Joint with Ryan Copus.
We use a newly constructed dataset of all civil rights cases between 1995 and 2016 in seven U.S. District Courts to reevaluate the differences in decision-making between judges appointed by Democratic presidents and judges appointed by Republican presidents. We make four main contributions. First, by analyzing the ultimate outcome of cases rather than the outcome at any particular decision-point, we preserve the causal identification opportunity provided by as-if random case assignment. Second, we confirm the common finding that outcomes do not change, on average, as a result of assignment to a Republican appointee. Third, we demonstrate that the low average treatment effects miss the fact that Republican appointees have positive effects in some cases and negative effects in others. Finally, we show that the effect of assignment to a Republican appointee can increase substantially if one focuses attention on the “hardest” cases. For example, in the hardest 25% of cases, we estimate that assignment to a Republican appointee decreases settlement rate by around 10% in some cases and increases it by around 4% in the other cases.