Tuesday, October 23, 2012

Parties and the Gender Gap

The following is a guest post from Christina Wolbrecht of the University of Notre Dame. Christina is the author of The Politics of Women's Rights: Parties, Positions, and Change (Princeton, 2000). She can be followed on Twitter at @C_Wolbrecht.

It’s that time of (presidential election) year when political observers once again turn their attention to the gender gap. Nate Silver at FiveThirtyEight recently noted that the “consensus of surveys” indicates that the gender gap—in this case, the difference between the percent women favoring a presidential candidate and the percent men doing so—is on a path to hit “historic highs” in 2012. Silver attributes the persistent gender gap to the issue of abortion and other social issues. In an election in which equal pay, contraception, and rape have been front and center in political discourse, it is easy to expect such “women’s issues” to drive women’s vote choice and contribute to a gender gap.

Yet political scientists who have looked closely at the gender gap in presidential elections over many decades have reached a different sort of consensus: That the gender gap has been a persistent feature of presidential elections for far longer than is widely recognized, that men, not women, bear much of the responsibility for the gender gap, and that economics and social welfare, not women’s issues like abortion, are the dominant cause.

Most observers focus on the apparent emergence of the gender gap in 1980. It would be more accurate, however, to say that the gender gap was discovered, rather than emerged, during the contest between Carter and Reagan. As Kathy Bonk has shown (in this book), the identification of a pro-Democratic gender gap in 1980 was in part the work of feminist activists seeking to defend the political power of women in anticipation of the defeat of the Equal Rights Amendment, the signature feminist cause of the 1970s. Under Reagan’s leadership, the Republican party had taken a number of distinctive stances on women’s issues in 1980; Silver rightly notes that the GOP first took a clear pro-life position in 1980, but far more attention-grabbing had been the removal of the ERA from the GOP platform after more than 40 years of official Republican endorsement. (Democrats not only maintained support of the ERA in 1980, but in an unusual move, pledged to only provide financial support to candidates who likewise endorsed the amendment). Given that context, it’s not surprising that the 1980 gender gap attracted considerable attention, and that women’s issues—the ERA, abortion, and so on—were given the lion’s share of blame, or credit, for the apparent emergence of partisan gender differences. Silver makes a similar argument, pointing to the 1973 Roe decision and the recurring debate over abortion in presidential elections since 1980 as responsible for the persistent, and he notes, growing gender gap in presidential elections.

Attributing the gender gap to abortion or so-called women’s issues in general has a certain compelling logic to it. Yet, as political scientists almost immediately pointed out, and have documented since, there are several fundamental problems with this line of reasoning. First, the gender gap did not emerge in 1980 but has been a recurrent, if not consistent, feature of presidential elections throughout the postwar era. As any time series analyst will tell you, the conclusions we draw depend a great deal on the data start and end points. The exit polls which Silver references are only available from 1972 onward, but the American National Election Studies survey begins in 1952. As those data show (see Figure 1, below), women favored Republicans more than men did in 1952, 1956, and 1960, but were more likely than men to vote for Democratic presidential candidates in virtually every election since 1964. From a broader scope of data, 1980 does stand out relative to the immediately preceding election, in which two fairly moderate major party candidates faced off and the gender gap disappeared. 1976 was the exception, however, and not the rule.

So the gender gap is a persistent feature of American presidential elections, and has favored the Democrats since 1964. Why? Repeated research has failed to uncover evidence that women’s issues, including abortion, cause the gender gap. Long story short, there are few consistent gender differences in attitudes on such issues, and limited evidence that women prioritize women’s issues such as abortion in their voting calculus to a greater extent than men do. If and when such issues do influence vote choice, we have reason to believe they may work in either a liberal or conservative direction. Finally, as I show in my book, the parties only diverged on women’s issues in the late 1970s, and as we’ve seen, the pro-Democratic gender gap emerged earlier than that. 

So if it’s not abortion and women’s issues, what causes the gender gap? One word: Men. Most explanations for the gender gap focus on women, implicitly assuming that men are the norm and any divergence from male behavior is an oddity to be explained. But a closer look at the data suggest that most of the relevant movement, at least initially, was on the male side, a point made originally by Wirls, and later supported by (among others) Norrander and Kaufmann and Petrocik. As Figure 2 shows, when we disaggregate male and female party identification (including leaners), the pattern is clear. Men and women were virtually equal in their propensity to identify with the Democratic party in 1964. In the 20 some years that followed, both men and women defected from the Democrats, but men at a far greater rate than did women. The movement of men away from the Democratic party opened up the gender gap in party identification and as a result, presidential vote choice.

Why did (some) men abandon the Democratic party? Why did (some) women stay? The answers to those questions are complex, and require careful attention to race, region, and religion, among other things. At the risk of over-simplification, however, a big part of the answer appears to be divergence in preferences for social welfare policies, such as aid to the poor and elderly, health care, and so on. Unlike attitudes on women’s issues, polls reveal consistent gender differences in preferences for social welfare, with women more likely to express liberal positions. Men’s and women’s different relationships to the social welfare state, employment patterns, and economic vulnerability are among the reasons given for these preference differences, as well as for the gender gap itself. (Some sources here, here, and here, and Erie and Rein in here.)

And what happened in the 1960s that might induce the emergence of a partisan gender gap based on social welfare attitudes? The War on Poverty, Great Society, civil rights, and other events sharpened and heightened party differences with regards to the size, scope, and purpose of the federal government. The Reagan Revolution further accentuated the parties’ divergent philosophies on the social welfare state. It’s not hard to argue that such differences continue to distinguish the parties and their candidates in the presidential election today.

While I have described the conventional wisdom, it’s also fair to say that important questions remain unanswered, and conditions often shift faster than we can publish about them. New issues hold the potential to upend the status quo; as Nate Silver notes, men and women differ consistently and substantially in their support for gay rights, and there is evidence that such social issues are increasingly important to women’s and men’s party ID. Recent gender gaps tend to be larger among those who are unmarried and those with college and especially post-graduate degrees. The distribution of those characteristics in the American population is changing, especially among women. The recent consistency in men’s and women’s party identification (in general and relative to each other) despite such demographic change might even suggest a puzzle. A full accounting of the gender gap in current presidential elections remains to be made, and the future is unclear, but so far the evidence suggests that the most obvious answers may not always be the right ones.


  1. Great summary, Christina! Will be sending to my students.

  2. This is a nice piece. Perhaps the other articles reference this, but there is no discussion here of the changing racial and ethnic makeup of the electorate over this time period--first the increase in the size of the black electorate and now the increasing Hispanic one. In addition, since the early '60s the government has spent more and more money on social programs, education, and health. All three professions provide large numbers of jobs to educated women, who are likely to vote, and likely to vote for the "government" party. More speculative is the possibility that the decrease in family size makes mothers less likely to vote for the "war" party. It's "easier" to risk losing one of four than one of one.

  3. Just sent this along to my American Government students! It is also discussed on Andrew Sullivan's blog!

  4. Hi to all,I am new person to the blog.I want to share about auto glass services through your blog.Keep update more things in the blog.
    Audi Auto Glass

  5. Great concepts on this page. This is really a nice post.I really like the way you start and include your thoughts. Thank you so much for this information. keep posting such good stuff.
    north providence auto glass