In a recent paper, John Lott Jr. claims to find evidence of anti-Trump fraud in the absentee counting procedure in Fulton County, Georgia, and Allegheny County, Pennsylvania. Using Lott's own data, we show that his claims are utterly baseless. Lott uses an unusual estimation strategy that suffers from a subtle but fundamental flaw: his conclusions about fraud in Fulton and Allegheny counties are entirely dependent on the completely arbitrary order in which pairs of precincts in other counties are entered in the dataset. When we rerun Lott's analysis using an alternative but equally justifiable coding rule, the evidence for anti-Trump fraud in these two counties entirely disappears. When we replace Lott's unusual specification with a more standard estimation strategy, we find absolutely no evidence of fraud. In short, Lott's (2020) analysis provides no evidence of anything distinctive or suspicious about the absentee ballot results in either Fulton County or Allegheny County.

by Justin Grimmer, Haritz Garro and Andrew C. Eggers

overlay image