aged 18 to 59 years. The YSR, YASR, and ASR are variations of the CBCL and have the same 3-point format (0 = not true, 1 = sometimes true or somewhat true, 2 = very true or often true). The YSR has been reported to have good reliability and validity for identifying adolescent behavior problems (Achenbach, 1997). Several studies have used these scales to examine the course and predictive value of self-reported problems among adolescents and adults (Hofstra, van der Ende, & Verhulst, 2000, 2001).
Although assessment of adult psychopathology relies extensively on self-report, meta-analyses of studies comparing self-report and an informant’s report revealed moderate correlations between cross-informant data for substance abuse, internalizing, and externalizing problems (Achenbach, Krukowski, Dumenci, & Ivanaova, 2005). Participants gave permission to have their parent/guardian or spouse/partner complete the Child Behavior Checklist (CBCL; Achenbach, 1991) for those participants less than 18 years or the Adult Behavior Checklist (ABCL; Achenbach & Rescorla, 2003) for participants 18 years or older. Forms were given to those parents or guardians that came to the clinic with the participant or sent by mail to those that did not accompany the participant. T scores for Internalizing, Externalizing, and Total Behavior Problems were derived from the informant forms. Of the 110 participants who completed the follow-up assessment, parent report data were missing for 16 participants, either because the adult participants did not give consent to have their parents complete questionnaires or the parents did not return the questionnaires.
184.108.40.206 Psychiatric Functioning
Selected modules of the Diagnostic Interview for Children and Adolescents (DICA; Herjanic & Reich, 1982) or the Diagnostic Interview Schedule (DIS, Version IIIA; Robins, Helzer, Croughan, & Ratcliffe, 1981) were administered depending on age at follow-up. Participants younger than 18 were administered the DICA (n = 64) and those 18 years or older were administered the DIS (n = 44). Psychiatric data were not available for two of the 110 participants who completed the face-to-face assessment.
The DICA is a semi-structured psychiatric interview for youth between 6 and 17 years to assess the presence or absence of psychiatric diagnoses based on the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III, American Psychiatric Association, 1980). The DICA was patterned after the DIS. Ten DICA modules were included in this study: (1) Oppositional Disorder, (2) Conduct Disorder (Aggressive and Non-aggressive), (3) Alcohol Abuse and Alcohol Dependence, (4) Major Depression, (5) Adjustment Disorder with Depressed Mood, (6) Mania, (7) Separation Anxiety, (8) Overanxious Disorder, (9) Marijuana Abuse, and (10) Other Drug Abuse and Drug Dependence. The total number of possible diagnoses a participant could meet criteria for on this measure was 13. The DICA has been demonstrated to have good reliability (Welner et al., 1987) and adequate validity (Carlson et al.1987; Ezpeleta et al., 1997).
The DIS is a standardized structured interview devised to assess the presence or absence of psychiatric diagnoses in adults based on the DSM-III (American Psychiatric Association, 1980). Fourteen modules were included in this study: (1) Depression (Major Depression [single episode], Major Depression [recurrent], Grief Reaction), (2) Dysthymic Disorder, (3) Obsessive-Compulsive Disorder, (4) Agoraphobia, (5) Social Phobia, (6) Simple Phobia, (7) Panic Disorder, (8) Agoraphobia with Panic Attacks, (9) Generalized Anxiety Disorder, (10) Alcohol Abuse and Alcohol Dependence, (11) Drug Abuse and Drug Dependence, (12) Anorexia Nervosa and Bulimia, (13) Schizophrenia, and (14) Bipolar Disorder . The total number of possible diagnoses a participant could meet criteria for on this measure was 17. The DIS has been found to have excellent inter-rater reliability (Anduaga, Forteza, & Lira, 1991; Hesselbrock et al., 1982). There have been some concerns about the validity of the DIS (e.g., Helzer et al., 1985) as studies have yielded mixed results with kappas ranging from low to moderate (for review, see Rogers, 2001). Following the publication of the DSM-IV, the Diagnostic Interview Schedule for DSM-IV (DIS-IV) was published (Robins, Cottler, Bucholz, & Compton, 1995) to correspond to the updated DSM diagnostic criteria. Test-retest reliability and validity of the DIS-IV was assessed in a sample of drug users (Dascalu, Compton, Horton, & Cottler, 2001; Horton, Compton, & Cotter, 1998). Test-retest reliability and validity ranged from excellent to fair for most disorders. However, test-retest reliability was considered poor for generalized anxiety disorder and specific phobia, kappa = .35 and .25, respectively (for more details on specific diagnoses, see Compton & Cottler, 2004).
For the DIS and the DICA, the total number of diagnoses each participant met criteria for was calculated.
220.127.116.11 Psychosexual Variables
The present study examined seven self-report measures of psychosexual outcome and one parent-report measure pertaining to sex-typed behavior.
18.104.22.168.1 Recalled Childhood Gender Identity and Gender Role Behaviors
Participants completed the Recalled Childhood Gender Identity/Gender Role Questionnaire (Zucker et al., 2006). This questionnaire consists of 23 items pertaining to various aspects of childhood sex-typed behavior and identification (e.g., peer preferences, toy preferences, roles in fantasy play, and feelings about being a male as a child), as well as relative closeness to mother and father during childhood. Items were rated on a 4-point or 5-point scale. For some items, however, an additional response option allowed the participant to indicate that the behavior did not apply (e.g., for the question about favorite playmates, there was the option “I did not play with other children”). Each participant was instructed to make ratings for their behavior as a child (“between the years 0 to 12”). Factor analysis identified two factors, accounting for 37.4% and 7.8% of the variance, respectively (all factor loadings ≥ .40). Factor 1 consisted of 18 items that pertained to childhood gender role and gender identity and Factor 2 consisted of three items that pertained to parent-child relations (relative closeness to one’s mother vs. father). Significant variation in factor scores between groups completing the RCGI was reported (e.g., heterosexual versus homosexual adults, adolescents with GID versus transvestic fetishism, women with CAH versus their sisters or female cousins), demonstrating that the questionnaire has good discriminant validity (Singh et al., 2010; Zucker et al., 1996a, 2006). For the present study, this questionnaire was used as a measure of recalled gender role and gender identity behaviors and, as such, the mean Factor 1 score was computed for each participant. Lower scores represent more recalled childhood cross-gender behaviors. The RCGI was added to the assessment protocol in 1993, after data collection had begun, so some data were missing for individuals who completed the follow-up prior to the inclusion of this measure.
