GameOver22 said:
Interesting. I never really payed much attention to gallup in particular. I just saw their numbers shoot up to around +7 at one point and hover around that area, so it caught my attention. We were discussing it in one of my classes, and one suggestion was that their likely voter model might be suspect. I guess they don't make their screening process public. Personally, I don't know much about it, but the poll numbers were odd. |
Actually, i double checked that and I was wrong...
it was Pew that was like that last tine. Sorry.
In 08, Gallup just like the majority of polls over estimated Obama's poll numbers. (Gallup by +2%.)
Which is really the warning that may come to bear. Polling tends to skew democrat... it's just rarely close enough to matter.
Still, it makes me think their problem is largely in overepresenting undecideds.
Either way, if you were to base elections based on previous accuracy, You'd use a mix of Rasmussen, Ruerters and GWU, and maybe Pew though their consistency is questionable.