An Empirical Comparison of Parametric and Nonparametric Methods Applied to the Measurement of Election Fraud

50 Pages Posted: 2 May 2022

See all articles by Kirill Kalinin

Kirill Kalinin

Stanford University - The Hoover Institution on War, Revolution and Peace

Date Written: April 1, 2022

Abstract

Two election forensics methods provide estimates of the magnitude of election fraud: Bayesian version of the finite mixture model developed by Walter Mebane and his students, and nonparametric approach based on vote-turnout histograms proposed by Sergey Shpilkin. While Mebane's approach offers rigorous statistical modeling based on the original Klimek model, Shpilkin's approach offers an intuitive and visually appealing algorithm. Using the data from Russian federal elections (2000-2021) this paper aims to compare both methods by conducting election forensics analysis and four validation studies from different elections. In addition, this paper comes up with two alternative methods based on univariate finite mixture models, as well as an algorithm designed to extract precinct-level estimates of election fraud from nonparametric method. The paper's main conclusion is that, despite the major differences in the approaches and identified biases, they can be relatively effective in measuring electoral fraud at both the precinct and aggregate levels.

Keywords: eforensics, nonparametric method, finite mixture model, Russian elections

Suggested Citation

Kalinin, Kirill, An Empirical Comparison of Parametric and Nonparametric Methods Applied to the Measurement of Election Fraud (April 1, 2022). Available at SSRN: https://ssrn.com/abstract=4073770 or http://dx.doi.org/10.2139/ssrn.4073770

Kirill Kalinin (Contact Author)

Stanford University - The Hoover Institution on War, Revolution and Peace ( email )

Stanford, CA 94305-6010
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
16
Abstract Views
136
PlumX Metrics