22.214.171.124.2 Concurrent Gender Identity
Concurrent gender identity was evaluated using interview and self-report data. During an audiotaped interview, each participant was asked to describe their current feelings about being a biological male. They were also asked to describe positive and negative aspects about their gender identity. For example, participants who reported a “male” gender identity were asked to describe positive and negative aspects of being male. Participants were also administered a semi-structured gender identity interview (see Appendix G) based on the adolescent and adult GID criteria outlined in the DSM-IV-TR (American Psychiatric Association, 2000). The interview contained five questions related to the Point A criteria (e.g., the stated desire to be a woman, passing as a woman, the desire to live or be treated as a woman) and four questions from the Point B criteria (e.g., a preoccupation with getting rid of their penis, belief that they should have been born a woman). Participants were asked to respond to these questions according to the last 12 months with No, Sometimes, or Yes.
Two self-report measures were also used to assess current gender identity and gender dysphoria: (a) The Gender Identity/Gender Dysphoria Questionnaire for Adolescents and Adults (GIQAA; Deogracias et al., 2007; Singh et al., 2010), or (b) the Gender Dysphoria Questionnaire (GDQ; Zucker et al., 1996a). The GDQ was developed prior to the GIQAA. As such, the GIQAA was introduced to the clinical battery subsequent to the GDQ and, as a result, the more recent participants were administered the GIQAA while earlier participants were given the GDQ. The GIQAA and GDQ differ in number of items and rating scale. They also differ in the wording of questions. In order to integrate the data collected from the GIQAA and GDQ, participants’ score on each measure was converted to a proportion score which ranged from 0-1. The proportion score, referred to hereafter as the gender dysphoria score, was used in all analyses instead of the mean scores on the GIQAA and GDQ. The gender dysphoria score reflects the proportion of the highest possible score that each participant received on either the GIQAA or the GDQ. For example, a participant with a proportion score of 1.00 on either the GIQAA or GDQ received the maximum score on that measure. A higher proportion score indicates more gender dysphoria. There were, however, some participants who completed the GIQAA and the GDQ (n = 17). In these cases, the GIQAA proportion score was used as it is a more psychometrically sophisticated measure of gender dysphoria compared to the GDQ.
The male version of the GIQAA (Deogracias et al., 2007; Singh et al., 2010) was completed. This 27-item questionnaire measures gender identity and gender dysphoria in adolescents or adults; participants over the age of 17 completed the adult version and younger participants completed the adolescent version. The adolescent and adult versions are identical in all regards except that, in the adult version, the words “man” and “woman” are used instead of “boy” and “girl.” The items were developed by the North American Task Force on Intersexuality (NATFI) Research Protocol Working Group (S. J. Kessler, H. F. L. Meyer-Bahlburg, J. M. Schober, and K. J. Zucker) based on prior measures, expert panels, and clinical experience, and were designed to capture multiple indicators of gender identity and gender dysphoria, including subjective (n = 13 items), social (n = 9 items), somatic (n = 3 items), and sociolegal (n = 2 items) parameters. Each item was rated on a 5-point response scale from Never (coded as 1) to Always (coded as 5) based on a time frame of the past 12 months. Item examples include the following: “In the past 12 months, have you felt unhappy about being a man?” and “In the past 12 months, have you had the wish or desire to be a woman?” Principal axis factor analysis identified a one-factor solution that accounted for 61.3% of the variance. All factor loadings were ³ .30 (median, .86; range, .34-.96). The GIQAA demonstrated evidence for discriminant validity and a high threshold for specificity (i.e., low false positive rate for non-GID individuals).
In a replication and extension of this study, Singh et al. (2010) compared adults and adolescents with GID to clinical controls (i.e., evaluated for reasons other than GID or transvestic fetishism). There was strong evidence for discriminant and convergent validity, with high sensitivity and specificity rates. Males and females with GID reported significantly more gender dysphoria than clinical controls, with large effect sizes (Cohen, 1988). Deogracias et al. (2007) suggested that the mean score on the Gender Identity Questionnaire for Adolescents and Adults (GIQAA) can be used to determine whether a participant was gender dysphoric, and, in their sample, used a mean score of ≤ 3 to indicate a “case” of gender dysphoria. Using this criterion for caseness, the measure was able to identify gender dysphoric adolescents with 93.3% sensitivity. For the clinical control participants, specificity (i.e., low false positive rate for non-GID individuals) was 100% (i.e., none of the clinical control participants met criteria for caseness). For adults, sensitivity was 87.8% for the GID participants and specificity was 100% for the clinical controls. In a subsequent study, Singh, McMain, and Zucker (2011) administered the GIQAA to 100 women diagnosed with borderline personality disorder (BPD), none of whom were ever evaluated in a gender clinic. Using the suggested GIQAA mean score of 3 as the cut-off for identifying gender dysphoria, none of the women with BPD met this criterion for caseness. Zucker et al. (2010) reported on 105 adolescent males with GID, of whom, 91.7% met criterion for caseness. Additional psychometric evidence for discriminant validity and clinical utility can also be found in Singh et al. (2011).
The Gender Dysphoria Questionnaire (Zucker et al., 1996a) contains 8 items pertaining to gender identity and gender dysphoria. Each item was rated on a 3-point or 5-point scale based on a time frame of the past 6 months. Item examples include the following: “In the past 6 months, how often have you wished that you had been a born a girl instead of a boy?” (with response options ranging from “never” to “a lot”) and “In the past 6 months, how have you felt about being a boy?” (with response options ranging from “very satisfied” to “very dissatisfied”). Factor analysis identified two factors, accounting for 31.4% and 12.5% of the variance, respectively (all factor loadings ≥ .45). Factor 1 consisted of 5 items pertaining to gender dysphoria and Factor 2 consisted of 3 items pertaining to gender role identification. For the present study, the mean Factor 1 score was computed for each participant.
Using the questionnaire data on concurrent gender identity, participants were classified as either gender dysphoric (“persisters”) or not gender dysphoric (“desisters”) at follow-up. The Results section outlines the specific criteria used to make this classification.
126.96.36.199.3 Concurrent Gender Role–Parent Report
The Gender Identity/Gender Role Questionnaire for Adolescents (GIGRQ-Ad) is a 13-item parent-report questionnaire pertaining to various aspects of concurrent sex-typed behavior (e.g., sex-of-peer affiliation preference, masculine vs. feminine interests, cross-dressing, the desire to be of the other sex; Zucker, Bradley, Owen-Anderson, & Singh, 2010b). Items were rated on a 5-point response scale, with the exception of a Peer Composite item, in which a difference score between the number of male and female friends was calculated. Factor analysis of the GIGRQ-Ad was based on a sample of 403 youth, including various comparison groups, such as siblings and clinical controls. A principal axis factor analysis identified a one-factor solution, accounting for 44.8% of the variance. Ten of the 14 items had factor loadings >.43 (range, .43-.82) and, for these items, a unit-weighted mean total score was derived. Cronbach’s alpha for this measure was .91. A lower score indicates more cross-gender behavior. Discriminant validity was reported in Zucker et al. (2010b).
