Gender differences in mathematics performance: a meta-analysis.
Level 1 - systematic review of randomized trials
Meta-analysis of 100 studies (by design analogy, not clinical CEBM)
PubMed 2138794 · doi:10.1037/0033-2909.107.2.139
What was done
The authors conducted a meta-analysis of 100 studies yielding 254 independent effect sizes and representing 3,175,188 participants to assess the magnitude of gender differences in mathematics performance across mathematical task types, age groups, sample selectivity, and publication year.
What was found
Across general population samples, the average effect size was d = -0.05 (negative values indicating female superiority). For computation, d was -0.14; for understanding mathematical concepts, d was -0.03; and for complex problem solving, d was 0.08. Girls showed a slight advantage in computation in elementary and middle school, with no problem-solving differences at those ages. Problem-solving differences favoring males emerged in high school (d = 0.29) and college (d = 0.32). Differences favored females or were smallest in general population samples, but favored males most strongly in highly selected and highly precocious samples. The overall gender difference declined over time from d = 0.31 in studies published in or before 1973 to d = 0.14 in studies published in 1974 or later.
Why it matters
This landmark synthesis demonstrated that general-population gender differences in mathematics performance are small to negligible, while pinpointing specific age- and task-related divergences in late adolescence.
Limits
The abstract does not provide confidence intervals or assess the methodological quality and heterogeneity across the included primary studies. Factors explaining the emergence of high school differences in problem solving cannot be determined from the summarized test data.
Cited by
- supports There are no learning science papers showing that middle school girls are incapable of learning middle school math and science.