Originally Posted by Requiem
ABOUT THE PEW POLL:
In a stunning reversal from a month ago, in Pew Research's poll
of likely voters, Mitt Romney now leads President Obama by 4 points. In September, Obama led by 9 points. Other polls have shown that Romney got a bump from the first debate, but most of them showed Obama still holding a narrow lead. The poll was conducted Oct. 4-7, entirely after the debate.
However, a close look
at the internals of the poll turns up something odd. In the October sample, 31% of the respondents self identified as Democrats (vs. 39% in September). Similar, there were 36% Republicans in October (vs. 29% in September).
While many people believe Romney "won" the debate, it is extremely unlikely that 21% of the nation's Democrats changed parties as a result of one debate. So there is a fair chance that the Pew poll is an outlier that undersampled Democrats and oversampled Republicans.
For comparison purposes, two other national polls hav also been done entirely since the first debate. Rasmussen has it as a tie at 48% to 48% and Gallup has it at 47% to 47%.
And FWIW, Rasmussen has Obama up in IA and CO.
Fair enough. I can't trust Rasmussen though, not after that amazing **** up they made in 2010 with the 2010 Colorado governors race.
I bet the best poll to believe for Colorado would be the Denver Post poll if/when they do one. I don't even know if they've done one yet but I believe them to be most accurate when dealing with Colorado.
When it comes to polling, the worst mistake that partisans make is to believe and push the poll that has their candidate winning. All it does is set yourself up for a letdown. Polling has to be tough. Trying to figure out what kind of sample size to use has to be really tough, because you have to predict the turnout.
I started this thread because Sullivan's reaction to the poll he chooses to believe gave me a chuckle. And I wanted to share it with you fine folks.