Gender Differences in Top-Performing Math Students’ Achievement and Motivation: An IPD Meta-Analysis

Background: Women’s underrepresentation in math-intensive fields such as science, technology, engineering, and mathematics (STEM) remains a concern of society (e.g., National Science Board, 2020). Previous research showed that students who did exceptionally well in standardized math tests (i.e., top-performing math students) are those most likely to major in and enter STEM fields (Halpern et al., 2007; Park et al., 2007). Several pre-college factors contribute to women’s underrepresentation in STEM. Among them are (a) students’ level of achievement, (b) their achievement profiles, and (c) their achievement motivation (e.g., Eccles, 1994; Wang et al., 2017). However, previous research has focused on U.S. samples and used data from special programs that solicit volunteers with unknown representativeness for the full student population (e.g., Lubinski & Benbow, 2006), which consequently has a potential for bias. Other studies used representative data to examine gender differences in top-performing math students’ achievement only (e.g., Baye & Monseur, 2016). Objectives: Capitalizing on representative individual participant data (IPD), the present IPD meta-analysis substantially enlarges the empirical evidence on gender differences among top-performing math students. Specifically, we investigated gender differences in top-performing math students (top 5% in their respective countries, N = 113,864, 82 countries) by drawing on international, representative, and unselective samples from well-defined populations––an approach considered the “gold standard” (Hedges & Nowell, 1995; Shrout, 2009) when studying gender differences. To do this, we used data from up to 343 independent student samples participating in six cycles of the Programme for International Student Assessment (PISA 2000–2015). Research questions: (1) How large are gender differences in achievement, achievement profiles, and achievement motivation in mathematics, reading, and science in the group of top-performing math students across countries? (2) To what extent do sociocultural factors (i.e., the level of gender equality in a country) moderate gender differences in the group of top-performing math students? Method: We applied the same analysis protocol to the six PISA cycles and integrated the results by using meta-analytical techniques. In accordance with the analysis strategy by Cheung and Jak (2016) for big data, we proceeded in three steps. In the first step, we computed effect sizes using the individual student data for each country and each PISA cycle. In the second step, we integrated the effect sizes to estimate (a) the average effect sizes for gender differences in achievement, achievement profiles, and achievement motivation and (b) the heterogeneity of effect sizes within and between countries. To do this, we used the R package “metaSEM” (Cheung, 2015) that implements random-effects models to allow the true effects to vary within and across countries. In the third step, we examined the extent to which moderator variables may explain the observed heterogeneity in effect sizes by using multivariate meta-regression models. Results: We found that in the group of top-performing math students, male students were overrepresented (mean female-to-male ratio 1:1.50, 95% CI [1:1.58, 1:1:43]). Furthermore, female students possessed better reading skills (dmean = –0.23, 95% CI [–0.25, –0.21]) and more positive reading attitudes (–0.64, 95% CI [–0.69, –0.60] ≤ dmean ≤ –0.38, 95% CI [–0.46, –0.30]). Male students had stronger math self-efficacy (dmean = 0.32, 95% CI [0.28, 0.35]), higher intentions to focus on math (dmean = 0.27, 95% CI [0.23, 0.31]), higher math self-concept (dmean = 0.15, 95% CI [0.12, 0.18]) and demonstrated mathematics-oriented achievement profiles, whereas female students’ profiles were more balanced across domains. Moreover, female students were more interested in organic and medical fields (–0.44, 95% CI [–0.48, –0.40] ≤ dmean ≤ –0.30, 95% CI [–0.34, –0.25]), whereas male students showed greater interest in physics-related topics (0.39, 95% CI [0.36, 0.43] ≤ dmean ≤ 0.54, 95% CI [0.50, 0.58]). Gender equality indicators moderated the proportion of female students in the top 5% in mathematics and explained variability in achievement profiles across countries. Conclusions and implications: The combination of IPD meta-analysis techniques and representative student data provided unbiased and precise empirical knowledge about the magnitude of gender differences in pre-college factors that may contribute to gender disparities in STEM. Our results demonstrate that the still existing preponderance of male students in the talent pool for STEM careers (i.e., at the top of the math achievement distribution) may partly explain women’s underrepresentation in STEM. Another potentially contributing factor may be male students’ more mathematics-oriented achievement profiles. Having one dominant academic strength is likely to promote higher self-concept in that domain and a clear goal to invest time, effort, and energy into pursuing mathematics-related fields in one’s future career. This lines up with our finding that male students reported on average slightly higher math self-concept, self-efficacy, and stronger intentions to choose additional math courses in school and beyond compared with female students. By contrast, having multiple academic strengths is likely to result in more diverse expectancies and self-concepts (Valla & Ceci, 2014) and, consequently, less specific career goals. This is more likely true for mathematically top-performing female students as they had more balanced achievement profiles and stronger verbal motivation than male students. Even if mathematically top-performing female students enter STEM careers, given their specific interests in organic sciences, they would be more likely to work in medical fields or biological sciences than in inorganic sciences. By contrast, top-performing male students would be more likely to enter inorganic STEM fields (e.g., physics or engineering), given their respective science interests. Furthermore, our results suggest that in societies that value higher education for women, more female students score in the top 5% in mathematics. Additionally, achievement differences in different domains are smaller for female students (and partially also for male students), the more women study at universities and the more women hold higher positions. Thus, the (realistic) perspective of attending a university and entering higher positions for female students might (a) motivate female students to develop mathematical talent and (b) motivate female and male students to develop skills in several areas at a more similar level.