Overview
- A University of Michigan political scientist produced a working paper that estimated the number of potentially fraudulent votes in Pennsylvania exceeded President Trump’s margin of victory, while noting those signals could come from other causes.
- A small Nevada nonprofit, the Election Truth Alliance, used the paper to push supporters to demand hand counts and later cited the analysis in a lawsuit seeking physical ballot reviews in Pennsylvania counties.
- Statisticians and election scholars told reporters they doubt the eforensics model can reliably prove fraud in U.S. elections because its signals can also arise from reporting quirks, turnout patterns, or data errors.
- The analyst relied on third-party data that he did not independently gather or fully verify, and he previously removed a 2020 South Korea working paper after acknowledging errors, which critics say shows the risks of using unvetted working papers in public advocacy.
- The episode underscores how technical, non–peer-reviewed analyses can be amplified by advocacy groups into legal and political pressure and highlights the need for public access to original data and peer review before analysts’ findings are used to challenge results.