Research

Using a natural experiment leveraging the 2024 U.S. presidential election, and a controlled online experiment, I show that beliefs about discrimination by third parties — such as an organization's clients — can cause hiring discrimination, and that these beliefs tend to be biased. A light-touch information intervention can correct these distortions.

We find discrimination against Black homebuyers in the U.S. real estate market in an audit experiment with 24,000 realtors across all 50 U.S. states, and find that "sponsorship" (intervention by a credible intermediary) does not attenuate this discrimination, even though it improves certain impressions of buyers.

Manuscript available on request

Across four pre-registered, incentive-compatible experiments (N = 12,000), we examine whether shortlist quotas — requirements to include a minimum number of underrepresented candidates in finalist pools — increase diversity in hiring through both direct choice set effects and positive signaling effects. We find little evidence of backlash.

* Joint first authors · Manuscript available on request

Using a comprehensive dataset of U.S. high school track times, we show that narrowly surpassing a performance goal in the present leads to lower performance and persistence in future periods — "slacking after success" — which cannot be explained by a standard model of reference dependence. We propose an alternative model that better fits the data.

Fresh starts — new weeks, months, or years — motivate goal pursuit by creating psychological distance from past failures. Using archival data from UK ParkRun participants and a controlled online experiment, we examine the flip side: whether failures during a fresh start period are weighted more heavily, producing outsized reductions in future effort.

We examine whether discrimination can arise from decision-makers' uncertainty about outgroup preferences: the sense that they cannot readily anticipate the tastes and wants of a buyer or worker from a marginalized group.