188.8.131.52.4 Sexual Orientation
As stated previously, in sexology research sexual orientation is often measured using two metrics: sexual orientation in fantasy and sexual orientation in (overt) behavior (Green, 1987). For the present study, participants’ sexual orientation in fantasy and behavior was assessed using a multi-method approach: face-to-face interview and self-report. Participants’ sexual orientation was classified in relation to their biological (birth) sex.
184.108.40.206.4.1 Sexual Orientation in Fantasy
Each participant’s sexual orientation in fantasy was assessed with specific questions from an audiotaped face-to-face interview and the self-report Erotic Response and Orientation Scale (EROS; Storms, 1980).
Questions posed in the audiotaped interview asked about four types of sexual fantasy: (1) crushes on other people; (2) sexual arousal to visual stimuli (e.g., strangers, acquaintances, partners, and individuals presented in the media [video, movies, magazines, internet]; (3) sexual content of night dreams; and (4) sexual content of masturbation fantasies. During the interview, participants were not asked directly about the gender of the person or persons who elicited sexual arousal, thus allowing time for the participant to provide this information spontaneously. Directed questions about the gender of the person(s) who elicited sexual arousal were asked only if the participant did not volunteer specific information about whether their arousal was directed to same-sex or opposite-sex partners, or both. By the end of the interview, each participant provided information about sexual arousal to both same-sex and opposite-sex individuals. Using the Kinsey scale criteria (Kinsey, Pomeroy, & Martin, 1948), the interviewer assigned Kinsey ratings that ranged from 0 (exclusively heterosexual in fantasy) to 6 (exclusively homosexual in fantasy) for each question. A dummy score of 7 denoted that the participant did not experience or report any fantasies. A global fantasy score was also derived based on ratings from the four questions. In the present study, only ratings for the last 12 months are reported since one goal of the study was to assess sexual orientation at the time of follow-up. During the interview, however, participants were asked about their “lifetime” (i.e., from age 13 until present) history of sexual fantasies. Kinsey ratings for sexual orientation in fantasy were available for 129 participants.
Inter-rater reliability on Kinsey ratings for sexual orientation in fantasy was examined for 29 participants, which were selected at random. The second scorer listened to the audio recordings of the semi-structured interview, with specific attention to the information collected on sexual orientation. The second coder was masked to participants’ group status at follow-up. The inter-rater agreement on the Kinsey Global Fantasy rating was very good (kappa = .95). For reference purposes, Appendix H lists the inter-rater agreement for each type of sexual fantasy evaluated during the semi-structured interview.
The Erotic Response and Orientation Scale (EROS) is a 16-item self-report measure assessing sexual orientation in fantasy over the past 12 months. Half of the questions pertained to heterosexual fantasy (e.g., “How often have you noticed you had sexual feelings (even the slightest) while looking at a woman?”) and the other half pertained to homosexual fantasy (e.g., “How often have you noticed you had sexual feelings (even the slightest) while looking at a man?”). Each item was rated on a 5-point scale for frequency of occurrence, ranging from “none” to “almost every day.” Mean homoerotic and heteroerotic fantasy scores were derived for each participant. Previous use of the EROS has shown good evidence of discriminant validity, such that it is able to discriminate between responses of self-identified heterosexual and homosexual men and women (Storms, 1980; Zucker et al., 1996a).
220.127.116.11.4.2 Sexual Orientation in Behavior
Each participant’s sexual orientation in behavior was assessed with specific questions
during the face-to-face interview and with a modified version of the Sexual History
Questionnaire (SHQ; Langevin, 1985).
In the interview, questions asked about five types of sexual behavior: (a) dating; (b) holding hands in a romantic manner; (c) kissing; (d) genital fondling or touching a woman on the breasts, and (e) intercourse (penile-vaginal and anal). Kinsey ratings for behavior in the past 12 months were made in the same manner as fantasy ratings.
Inter-rater reliability on Kinsey ratings for sexual orientation in behavior was examined for 29 participants, which were selected at random. The second scorer listened to the audio recordings of the semi-structured interview, with specific attention to the information collected on sexual orientation. The second coder was masked to participants’ group status at follow-up. There was perfect inter-rater agreement on the Kinsey Global Behavior rating (kappa = 1.0). For reference purposes, Appendix H lists the inter-rater agreement for each type of sexual behavior evaluated during the semi-structured interview.
The modified Sexual History Questionnaire (SHQ) consists of 20 questions. Ten questions pertained to heterosexual experiences (e.g., “How many women have you kissed on the lips in a romantic way?”) and 10 questions pertained to homosexual experiences (e.g., “How many men have you kissed on the lips in a romantic way?”). Participants who were 18 years and older completed the adult version and younger participants completed the adolescent version. The adolescent and adult versions are similar in all regard except that, in the adult version, the words “man” and “woman” are used instead of “boy” and “girl.” Each item was rated on a 5-point scale for frequency of occurrence, ranging from none to 11 or more, based on a time frame of the past 12 months. Mean total scores for heterosexual and homosexual experiences were derived.
18.104.22.168.4.3 Sexual Orientation Group Classification
On the basis of Kinsey ratings, participants who completed the face-to-face interview were classified into the following three sexual orientation groups for both fantasy and behavior: (1) heterosexual (Kinsey global ratings of 0-1); (2) bisexual/homosexual (Kinsey global ratings of 2-6), and (3) no sexual fantasy or behavior. Participants’ sexual orientation was classified in relation to their birth sex.
