Prepared for presentation at the 2013 Summer Meeting of the Political Methodology Society, University of Virginia, July 18–20.
ISSN/ISBN: Not available at this time. DOI: Not available at this time.
Abstract: Very recently there has been controversy about a method suggested for detecting election fraud: Pericchi and Torres argue that Benford’s Law applied to the second digits of vote counts can be a standard for detecting fraud, while Deckert, Myagkov and Ordeshook argue that Benford’s Law is useless for this purpose. Using data from elections from several countries and election systems I show that with precinct- or polling station-level vote counts, the so-called second-digit Benford’s Law distribution (2BL) describes very few of the empirical distributions. Contra Pericchi and Torres, however, it is not that fraud is rife in all these elections. Instead the digits in vote counts can help diagnose both the strategies voters use in elections and nonstrategic special mobilizations affecting votes for some candidates. Using excerpts from my book manuscript Election Forensics, which examine data from elections in the United States, Germany, Canada and Mexico, I assess the performance of tests based on the second significant digits of precinct-level vote counts. The claim that deviations in vote counts’ digits from the distribution implied by Benford’s Law is an indicator for election fraud generally fails. With precinct vote counts, the second significant digits are sensitive both to imbalances in district preferences and to the strategies voters are using in the election as well as to other special mobilizations. All these produce systematic deviations from the distribution implied by 2BL. Similar patterns are observed in many elections in many countries when there is virtually no fraud.
Bibtex:
@unpublished{,
AUTHOR = {Walter R. Mebane, Jr},
NOTE = {Prepared for presentation at the 2013 Summer Meeting of the Political Methodology Society, University of Virginia, July 18–20},
TITLE = {Election Forensics: The Meanings of Precinct Vote Counts’ Second Digits},
YEAR = {2013},
URL = {https://pages.shanti.virginia.edu/PolMeth/files/2013/07/Mebane.pdf},
}
Reference Type: Conference Paper
Subject Area(s): Voting Fraud