Fraudulent Democracy? An Analysis of Argentina'sInfamous DecadeUsing Supervised Machine Learning

Abstract

In this paper, we introduce an innovative method to diagnose electoral fraud using vote counts. Specifically, we use synthetic data to develop and train a fraud detection prototype. We employ a naive Bayes classifier as our learning algorithm and rely on digital analysis to identify the features that are most informative about class distinctions. To evaluate the detection capability of the classifier, we use authentic data drawn from a novel data set of district-level vote counts in the province of Buenos Aires (Argentina) between 1931 and 1941, a period with a checkered history of fraud. Our results corroborate the validity of our approach: The elections considered to be irregular (legitimate) by most historical accounts are unambiguously classified as fraudulent (clean) by the learner. More generally, our findings demonstrate the feasibility of generating and using synthetic data for training and testing an electoral fraud detection system. 2 See Lehoucq (2003) for an extensive review of the literature on electoral fraud. 3 Digit analysis using Benford's Law is an example of such a method. Benford's Law specifies that in a collection of numbers, the first possible digits should not occur with equal frequency. Widely applied to financial auditing (Drake and Nigrini 2000), conformity with Benford's Law has also been used to detect manipulation of economic indicators (Nye and Moul 2007), campaign

Forensic Analysis of Venezuelan Elections during the Chávez Presidency

“…A recent study based on authentic and synthetic election data reports that the non compliance of the law is associated with fraud at least in 50% of cases [18]. With the exception of the 2005 elections, we cannot provide an explanation of why the law fails for the referenda and elections held between 2004 and 2012.…”

Who monitors the monitor? Effect of party observers on electoral outcomes

“…For instance, Mebane (2008) looks for distributions in the last digits of electoral reports that deviate from an expected distribution (or the law of digits' distribution), the so-called Benford's "law". Beber and Scacco (2012) uses some variations of this law to detect corruption in Nigeria, and a related synthetic approach is used in Cantú and Saiegh (2011) to study fraud in Argentina in the 1930s. A different approach within electoral forensics looks for "odd patterns" of turnout and their relation-4 ship with incumbents' vote share (Myagkov et al (2009)).…”

Statistical detection of systematic election irregularities

“…The digit 0 would then occur much more often as the last digit in the vote counts compared with uncorrupted numbers. Voting results from Russia (6), Germany (7), Argentina (8), and Nigeria (9) have been tested for the presence of election fraud using variations of this idea of digit-based analysis. However, the validity of Benford's law as a fraud detection method is subject to controversy (10,11).…”