Same-Sex Role Model Effects in Education

Abstract
We study same-sex role model effects of teachers with a meta-analysis and our own study of three million students in 90 countries. Both approaches show that role model effects on performance are, on average, small: 0.030 SD in the meta-analysis and 0.015 SD in our multi-country study. Going beyond test scores, our multi-country study documents larger average role model effects on job preferences (0.063 SD), subject enjoyment (0.089 SD), and subject confidence (0.050 SD). We then shed light on an important question that has not yet been studied: How universal are same-sex role model effects? We address this question by estimating the distribution of role model effects based on our country-level estimates. Our results show that same-sex role model effects on test scores are universally small. For non-test score outcomes, in contrast, we find near universally positive role model effects as well as meaningful cross-country variation. We explore what explains this variation and find that role model effects on job preferences are larger in countries with larger gender gaps. These results are consistent with same-sex role models inspiring students to overcome stereotypes and pursue a STEM career. However, in countries with negligible gender gaps, same-sex role models do not seem to have this equalizing function.

This is joint work with Alexandra de Gendre, Jan Feld, and Nicolás Salamanca.