|
Not biased but on the whole a bit further off the numbers than others when looked at retroactively. E.g. Zogby (relatively good) tends to be off by 1% or less but will give a real outlier every now and then. Pew will be 2% off, in either direction, but it does fewer illogical outliers.
Always check for 'registered voters' vs 'likely voters'. Most polls model in a bit more Republicans than they probably should, and non-voting people tend to skew with family/environment- model them as 38% Kerry/Democratic (plus a little bit to reflect Bush-antagonistic swing voter households) and give Bush/Republicans his current approval rating (which bounces around the Republican-leaning 43%-45%).
About half of the rest of nonvoters (8%-10%) tell pollsters they're leaning Nader where about 3% of 'likely voters' will probably vote Nader or Third Party- that's why Nader gets reported as getting ~3% in some polls (likely voters) and 6%-7% (registered voters).
All the numbers involving likely voters I've seen project to Kerry ~52%, Bush ~45%, Nader ~3%, plus or minus about 1%. That's starting to push Republican states by margins of 5% or under in 2000 into 51/49 territory. As it stands Kerry stands to win six or so states (Arizona is a tough call) beyond Gore's and is putting a scare into Republicans in three or four more. The 45% that are voting Bush can't be dissuaded this year, though, but the very soft Religious Right passive voters they're hoping to 'reactivate' this year have to be driven off. (Not that I'm sure the three million of them that Rove proposes exist do; the Christian Right voter decline fits pretty closely to the numerical trend numbers of geriatic right wing Republican voters dying out. He may be talking about a demographic mostly in the graveyards.)
|