When does Regression Discontinuity Design Work? Evidence from Random Election Outcomes

We use elections data in which a large number of ties in vote counts between candidates are resolved via lottery. We benchmark non-experimental RDD estimates of personal incumbency advantage against the estimate produced by this experiment that takes place exactly at the cutoff. The experimental estimate suggests that there is no personal incumbency advantage. In contrast, standard local linear RDD estimates suggest a moderate and statistically significant effect. Bias-correction and undersmoothing procedures however bring the RDD estimate(s) in line with the experimental estimate. Therefore, careful implementation of RDD can meet the replication standard in the context of close elections. 'This version (September 21, 2015) replaces the November 20, 2014 version.'