The 2004 election was one of the closest and most bitterly fought in recent years. Though George W. Bush won by the smallest margin for an incumbent since Woodrow Wilson, he nonetheless went from a 543,000-vote deficit in 2000 to a winning margin of just over 3,000,000 votes in 2004. What accounts for this small but significant shift in electoral support for Bush? In this article, I argue that contrary to much speculation at the time, social and moral issues played only a small role and that foreign-policy issues were central to the outcome of the election largely because of significant partisan polarization on the issue of Iraq. This partisan polarization, though extremely high compared to previous wars, is not, however, the result of strikingly different attitudes between Democrats and Republicans toward foreign policy, but rather the result of polarized attitudes toward George W. Bush.
Soon after the election, many commentators claimed that Bush's victory resulted from a massive increase in turnout by culturally conservative voters and especially white evangelical Christians. In particular, many speculated that the presence of anti-gay marriage referendums in eleven states spurred religious conservatives to the polls.
Much of this speculation was based upon the results of the National Election Pool (NEP) exit poll showing that 22 percent of voters cited "moral values" as the most important issue in the election, more than the economy and jobs (20 percent), terrorism (19 percent), or Iraq (15 percent). Others pointed out that the percentage of evangelical Christians in the electorate jumped from 14 percent to 23 percent. But both of these findings were extremely problematic. For example, as experts quickly noted, the category of "moral values" is extremely vague and might have artificially inflated the number of people choosing it over a more specific issue. (1) In addition, the presumed jump in the evangelical vote was due largely to a change in question wording. In 2000, the exit poll asked whether respondents considered themselves "part of the conservative Christian political movement, also known as the religious right." In 2004, however, the NEP asked, "Would you describe yourself as a born again or evangelical Christian?" The latter question is clearly much more inclusive and less politically charged, and thus the likely reason for the larger percentage of voters who chose it in 2004.
The actual evidence of a surge in support for Bush by evangelicals or around socially conservative issues was rather thin. For example, Bush did not, as first speculated, run any better in states with anti-gay marriage referenda on the ballot. Political scientist Alan Abramowitz found that Bush's vote increased by 2.6 percentage points in states with gay marriage ballot measures, lower than the 2.9 percent in states without such measures. (2) In addition, Bush's vote among churchgoers changed very little from 2000 to 2004. Among those who attended church weekly or more, Bush's support went from 59 to 61 percent, but this slight increase was offset by a drop in these voters from 42 to 41 percent of the electorate.
The 2004 National Election Study (NES) survey provides a fuller picture of the 2004 vote. The NES does not include a specific question about evangelical status, but it has for many years asked several questions about the nature of respondents' religious beliefs, including how important religion is in their life and whether or not they believe that the Bible is the word of God and should be taken literally (probably a close proxy for evangelical status). To measure the impact that these groups had on the Bush vote in 2004 versus 2000, I calculated the Republican performance for whites in both groups. Performance is defined as a group's percent of the electorate multiplied by its level of support for a candidate. For example, if a particular group made up 50 percent of the electorate and gave a candidate 60 percent of its vote, that group's performance would be 30 percent (.50 x .60 = .30). A group's performance can therefore rise or fall between elections depending on two factors: its turnout relative to the rest of the electorate and how strongly it supports a candidate. In this case, if religious voters were crucial to Bush's victory, we should expect to see a strong increase in their Republican performance because of a large increase in their turnout, a large increase in the percentage voting Republican, or both.
As Figure 1 shows, there is little evidence that religious whites surged to the polls in 2004. Whites who said that religion was important in their lives and whites who interpreted the Bible literally saw a small 2-point increase in their Republican performance between 2000 and 2004. Because Bush's overall vote increased by 3 points over 2000, these religious voters were no more (and perhaps a bit less) supportive of Bush in 2004 over 2000 compared to the rest of the electorate.
[FIGURE 1 OMITTED]
The significance of these increases in Bush's performance among religious voters depends upon how they compare to changes in performance on other issues, particularly foreign policy, the other big issue in the campaign. Unfortunately, the NES did not ask many questions about foreign or military policy in the 2000 survey, so direct comparisons are tougher. But two relevant questions did appear on both surveys--first, whether or not respondents believed the United States should increase or decrease defense spending and, second, whether or not they agreed that it was better if the United States just stayed home and did not concern itself with problems in other parts of the world. Figure 1 also shows the Republican performance of those who favored increasing defense spending and those who believed that the United States should not just stay home. Among those who believed in increasing defense spending, a combination of increased turnout and support for George W. Bush boosted the Republican performance of this segment of the electorate by 5 points, and among those who believed that the United States should not just stay home, Republican performance jumped by 8 points. These findings suggest that foreign-policy concerns were much more crucial to Bush's reelection than were religious ones.
Stronger evidence of the importance of foreign-policy issues comes from the NES question about voters' most important issue. Unlike the NEP, which asked voters to choose the most important issue from one of seven categories, the NES asked the open-ended question, "What do you think has been the most important issue facing the United States over the last four years?" According to the NES, fully 43 percent of voters cited terrorism as the most important issue. Coming in a distant second was the war in Iraq with 17 percent, followed by the economy at 15 percent. In contrast, the cluster of moral and social issues (abortion, family values, gays and gay marriage, etc.) made up less than 3 percent of responses.
As Table 1 shows, rarely has one issue so dominated an election. In fact, the last time that an issue was cited by this many voters was in 1968, when 44 percent said that the Vietnam War was the most important issue. The next highest was inflation in 1980 (34 percent) and the budget/deficit in 1984 (34 percent). In fact, in any election when there was a shooting war going on (1964, 1968, 1972, and 2004), it rose to the top of voters' minds, suggesting the power of such events to dominate the political agenda. (3)
Not only did the terrorism issue dominate the election, but President Bush dominated on the issue of terrorism. Among those who cited terrorism as the most important issue, 70 percent voted for Bush. That is the highest percentage supporting a candidate on the most important issue in any election since the NES first asked the question in 1960. The next closest was 68 percent for Jimmy Carter on the issue of unemployment in 1976 and 66 percent for Richard Nixon on Vietnam in 1972. But the overall advantage received by those candidates was lower than Bush's in 2004 because the percentage of people citing those issues was much lower (only 30 percent citing unemployment in 1976 and 27 percent citing Vietnam in 1972) than the 43 percent citing terrorism in 2004. Thus, Bush's performance on terrorism in 2004 was .30 (.43 x .70) compared to .20 (.30 x .68) for Carter in 1976 and .17 (.26 x .66) for Nixon in 1972. As a result, Bush's advantage on the most important issue in 2004 was huge compared to any election since 1996. (Advantage is defined as the performance for the Republican party minus the performance of the Democratic party. Negative numbers indicate an advantage for the Democrats.) In 2004, Bush had an advantage of 18 points over...