Geography in Election Forensics

Abstract

We begin to investigate how geography affects the ability to draw inferences - based on reported vote counts - about election frauds or about other election features such as voters' strategies. Can we localize places where anomalies occur? It can be important to take into account geographic features such as local jurisdictional boundaries, but beyond that is proximity in physical space important? Should physically neighboring precincts be analyzed as having similar probabilities of frauds or of voters acting strategically, say, independent of whatever other covariates may be used in some model? We begin to study these concerns using data from several countries (Germany, Mexico, Russia).

Keywords: election forensics,geography,GIS,second digit tests,elections

JEL Classification: C14