But while it is convenient for analysts and campaign advisers (us included) to talk about women as a group of voters, and focus on how much candidate X trails candidate Y among women, the reality is more complicated. It's more accurate to think of the gender gap as the accumulation of many smaller margins among different demographic groupings of women, the totality of which lean toward the Democratic column. Within each of these margins are different influences, experiences and socioeconomic realities that, taken together, make the common reference to the gender gap a clear oversimplification.
There are reasons analysts tend to look at it this way, even while many are aware of the complexities it masks. One such reason is that, in most polls, there is simply not enough data to drill too far down into narrower subgroups. A single poll of 500 Massachusetts voters will never give us enough Republican women, for instance, since they make up only about 6 percent of the electorate. When larger data sets are available, such as the national presidential exit polls, enough detail is available to allow analysts to dig deeper.
This year, we have enough data for a closer examination here in Massachusetts. Combining all of the 2,700 interviews back to the beginning of the WBUR tracking poll together yields enough data to break the female vote down into its constituent parts.
Specifically, we were looking for which groups of women tend to favor Democrats by larger margins, and which groups tend to favor Republicans more. In order to account for the fact that the overall margin between Republican gubernatorial candidate Charlie Baker and Democrat Martha Coakley has shifted over the course of the tracking poll, we looked at how much more Democratic or more Republican each groups skews when compared to female voters as a whole. This allows us to see how spread across the political spectrum the different subgroups of women actually are.
The results make a compelling case against the notion of a single "gender gap." Far from being a monolith, there are large differences among women voters based on party, race, geography and socioeconomic status.
- Party: As one might expect, each candidate is winning the women of their respective party. Baker is doing a whopping 74 points better with GOP women than with women overall. But because there are nearly four times as many female Democrats as female Republicans, Coakley's advantage among Democrats carries more weight. Baker also holds a 14-point advantage with unenrolled women, which comprise the largest share of the electorate, but even that lead is overcome by Coakley's Democratic margin.
- Race: Baker does better among white women than women as a whole, mirroring national political dynamics. In 2012, Mitt Romney beat Barack Obama among white women (56 percent to 42 percent) but was undone by massive margins for Obama among nonwhite women. The net result was that Obama won "women" by 11 points, which is all some analysts mention. A similar split is present here in Massachusetts, though not quite as large at the moment. Coakley's margin among nonwhite women is 25 points better than women as a whole and serves as a big part of her overall strength among women.
- Geography: Women in Boston and the inner suburbs are much more likely to be with Coakley than women elsewhere. This is deeply interwoven with Democrats' recent successes in Boston and also reflects dynamics related to race, party affiliation and socioeconomic status.
- Socioeconomic status: Women across the socioeconomic spectrum also vote differently. Baker does better among the wealthiest women, but Coakley has the edge with the other income brackets. Coakley leads among the most educated women, while those with less than an advanced degree favor Baker slightly.
One dynamic we have not explored in our own polling is the role of marriage. But all of the factors we have discussed so far are tightly wound in with the so-called "marriage gap," or the tendency of married women to vote more conservatively than unmarried women.
While we have not asked a question of marital status in our tracking polls, others have done so and have speculated about the needed performance for each candidate among married vs unmarried women. By way of example, Obama won among unmarried women by 36 points, while losing married women by 7 to Romney, according to 2012 exit polls. So the gap is large enough that it can easily be the difference between winning and losing.
While some part of these differing voting patterns may have to do with marital status, there are also demographic factors at play. Kay Hymowitz of City Journal summarized it as follows:
Analysts offer a number of theories about the marriage gap: married women are more financially stable and therefore less reliant on government assistance; they care less about reproductive issues than about their pocketbooks and security; when they marry, they adopt their husbands’ political preferences. But the obvious reason for the marriage gap is that for several decades now, married women have become likelier to be white, educated, affluent, and older—demographic groups that leaned Republican in this election.
Similarly, unmarried women voted just the way you’d expect them to, considering their age, income, education, race, and ethnicity.
So is the gender gap as important as pundits make it out to be? Does it even exist? Different subgroups of women vote differently, influenced by their own situations and values. The totality of these patterns do lean Democrat, yes. But it's at least questionable whether the sum is more important than the constituent parts. The overlaps between the demographics of each subgroup make parsing out the truly important factors difficult. But whatever is driving women's choices, the key takeaway is that women do not vote in a single bloc. Campaigns that forget that and try to appeal to women as women, instead of as voters, do so at their own peril.
Steve Koczela is the lead writer for Poll Vault and president of The MassINC Polling Group. Rich Parr is research director of The MassINC Polling Group.