Presentation Title

A Statistical Analysis of Alleged Fraud in the Election of 2016

Faculty Mentor

T.L. Brink

Start Date

17-11-2018 8:45 AM

End Date

17-11-2018 9:00 AM

Location

C151

Session

Oral 1

Type of Presentation

Oral Talk

Subject Area

behavioral_social_sciences

Abstract

This research examines the possibility of fraud in the last U.S. presidential election (2016) through the application of Benford’s Law and validates the results using Kolmogorov-Smirnov statistical testing. Benford’s Law is an accounting concept used to detect fraud in a variety of different applicable data sets, including accounting ledgers and other financial uses, based on the frequency of the leading digits of the numbers in the sample. Any data set that follows the Benford’s Law curve is evidence of the improbability of fraud for large data sets. The purpose of the present study was to apply this heuristic to election results. By determining the frequency of the leading digits in our sample of individual vote counts by county (N=6,226) and comparing them with the expected frequency, as defined by Benford’s Law, we were able to determine that they follow the Benford’s Law curve. After performing the Kolmogorov-Smirnov test and the chi-square test for the whole sample as well as an additional Kolmogorov-Smirnov test for the most contested states, we found that the observed frequencies did not differ significantly from those expected by Benford’s Law. Therefore, there does not appear to be evidence of fraud in the election of 2016.

Summary of research results to be presented

This study uses statistical analysis to determine whether fraud took place in the election of 2016. We used the frequency distribution of the leading digits in our data entries for comparison to Benford’s Law. We then used both the Kolmogorov- Smirnoff and chi-square tests to validate our findings. The results of both tests provide substantial evidence that the allegations of fraud in the election of 2016 may be invalid, especially when compared to the 2012 election, in which fraud (or allegations of such) were not present.

This document is currently not available here.

Share

COinS
 
Nov 17th, 8:45 AM Nov 17th, 9:00 AM

A Statistical Analysis of Alleged Fraud in the Election of 2016

C151

This research examines the possibility of fraud in the last U.S. presidential election (2016) through the application of Benford’s Law and validates the results using Kolmogorov-Smirnov statistical testing. Benford’s Law is an accounting concept used to detect fraud in a variety of different applicable data sets, including accounting ledgers and other financial uses, based on the frequency of the leading digits of the numbers in the sample. Any data set that follows the Benford’s Law curve is evidence of the improbability of fraud for large data sets. The purpose of the present study was to apply this heuristic to election results. By determining the frequency of the leading digits in our sample of individual vote counts by county (N=6,226) and comparing them with the expected frequency, as defined by Benford’s Law, we were able to determine that they follow the Benford’s Law curve. After performing the Kolmogorov-Smirnov test and the chi-square test for the whole sample as well as an additional Kolmogorov-Smirnov test for the most contested states, we found that the observed frequencies did not differ significantly from those expected by Benford’s Law. Therefore, there does not appear to be evidence of fraud in the election of 2016.