A comment is warranted on the decision to combine the bisexual and homosexual participants. This validity of this procedure has been questioned (e.g., MacDonald, 1983; Paul, 1993) on the basis that combining bisexuals with homosexuals in research samples, though standard among sex researchers, confounds the research on both groups. In the present study, and similar to Green’s (1987) follow-up study, participants who were categorized as bisexual or homosexual on the basis of their Kinsey ratings were combined for a number of reasons. On the basis of extensive clinical experience, Zucker and Bradley (1995) stated that it is common for homosexual men to recall a period of bisexual behavior during adolescence. It is, therefore, conceivable that some adolescent participants who reported bisexual behavior and/or fantasies will move toward an exclusively homosexual sexual orientation in adulthood. Research on sexual identity development has demonstrated that men who adopt a homosexual identity might go through a stage in which they identify as bisexual (e.g., Lever, 1994; Stokes, Damon, & McKirnan, 1997). Recent research on the sexual arousal patters of men who identify as bisexual has found that most were more strongly aroused by homosexual stimuli than by heterosexual stimuli and, therefore, appear to be homosexual with respect to their genital arousal (Rieger, Chivers, & Bailey, 2005). Moreover, decades of research on “coming out” as gay or homosexual has shown that the period of first disclosure can range from during mid-adolescence, around 16 or 17 years of age (e.g., D’Augelli, Hershberger, & Pilkington, 1998; Grov, Bimbi, Nanin, & Parson, 2006) to late adolescence and early 20s (e.g., McDonald, 1982; Savin-Williams, 1998; Savin-Williams & Ream, 2003) and depends on numerous factors, including ethnicity, relationship with parents and their anticipated reaction, and person to whom the individual is disclosing–individuals with same-sex attractions typically come out to a supportive friend before coming out to parents (Beaty, 1999; Savin-Williams & Dube, 1998). Finally, it was not a goal of the present study to embark on a comparison of exclusively homosexual versus bisexual participants.
22.214.171.124 Social Desirability
Social desirability refers to the desire to cast a favorable impression on others. It can threaten the validity of self-report scales if in answering questions respondents seek social approval or try to represent themselves in a favorable manner (King & Brunner, 2000; Tan & Grace, 2008). People scoring high on social desirability tend to provide socially acceptable answers regardless if their response accurately describes them. As such, researchers have recognized that, particularly when assessing attributes of a personal or sensitive nature, such as an individual’s sexual history, the respondents’ propensity to give socially desirable responses should be measured (e.g., Wallien & Cohen-Kettenis, 2008). Participants 18 years and older were given the Marlow-Crowne Social Desirability Scale (M-CSDS; Crowne & Marlowe, 1960), which consists of 33 true-false items. The scale consists of 18 culturally acceptable but unlikely statements keyed in the true direction and 15 socially undesirable but probable statements keyed in the false direction for a maximum possible score of 33. Participants 17 years and under, were given a shorter version of the M-CSDS (Strahan & Gerbasi, 1972), containing 20 items that consists of 12 culturally acceptable but improbable statements keyed in the true direction and 8 socially undesirable but probable statements keyed in the false direction for a maximum possible score of 20. For the present study, the percentage of endorsed socially desirable items was calculated for each participant. In order to integrate the data from both versions of the M-CSDS, participants’ percentage score on each measure was converted to a proportion score which ranged from 0-1, which was used in all analyses. A higher proportion score indicates a greater propensity to give socially desirable responses. Several studies have found that the MCSDS is a reliable and valid measure of social desirability (Crowne & Marlowe, 1960; Holden & Feeken, 1989; Silverthorn & Gekoski, 1995).
126.96.36.199 Suicidality Experiences
Suicidal experiences were assessed with a Suicidality Questionnaire (see Appendix I) that consists of 13 items derived from Centre for Disease Control and Prevention Survey on Youth Risk Behavior Surveillance System (Brener et al., 2002; CDC, 2002) and questionnaires given in other studies (D’Augelli et al., 2002; Savin-Williams & Ream, 2003). Six of the questions pertained to “lifetime” experiences (since the age of 13) and 7 pertain to suicidal thoughts and/or experiences within the past 12 months. Frequencies of suicidality experiences were derived. Participants’ responses on this questionnaire were reviewed before they left the clinic. A procedure was set in place such that participants who expressed suicidal thoughts would be asked additional questions as part of a risk assessment; however, none of the study participants were at imminent risk for suicide. The Suicidality Questionnaire was introduced to the study protocol after data collection had begun; therefore, some participants did not complete this measure. In these instances, data on suicidality were extracted from their psychiatric interview (i.e., the Diagnostic Interview Schedule or the Diagnostic Interview for Children and Adolescents).
188.8.131.52 Victimization Experiences
Victimization experiences were assessed through a 12-item Victimization Survey (see Appendix J), which was adapted from previous studies on sexual orientation victimization of lesbian and gay youth and adults (D’Augelli et al., 2002; D’Augelli & Grossman, 2001; Herek, Gillis, Cogan, & Glunt, 1997). The survey was modified to target victimization due to cross-gender behavior. Questions ask participants’ frequency of victimization experiences over the past 12 months and since the age of 13 (i.e., “lifetime” victimization experiences). Seven types of victimization experiences were asked, including three verbal and four physical types. Verbal victimization included: (a) verbal insults, (b) threats of violence, and (c) threats by other to disclose gender identity. Physical victimization consisted of: (a) objects being thrown at the individual, (b) physical assault (e.g., being punched, kicked, or beaten), (c) threat of an attack, using a knife, gun, or weapon, and (d) sexual assault. Participants responded to each question based on the frequency of occurrence for each type of victimization with 0 = “Never,” 1 = “Once,” 2 = “Twice,” and 3 = “Three or more times.” An average verbal victimization score, an averge physical victimization score, and an average total victimization score was computed (the sum of all items divided by the relevant number of items) for ratings based on lifetime experiences as well as the past 12 months (D’Augelli, Pilkington, & Hershberger, 2002). Participants were also asked about the location of their victimization experiences (e.g., home, school, neighborhood, work place) and the type of relationship they had with the assaulter (e.g., parent, sibling, significant other, peer, or stranger). The Victimization Survey was added to the assessment protocol in 2008, after data collection had begun, so data were missing for individuals who completed the follow-up prior to the inclusion of this measure.
3.1 Participants vs. Non-Participants
The non-participants represent three groups: (1) patients who were eligible to participate in the study but were not contacted (n = 163), (2) patients who declined to participate in the study (n = 6), and (3) patients who were not successfully traced (n = 19). Two sets of analyses were conducted to compare the study participants with the non-participants. First, the study participants were compared to the boys who were eligible but were not contacted for research follow-up. Second, the study participants were compared to those who refused to participate in the study and to those where contact was attempted but the families were not successfully traced. Group comparisons were conducted on demographic variables (age at assessment, IQ, social class, marital status, ethnicity), CBCL behavior problems (Internalizing T score, Externalizing T score, Total T score), and nine measures of childhood sex-typed behavior.
Table 5 shows the childhood assessment data (demographics and CBCL behavior problem ratings) of the 139 boys who participated in the study compared to the 163 boys who were eligible to participate but were not contacted due to time constraints and lack of study resources. There were no significant differences between the participants and non-participants on the demographic variables of age at assessment, social class, ethnicity or marital status (ps > .05). However, the comparison on childhood IQ was significant, t(289) = 2.01, p = .046, with the participants having a higher IQ than the non-participants. The effect size for this comparison was small (unpooled d = .22). With regard to parent-report of behavior problems on the CBCL, there were no significant differences between the
participants and non-participants on Internalizing problems, Externalizing problems, and Total problems (ps > .05). The participants and non-participants were also compared on their diagnosis in childhood and their scores on measures of sex-typed behavior (Table 6). There was no significant difference between the participants and non-participants in terms of
whether they were diagnosed with GID in childhood or were subthreshold for the diagnosis, χ2(1) < 1. There were also no significant differences between the participants and non-participants on any of the eight measures of sex-typed behavior (ps > .05). In sum, of the 17 analyses conducted to examine the differences between the participants and those who were eligible to participate but were not contacted, only one comparison (childhood IQ) was
Using one-way ANOVA or chi-square, the study participants were also compared to the 6 cases where either the parents or the potential participant himself refused to participate and to the 19 cases where the families could not be traced. Group comparisons were conducted on demographic variables, CBCL behavior problems, and measure of sex-typed behavior. There were no significant group differences on the demographic variables of age, social class, and ethnicity (ps > .05). The ANOVA on childhood IQ approached significance, F(2, 162) = 2.99, p = .053. Duncan’s multiple range test for unequal Ns showed that the participants did not differ significantly from the other two groups; however, the non-participants who refused had a higher IQ in childhood than those who could not be traced. The three groups differed significantly on marital status, χ2(2, N = 164) = 9.02, p = .011. Post-hoc analyses were done using chi-square or Fisher’s Exact Test. The participants did not differ significantly from the non-participants who refused; however, they differed significantly from the cases that could not be traced, χ2(1, N = 158) = 6.39, p = .012. The participants were more likely to have originated within a two-parent household while those who could not be traced were more likely to have come from a family composition other than two-parent (e.g., single parent, living with relatives). The comparison between the non-participants who refused and those who could not be traced approached significance (p = .056, Fisher’s exact test). Again, the non-participants who could not be traced were more likely to have come from a family composition that was not two-parent. With regard to parent-report of behavior problems on the CBCL, there were no significant differences between the three groups on Internalizing problems, Externalizing problems, and Total problems (ps > .05). The three groups were also compared on their diagnosis in childhood and on measures of sex-typed behavior. There was no significant difference between the three groups on their diagnosis in childhood; the participants were as likely to have met full diagnostic criteria for GID as the non-participants. There were also no significant group differences on any of the measures of sex-typed behavior (ps > .05). In sum, of the 17 analyses conducted to compare the participants to those who refused and to those who could not be traced, one (marital status) was significant and one (childhood IQ) approached significance. The other 15 group comparisons were not significant.
3.2 DSM Diagnosis for Gender Identity Disorder in Childhood
Of the 139 participants, 88 (63.3%) met diagnostic criteria for GID in childhood and the remaining 51 (36.7%) were subthreshold for the diagnosis. These two groups were compared on demographic variables, CBCL behavior problems, and measures of sex-typed behavior. There were no significant group differences on the demographic variables of IQ in childhood, marital status or ethnicity (ps > .05). However, the threshold participants were significantly younger in age at the time of the childhood assessment, t(108) = -6.31, p < .001, and originated within a family of higher social status compared to the subthreshold participants, t(108) = 2.31, p = .023. The effect size for these differences was medium, .34 and .46, respectively.
3.2.1 Childhood Behavior Problems as a Function of Diagnostic Status for Gender Identity Disorder
One-way ANCOVAs, with age and social class covaried, were conducted to evaluate whether the two diagnostic groups differed on CBCL behavior problems and sex-typed behavior in childhood. There were no significant differences between the threshold and subthreshold participants on Internalizing T score and Total T score (both ps > .05). The subthreshold participants were rated by their parents as having, on average, significantly more externalizing difficulties (M = 61.02, SD = 11.78) compared to the threshold participants (M = 60.77, SD = 10.48), F(1, 138) = 7.86, p = .006, partial η2 = .06. Visual inspection revealed modest differences in Externalizing T scores between the two groups.
3.2.2 Childhood Sex-Typed Behavior as a Function of Diagnostic Status for Gender Identity Disorder
Table 7 shows the means and SDs (for continuous variables) or percentage scores (for dichotomous variables) for the threshold and subthreshold participants on measures of childhood sex-typed behavior. With age and social class covaried, the threshold participants had, on average, more cross-gender behavior in childhood than did the subthreshold participants. There were significant group differences on five of eight measures of childhood sex-typed behavior: (1) the Draw-a-Person test (p < .001), (2) free play, F(1, 120) = 28.94, p < .001, partial η2 = .19, (3) Gender Identity Interview, F(1, 84) = 9.07, p < .001, partial η2 = .10, (4) Gender Identity Questionnaire for Children, F(1, 96) = 19.13, p < .001, partial η2 = .17 and (5) cross-sex toy choice on the Playmate and Play Style Preferences Structured Interview, F(1, 40) = 9.72, p = .004, partial η2 = .21. The groups did not differ on temperament/activity level, cross-sex peer preference on the Playmate and Play Style Preferences Structured Interview, and difference between cross-sex and same-sex responses on the Rorschach (ps > .05).
3.3 Psychosexual Differentiation
3.3.1 Gender Identity at Follow-up
A summary of the psychosexual differentiation follow-up data, including gender
identity and sexual orientation for each participant is shown in Appendix K.
184.108.40.206 Criteria for Persistence of Gender Dysphoria
Classification as either a persister or desister with regard to gender dysphoria was based on participants’ mean scores on the dimensional measures of concurrent gender identity, either the Gender Identity/Gender Dysphoria Questionnaire for Adolescents and Adults or the Gender Dysphoria Questionnaire, depending on which measure was administered.
Some participants completed both measures and, in these cases, the Gender Identity/Gender Dysphoria Questionnaire for Adolescents data were used as the basis for classification as it is a more psychometrically sophisticated measure of gender dysphoria compared to the Gender Dysphoria Questionnaire.
Participants were classified as having persistent gender dysphoria if their mean score on the Gender Identity/Gender Dysphoria Questionnaire for Adolescents was ≤ 3.00 (Deogracias et al., 2007). For participants who did not complete the Gender Identity/Gender Dysphoria Questionnaire for Adolescents, the Gender Dysphoria Questionnaire was used to make the classification. A participant was classified as a persister if he endorsed two or more of the following 5 items on the Gender Dysphoria Questionnaire: wish to have been born a girl (Item 1), wish to have surgery to change body (Item 2), feel more like a girl than a boy (Item 3), wonder if would be happier as a girl (Item 4), and very or somewhat dissatisfied with being a boy (Item 5).
