The Limited Power of Post-Election Samples to Detect Incorrect Election Results in the Absence of Similarity Assumptions

Resolving Contested Elections: The Limited Power of Post-Vote Vote Choice Data (with Thomas S. Richardson and Mark S. Handcock)

In close elections, the losing side has an incentive to obtain evidence of an incorrect election result. Sometimes this evidence comes in the form of court testimony from a subsample of invalid voters (Borders v King County, 2005; Belcher v Mayor of Ann Arbor, 1978). In this paper we show that even if such testimony is accurate, it has limited power to detect incorrect election results without precinct level polarization or high response rates. Therefore in U.S. election disputes, even high quality post-vote vote choice data will often not be sufficient to resolve contested elections, and court decisions will hinge on debates over modeling assumptions.

Submitted to the Journal of the American Statistical Association pdf