|
Edited on Wed Apr-23-08 12:48 AM by FlyingSquirrel
But Zogby wasn't as good as you're thinking.
Yes, their final poll got it right - but just a two days before that they had Clinton up by only 3. If you average all four of their polls from March 22-April 22, they are second to last behind PPP.
Their average was a 5.25% win by Clinton.
Of pollsters that did multiple polls in PA:
PPP - Obama by 1.25% Zogby - Clinton by 5.25% Rasmussen - Clinton by 6.17% Quinnipiac - Clinton by 7% Strategic Vision - Clinton by 7.25% InsiderAdvantage - Clinton by 7.5% ARG - Clinton by 11.25% SurveyUSA - Clinton by 12.5%
Looks like ARG came closest, followed by InsiderAdvantage and SurveyUSA. Honorable mention to Strategic Vision and Quinnipiac.
Zogby perhaps got lucky in its final poll but that ain't worth much to me. None of the others predicted more than a 7-point lead for Clinton in their final poll, with the exception of ARG which gave her 13 points.
If you averaged all the pollsters above except PPP over the past month, you'd have been predicting an 8.1-point margin.
If you only averaged the final two polls from each of the pollsters above except PPP, you'd come up with an 8.6% margin.
If you averaged the final poll from EVERY pollster except PPP, you'd get a 7.3% margin.
If you averaged every poll taken within the past two weeks by ANY pollster except PPP, 7.7%
-----
So here's the deal. Either NONE of the pollsters were all that accurate, (even ARG gave her a 20-point margin on its second-to-last poll, just after predicting a 45-45 tie the week before), or something happened in the last few days because that extra 2-3% must have come from somewhere. Either undecided voters swung to Clinton in a BIG way, or........
:shrug:
Looks like the best method might be to average the final two polls from each pollster that does multiple polls in the state. With the exception of the obvious outlier. The exit polls did show undecideds going to Clinton by about 60%.
|