Puberty Does Not Follow the Gender Binary, Large Retrospective Study Finds

Most people assume that the different strands of gender line up neatly during adolescence: a child registered female at birth will feel like a girl, behave in feminine ways, be attracted to boys, and welcome the bodily changes of puberty without conflict. A new study from Tel Aviv University puts that assumption to a rigorous test — and finds it wanting. Drawing on detailed recollections of puberty from 1,916 young Israeli adults, including 1,797 cisgender and 119 transgender and non-binary participants, researchers Aviv Sion and Daphna Joel show that gender-related experiences during adolescence form highly individual mosaics rather than the coherent, binary packages that both popular culture and much of psychology have long presumed.

The study, published in Frontiers in Social Psychology, introduced a new instrument called the Gender Multi-dimensional Questionnaire—Puberty, or GMQ-P. Because puberty can begin as early as age nine, making real-time assessment difficult, the researchers opted for a retrospective design restricted to adults aged 18 to 30. Participants answered questions about four core facets of gender during their adolescent years: gender identity, behaviors typical of their own or the ‘other’ gender, sexual orientation, and their feelings toward the development of sex characteristics such as breasts, facial hair, menstruation, and voice change. Every participant saw every item, including questions framed for both binary categories, which allowed the researchers to detect experiences that cut across conventional categories.

Before testing any hypotheses, the team put the questionnaire through an extensive psychometric workup. An exploratory factor analysis on half of the cisgender sample, followed by confirmatory factor analysis on the remainder plus the transgender and non-binary groups, yielded seven distinct factors: feelings and behaviors related to pubertal bodily changes; body appearance concerns; own gender-typical behavior; other gender-typical behavior; sexual orientation; gender identity; and sex characteristics dysphoria — the wish to have or not have particular sex characteristics. All factors showed acceptable internal consistency, and the questionnaire demonstrated measurement invariance across cisgender men, cisgender women, and the combined transgender and non-binary group, meaning the same constructs were being measured in each. Known-groups validity was also confirmed: the factors reproduced previously documented differences, such as higher sex characteristics dysphoria among transgender participants, with effect sizes ranging from moderate to very large.

The first major finding concerned variability within groups. On every factor, the scores of cisgender men and cisgender women spanned nearly the entire range and overlapped substantially with the scores of transgender and non-binary participants. Group-level differences — for instance, that cisgender women on average reported more negative feelings toward pubertal bodily changes than cisgender men — concealed enormous diversity. A considerable minority of cisgender men reported negative feelings toward their developing bodies, and a minority of cisgender women did not. Variability and overlap were greatest for the two body-related factors and smallest for gender identity and sex characteristics dysphoria, the two components most closely tied to the transgender experience, though even there the transgender and non-binary groups showed wide internal spread.

The second major finding addressed the central theoretical question. The binary framework predicts congruence: someone atypical in one facet of gender should be atypical in the others. The competing mosaic framework, developed by Joel and colleagues in earlier work on brain and behavior, predicts that individuals carry unique combinations of both ‘normative’ and non-normative characteristics. Three lines of evidence favored the mosaic view. First, the seven factors were largely distinct, with pairwise correlations mostly small to medium — below 0.50 — with one exception: the two body-related factors correlated at 0.60 in the cisgender sample, indicating overlap but also that distress about pubertal changes is not reducible to appearance concerns. Second, an unsupervised cluster analysis using the K-means algorithm identified six profiles, each mixing ‘normative’ and non-normative mean scores in different combinations. Third, and most strikingly, an individual-level mosaic analysis showed that combinations of ‘normative’ and non-normative scores were more common than internally consistent profiles in nearly every group.

