Here’s the statistical problem with the Census Bureau’s 24,000 noncitizen voter claim:
When the thing you’re measuring is extremely rare, even tiny matching errors can pollute the well.
Bad data work can manufacture a scandal out of noise. More on Substack: