Exit polls

[dougo]

CNN.com has full exit poll results. (Yes, the exit polls during the day were grossly inaccurate, but they became accurate by the end.) There's plenty to chew on here, but I'm not going to right now, except to point out that the "moral values" thing is really overblown—it seems clear that concerns about terrorism had a much bigger effect on the election. I think there's really just a huge number of single-issue voters like Dennis Miller and Ron Silver: socially and economically liberal, but "killing the bad guys" trumps everything else.

(Thanks to chrismwage and wonkette for the links.)

Comments

[ketzl.livejournal.com]

As temvald noted, the exit polls themselves were still showing a Kerry victory, but after actual results were coming in they averaged the results with the polling numbers broadcast so of course they tended to converge. I have heard additional speculation that in states with Diebold voting and no paper trails, the gap between actual vote & exit polls was significantly higher than in other states just on the Presidential vote, but that exit polling was much more in agreement with vote results for other elections. I find myself speculating how easy it would be to insert a special case in these things which wouldn't be caught by testing procedures, something like "if date=11/2/2004 and candidate.lastname='bush' and testmode=false and state in ('OH','NV', etc) and random(1)>0.95 then". But of course that's just depression talking. If it hadn't been close, such potential cheating couldn't be concealed.