The Women's Vote Revisited

That's why they call them swing voters.

When I wrote about the women's vote in dead-tree TIME a few weeks back, in the days after the GOP convention, our polling was showing that there was indeed a decided "Palin effect." John McCain and Barack Obama were running dead-even among women voters overall, and among older, non-college-educated white women--who are classic swing voters, and a key demographic that both parties are watching closely this year--McCain enjoyed a commanding 18-point advantage.

But no longer. In the latest TIME poll, which is part of our "How America Decides" series (it is featured in the upcoming dead-tree issue, though I don't think there's a link on TIME.com yet), Obama has opened up a 19-point lead (56-37%) over McCain among likely female voters. He even leads narrowly (48-45%) among white women--a group that George Bush won by 11 points in 2004. Among married women, whom Bush won 57-42% over John Kerry, Obama is ahead 51-42%. Our pollster Mark Schulman notes that no Democratic presidential candidate in recent history has had numbers that strong with married women and white women.

Our poll, by the way, was conducted October 3-6, among a random sample of 1,053 likely voters, and has a margin of error of +/- 3 percentage points. *

UPDATE: Commenter Depressed by Tribal Warfare points out something that should have occurred to me when I posted that description of the poll.

The margin of error for the poll overall (men and women, with a sample size 0f 1,053 likely voters age 18 and older throughout America) is +/-3. Because these subsets (women, white women, married women) have smaller sample sizes, the MOE for them is greater, though I don't have that figure precisely. Also, commenter Trifecta asks for the topline result of the entire poll. In the overall poll, Obama leads McCain 50-44%, which is pretty much unchanged from the previous week.

And while we're at it: The sample's partisan distribution was 37% Democrats, 29% Republicans, 27% independents. Likely voters are identified based on their registration status, interest in the campaign, self-reported likelihood to vote, and previous voting record, unless they are new registrants. Those who have already "early voted" are included in the likely voter pool.