Event Recap: Experimental Economics Workshop on Discrimination and Prejudice

Louis-Pierre Lepage (Stockholm University) presented joint research on how anticipated discrimination in the labor market impacts women's choice of college major, leading to their underrepresentation in STEM, business, and economics. Using transcript data from a large U.S. university and a novel survey of students, the researchers found that women anticipate facing relatively more bias in STEM careers—from lower pay and fewer job offers to being taken less seriously, dealing with hostility, or facing sexual harassment—while men expect little or no such treatment. They then showed that these expectations predict women's course-taking and major choices, but not men's, and can account for up to 30% of the gender gap in STEM fields. The results suggest that anticipated discrimination distorts educational decisions, deterring women from entering high-paying fields and contributing to persistent gender gaps in the labor market.

Anne Boring (Erasmus University Rotterdam) and coauthors examined how blind hiring processes affect the application behavior of counter-stereotypical candidates. While most research on blind hiring practices focuses on the demand-side, or the employers, it focuses on the supply-side. They conducted two large, controlled experiments using candidates and recruiters, and measured candidates’ “willingness to apply” (WTA) under both a blind and non-blind application. They found that while blinding an application process initially reduces age and gender gaps by increasing WTA in counter-stereotypical candidates, a rejection after a blind application process has a significantly stronger deterrent effect on future applications for those candidates than a non-blind process.

Alejandro Martínez Marquina (University of Southern California) presented joint research on the negative spillover effects of gender-based quotas. They found robust evidence of a female quota backlash - they raised the bar for managers evaluating women not targeted by the quota. There was no backlash when the performance of the women targeted by the quota was very high. Overall, however, the backlash measured in the experiment was strong enough to render the quota ineffective in improving gender equity. This suggests that quotas may be more effective when targeting high performers in competitive markets.

Maria Kogelnik (Yale University) presented research with coauthor Philipp Strack (Yale University) on how people (mis-)understand the effect of quotas. They conducted an experiment where STEM students took a math test and were randomly assigned to groups of 10 to be evaluated as candidates for a job. There was no gender gap in the math performance, and therefore, the effect of the quota was minimal. However, they found that people massively overestimate the effect of quotas, both on their own outcome and on others. Men who were not selected grossly overestimated their likelihood of being hired without the quota. Women who were selected grossly overestimated the role of the quota in increasing their likelihood of being hired. Multiple treatments allowed the researchers to document that the mistakes that people make when reasoning about quotas are driven partly by ego-related biases and partly by cognitive errors related to contingent reasoning. They found that quotas can causally increase biases towards women, and may reduce their chances of being promoted later.

Arkadev Ghosh (Duke University) and his coauthors study stereotypes and misperceptions about outgroups, and how contact with outgroups changes these beliefs. Particularly, they identify evidence of effects of intergroup contact on generalization beyond a person whom they had met. Through an experiment in West Bengal, Hindu and Muslim workers were randomized to broad contact (brief interactions with many outgroup members) or deep contact (repeated work with one outgroup partner). The researchers found that broad exposure led to greater generalization by averaging out individual heterogeneity across people, but this was largely based on beliefs about easily observable traits (e.g., competence traits like productivity). In contrast, deep contact between groups revealed more low-visibility traits (e.g., friendship potential), but also led to lower generalization and can even polarize beliefs. They did not find evidence of generalized behavior change toward the outgroup, which may require contact that is both deep and broad simultaneously. Overall, this suggests a fundamental tradeoff in intergroup contact: deep contact builds strong individual ties, while only broad contact corrects misperceptions about the outgroup at large.

Adrian Blattner (Stanford University) presented research on the effect of cross-partisan contact among those who avoid exposure to differing political viewpoints and whether those choices maximize their own benefit. In a field experiment, participants recruited through apolitical social media ads engaged in structured video-call conversations with political outgroup members, with treatment assignment exploiting variation in willingness-to-accept to target those who specifically avoid differing political views. The researchers found a small but meaningful reduction in discrimination towards the outgroup, particularly among those with the highest baseline hostility. With reductions in polarization and increased support for future interaction persisting a month later, the findings suggest that forcing initial cross-partisan contact may break cycles of avoidance and reduce political hostility.

Milena Djourelova (USC) presented research on the impact of immersive experiences packaged with information in building empathy. Through a unique field experiment simulating migrants’ border-crossing experiences through VR, they randomized whether visitors received relatable information (shared attributes) about unauthorized migrants before or after the experience. The researchers found that the VR experience alone increased pro-immigration attitudes, but the effect was far larger when the information came first. Follow-up experiments replicate the effect of ordering for charitable giving, demonstrating that the boost in giving is weaker when the information is unrelatable to respondents, highlighting how attention and relatability shape empathy. The findings suggest that empathy is most effectively generated when information primes individuals to see shared attributes before encountering others’ experiences, rather than after.

Matteo F. Ferroni (University of Missouri) presented research exploring why Americans disagree on issues of racial inequality. Using large surveys of adults and teenagers, Ferroni and his coauthors found broad awareness of gaps in Black–White income and mobility, yet sharp disagreement over whether these gaps come from individual or systemic factors. Thus, disagreement occurs more over the root causes of racial gaps than the existence of such gaps. Ferroni also found that these fundamental beliefs were the strongest predictors of support for race-targeted policies, while support for redistributive policies was more determined by political affiliation rather than race. With an additional information treatment, Ferroni found that emphasizing systemic racism created the largest shifts in perception and policy support among White respondents, underscoring the study’s findings that shifting fundamental beliefs are the most politically consequential.

Christine Exley (University of Michigan) presented research on societal views of men and women who exhibit similar traits and behavior. Using measures of traits such as confidence and competitiveness, the researchers elicited beliefs about how perceived societal views of men or women affect varying traits and choices. The researchers found that men face a restricted path: conforming to traditionally “male” traits maximizes expected praise and minimizes criticism. In contrast, women face what Exley calls a “mission impossible”: adopting these same traits increases expected praise but also increases expected criticism, leaving no clear strategy to avoid social penalties. Follow-up studies on workplace competition and gender-role expectations reveal similar “mission impossible” dynamics, highlighting how expected societal judgment may deter women from pursuing behaviors that are otherwise rewarded in the labor market.

The broad range of research presented at the Experimental Economics Workshop demonstrates how behavioral and experimental methods are uniquely suited to decipher some of the most persistent and complex social challenges. Drawing on lab and field experiments across diverse contexts—from US university classrooms to labor markets in India—scholars showcased how beliefs, expectations, and social norms shape behavior long before formal policies take effect. The findings highlight that well-intentioned interventions can generate unintended consequences, making design details crucial. Likewise, innovations in immersive technologies and large-scale surveys demonstrate how Experimental Economics continues to answer nuanced questions and is a powerful tool for diagnosing the roots of discrimination and prejudice.