Wednesday, January 29, 2020
Andrew Dustan, Kristine Koutout & Greg Leo, Beliefs About Beliefs About Gender
Do women believe that leaders in science, technology, engineering, and math (STEM) fields believe that women are bad at doing science? Such beliefs about beliefs—second order beliefs—could drive women to sort out of STEM fields, leading to the observed gender gap in employment (Beede et al., 2011). Importantly, this belief-driven sorting could occur regardless of leaders’ true beliefs about women’s scientific abilities. When historically persistent beliefs about the differences between men and women—first-order beliefs—cause disparities, they may generate second-order beliefs that perpetuate those disparities even once first-order beliefs change. To facilitate investigating questions of this nature, we develop an incentive-compatible
experimental framework for measuring first- and second-order beliefs about the difference in any quantifiable characteristic between any two populations. We implement this procedure in a lab experiment to elicit beliefs about men’s and women’s performance on a timed math task and choices in an abstract bargaining task.
We find an interesting contrast between first- and second-order beliefs. There is no evidence that men’s and women’s first-order beliefs differ; however, both men and women believe that such differences exist. While a large majority of people believe that most men believe men outscore women on the math task, the majority also believe that most women do not share this belief. In the bargaining task, we again find that people believe that men and women hold different first-order beliefs even though we observe no such differences in the data. In summary, even when men and women have similar first-order
beliefs, second-order beliefs about men and women can vary substantially.