Richard C. Eichenberg: Working Papers

Gender, National Security, and the Election of 2004

Richard C. Eichenberg, International Relations, Foreign Policy, Western Europe, Political Behavior
Department of Political Science - Tufts University
September 30, 2004

Update: October 13, 2004 (PDF)
Update2: October 26, 2004 (PDF)
Update3: October 31, 2004 (PDF)

If anything should have been true about the election of 2004, it was that the continuing violence in Iraq would confer an advantage to the democratic candidate among female voters, a group that has been crucial to the party's voter base in recent elections. As is well-known, in the 2000 election Al Gore outpolled George Bush among women by 54 to 43 percent, a pattern that matched the disproportionate success among women that the democrats had enjoyed in the 1992 and 1996 (and earlier) elections.

As we will see below, John Kerry maintained a similar lead of over 10 points among women as late as July 25th, but recent survey results suggest that the margin has all but disappeared. In a spate of late September polls, Kerry is essentially tied with Bush among women voters, and he may even be slightly behind (CBS/NYT; IPSOS-AP; ABC/WashPo; Pew Center; LATimes).

This is surprising, for if there is one consistent finding of scholarly studies, it is that women are disproportionately sensitive to and critical of the human cost of armed conflict (see the additional sources listed at the end of this essay). Writing in late spring 2003, I myself generalized from this evidence by predicting that "Should establishment of stability [in Iraq] prove a lengthy and inconclusive process, and especially if US forces continue to suffer casualties, we would expect public support for the mission in Iraq to decline, and based on the evidence...gender differences would be no small part of the loss of consensus." (p. 141 of my study listed at the end of this essay).

True, candidate Bush still enjoys a much larger margin among men than his 50 percent standing among women; as some have pointed out, a "gender gap" still exists. Yet it remains true that no democratic candidate is likely to win a presidential election unless he or she can find the female votes to offset what appears to be a permanent advantage of republicans among men.

At the moment, Mr. Kerry is not doing that. In the remainder of this short essay, I raise 3 questions:

  • When did Kerry lose his advantage among women?
  • Have women's views of national security issues changed over the same period?
  • Why?

As we will see, the answers to the first two questions are fairly clear, but the answer to the third remains open to debate and speculation.

When Did John Kerry Lose His Lead Among Women?
We can answer this question because the Washington Post has made it possible to tabulate detailed demographic breakdowns of its recent poll results (find it here). In the following analyses, voter preference for Bush and Kerry is based on samples of likely voters in the ABC/Post polls (usually about 750 of the total sample), while other questions on policy issues are based on the full national sample of the adult population. The voter preference question is a three-way "horse race" that also offers Ralph Nader as a choice.

As we can see in the graphic on the left, on July 25, Kerry led Bush among women by about the same margin that Gore had enjoyed in the 2000 election, but he has been slipping ever since. It is not clear from the graphic if any single event drove the change (Bush actually dropped back among women after the republican convention, but Kerry did not gain). What seems most significant is that Kerry appears to have been dropping among women throughout August and September, while the gains for Bush are most noticeable during September. Why this might be the case is something I discuss below.

Interestingly, the same pattern is not true of voting preferences among men. As shown in the chart immediately below, Kerry's support among men has fluctuated above and below 40 percent since mid-July. Bush's support among men has fluctuated a bit more widely, but like Kerry's, his support among men on September 26 was not dramatically different from what it had been on July 11 or July 25.

What this tells us, of course, is that the lead that Bush has opened over Kerry during the last month especially is due entirely to gains among women voters.

Women and National Security

I mentioned above that I would not try to explain these changes in voter preferences until the third section of this essay. But recent popular discussion requires that we at least take an intermediate step, for a great deal of recent speculation - at least in the media - focus on the role of national security issues as the cause of shifting female preferences. There are two variants. The first, best expressed by Scott Lehigh in the Boston Globe, is that the Bush campaign has successfully equated the war in Iraq with the war against terror as retribution for the attacks of September 11, 2001. The second variant is that, quite apart from Iraq, the Bush campaign has succeeded in making the "war against terror" the central preoccupation of voters, and women in particular are sensitive to the argument that security against terrorism is the central challenge that the country faces (two prominent press discussions are in the NYT and the WashPo).