The six clusters painted a nuanced portrait of adolescent gender. One cluster, ‘normative’ on all seven factors, contained 68 percent of straight cisgender men but also 34 percent of LGB cisgender men and 23 percent of straight cisgender women. Another, matching the stereotype of the appearance-anxious adolescent girl, held only 38 percent of straight cisgender women. A ‘gender non-conforming’ cluster — rare own-gender typical behavior, frequent other-gender typical behavior, but otherwise conventional — included 24 percent of straight cisgender women. A fourth cluster was ‘normative’ on everything except sexual orientation. A fifth deviated on all factors except sex characteristics dysphoria, and a sixth, non-normative on all factors, contained most transgender men (94 percent), transgender women (82 percent), and non-binary individuals (66 percent) — but also 1 to 7 percent of each cisgender group.

The mosaic analysis quantified just how common mixed profiles are. Using the strictest cutoff, 65 percent of straight cisgender women, 72 percent of LGB cisgender men, 64 percent of LGB cisgender women, and 34 percent of straight cisgender men had mosaics of ‘normative’ and non-normative scores, compared with only 10 percent of straight cisgender men who were consistently ‘normative’ across all seven factors. Even within the most ‘normative’ cluster, 37.6 percent of individuals had at least one non-normative score; within the most non-normative cluster, nearly half had at least one ‘normative’ score. Notably, 43 percent of transgender women and 58 percent of non-binary participants also showed mosaics, underscoring that even among transgender and non-binary people, gender dysphoria and identity do not always travel together.

Several correlation patterns differed between cisgender men and women in ways the authors interpret through the lens of social pressure. Other-gender typical behavior correlated strongly with gender identity questioning in cisgender men (r = 0.69) but only moderately in cisgender women (r = 0.37), while in women it correlated more with own-gender typical behavior. The authors suggest that because feminine behavior in boys is less tolerated, boys who display it may be more likely to question their gender identity — or may under-report such behaviors, flattening other correlations. They caution against the increasingly common use of ‘gender expression’ as a label for gendered behavior, arguing that gender-atypical behavior does not necessarily reflect gender identity at all, a finding that contradicts the classic assertion that gender role is the public expression of gender identity.

The study’s implications reach beyond theory. The authors argue that educators, clinicians, and parents should recognize that being gender-typical in some characteristics and atypical in others is itself the statistical norm, and that adolescents whose feelings diverge from expectations are not abnormal. For clinicians working with transgender youth, the findings caution against inferring one aspect of gender from another: body dysphoria is not a universal component of the transgender experience, and a strong desire to alter sex characteristics is not always accompanied by identifying with the ‘other’ gender. The authors explicitly state that the results should not be used to undermine transgender identities or gender-affirming care, but rather to support a person-centered approach.

The researchers acknowledge important limitations. The retrospective design relies on memories that may be colored by current feelings, though restricting the sample to young adults mitigates this. Self-selection bias is possible, since participants knew the study concerned adolescent experiences, and some were deliberately recruited through body positivity and LGBTQ community groups to increase diversity — a strategy that, while enriching the data, prevents estimating true population prevalence. The study was conducted in Israel, which may limit generalizability, though the authors note consistency with prior international findings. Even with these caveats, the message is clear: the tidy binary story of puberty fits far fewer people than assumed, and the reality is a diverse, multi-dimensional landscape in which nearly everyone is typical in some ways and atypical in others.

Subject of Research: Retrospective study of gender identity, gender typicality, sexuality, and feelings toward pubertal sex characteristics in cisgender, transgender, and non-binary young adults

Article Title: Beyond binary models of pubertal development: gender identity, gender typicality, sexuality, and attitudes toward sex characteristics

Article References: Sion, A., & Joel, D. (2026). Beyond binary models of pubertal development: gender identity, gender typicality, sexuality, and attitudes toward sex characteristics. Frontiers in Social Psychology, 4, Article 1904592. https://doi.org/10.3389/frsps.2026.1904592

Image Credits: AI Generated

DOI: 10.3389/frsps.2026.1904592

Keywords: puberty, gender identity, gender typicality, sexual orientation, sex characteristics dysphoria, mosaic hypothesis, gender binary, adolescence, retrospective study, psychometrics, cisgender, transgender and non-binary