The most economically prosperous and gender-equal societies often display stronger gender segregation in educational choices – a pattern known as the ‘gender equality paradox’. This column explores this paradox using data on half a million children in 37 Western countries and 10,000 British families over two decades. The findings suggest that gender segregation in education stems from gender segregation in childhood activities and friendships, which are shaped by family income and parenting styles. Policymakers may need to move beyond promoting formal equality and focus on family, community, and school interventions that encourage mixed-gender interactions.
Around the world, women are underrepresented in STEM fields (science, technology, engineering and mathematics), while men are minorities in areas such as education and health. Paradoxically, the most economically prosperous and gender-equal societies often display stronger gender segregation in educational choices – a pattern known as the ‘gender equality paradox’. Our new research (Bagues and Zinovyeva 2025), based on data from half a million children in 37 Western countries and a longitudinal panel of 10,000 British families over two decades, offers a novel explanation: gender segregation in education stems from gender segregation in childhood activities and friendships, which in turn is shaped by family income and parenting styles.
Time does not solve the problem
Over the last two decades, societal attitudes toward gender roles have become increasingly egalitarian and women have made substantial gains in labour market participation, politics, and educational attainment. However, gender segregation in higher education has remained surprisingly stable (Bozzano and Bertocchi 2020, Delaney and Devereux 2019, Sevilla et al. 2025, Tinsley et al. 2017). In Western countries, the Duncan segregation index remains close to 0.30, indicating that 30% of students would need to switch majors to achieve equal gender distribution across all fields of study (Figure 1).
Figure 1 Evolution of gender segregation in higher education and gender inequality
Notes: Data cover 37 European and North American countries. The Duncan Segregation Index measures the minimum proportion of graduates of either gender who would need to change fields to achieve equal gender distribution in higher education, using the UNESCO 1-digit level classification system. The Gender Inequality Index captures disparities in reproductive health, empowerment, and labour market participation, with lower values indicating greater gender equality.
The ‘paradox’ of gender equality
Even more surprisingly, this segregation is stronger in the most prosperous and egalitarian societies. The data reveal a counterintuitive but consistent correlation: in the richest countries with the greatest gender equality, women have a lower propensity to graduate in STEM and there is greater educational segregation overall. For instance, during 2000–2021, a Finnish female graduate was four times less likely than her male counterpart to have studied STEM; in Spain, the ratio was three to one, and in Portugal, around two to one, despite lower overall gender equality scores in Spain and Portugal compared with Finland (Figure 2).
Figure 2 The gender equality paradox in higher education
Notes: Data cover 37 European and North American countries. All variables are averaged over the 2000–2021 period. Each point represents a country average; dashed lines show fitted linear regressions. ‘Female propensity to STEM’ is the ratio between the share of female graduates in STEM and the share of male graduates in STEM. The Gender Inequality Index captures disparities in reproductive health, empowerment, and labour market participation, with lower values indicating greater gender equality.
Previous studies in sociology have attributed the greater educational segregation in more prosperous societies to a cultural shift toward emphasising personal fulfilment and self-expression (Charles and Bradley 2009). When basic economic security is assured, this theory suggests, people feel freer to pursue educational fields that align with their internalised preferences. For example, when college students are asked to focus on careers that excite them rather than those with the best economic prospects, gender-typical patterns widen (Siy et al. 2023). Economic factors, such as differences in occupational wage gaps and the size of service sectors across countries, may also contribute to these patterns. While these explanations are compelling, they focus primarily on the moment of educational choice in late adolescence. Our research suggests the roots of this paradox may lie much earlier – in childhood itself.
Tracing segregation back to childhood
Data on over half a million 11- to 15-year-olds from the Health Behaviour in School-Aged Children shows that childhood friendship patterns mirror educational segregation. Overall, children aged 11 report that only 27% of their close friends are of the opposite sex, though the proportion varies dramatically by country. In Finland, where relatively few women graduate in STEM, children report that just 16% of their friends are of the opposite sex. In Spain and Portugal, where women graduate in STEM at substantially higher rates, 30% and 39% of children’s friends, respectively, are of the opposite sex (Figure 3).
Figure 3 Gender segregation in higher education and share of opposite-sex friends in childhood
Notes: Both variables are averaged over the 2002–2010 period. Each point represents a country average; dashed lines show fitted linear regressions. ‘Share friends opposite-sex’ is measured at age 11 as the ratio of opposite-sex friends to total friends, using data from children with uncensored information on friend counts. ‘Female propensity to STEM’ is the ratio between the share of female graduates in STEM and the share of male graduates in STEM.
