Scopri IRVAPP

This seminar proposes a new manipulation test for regression discontinuity designs based on Benford's Law. Since manipulation often creates asymmetry in the running variable's density near the cutoff, the authors use Benford's Law—well known in fraud detection—to detect it. Unlike McCrary-type tests, this approach removes researcher-chosen parameters, using predetermined thresholds from Nigrini (2012) instead. A new bandwidth-selection method and probability-based thresholds extend Benford's Law beyond digit analysis, with empirical examples included.