Based on the exit polls, race and ethnicity are a better predictor of candidate preference than gender.
Of all the big winners on Election Day, one of the biggest may have been a concept: the gender gap in American politics.
From President Barack Obama's 11-point edge with women over Mitt Romney in exit polls to Republicans losing two senate seats over troubling statements about rape, 2012 seemed to further the idea that gender is the leading definer of Democratic voters: double x marks the spot.
But lumping more than 50 percent of the population into a group and talking about it as a single unit can oversimplify things a lot. Go deeper into the 2012 exit poll numbers to look at the women's vote and picture begins to change.
To be clear, the gender gap in America is not a myth—the numbers show it's real—but it's also very complicated. It can grow or shrink depending on a host of factors: race, age, marital status, even geography.
Let's start with one of the biggest story lines of 2012: that the gender gap was an epic problem for Romney and the GOP in general. It started when the Republican primaries featured conversations about the morality of birth control, which ended up inspiring a tidal wave of women to push Obama over the top.
When everything was tallied from 2012 this turned out not to be true. This year's 11-point margin was big, but it wasn't bigger than it was four years ago. According to the exit polls, the margin was actually bigger in 2008—13 percentage points. Obama won women 56 percent to 43 percent in 2008. He won them 55 percent to 44 percent on Tuesday.
More important in the gender gap discussion is how the significance of the gender gap is relative to the population you are looking at as a whole.
For instance, in the 2012 exits Obama did better with white women than with white men—seven percentage points better—but that's smaller than the 11-point divide in the electorate as a whole. And overall Obama still lost white women. The president captured only 42 percent of the white women's vote, Romney captured 56 percent of it. So, white women actually were an area of strength for Romney.
Obama gets a huge edge when you shift to look at black women or Hispanic women. He won 96 percent and 76 percent of their votes respectively. That's even better than Obama did with black men and Hispanic men.
The point? Obama, and Democrats in general, consistently do better with women than men when you compare voters within a racial or ethnic group. That's the "gap." But simply grouping all women together and thinking they are reliably Democratic is wrong. If you were to list gender and racial/ethnic groups by their Democratic vote on Tuesday, the list would go like this: black women (96 percent), black men (87 percent), Hispanic women (76 percent), Hispanic men (65 percent), white women (42 percent) and white men (35 percent).
In other words, if you want to place a bet on how someone will vote and you have to choose between knowing that person's gender or their race/ethnicity, you're better off learning their race or ethnicity. That marker is more telling.
The "marriage gap" was also more important than gender in determining votes in 2012, according to the exit polls. Obama won voters who were not married by 20 percentage points. There was a gender gap here—Obama won 67 percent of unmarried women and 56 percent of unmarried men—but he won single women and men.
Romney won married voters of both genders. He won 53 percent of married women.
And when you add all the various elements of the women's vote together—race, ethnicity, marital status, age, etc.—in the context of the presidential you get some very interesting gender gap differences on the Electoral College map.
Obama won a remarkable 68 percent of the women's vote in New York State, where the electorate in general is younger and more diverse. But he won only 50 percent of the women's vote in Arizona, where there are a lot of Hispanics, but also a lot of older white voters. In Indiana, where the electorate is also older and less diverse, Obama actually lost the women's vote—he won only 48 percent of it.
As director of the Patchwork Nation project, which uses demographic and consumer data to break U.S. counties into 12 types of communities, I've looked at this issue before. When you add in the various cultural, educational and economic factors we study, the women's vote moves a lot.
This fall, we looked at women in the 2012 Wall Street Journal/NBC polling data and found that in small-town rural locations (places we call Service Worker Centers) and exurban locales (places we call Boom Towns), Obama had only small leads with women—about 2 percentage points. And on Election Day, he lost those counties in Patchwork Nation. He had a huge lead in collegiate communities (places we call Campus and Careers) and he won those places handily.
Maybe most important, that same polling data showed Obama was leading with women in wealthy suburban counties (places we call the Monied Burbs) by 10 percentage points while he was even with men in them. Those places hold large chunks of population in key states like Ohio, Colorado and Virginia and they were crucial for Obama on Tuesday. Even though they are overwhelmingly white, he still won them by seven points.
And when we get done digging though the 2012 exit poll numbers, that may wind up being the real significance of the gender gap in this election.
The gender gap is not an easy thing to label. There are big differences in what the women's vote looks like depending on what other data points you use to slice and dice it. But it seems when you add all those other little data points together it equals an edge for the Democrats not only in the national polls but in key places on Election Day.