Both arguments might suggest that women's assessments of the Iraq war would change, either because it is equated with combating terrorism, or because the increased salience of terrorism might simply divert attention from Iraq.

The polling data from the Washington Post do indeed suggest that something has happened to women's views of the situation in Iraq. The graphic at left, for example, shows the net level of approval of the President's handling of the situation in Iraq (approval minus disapproval). In April, when things were very bad indeed in Iraq (135 American deaths), there was a yawning 40 point gender gap on the war: a net 13 percent of men approved of the President's handling of Iraq, versus a net -27 percent of women. By September 26, that gap had been reduced to 16 percent: net 6 percent of men approving, and -10 women approving.

What is more, the most significant movement in the numbers among women occurred between July 25 and September 26, that is, after the intense republican campaigning on Iraq (and terrorism) during August that continued in their convention and after. Women's views of Iraq improved by about 10 points during this period, while men's views moved hardly at all.

Finally, if we compare these views on Iraq to the first graphic shown above (voter preference among women only), we see that it was precisely during this period that Bush gained among women in the horse race and Kerry declined. Between July 25 and September 26, Bush gained 8 points among women; Kerry lost 6 points.

And what of terrorism? Did the Bush campaign convince women voters in late summer that terrorism -and not the war in Iraq - was the most important issue facing the country? The graphic at right suggests that it certainly increased the salience of terrorism among women. The graphic shows answers to the WashPo question that asks "What will be the SINGLE most important issue in your vote for president this year: The U.S. campaign against terrorism, the war in Iraq, the economy and jobs, education, health care, or something else?". On July 25, just before the democratic convention, the Iraq war was a very important issue for women: it overshadowed "the US campaign against terrorism" by 7 percentage points, but perhaps even more important, it was also more important than education and healthcare (combined in the graphic), issues of demonstrated concern to women. By August 29, not much had changed; if anything, women were slightly more concerned about healthcare. The choice of "the economy and jobs" (not shown) was unsurprisingly high for both men and women; for men, it registered 24, 34, and 31 percent for the three time points shown, and for women 27, 31, and 28.

The most noticeable change came in the numbers after August 29. By September 26, after the republican convention and post-convention campaign that dwelt intensively on terrorism, Iraq, and the connection between the two, women placed the war against terrorism near the top of their list of concerns (after the economy), an increase of 7 percent (the increase between August and September is particularly significant given that the question offers 6 response alternatives, thus offering a number of ways to distribute an answer). The importance of the war in Iraq also declined noticeably. Curiously, this was not the case among men (not shown). Among men, the percent listing terrorism as the most important electoral issue actually declined by 3 points from August 29.

In any case, the net result of all these changes for the views of women is this: on July 25, the economy was the top-ranked issue among women, the war in Iraq ranked second, and the war against terrorism ranked a distant fourth. By September 26, the economy still ranked first among women, but terrorism was now a close second, and the Iraq war had dropped to fourth.

We can summarize these questions on Iraq and important electoral issues as follows. First, women's disapproval of the President's handling of Iraq declined significantly during August and September. Second, during September, the war against terrorism increased noticeably in importance to the vote choice of women in particular. Third, the importance of the war in Iraq declined in importance among women.


I have established a correlation: women's vote preferences have shifted along with their evaluation of the situation in Iraq and the importance of terrorism as an election issue in ways that men's views did not.

But the question remains: why did this happen?

Two plausible explanations come to mind.

  1. The Security Moms. One explanation, popular with the press, is that women are disproportionately fearful of terrorism and have therefore shifted their preferences to the President because of his demonstrated leadership since September 11th and his success in employing the imagery of that event. The crux of this argument is that women are more receptive to this message than men (see NYT and WashPo for two recent examples).