British longitudinal data confirm this connection at the individual level. Children who had fewer opposite-sex friends at age 7 are more likely to select subjects dominated by their own gender in high school, even after controlling for cognitive ability, personality traits, family background and parental attitudes. These children also tend to hold more traditional gender role views in adolescence and aspire to gender-stereotypical careers.
These findings suggest that early social environments – long before career decisions – help shape tastes, identity, and norms about appropriate fields of study. Peer influence within gender-homogeneous groups appears to generate self-reinforcing feedback loops: children in same-gender groups may reinforce each other’s gender-typical interests, gradually steering them toward gender-typical activities and academic tracks.
Norms, income, and childhood segregation
Why do children have more opposite-sex friends in some countries? At first glance, we observe a strong inverse correlation with traditional measures of gender equality: the percentage of opposite-sex friends is lower in countries that are more egalitarian according to traditional gender equality indices (Figure 4). However, this correlation does not necessarily represent a causal relationship, as these countries also differ in numerous other dimensions. Most notably, we observe that children tend to have significantly fewer opposite-sex friends in richer countries and in countries where parents place greater emphasis on children’s self-realisation and self-expression.
Figure 4 Gender inequality and proportion of opposite-sex friends in childhood
Notes: Both variables are averaged over the 2002–2010 period. Each point represents a country average; dashed lines show fitted linear regressions. ‘Share of friends opposite-sex’ is measured at age 11 as the ratio of opposite-sex friends to total friends, using data from children with uncensored information on friend counts. The Gender Inequality Index captures disparities in reproductive health, empowerment, and labour market participation, with lower values indicating greater gender equality.
To disentangle norms from income, we exploit within-country variation. In 34 of the 37 countries, children from higher-income families report fewer opposite-sex friends. In the UK cohort, for instance, children in the top income quintile are 36% less likely to have mixed-gender friendship groups by age 11 than those in the bottom quintile. Yet children whose parents hold more egalitarian attitudes have more opposite-sex friends, suggesting that family income – not parental attitudes toward equality – is the key driver of the country-level patterns.
How prosperity creates segregation
Our research identifies two complementary mechanisms through which economic prosperity inadvertently generates greater gender segregation in childhood.
First, affluent families often embrace parental values that emphasise children’s self-expression, while simultaneously enrolling their children in organised activities that tend to be gender segregated. This combination of progressive ideology with structured socialisation may foster environments where children can freely express typical gender preferences. In our data, girls account for about 80% of after-school dance class participants and 66% of music lesson attendees. Sports clubs, more common among higher-income families, also tend to be organised along gender lines. By contrast, children from less affluent families spend more unsupervised time with relatives and neighbourhood peers, creating more opportunities for spontaneous cross-gender interactions.
Second, children from higher-income families typically exhibit better emotional regulation, stronger cooperation skills, and fewer behavioural difficulties – traits that tend to correlate with fewer opposite-sex friendships. The number of opposite-sex friendships typically increases as children approach adolescence, but children with better emotional regulation and fewer behavioural difficulties may be more content with their existing same-gender groups, delaying this natural transition and maintaining for longer the predominantly same-sex friendship groups characteristic of early childhood.
Together, these mechanisms indicate that the very advantages that affluent families provide their children – structured activities, emotional stability, and strong family relationships – might paradoxically contribute to the persistence of educational gender segregation.
Rethinking policies for gender balance
While our analysis with observational data cannot definitively establish causation, the observed patterns strongly suggest that gender segregation in higher education and labour markets has important roots in childhood social dynamics. Our research shows how childhood friendship patterns, shaped by family income and parenting styles, set children on gender-typed trajectories long before they make educational choices in middle and high school.
These findings indicate that eliminating formal barriers and promoting gender-equal values, while important steps, may not be sufficient to reduce educational gender segregation. The risk is that peer influence within segregated networks may deter individuals from pursuing fields that match their abilities rather than their social environment. If affluent parenting styles unintentionally reinforce gender-segregated peer groups, more nuanced interventions are needed to counteract the pull of homogeneous friendships.
Addressing this potential inefficiency may require a life-course approach that integrates family, community, and school interventions. Potential policy options include encouraging mixed-gender activities in after-school and community programmes; launching parental guidance campaigns to design inclusive structured activities; and implementing early counter-stereotyping initiatives that create spaces for children with non-stereotypical interests to interact with same-sex peers who share those interests. The goal is to ensure that children’s social worlds do not inadvertently limit their academic and career horizons.
Source : VOXeu
