There’s plenty of talk about the Republican CNN debate in Nevada, asking who won and who lost. I’ve enjoyed using the polling to make those determinations, because the debates have tended to swing the numbers. But, to know where the swing is, we have to know where we started.
The new Associated Press/GfK/Roper poll helps tell us just that.
It’s a telephone poll, mobile and landline handling. 431 Republican adults. MoE around 5.
The key result: I’d love to have more favorability polling among Republicans, but AP didn’t do that. They only got favorability readings for the entire sample of adults. They did pare down to Republicans for the horse race poll, which Mitt Romney won at 30 to Herman Cain’s 26, and Rick Perry’s 13. This is not a surprising result at all, as it matches well with other results we’ve seen lately. Perry’s clinging on, Romney’s just barely held off Cain, and Cain’s still well in the thick of it.
So now we wait eagerly for the next batch of polling, to take into account a debate whose consequences are being debated hotly across the Internet. For once, I’m seeing no punditocracy consensus on winners and losers, which I find to be an exciting development.
House of Representatives Swingometer
Electoral College Swingometer
I’m no statistician, but 431 people pared down sounds like an awfully small sample size. Am I wrong to question the methodology on this? My gut tells me that the margin of error must be a ton higher in reality than 5, just by virtue of the assumptions that went into determining the sample in the first place. But my gut has been wrong before…
431 mathematically is a lot more people than you think.
Question the methodology if you like, but the sample size I believe is reasonable.