Women score lower on socio-emotional skills. Or do they? New evidence from Tanzania

Across much of the research on low- and middle-income countries, a familiar finding keeps showing up in the data: men appear to have stronger socio-emotional skills than women. New evidence from Tanzania suggests this pattern may be a product of how skills are measured, rather than a reflection of how they are actually distributed. A closer look at socio-emotional skills in Tanzania Most evidence on socio-emotional skills relies on self-reported surveys, where individuals rate their own abilities. But self-reports can capture more than skills alone, they also reflect confidence, social desirability, and gendered beliefs about competence. To disentangle these factors, the Africa Gender Innovation Lab (GIL), in collaboration with Innovations for Poverty Action (IPA), compared self-reported measures with behavioral assessments of socio-emotional skills among young people in Tanzania. The study surveyed more than 4,400 young women and men who were not in education, employment, or training (NEET) across three Tanzanian cities: Dar es Salaam, Dodoma, and Iringa. NEET youth are a key policy target in Sub-Saharan Africa, and while they may differ from other young people in ways such as economic stress or mental health, these differences are unlikely to drive the gender patterns documented here since the analysis compares men and women within this group. Participants were assessed on 14 socio-emotional skills using the ESTEEM framework, ranging from awareness skills such as empathy and self-awareness, to management skills such as perseverance, problem-solving, and collaboration. Each skill was measured using self-reported scales, and through behavioral assessments, including situational judgment tests and task-based exercises designed to capture how individuals respond in concrete scenarios. This dual approach allowed a direct comparison of how people perceived their own skills (self-reported) versus how they perform when those skills are put to the test. Self-reports show a male advantage, but behavioral measures do not Using self-reported measures alone, the results appear striking. Men report significantly higher levels of socio-emotional skills than women across nearly all skill categories, with an average advantage of about 0.20 standard deviations — meaning the average man would outscore roughly 58 out of 100 women and be outscored by the remaining 42. This pattern reinforces the common narrative that women lag behind men in these domains. However, when socio-emotional skills are measured through behavioral assessments, gender differences largely vanish. Out of the 14 skills assessed, only a handful show small gender gaps — most notably active listening, emotional regulation, perseverance, problem-solving, and networking — and these differences are modest, typically no larger than 0.08 standard deviations — barely distinguishable from a coin flip. Taken together, these results point to a sharp contrast: while self-reported measures suggest a substantial male advantage, behavioral measures, often less biased, indicate little to no systematic gender difference in skill levels. Overestimation or underestimation? If women are not performing worse than men, why do self-reported gaps appear so large? The analysis points to a clear answer: men tend to overestimate their socio-emotional skills more than women underestimate theirs. The gap between what men report and what their behavior demonstrates is significantly larger than the equivalent gap for women. Among men, this inflation in self-reported skills is strongly associated with two factors: • Social desirability bias, meaning the tendency to respond in ways that align with what is socially valued or expected, and • Gendered beliefs about competence, particularly the belief that men are naturally better at problem-solving and decision-making. Men who hold these beliefs most strongly show the largest gap between what they report and how they actually perform. These patterns are not observed among women. Women’s self-reports are not tied to social desirability or beliefs of gendered competences. The implication is significant: what looks like a skills gap between men and women may, in large part, be a perception gap - and it is running in the opposite direction from what is commonly assumed. Looking ahead These findings have three direct implications for labor market policies, training programs, and hiring practices, across low- and middle-income countries. • Programs that diagnose gender gaps using self-reported socio-emotional skills should be interpreted with caution. If men systematically overstate their abilities, self-report-only approaches will exaggerate women's disadvantages and direct resources toward the wrong problem. Funders and program designers should treat self-reported gender gaps in these domains as a starting point for investigation, not a confirmed baseline. • Behavioral assessments should become a standard component of skills measurement where accuracy matters. The ESTEEM tools used in this study required more time and enumerator training than self-reports, but that cost is manageable for programs where accurate skills measurement is a priority. • Hiring and selection processes should be redesigned to reduce the advantage that overconfidence currently confers. Employers who rely on interviews or CVs may be systematically rewarding overconfidence rather than ability. Structured assessments that test performance directly — rather than deferring to self-description — would level the playing field in ways that benefit employers and candidates alike. Getting measurement right The implication are important: some of the gender gaps that have shaped a decade of training programs and hiring policies may not reflect reality. Women are not underperforming — they are being measured by tools that inflate the appearance of male advantage. As investments in youth employment and women's economic empowerment continue to scale, getting the diagnosis right is not a technical detail