The meme du jour is that women are flocking to Obama, based on a new USA Today/Gallup poll in the swing states showing Romney’s support among women there has dropped precipitously.
Polls at this point are almost ludicrously unimportant, except to show two things: (1) shifting trends of the moment; and (2) how the media and pollsters operate.
I can’t find any link to the actual stats for the poll; just the article on it (if you can provide a link I’d appreciate it). But reading between the lines I note a couple of things. The first is that there’s no breakdown of results swing state by swing state, so we don’t know if the changes mentioned are across the board in all the swing states or a skewed by just a couple. Nor do we know if some states were over- or uner-represented, and if so which ones.
The second is this, and it’s a whopper:
While women typically are more likely to identify themselves as Democrats than men are, that difference widens to a chasm in the USA TODAY poll. By 41%-24%, women call themselves Democrats; men by 27%-25% say they’re Republicans.
So here we have a poll in which, among women, self-professed Democrats outnumber Republicans by 17 points (we can imagine that the other 35% of the women are Independents). In that same poll, Obama is outdoing Romney by 14 points. Hmmm. Sounds like Romney is pulling in some of the Democratic women and at least half of the Independents, and Obama isn’t getting any of the Republican women.
And how does this breakdown of women by party affiliation compare to that of previous USAToday/Gallup polls on the same subject? Darned if I can find the answer, and I’ve tried (maybe you can).
Oh, and one other thing: the article does mention that the number of people questioned in the poll was 933 registered voters, and the number of swing states involved was 12. That works out to approximately 78 people sampled per state, and my guess is that about half of them would be men. So this big female shift is based on a poll that sampled an average of 39 women per state???

