I read one of the recent papers. It wasn't written for clarity. It needed to clearly define what would be coded as fraud, and didn't. It wouldn't have made it 5 minutes into a job market seminar where I went to school. BTW I live in PA and I'm an elected local official; fraud here nearly impossible.

Ballots vary considerably from township to township over very small geographies. Ballot blanks are numerically controlled. We use paper ballots. We match signatures in voter rolls at the polling place. Fraud at scale would be incredibly difficult.

Mebane's approach strikes me as a more sophisticated version of Navarro's claim that, essentially, an election is an aggregation of flips of a coin with a bias, so elections look rigged to him because he expects the Trump 2020 election to have used the Trump 2016 coin. But votes aren't coin flips.

To be fair, I think the other papers presume familiarity with the paper below. I really want to see him backtest on US 1980 and US 1992. websites.umich.edu/~wmebane/mea...