    The problem with this argument is that the evidence to support it is mixed. The NYT article linked immediately above does include claims from some pollsters that their results show a disproportionate fear of terrorism among women. But in other surveys, the views of men and women on the threat of terrorism and how to deal with it are in fact quite similar. It is true that women see terrorism as a threat. In a recent survey by the German Marshall Fund (June 04), 94% of women said that "the threat of a terrorist attack with weapons of mass destruction" was an important threat to the United States. But the figure for men is exactly the same 94%. In the same survey, 93 percent of men and 90 percent of women approved of using military force to prevent a terrorist attack. Asked about the American attacks against the Taliban regime after 9/11, 71 percent of women were in favor -but so were 80 percent of men (the average of 19 surveys reported in my article listed in the resources at the end of this essay).

    In general, both men and women are supportive of forceful actions to deal with terrorist acts against the United States, and that is precisely the point in attempting to explain the recent movement of opinion in the Presidential race. If both genders have similar views on terrorism and how to deal with it, why is it that the vote preferences of women -in particular- have shifted? In addition, as concerns the war in Iraq specifically, there is a complication: of those who expressed a clear opinion, 73 percent of women said that the war had "increased the threat of terrorism in the world", compared to 57 percent of men (in the Marshall Fund Survey linked above). Even on the issue of terrorism, the President had an "Iraq problem" among women.

    In summary, the problem with the "security Mom" explanation is that it works equally well for "security Dads". Simply raising the fear of terrorism in general is likely to provoke similar reactions from men and women. If we are to explain why women, highly critical of the Iraq war, nonetheless switched their vote preference to Bush, we have to base that explanation on an argument that differentiates the views of men and women.
  2. Diversion and Redefinition. In my article on gender differences listed at the end of this essay, I discuss several factors that (disproportionately) lower women's support for military operations when they are mentioned in survey questions: military or civilian casualties; involvement in civil wars; and the combination of casualties and a failed military mission (such as in Somalia). It is also worth noting that women are more supportive of social spending programs than men, and slightly less supportive of defense spending. Simply listing these factors helps us to understand why women's disapproval for the war in Iraq was so high by the summer of 2004: casualties in Iraq continued in what had essentially become a civil war, the prospects for success looked low indeed, and the financial cost was very high.

    I would argue that by placing terrorism at the center of its rhetoric, the Bush campaign reversed the trend of these sentiments in two ways: First, it diverted attention from Iraq and returned it to terrorism. Perhaps more importantly, the rhetoric insistently -and persistently- redefined the situation in Iraq from an issue that is unpopular with women (involvement in an increasingly intractable civil war) to one that is very popular (fighting terrorists). The concatenation of arguments about terrorism, Iraq and September 11, especially at the republican convention and after, presumably had an even more pointed effect (reinforced by the anniversary of September 11th). The administration may also have received an assist from the media, which some argue reduced its coverage of Iraq during the summer (here is one such argument). We do know that the public's attention to the news from Iraq declined during the summer, although there may have been less news to attend to.

    Of course, there was nothing Bush could do about the casualties that had been suffered (other than to claim increasingly that they were lost so the fight against terror would not be brought to American shores). Nonetheless by redefining the war in Iraq as part of the post-September 11 struggle against terrorism, Bush shifted the focus from a situation that is disproportionately unpopular with women to one on which the views of men and women are similar (see above).

    Perhaps the most convincing testimony to the plausibility of the second explanation is that it reads like a playbook for John Kerry's renewed criticism of the war in Iraq. Beginning with the speech at NYU on September 20, Kerry has criticized both the human and financial costs of the war and charged that the administration's claims of success are a "fantasy". Of course, these arguments could influence the views of both men and women, but as I have noted, they are of particular concern to women. In addition, the Kerry rhetoric has had the predictable effect of increasing news coverage of Iraq and the problems there (the public's attention to news about Iraq increased in September to a level not seen since April)

    If the "divert and redefine" explanation for Bush's improving poll numbers is correct, we would expect Kerry's rhetoric to have the effect of winning back some female voters, especially if amplified before a huge audience in the presidential debates. However, the problem for Kerry -in the debates and beyond - is that it is not entirely clear that his own policies on Iraq would bring fewer casualties, lower costs, or improved chances for success.

Additional Resources

For helpful comments on this essay, I am grateful to Jeffrey Berry, Kent Portney, Virginia Sapiro, and Deborah Schildkraut.