Study design's influence on empirical findings in economic education

We use Meta-Regression Analysis (MRA) to investigate how study design influences the finding of a gender gap in the economics education literature. We survey 65 studies published in the <italic>American Economics Review Papers and Proceedings</italic> and the <italic>Journal of Economic Education</italic> over the past 30 years, containing 279 distinct regressions. We find that studies with more explanatory variables, include statistics courses taught by economists, or sample preadolescent children are associated with a reduced likelihood of observing a gender gap; in contrast, studies that employ a stock model methodology or use course grade as the dependent variable are significantly more likely to find evidence of a gender gap. Estimation strategy seems to have little impact. We conclude that much of the conventional view of a gender gap in economics performance can be attributed to study design.