this kind of post is beginning to make me frustrated. Not for the poster's lack of effort, but from the limitations of two-way cross tabulations.
Everyone is analysing tabulations of voting intention by one other variable, leading to such apparent paradoxes as Bush picking up votes from high income people, while losing votes amongst the better educated.
someone out there must have actual data from exit polls! why aren't they telling us about what affects voting after adjusting for all these confounding variables?
Posted by robe0419 at November 10, 2004 05:19 PM | TrackBackThis is why I really wish agencies would make their raw data available. Or at least, something closer to it than what they do release. We could have all been spared some let-down last Tuesday if the exit poll raw data had been there, rather than a number that may or may not have been adjusted for projected attendance at each precinct.
One thing I know could be done better - if CNN has county-wide numbers for each candidate, then someone must have more local numbers. In the Democratic primaries, I know Al Sharpton got a delegate from one of Delaware's districts, but they wouldn't even tell us if Delaware has 10 districts each sending one, or two each sending five, or what, and none of the three counties had more than a tiny percentage voting for Sharpton.
Posted by: Kenny Easwaran at November 10, 2004 06:32 PM