Using Numerical Anomalies to Test for Fraud in Colonial New South Wales Elections

Abstract

When fabricating numbers, humans tend to make systematic errors, favouring some numerals over others. Analysing twenty-first-century elections in which fraud has been alleged, researchers have observed anomalous patterns in the last digits of vote counts. I show that vote counts for the 2022 Australian election exhibit no anomalies, and then turn attention to nineteenth-century colonial elections in New South Wales. Pooling data for elections from 1843 to 1887, I find a less uniform pattern in final digits. Even so, formal statistical tests cannot reject the hypothesis that vote counts in these elections were unaffected by fraud.