“The first thing we started to do was look at houses with more than six voters in them" Engelbrecht said, because those houses were the most likely to have fraudulent registrations attached to them. "Most voting districts had 1,800 if they were Republican and 2,400 of these houses if they were Democratic . . .