Information regarding participants’ gender identity/gender dysphoria was also collected during the semi-structured clinical interview and, therefore, allowed for cross-validation of these questionnaire data. For those participants who did not complete the face-to-face interview, clinical information regarding gender identity/gender dysphoria was obtained through self- or parent-report or chart review. Across the entire sample, the Gender Identity/Gender Dysphoria Questionnaire for Adolescents and Adults was used to classify persistence or desistence for 64 participants, the Gender Dysphoria Questionnaire for 42 participants, and interview/parent report/chart data for 33 cases.
220.127.116.11 Rate of Persistence and Desistence
Of the 139 participants, 17 (12.2%) were classified as persisters and the remaining
122 (87.8%) were classified as desisters at follow-up. For the 42 participants where the Gender Dysphoria Questionnaire was used to determine gender identity status at follow-up, 38 were classified as desisters and 4 were classified as persisters. Of the 38 desisters, three endorsed one item on the Gender Dysphoria Questionnaire –one participant endorsed Item 4 and two participants endorsed Item 3. The four participants classified as persisters using the Gender Dysphoria Questionnaire endorsed three or more items. Specifically, one persister endorsed three items, one endorsed four items, and two endorsed five items. In regard to the specific items endorsed, all four persisters endorsed Item 1, three endorsed Item 2, four endorsed Item 3, four endorsed Item 4, and two endorsed Item 5.
For the 64 participants where the Gender Identity/Gender Dysphoria Questionnaire for Adolescents and Adults was used to determine gender identity status at follow-up, 12 were classified as persisters and the remaining 52 were classified as desisters. All 52 desisters had a mean score > 3.00 on the Gender Identity/Gender Dysphoria Questionnaire for Adolescents and Adults. Of the 12 persisters, 10 had a mean score ≤ 3.00 and two had mean scores that were > 3.00. In spite of having mean scores on the Gender Identity/Gender Dysphoria Questionnaire for Adolescents and Adults that were above the recommended cutoff for caseness (Deogracias et al., 2007), these two participants were considered persisters because their responses on the Gender Dysphoria Questionnaire as well as clinical interview data indicated that they were experiencing significant gender dysphoria. Thus,
clinical judgment was used to make the final classification for these two participants.
18.104.22.168 Persistence of Gender Dysphoria as a Function of GID Diagnosis in Childhood
The persistence rate of gender dysphoria was examined as a function of participants’ GID diagnostic status in childhood, that is, whether they met full diagnostic criteria or were subthreshold for the diagnosis. Of the 88 participants who met the full diagnostic criteria for GID in childhood, 12 (13.6%) were gender dysphoric at follow-up and the remaining 76 (86.4%) were no longer gender dysphoric. Of the 51 participants who were subthreshold for the GID diagnosis in childhood, 5 (9.8%) were gender dysphoric at follow-up and the remaining 46 (90.2%) were not. A chi-square analysis revealed that these rates of persistence across subthreshold and threshold groups did not differ significantly, χ2(1) < 1.
22.214.171.124 Summary of Gender Dysphoric Participants
Table 8 summarizes information on some domains of gender role outcome for the 17 participants classified as having persistent gender dysphoria. There was notable variation within the group of persisters in the extent to which they had socially and medically transitioned to the female gender role.
At follow-up, the majority of the gender dysphoric participants (n = 9) were not receiving any medical/biologic treatments to either suppress physical development of secondary sex characteristics/masculinization of their body or to feminize their body. The remaining 8 (47.1%) participants were taking puberty suppressing hormones or were on cross-sex hormonal therapy to feminize their physical appearance. Of the 9 participants who were not taking puberty suppressing or cross-sex hormones, 6 did not report any attempt to
present socially in the female gender role (e.g., using a female name, living in the female role). Of the remaining 3 participants, 2 were attempting to live socially in the female gender role and data were incomplete for 1 participant. None of the 17 gender dysphoric participants had received any type of surgery to feminize their appearance (e.g., breast implants, facial feminization surgery, genital reconstruction).
Among the 17 persisters, 11 were using a female name. Of these 11 participants, 2 had legally changed their name on official documents (e.g., driver’s license) and 1 was in the process of pursuing a legal name change. Nine of the gender dysphoric participants were living in the female gender role and 4 of them were living in the male gender role. Of the remaining 4 participants, 1 was living partially in the female role (e.g., would sometimes wear make-up and at others times presented androgynously, but did not wear stereotypic female clothing), 2 presented androgynously, and data were not available for 1 participant.
None of the 122 participants classified as desisters desired cross-sex hormones or sex reassignment surgery to feminize their bodies nor did they express a desire to get rid of their male sex characteristics. None of the desisters presented socially as women nor did they express the desire to socially transition to the female gender role (e.g., name change, clothing choice).
126.96.36.199 Odds of Persistent Gender Dysphoria
Formal epidemiological studies on the prevalence of GID in adolescents and adults have not been conducted. Since the 1960s, a number of studies have reported estimated prevalence rates (for a review, see Zucker & Lawrence, 2009). Rates have varied, in part, depending on the inclusion criteria (e.g., including individuals who have had, at least, hormonal treatment vs. only including individuals who have had sex reassignment surgery). For example, De Cuypere et al. (2007) estimated that 1 in 12,900 biological adult males have GID. Weitze and Osburg (1996) estimated a prevalence rate of 1 in 42,000. The estimated prevalence rate in most other studies have fallen within this range (i.e., 1/12900-1/42,000). Using the prevalence values estimated by De Cuypere et al. and Weitze and Osburg, the odds of persistent gender dysphoria (12.2%) in the present sample was 1794-5840 times higher than it is in biological males in the general population.
3.3.2 Sexual Orientation at Follow-up
On the basis of Kinsey ratings, participants who completed the face-to-face interview were classified into the following three sexual orientation groups for both fantasy and behavior: (1) heterosexual (Kinsey global ratings of 0-1); (2) bisexual/homosexual (Kinsey global ratings of 2-6), and (3) no sexual fantasy or behavior. Throughout this thesis, sexual orientation was classified in relation to birth sex, rather than the participants’ subjective sexual identity at follow-up. This is a particularly relevant issue for those participants who had persistent gender dysphoria. For example, if a biological male participant identified as female at follow-up, was sexually attracted to other biological males and self-labeled their sexual identity as heterosexual, they would be classified in the study as homosexual in relation to their birth sex.
