Our original article here is by Thompson.
Use the replication data.
Use rdrobust to estimate the incumbency advantage for sheriffs at the 50% vote threshold. Report the conventional, bias‑corrected, and robust estimates. Plot the RD graph using rdplot.
Conduct a density test using rddensity. Is there evidence of sorting around the cutoff? Interpret the plot and p‑value.
Choose two placebo cutoffs (e.g., 0.4 and 0.6) and re‑run rdrobust. Do you find significant effects? What does this tell you about the validity of the design?
Test whether a pre‑treatment covariate (e.g., past election margin) is smooth at the cutoff using rdrobust with that covariate as the outcome. Report the result. If you find imbalance, what does that imply?