Fact check: What were the findings of the 2020 election audits?

1. What auditors actually reported — concise claims that matter

Post-election reviews and academic audits converged on the claim that the net error rate in counting 2020 presidential votes was tiny, on the order of thousandths of a percent, and that audits shifted the net presidential count by about 0.007%, a margin far smaller than margins of victory in contested states ^{[1] [2]}. State-level audit reports, including Michigan’s official post-election audit, reported no examples of fraud or intentional misconduct by election officials and no evidence that tabulation equipment failed to function properly ^[4]. Georgia’s statewide risk-limiting audit, conducted under law, confirmed Biden’s win there with a variation from the original count of about a tenth of a percent, reinforcing that tabulation accuracy held at scale ^[3].

2. The national dataset that underpins the headline numbers

A nation-scale audit dataset compiled for the Proceedings of the National Academy of Sciences study included 856 regional governments across 27 states, covering 71,702,471 individual votes and 1,210,528 ballots audited, forming the empirical basis for the reported thousandths-of-a-percent error rates ^[1]. The study’s authors aggregated county-level and higher audits to measure how recounts and hand tallies changed vote totals, and they reported a net shift of about 0.007% in the presidential vote after audits — a statistically and practically negligible amount relative to total votes cast ^{[2] [1]}. The dataset’s scale gives weight to the finding that errors were dispersed and minuscule rather than concentrated in a way that would flip outcomes ^[1].

3. State-level confirmations that reinforce the national picture

Individual states’ post-election procedures echoed the national finding: Michigan’s Bureau of Elections and county clerks documented no fraud, intentional misconduct, or equipment malfunction in their post-election audits of the November 3, 2020 election ^[4]. Wisconsin’s post-election voting equipment audit hand-tallied more than 145,000 ballots and found no evidence that voting equipment changed votes, mis-tabulated, or altered totals, reinforcing that machine and paper systems produced reliable counts ^[5]. Georgia’s rapid, statewide risk-limiting audit of 41,881 ballot batches across 159 counties validated its presidential outcome and showed only about a tenth of a percent variation, demonstrating that legally required audit methods can confirm results quickly and at scale ^[3].

4. How audits were conducted and what "error" means in this context

Audits compiled in the national study combined a variety of audit types — routine post-election audits, hand recounts, and risk-limiting audits — each comparing machine tabulation to independent hand counts or statistical checks to detect discrepancies ^[1]. The reported net error rate and the 0.007% shift represent the aggregate difference between original tabulated totals and audited counts across millions of ballots; this metric reflects net direction and size of discrepancies rather than isolated anomalies. Risk-limiting audits specifically are designed to provide statistical confidence about the winner without recounting every ballot, and Georgia’s 2020 implementation showed such audits can confirm outcomes within operational timelines ^[3].

5. What audits found and emphatically did not find

Across the nationwide aggregation and state reports, audits found very small, localized discrepancies but no pattern of systematic miscounts that would change presidential results; the PNAS study and state audits both emphasize that errors were minuscule and scattered ^{[2] [1]}. Auditors did not uncover evidence that tabulation equipment intentionally altered votes or that widespread fraud affected the outcome; Michigan’s official report and Wisconsin’s equipment audit explicitly reported no evidence of equipment malfunction or vote-switching ^{[4] [5]}. The Georgia risk-limiting audit’s confirmation of the state result demonstrates that properly executed audits can detect and rule out significant tabulation errors ^[3].

6. Limits, remaining questions, and how different stakeholders use these findings

The comprehensive audit dataset and state reports provide strong empirical support that 2020 presidential vote counts were accurate, yet audits vary by method, scope, and legal requirements across jurisdictions; the national study aggregates these diverse procedures to produce an overall estimate but cannot substitute for state-specific legal adjudication ^[1]. Stakeholders have used these findings in different ways: officials and many election experts point to the audits as confirming legitimacy, while some critics emphasize isolated irregularities or call for broader standardization of audit methods — arguments that the aggregated results do not substantiate as indicators of widespread fraud. The documented evidence remains that audited ballots and equipment checks overwhelmingly showed no systemic errors sufficient to alter the election ^{[2] [4]}.