Table 9 shows the frequency of ratings for sexual orientation in fantasy. Appendix K lists the individual global fantasy ratings for all participants for whom these data were available. Data were not available for 10 participants, all of whom were desisters with regard to gender dysphoria. Based on the global ratings for sexual orientation in fantasy, 43 (33.3%) participants were classified as heterosexual in fantasy and 82 (63.6%) were classified as bisexual/homosexual in fantasy. In the remaining 4 (3.1%) cases, the participants were classified as having no sexual fantasies and, therefore, a Kinsey rating could not be assigned. In all 4 cases, the participants were desisters.
Of the 17 participants classified as gender dysphoric, 1 (5.9%) was heterosexual in fantasy and 16 (94.1%) were bisexual/homosexual in fantasy. Chi-square analysis revealed that the rates of sexual orientation in fantasy across this study, Green (1987), and Wallien and Cohen-Kettenis (2008) did not differ significantly, χ2(2) = 2.82, p > .05.
Table 10 shows the frequency of ratings for sexual orientation in behavior. Appendix K lists the individual global behavior ratings for all participants for whom these data were available. Data were available for 108 participants. Based on global ratings for sexual orientation in behavior, 29 (26.9%) participants were classified as heterosexual and 51 (47.2%) were classified as bisexual/homosexual. The remaining 28 (25.9%) participants did not report any sexual behaviors in the 12 months preceding the follow-up assessment.
For those participants who could be assigned a Kinsey rating (i.e., excluding those participants who did not report any sexual fantasies or behavior or for whom data were not available), the correlation between Kinsey global fantasy and global behavior ratings was very strong, r(78) = .92, p < .001.
188.8.131.52 Group Classification as a Function of Gender Identity and Sexual Orientation in Fantasy at Follow-up
Combining gender identity (i.e., persister or desister) and sexual orientation in fantasy (i.e., heterosexual or bisexual/homosexual) at follow-up, the participants were classified into one of the following four outcome groups: (1) persistence of gender dysphoria with bisexual/ homosexual sexual orientation (n = 16); (2) desistence of gender dysphoria with
bisexual/homosexual sexual orientation (n = 66); (3) desistence of gender dysphoria with heterosexual sexual orientation (n = 42); and (4) persistence of gender dysphoria with heterosexual sexual orientation (n = 1). The participants who reported no sexual fantasies (n = 4) could not be included in this outcome classification. Given that only 1 participant was classified as gender dysphoric with a co-occurring heterosexual sexual orientation (Group 4), this category was excluded from subsequent analyses that compared these outcome groups.
3.4 Demographic Characteristics
3.4.1 Demographic Characteristics in Childhood as a Function of Gender Identity and Sexual Orientation in Fantasy
Table 11 shows the demographic variables in childhood as a function of group.
One-way ANOVAs and chi-square were conducted to evaluate whether the outcome groups differed on demographic variables in childhood. The groups differed significantly on four of the five childhood demographic variables (ps < .05). The group comparison on ethnic background was not significant.
Duncan’s multiple range test for unequal Ns showed that the bisexual/homosexual persisters were, on average, significantly older at the time of the childhood assessment than both the heterosexual desisters and the bisexual/homosexual desisters, who did not differ significantly from each other. Regarding IQ, the bisexual/homosexual desisters had, on average, a higher IQ than the bisexual/homosexual persisters but did not differ significantly from the heterosexual desisters. There was no significant difference in childhood IQ score between bisexual/homosexual persisters and heterosexual desisters. The bisexual/
homosexual persisters were significantly more likely to come from a lower social class background compared to the heterosexual desisters and the bisexual/homosexual
desisters, who did not differ significantly from each other. The bisexual/homosexual desisters were more likely to be living with both parents compared to the bisexual/ homosexual persisters. There was no significant difference on marital status between the two desister groups.
As shown in Table 12, the demographic variables on which the three groups differed–age at assessment, IQ, social class, and marital status–were significantly correlated. To evaluate the influence of these variables on group outcome at follow-up, a multinomial logistic regression was performed. It can be seen from Table 13 that only social class had a significant contribution to the prediction of group outcome at follow-up. The bisexual/ homosexual persisters had a 13% increase in odds of coming from a lower social class background compared to the bisexual/homosexual desisters. However, social class did not predict outcome when the two desister groups were compared. Figure 1 shows the distribution of social class across the outcome groups.
Figure 1. Distribution of social class for the outcome groups at follow-up.
1 = Bisexual/homosexual persisters (n = 16; M = 23.76, SD = 10.22)
2 = Bisexual/homosexual desisters (n = 66; M = 44.97, SD = 13.64)
3 = Heterosexual desisters (n = 42; M = 39.44, SD = 15.91)
3.4.2 Demographic Characteristics at Follow-up as a Function of Gender Identity and Sexual Orientation in Fantasy
Table 11 also shows the demographic variables of age and IQ at follow-up as a function of group. One-way ANOVAs revealed that both variables differed significantly among the three groups (ps < .01). Duncan’s multiple range test for unequal Ns showed that the heterosexual desisters were, on average, younger than both the bisexual/homosexual persisters and the bisexual/homosexual desisters, who did not differ significantly from each other.
Regarding IQ at follow-up, the results were similar to those for IQ in childhood. The bisexual/homosexual desisters had, on average, higher IQ than the bisexual/homosexual persisters but did not differ significantly from the heterosexual desisters. There was no significant difference in IQ between the bisexual/homosexual persisters and the heterosexual desisters.
Table 11shows the mean proportion score on the measure of social desirability as a function of group. A one-way ANCOVA for Group (with age at assessment, age at follow-up, IQ in childhood, IQ at follow-up, social class, and marital status6 covaried) was conducted to evaluate the proportion of socially desirable responses on the Marlow-Crowne Social Desirability Scale for participants according to their group classification at follow-up. There was no significant difference in the proportion of socially desirable responses on the Marlow-Crowne Social Desirability Scale as a function of gender identity and sexual orientation in fantasy (p = .089).
3.6 Childhood Sex-Typed Behavior
3.6.1 Childhood Sex-Typed Behavior as a Function of Gender Identity and Sexual Orientation at Follow-up
Table 14 shows the means and SDs (for continuous variables) or percentage scores (for dichotomous variables) of the childhood sex-typed variables obtained at the assessment as a function of the three outcome groups. ANCOVAs (with age at assessment, IQ, social class, and marital status covaried) or chi-square were used to examine whether the groups differed on any of the reported variables. The corresponding F or chi-square values are also shown.
Of the 9 sex-typed measures, there was a significant difference between the groups on four child-report measures (first drawn person on the Draw-a-Person, free play, Gender Identity Interview, and cross-sex peer preference on the Playmate and Play Style Preferences
Structured Interview, and one parent-report measure (Gender Identity Questionnaire
for Children). The ANCOVA for the cross-sex toy preference on the Playmate and Play Style Preferences Structured Interview approached significance (p = .096). The significant one-way ANCOVAs were followed up with post hoc analyses using lmatrix commands and the significant chi-square was followed-up with pair-wise chi-square comparisons.
