November 13, 2004
More On Freeman's Exit Poll Study
Thank god someone a lot more well-known than me is saying the same thing I've been saying about these exit poll "studies".
Plus, he makes the awesome point that according to these exit poll summaries that are supposedly so predictive of election results, Dukakis beat Bush in 1988.
Remember, the lessons aren't that exit polls are useless or "wrong" - they're just not designed to be used for this purpose.
The detail level of the exit polls (that we don't have access to) is useful for projecting precinct results, and, in turn, the state results. But the summary level (that we see) is not predictive of statewide results, and is not supposed to be.