Predictive Power of Biological Sex and Gender Identity on Economic Behavior

Behavioral differences by biological sex are still not fully understood, suggesting that studying gender differences in behavioral traits through the lenses of continuous identity might be a promising avenue to understand the remaining observed gender gaps. Using a large U.S. online sample (N=2017) and machine learning, we develop and validate a new continuous gender identity measure consisting of separate femininity and masculinity scores. In a first study, we identify ninety attributes from prior research and conduct an experiment to classify them as feminine and masculine. In a subsequent study, a different group of participants completes tasks designed to elicit behavioral traits that have been previously documented in the behavioral economics literature to exhibit binary gender differences. Data for the second study are collected in two waves; the first wave serves as a training sample, allowing us to identify key attributes predicting behavioral traits, create candidate identity measures, and select the most effective one, comprising sixteen attributes, based on predictive power. Finally, we use the second wave (test sample) to validate our gender identity measure, which outperforms existing ones in explaining gender differences in economic decision-making. We show that confidence, competition, and risk are associated with masculinity, while altruism, equality, and efficiency are with femininity, providing new possibilities for targeted policymaking.