How Dark Is Dark? Bright Lights, Big City, Racial Profiling

Grogger and Ridgeway (2006) develop a feasible test for racial profiling "behind a veil of darkness" that exploits the exogenous variability in the visibility of driver race caused by the daylight savings time shift. The idea is that police are less able to use race as a criteria in traffic stops when it is dark, so differences in stop rates for African-Americans between dark and light (while controlling for time of day) may be evidence of racial profiling. However, in urban environments streets may be well-lit at nighttime, leading to low power tests and false inferences. Using satellite imagery of ambient streetlight, we refine their "veil of darkness" methodology and test for racial profiling in Syracuse, NY from 2006 to 2009 (while controlling for time of day and location). We find that ignoring ambient lighting leads to a conclusion of no racial profiling against African-Americans, while accounting for ambient lighting leads to the opposite result.