I posted this a couple days ago and got no response from Marv.
This study applies Benford’s law to detect anomalies in county-level vote data for the 2020 US presidential election. Most prominent distribution violations are
papers.ssrn.com
3. ResultsTables 1 and 2 below present the estimation results for the whole sample and across subsamples based on current election outcome (Democratic and Republican wins) and state category (blue,red, and swing states). In the baseline results for the first-digit test, there are multiple anomalies seemingly detected. For example, all vote counts significantly deviate from Benford’s law in the whole sample and in states won by Democrats. However, only a few are consistent across all estimations and robustness checks.
Statistically significant irregularities are observed in all five tests for Democratic vote counts in states won by the Democratic candidate, for Republican vote patterns in safely blue states (such as Massachusetts or California),
and, perhaps most importantly, for Democratic votes in key battleground states. Notably, no anomalies are detected for any of the tests in states won by the Republican candidate, and only one out of 15 estimations is significant in safely red states.
4. Conclusion
This study has applied Benford’s law to detect irregularities in county-level voting data for the
2020 US Presidential election.
The distribution of digits for voting counts is consistently
anomalous across all estimations and robustness checks for Democrats in states their candidate
has won, including key swing states, and for Republicans in safely blue states.
The observed
voting patterns are significantly different from first-digit and two-digit Benford’s law, the
empirical distribution observed in past elections, and Monte Carlo simulated distributions.
The contribution of this piece of research is two-fold. First, it evidences the applicability
of Benford’s law to election analysis and develops a set of novel robustness tests that can be
applied in future studies on the topic. Second,
it provides some consistent evidence for
anomalous behaviour of vote counts in the 2020 US Presidential election, most importantly in
key swing states, which potentially warrants further research and academic investigation.