The Florida exit poll had a
sample size of 1,829. Nader’s support in Florida was 1.63 percent, meaning the pollsters only found approximately 30 Nader voters -- a sample too infinitesimal from which to extrapolate. (Consider that the official margin of victory was 0.009 percent. One voter in the Florida exit poll sample amounted to 0.055 percent, more than the margin.)
In fact, some argue that the national exit poll yielded too few Nader voters for the purposes of analysis. So in 2006, professors Michael C. Herron and Jeffrey B. Lewis conducted a granular
“ballot-level” analysis of 3 million Florida ballots, because “ballot images directly reveal voting behavior in its most raw form, unmitigated by hindsight, social desirability, or other intervening affects.”
By looking at the partisan nature of the down-ballot choices made by Nader voters, the two scholars estimated that the Gore-Bush breakdown would have been about 60-40. That’s a slightly smaller ratio than found in the national exit poll, but nonetheless a clear lean toward Gore. Herron and Lewis note this means Nader voters were not all left-wing, yet they still conclude, “Nader spoiled Gore’s presidency only because the 2000 presidential race in Florida was unusually tight.”