Taking A Different Look At The Gender Gap

by Stuart Rothenberg April 10, 2012 · 11:58 AM EDT

I have found over the years that when a narrative works its way into the collective wisdom, there is no way of changing it. So my goal here is quite modest: to get at least a handful of people to pause, take a deep breath and simply chew over the data a bit before using it to draw unshakable conclusions.

In this case, the conclusions involve two different Gallup polls and an interpretation about the role of women in President Barack Obama’s standing in a general election ballot test against likely GOP presidential nominee Mitt Romney.

USA Today Washington bureau chief Susan Page, with whom I sometimes team up on PBS’ “NewsHour,” noted in a recent piece about the March 20-26 Gallup/USA Today Swing States Poll that Obama leads Romney in 10 crucial states by 9 points, a dramatic reversal from Romney’s 2-point lead a month ago. (Gallup’s Frank Newport also writes about the data in his blog, “Polling Matters.”)

“The biggest change came among women under 50,” Page wrote with her usual accuracy. “In mid-February, just under half of those voters supported Obama. Now more than six in 10 do while Romney’s support among them has dropped by 14 points, to 30 percent.”

I have been struggling with Gallup’s numbers on women, as well as with the conclusions drawn from them, because the firm’s national polling and Swing States Poll don’t fit together snugly. In fact, some of the data appear contradictory.

First, let’s start with the obvious. There is a gender gap and has been one for years.

According to exit polls, women were 7 points more likely to favor Obama in 2008 than were men (56 percent to 49 percent), 7 points more likely to vote for John Kerry in 2004 than were men (51 percent to 44 percent) and 12 points more likely to support Al Gore in 2000 than were men (54 percent to 42 percent).

Men and women often preferred different candidates, but even when they favored the same candidate, they did so by varying degrees.

One of Gallup’s blogs, “The Queue,” incorporates a couple of interesting tables in its April 3 posting (“More Data on the Women’s Vote”), including one that presents the data from five national polls and five Swing States polls from October 2011 to March 2012. One table includes data on women (in general and in two broad age categories).

As Page noted, the Swing States Poll showed that women ages 18 to 49 surged during the past month for Obama. He drew 49 percent of them in late February but 61 percent in late March. Romney’s support among those same women plunged, going from 44 percent to 30 percent.

But if you look at the national Gallup poll numbers for women ages 18 to 49, in surveys conducted at virtually the same time as the February and March Swing States polls, you find that Obama’s numbers slipped — yes, slipped — by 4 points, from 59 percent in February to 55 percent in March. And Romney’s support among that same demographic group dropped by only a single point, from 37 percent to 36 percent.

Of course, it is the Swing States Poll that got all of the attention, and much of the commentary about it has concluded that the recent political discussion of birth control and funding for Planned Parenthood alienated women, who have fled Romney for Obama on the ballot test.

Intuitively, that seems right to me, but that doesn’t stop me from asking obvious questions. Why don’t the national polls show the same trend as the Swing States polls, particularly among female registered voters ages 18 to 49, as they are all conducted by Gallup? And if something really big happened during March, why is Obama’s support among women in Gallup’s late March national poll identical to his support from them in the firm’s January survey?

I can’t explain the different results, and I’m not trying to. But I do think the contradiction is worth noting and considering.

Anyway, all of the attention on the changing views of women in this age group during the past month might miss a much larger, more important development that has gone largely ignored. Both of Gallup’s national polls and Swing States surveys for USA Today confirm that during the past six months, the more dramatic change in presidential preference has been among men, not women.

In October, Gallup’s national poll found that Obama led Romney by 14 points among women (54 percent to 40 percent). That margin shrunk only slightly, to 12 points, in March. But among men, Romney’s 16-point advantage in October shrunk to just 3 points in March.

In Gallup/USA Today Swing States polling, among women, Obama drew 51 percent in October and 54 percent in March, a gain of 3 points. Romney, in contrast, lost 6 points during the same period, dropping from 42 percent to just 36 percent.

Clearly, Romney can’t win the White House if he is winning only 40 percent of female voters nationally or 36 percent of female voters from the 10 swing states. But it’s equally true that Romney can’t defeat Obama if the Republican carries men by only 3 points (as he does in Gallup’s most recent national poll) or by a single point (as he does in the most recent Swing States survey).

Why have we heard so much about female voters and little or nothing about men? I’d guess that it is because the narrative has been set (about the Republican “war on women”), so journalists look for data and anecdotes that fit into it. Certainly, some of it has to do with the reach of USA Today.

We will see what other surveys show and whether opinions — both men’s and women’s — change over time. I expect they will. Whatever happens, it is best to be cautious when interpreting polling data. Sometimes, things aren’t as obvious as they seem.