re: #243 Timothy Watson
Is white men a larger percentage of the people polled versus polls?
And exit polling does tend to not be entirely representation since it’s hard to poll a large number of precincts.
What’s working, too, is the working assumptions of the models pollsters use. Recall this article: Same raw data given to four pollsters give four different results.
Some of this is really bad methodology (only 10 non-white? No people under 40?); some of it is very questionable modeling (non-white voting falls back to 2004, despite demographic changes?). Like I said before, I’m beginning to think there are enough really bad polling being done that even the aggregators are being affected.