On the Draw-a-Person, there was one significant post-hoc contrast. The bisexual/ homosexual persisters were, on average, significantly more likely to draw a female first compared to the heterosexual desisters (p = .04). The comparison between the bisexual/ homosexual desisters and heterosexual desisters approached significance (p = .09), with the bisexual/homosexual desisters showing a greater tendency to draw a female first. The comparison between the bisexual/homosexual persisters and bisexual/homosexual desisters was not significant (p > .05). On the free play measure, all post-hoc contrasts were significant. The bisexual/homosexual persisters displayed, on average, more cross-sex play than did the bisexual/homosexual desisters and the heterosexual desisters. The latter two groups differed significantly from each other; the bisexual/homosexual desisters displayed, on average, significantly more cross-sex play than did the heterosexual desisters. On the Gender Identity Interview, a semi-structured interview that assesses gender identity, feelings of gender confusion, and gender dysphoria, the bisexual/homosexual persisters reported, on average, significantly more gender confusion than did the bisexual/homosexual desisters. The post-hoc comparison between the bisexual/homosexual persisters and the heterosexual desisters approached significance (p = .054), with the bisexual/homosexual persisters reporting more gender confusion than the heterosexual desisters. The comparison between the bisexual/homosexual desisters and heterosexual desisters was not significant. There was also significant group difference on the peer preference domain of the Playmate and Play Style Preferences Structured Interview. The bisexual/homosexual persisters had, on average, significantly more cross-sex peer preference compared to the bisexual/homosexual desisters and the heterosexual desisters, who did not differ significantly from each other. On the Gender Identity Questionnaire for Children, a parent-report measure of cross-gender interest and identification, the bisexual/homosexual persisters were reported as significantly more cross-gendered than the bisexual/homosexual desisters and the heterosexual desisters. The latter two groups did not differ significantly from each other.
As can be seen from Table 15, the childhood sex-typed behavior measures on which the groups differed were significantly correlated. From these six measures (first drawn
person on the Draw-a-Person, free play, Gender Identity Interview, cross-sex peer preference on the Playmate and Play Style Preferences Structured Interview,21 cross-sex toy preference on the Playmate and Play Style Preferences Structured Interview, and the Gender Identity Questionnaire for Children), a composite score of childhood sex-typed behavior was derived for each participant by taking the average of the six variables (each expressed as z-scores). Thus, the composite score was expressed as a z-score. A higher composite z-score indicates more cross-gender behavior at assessment (Fig. 2).
Figure 2. Distribution of the mean composite z-score for the outcome groups at follow-up.
1 = Bisexual/homosexual persisters (n = 16; M = .36, SD = .60)
2 = Bisexual/homosexual desisters (n = 66; M = .03, SD = .77)
3 = Heterosexual desisters (n = 42; M = -.32, SD = .90)
To evaluate the influence of childhood sex-typed behavior and demographic variables on group outcome at follow-up, a multinomial logistic regression was performed using the composite score and the demographic variables on which the groups differed–age at assessment, IQ, and social class–as predictor variables. It can be seen from Table 16 that both social class and the composite score of childhood sex-typed behavior were significant predictors of group outcome at follow-up in the first model which compared the bisexual/ homosexual persisters to the bisexual/homosexual desisters.
The bisexual/homosexual persisters had a 274% increase in odds of having a higher composite score (i.e., more childhood cross-gender behavior) and 11% increase in odds of coming from a lower social class compared to the bisexual/homosexual desisters. Age at childhood assessment and IQ did not have a significant effect on group outcome (both ps > .05). In the second model, which compared the heterosexual desisters to the bisexual/ homosexual desisters, the only significant predictor of group outcome was the composite measures of sex-typed behavior. The bisexual/homosexual desisters had a 48% increase in odds of having a higher composite score compared to the heterosexual desisters.
3.6.2 Childhood Sex-Typed Behavior and Year of Assessment
The childhood assessment data were collected over the course of three decades, 1975-2004. Using correlation analysis, I assessed whether a relationship existed between year of assessment and scores on measures of sex-typed behavior. Across the entire sample, there was no significant correlation between year of assessment and any of the nine measures of childhood sex-typed behavior, all ps > .05. However, three correlations approached significance: free play (p = .05), the Gender Identity Questionnaire for Children (p = .08), and the cross-sex peer preference on Playmate and Play Style Preferences Structured Interview (p = .08). Across all three measures, there was a trend for participants with a later year of assessment (i.e., assessed more recently) to have had, on average, more childhood cross-gender behaviors compared to participants assessed earlier in the data collection period.
3.7 Peer Relations in Childhood
In the present study, we examined the quality of participants’ childhood peer relations by computing the Peer Relations Scale (Zucker, Bradley, & Sanikhani, 1997). In an effort to identify the relationship between peer relations and age across the entire sample of boys, participants’ age at childhood assessment and score on the Peer Relations Scale was correlated. The correlation was significant, r(121) = .58, p < .001, with older boys having poorer peer relations than younger boys. To examine the relationship between peer relations and behavior problems in childhood, participants Peer Relations Score was correlated with the sum of items rated as 1 or 2 on the CBCL, with the three items contributing to the Peer Relations Scale removed to prevent artificial inflation of the sum score. The correlation was significant, r(121) = .43, p < .001.
A one-way ANCOVA (with age at assessment, IQ in childhood, social class, and marital status co-varied) was conducted to examine whether the three outcome groups at follow-up differed in the quality of their peer relations in childhood. The ANCOVA was not significant, F(2, 120) < 1.
3.8 Childhood Behavior Problems as a Function of Gender Identity and Sexual Orientation at Follow-up
Table 17 shows the maternal ratings on the Child Behavior Checklist (CBCL) as a function of group. ANCOVAs (with age at assessment, IQ, social class, and marital status co-varied) were used to examine whether the groups differed on any of the reported childhood variables. The corresponding F values are also reported.
One-way ANCOVAs for Group revealed that there were no significant differences between the three groups on Internalizing T score, Externalizing T score or Total Problems T score at the childhood assessment (ps > .05). Table 17 also shows the percentage of participants in each group who fell within the clinical range (T score > 90th percentile) according to their Internalizing, Externalizing, and Total Problems T scores. For the bisexual/homosexual persisters, 62.5%-68.8% fell within the clinical range depending on whether they were classified according to their Internalizing T score, Externalizing T Score or Total T score. For the bisexual/homosexual desisters, 40.9%-42.4% were within the