Can an Accounting Tool Detect Election Fraud?

By now, you might have read on Facebook or Twitter that a tool used to detect accounting fraud has supposedly exposed cheating in the 2020 presidential election.

The way the story goes is that in some counties, the distribution of Joe Biden’s vote totals across precincts fails an accepted test for catching fraud.

The tool in question is Benford’s Law, which observes that in many naturally occurring data sets, more numbers start with one than any other figure, followed by two, then three and so on.

The proportions are remarkably consistent.

If a series of numbers conforms to Benford, when you strip off the first digits and count them, 30.1% will be the number one; 17.6% will be the number two; and 12.5% will be the number three. Each successive digit represents a progressively smaller proportion, with nines coming in last, at 4.6%.

*** This article has been archived for your research. The original version from The Wall Street Journal can be found here ***