Women consistently lean toward literacy as their strongest cognitive skill, while men tilt toward numeracy, and those patterns hold from early adulthood well into the 50s and beyond, according to a peer-reviewed study published in the Journal of Intelligence that analyzed adult skills data from roughly 30 countries.
The research, which drew on the OECD’s Programme for the International Assessment of Adult Competencies (PIAAC) Cycle 2 dataset released in December 2024, did not simply compare men’s and women’s average test scores. Instead, it measured each person’s performance in literacy, numeracy, and adaptive problem-solving against their own cross-domain average, a technique that reveals what each individual does best relative to themselves rather than relative to other test-takers.
The result is a detailed map of cognitive specialization across five age bands (16 to 24, 25 to 34, 35 to 44, 45 to 54, and 55-plus) and dozens of nations. As of May 2026, it represents one of the most comprehensive looks at how gendered cognitive profiles persist, shift, or intensify over the adult lifespan.
What the study actually found
When the researchers ranked each participant’s three domain scores against their own personal average, a clear pattern emerged. Women were more likely than men to show literacy as a relative strength, meaning their reading scores outpaced their own math and problem-solving marks. Men, conversely, were more likely to show numeracy as a relative strength. Problem-solving in technology-rich environments fell somewhere in between, though men showed a slight edge in that domain as well.
These patterns appeared in the youngest age group (16 to 24) and remained visible in every subsequent band through age 55 and older. The persistence is notable because adults in their 50s grew up in vastly different educational and economic conditions than today’s twentysomethings, yet the broad shape of the sex-linked profiles looked similar across generations.
The differences were not enormous in absolute terms. Men and women overlapped substantially on all three domains. But the relative tilt, the question of which skill each person did best at compared to their own baseline, showed a consistent gender split that did not vanish in any age group or in any broad regional cluster of countries.
How the data were collected
PIAAC is sometimes called the “adult PISA.” Administered by the OECD, it tests nationally representative samples of adults aged 16 to 65 in literacy, numeracy, and (in Cycle 2) adaptive problem-solving. The second cycle’s first round of results, covering approximately 30 countries and economies, was publicly released in December 2024. The U.S. National Center for Education Statistics confirmed the release timeline, and the full public-use microdata files are freely downloadable in CSV, SPSS, and SAS formats, meaning any researcher with statistical software can replicate or challenge the study’s findings.
The authors used the OECD’s standard procedures for weighting and variance estimation, which account for the complex sampling design and the use of multiple “plausible values” for each test-taker’s scores. They also ran robustness checks with alternative score cutoffs and after excluding participants with missing background data. The main patterns held across those variations.
A full open-access version of the paper, including supplementary tables and figures, is available through the National Library of Medicine.
The gender-equality paradox, revisited
One of the study’s most provocative threads connects to a long-running debate in psychology known as the gender-equality paradox. The paradox, in short, is this: countries that score highest on measures of gender equality often show larger, not smaller, sex differences in certain outcomes, including STEM career choices and cognitive-profile patterns.
A related paper by overlapping authors, published in Psychological Science, applied the same intraindividual-strengths method to multiple waves of the PISA assessment for teenagers. That work found that sex-typed cognitive profiles (girls tilting toward reading, boys toward math) did not shrink in more gender-equal nations. The new PIAAC study extends that finding from adolescents into the full working-age population, suggesting the paradox is not just a feature of schooling but persists into adult life.
Why this happens remains genuinely contested. Competing explanations include the idea that greater freedom of choice allows people to lean into pre-existing preferences, that subtle forms of gendered socialization survive even in egalitarian societies, and that measurement artifacts in equality indices distort the picture. The current study does not resolve that debate, but it adds a significant new data point by showing the pattern across adult age groups, not just among 15-year-olds.
What the study cannot tell us
PIAAC Cycle 2 tested adults at a single point in time, which means the age-band comparisons reflect differences between generations, not changes within the same individuals as they grow older. A 55-year-old tested in 2022 or 2023 came of age in a different world than a 20-year-old sitting for the same assessment. Differences in schooling quality, labor-market access, and exposure to digital technology could all shape cognitive profiles independently of sex.
The study also cannot pinpoint causes. It describes patterns but does not directly measure whether they stem from biology, classroom practices, occupational sorting, cultural expectations, or some combination. Two adults with identical overall ability levels can show opposite profiles, one literacy-dominant, the other numeracy-dominant, and the data alone cannot say why.
Country-level variation adds further complexity. While the broad sex-linked patterns appeared widely, they were not identical everywhere. The degree to which national culture, economic structure, or education policy amplifies or dampens these differences is an open question. Roughly 30 nations is a large sample, but it still leaves out much of Africa, South Asia, and Latin America, limiting how far the conclusions can be generalized.
Finally, without longitudinal data tracking the same people over years or decades, it is impossible to know how stable an individual’s cognitive profile is. Extended work in a numeracy-heavy job might shift someone’s relative strengths, but the cross-sectional snapshot cannot capture that kind of change. Future PIAAC waves or dedicated panel studies would be needed to test how malleable these profiles really are.
What this means for education and workforce policy
The most practical implication is not that one sex is “better” at any domain. The overlap between men and women on all three skills is large. Rather, the findings suggest that men and women, on average, tend to differ in how their abilities are distributed across literacy, numeracy, and problem-solving, and that this distribution is remarkably consistent across age groups and national contexts.
For educators and workforce trainers, that points toward diagnostic approaches: identifying each learner’s relative strengths and weaknesses rather than assuming a uniform profile. Programs that build on a person’s strongest domain while systematically supporting weaker ones may outperform one-size-fits-all curricula, regardless of whether the underlying sex differences trace to biology, culture, or both.
At the same time, the persistence of these patterns across countries with varying levels of gender equality suggests that policy changes alone are unlikely to rapidly erase all group-level differences in cognitive specialization. That is not an argument against equity efforts. It is a caution against using the disappearance of all statistical sex differences as the sole benchmark for success. Understanding which aspects of cognitive specialization are flexible, and under what conditions they shift, remains one of the more consequential open questions in education research.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
