I found Brad Friedman’s BradBlog after the 2004 election. He’s got a good piece up about
“
Yes, once again, according to several analyses, the disparities between Exit Polling and Election Results, as reported by unverified computer tabulators, suggest that one or the other of those two sets of data may be wrong. The first Exit Polls released by the media consortium on Election Night suggested that Hillary Clinton won in key battleground states, such as Wisconsin, Pennsylvania, Florida, Michigan and North Carolina, that she ended up reportedly losing.
Figuring out what did or didn't happen, of course, remains very important. Had just about 50,000 votes across the three states of WI, PA and MI alone been recorded for Hillary Clinton instead of Donald Trump, she would be the President-Elect today.”
bradblog.com/...