There was a significant difference in age between the three groups (F_2,158=30.5, p < 0.0001), with the transmen being younger than the lesbian women and the reference group. The lesbian women and the reference group had attained significantly higher levels of education (c²= 42.9, df = 4, p < 0.0001) and were more likely to have a partner (c² = 16.6, df = 2, p < 0.001) compared to the transmen. The socioeconomic status of the household in which the participants were raised was quite similar across groups, with the exception that the lesbian women were slightly more likely to come from a lower socioeconomic status environment (c²= 9.7, df = 4, p < 0.05).

The mean age in which the transmen first considered transition was 16.9 years; the mean age at which these individuals initiated transition was 23.4 years. The transition occurred at approximately the same time as the initiation of hormones (23.5 years). All of the individuals in this group had socially transitioned. All but four were receiving hormones. Twenty of the transmen underwent mastectomy; eight underwent hysterectomy; four underwent metoidioplasty; three participants had undergone vaginectomy, and one individual had undergone phalloplasty.

Participants were asked whether they had been regarded as tomboys during childhood and adolescence. Thirty percent of the reference group endorsed having received a label of “tomboy,” compared to 73% of lesbians and 90% of the transmen. All three comparisons were significant at (p<.001) Figure 1. There were no significant differences in the physical activity levels between the transmen and the lesbians, however both groups engaged in more intense physical activity in childhood than did the reference group (F_2,158=5.3, p=0.006).

There was a highly significant difference between the reference group compared to the lesbian women and the transmen in all areas related to interest in babies, maternal role-play, and doll play (Figure 2). Between transmen and lesbian women there were significant differences in maternal role play (χ²=9.2, df=1, p=0.002), paternal role play (χ²=12.4, df=1, p=0.0004) and play with cars and trucks (χ²=4.9, df=1, p=0.03), with the lesbian women having spent significantly more time playing “mommy” and the transmen preferring cars and trucks to dolls. There were no differences between lesbian women and transmen in desire to become a parent, interest in babies, or in enjoyment of babysitting.

Three questions pertained to childhood clothing preferences. There were significant differences between the reference group, the lesbian women and the transmen in regards to their clothing choices in childhood (Figure 3). Thirty-seven percent of the reference group preferred plain clothing to fashionable clothing. This is in contrast to both the lesbian women (82%) and the transmen (80%): both groups preferred plain rather than stylish or “fancy” clothing as youngsters. Interestingly, there were highly significant differences in the desire to wear boys’ shoes and boys’ underwear. None of the reference group, and only 9% of the lesbian women stated the desire to wear boys’ underwear during childhood. This was in contrast to 78% of the transmen who preferred boys’ underwear (χ²=45.7, df=1, p=1.4x10^-11). Six percent of the reference group, compared to 49% of lesbian women preferred to wear boys’ shoes as a child. This is in contrast to 90% of the transmen who had a marked preference for boys’ footwear (χ²=17.3, df=1, p=3.1x10^-5).

A continuous scale measuring the preference for short hair compared to long hair was highly significantly different between the three groups (F_2,158=53.0, p=2.0x10^-16). The reference group preferred long hair at a scale of 70.1, compared to 36.2 and 23.6 for the lesbian women and transmen, respectively. Contrasting the lesbian women to the transmen on this variable also proved significant. (t=2.4, df=90, p = 0.02).

While many children who display gender atypical behavior do not identify as transgender in adulthood, there are few studies that examine sexual and gender identity trajectories in these children as they reach adulthood. In 1979, Erhard et al. reported that certain variables distinguished between two groups of adults who had been assigned female at birth, but exhibited gender atypical behavior in childhood. One group identified as lesbians, the other group identified as transgender. Using a retrospective analysis of specific childhood variables, the investigators concluded that childhood cross-dressing was strongly associated with gender transition at adulthood.

There has been a tectonic change in our knowledge and understanding of sexual orientation and gender identity in the ensuing four decades. In spite of these contextual changes, the current study, employing a much larger sample and a reference group, replicated the most salient findings of the 1979 study: Namely, amongst a group of children who were labeled tomboys, the preference for boys shoes and boys underwear in childhood was significantly associated with gender transition at maturity. In fact, none of the reference group had a desire to wear boys underwear, and only 9% of the lesbian women desired to do so, in contrast to 78% of the transmen. It is reductionistic to conclude that specific childhood behaviors or preferences, in and of themselves, can predict future outcomes. But the robustness of the data indicating that desire to wear boys underwear is consistently related to gender transition is indeed striking.

Children become aware of gender categories at an early age 2. At two years, toddlers begin to use gender stereotypes in play and by age five, preference for stereotypic toys is fairly well established 4. Three-year-olds recognize that there are two distinct groups, and certain clothing styles and/or colors, hairstyle and/or hair length are social signifiers that denote gender group membership 5. Often, the display itself serves as a proxy for gender, causing very young children to arrive at unique interpretations. For example, a four-year-old child may announce that an androgynous appearing teacher is a girl because he wears earrings. Similarly, four to six-year olds understand gender scripts. Instead of grouping items, they begin to categorize activities into gender categories. So, a person putting on make-up is deemed female. By age six to seven, youngsters recognize that gender is constant, and a man putting on make-up is a man.

In tandem with the awareness that males and females can be differentiated based on external presentation, the child has the burgeoning phenomenological sense of comfort or discomfort with their own assigned category, i.e. the development of gender identity. One participant in the lesbian group commented, As a child I was happy if someone thought I was a boy. Sometimes I wished I was a boy, but I didn t really want to be a boy once I got a little older. This sentiment is consistent with the Steensma et al observations. They found that gender non-conforming girls who in early childhood claimed that they felt they were boys were more likely to transition than those that only wished they were boys 678.

For the child who experiences gender incongruity, adoption of the emblems, representations and play activities that cue the relevant gender can be intensely pursued. The literature documents that those children who were assigned female at birth and later transition are more likely to experience intense gender dysphoria in childhood 678. The presence of such strong convictions early in life, coupled with a growing assemblage of research, points to genetic and neurodevelopmental explanations for this cross-sex development of gender identity 9101112131415161718.

Given that external displays equate to gender categorization, why is it that underwear, which is non-visible, is retrospectively highly associated with a transgender identity in both studies? One explanation is that these displays of gender non-conformity often arouse the shame affect. Children who exhibit gender atypical behaviors or expressions are often rebuked by family, peers or community 51920, although recent years have ushered in greater understanding and acceptance of gender diversity. However, in the context of less accepting environments, underwear may be a private declaration of identity that reduces the potential for shame. Additionally, with the availability of more gender-neutral clothing, underwear remains quite gender specific. Furthermore, young children often employ self-soothing and tactile talismans, and these concrete items can be more satisfying than role-playing or fantasy activities wherein the child assumes the desired gender. Perhaps most importantly, undergarments obscure and are proximate to the genitals, an area of the body that can be visually disturbing for a gender dysphoric child. One intensely dysphoric six year old, assigned female at birth, cried whenever she was naked.

Future longitudinal studies that examine these variables prospectively will help elucidate and potentially parse the complexities of identity formation, as well as the specificity of our findings. However, large population-based studies would be required to achieve an adequate sample size to test these factors within the general population. While much remains unknown, what is known is that the overarching goal is reducing stigma, fostering resilience and providing support and care for these children and their families.