Report for the Nature Human Behaviour Mass Reproduction Initiative: Seeing Racial Avoidance on New York City Streets

This report investigates the computational reproducibility and robustness of the paper "Seeing Racial Avoidance on New York City Streets" (Dietrich and Sands 2023). These reproduction efforts are part of the mass reproduction initiative for articles published in Nature Human Behaviour (NHB), which is jointly organized by the Institute for Replication and NHB. In the original NHB paper, Dietrich and Sands analyze a field experiment in New York City, finding that "pedestrians deviate by, on average, 3.43% of the sidewalk width [...] or around 4 inches, in the presence of black confederates" (compared to white confederates), signalling a statistically significant racial avoidance of black people. For this report, I first conduct a step-by-step reproduction of the original replication materials, followed by robustness checks including 1) an analysis without outliers, 2) analyses with alternative seeds for the bootstrapped standard errors, and 3) an analyis with non-bootstrapped standard errors. I find that the original results are fully reproducible and that they are robust to many, but not all, alternative specifications.