Means, Tails, and Fairytales – Ways to Investigate Gender Effects

Abstract

This methodological work focuses on two shortcomings of gender research in education: Firstly, most studies focus on a restricted range of methods to test for gender effects with an unjustified focus on differences. Secondly, most studies assume a binary operationalization of gender, marginalizing individuals with non-binary and non-conforming gender identities, raising ethical problems and validity issues (Kube et al., 2022; Meier & Diefenbach, 2020). In principle, three core perspectives on gender effects exist: the gender difference hypothesis (Hyde, 2005), the gender similarities hypothesis (Hyde, 2005), and the greater male variability hypothesis (Shields, 1982). Different methods exist to test the three hypotheses, such as t-tests and Cohen’s d for differences, equivalence tests and overlap measures for similarities, and variance ratios and male-female ratios for variances. To investigate how these perspectives and methods are employed, we conducted a literature review of gender studies conducted with PISA-data published until 2020. Of 69 articles, 61% reported methods focusing on differences and 23% on variances. Only 36% of the articles applied several methods. This review demonstrates the prevalent focus on differences. We argue that the decision for one perspective has tremendous consequences for how data is depicted, and that multiple methods should be applied in order to fully understand comparisons based on gender. Furthermore, all perspectives have the shortcoming that they assume a binary perspective on gender, as they are only defined for two groups. Hence, we discuss advantages and disadvantages of assessing gender-identity on a continuum and categorically, and list methods that overcome this shortcoming, such as ANOVA, regression analysis, and visualization techniques.