We introduce a method to statistically characterize compliers and non-compliers in an (quasi-) experiment in which some subjects are assigned to take a treatment but free to choose whether to comply or not with this assignment.
We show that hypothetical choices measured in a conjoint survey experiment are driven by the same structural determinants of the actual choices made in the real world.
Using lottery tie-breakers in local elections as experimental benchmarks, we find that different RDD estimators accurately recover experimental benchmarks when the conditional expectation function of outcomes near the cutoff is close to linear. When approximating curvature near the cutoff is more challenging, bias-corrected estimators with robust inference outperform alternative methods.