Research

Stereotypes and Memory (draft available upon request)

(joint work with Zhuokun Liu)

Imperfect memory can sustain widely held stereotypes that are inaccurate. To establish causal evidence, we conduct an online experiment where subjects grade a gender-stereotyped quiz and later predict the performance of the same group on a similar quiz. We vary (1) reliance on memory by manipulating access to grading records and (2) the information about the group by manipulating the sample of quiz-takers with different gender performance gaps. Defining the perceived gender performance gap as stereotypes, we find that stereotypes persist more strongly when subjects rely on memory, but only when the gap in the quiz is in the opposite direction of the stereotypes. Then, we extend the standard Bayesian framework to incorporate imperfect memory and find evidence of memory attenuation: subjects place less weight on graded quiz performance by 50%, due to uncertainty and potential contamination in recall, with contamination accounting for 72% of the effect.

Correlation Neglect and Information Demand