Political Behavior

, Volume 29, Issue 2, pp 175–195

Mission Accomplished: The Wartime Election of 2004


    • Stony Brook University
  • Andrew H. Sidman
    • Stony Brook University
Original Paper

DOI: 10.1007/s11109-007-9036-7

Cite this article as:
Norpoth, H. & Sidman, A.H. Polit Behav (2007) 29: 175. doi:10.1007/s11109-007-9036-7


The war in Iraq, so the widely accepted view, hurt the reelection of George W. Bush. We contend, to the contrary, that the war helped him get reelected. First, we show that his victory fits the dominant pattern of wartime elections in American history. Second, we find that Bush’s approval ratings benefited from a complex rally where the Iraq war prolonged rather than diminished the 9/11 effect; most Americans affirmed rather than disputed a link between the war in Iraq and the war on terror. Third, while Bush’s approval proves sensitive to U.S. casualties in the Iraq war, any damage to his standing prior to the election was mitigated by sufficient popular support for that war. And finally, on Election Day, Bush was able to garner the vote of two critical blocks with favorable feelings about the Iraq war, be it the decision to invade or the prospect of success.


Presidential electionsWartime electionsRally effect

In winning reelection in war time, George W. Bush emulated the success of some illustrious predecessors, including Abraham Lincoln and Franklin Roosevelt, while avoiding the failure that drove others or their party from the White House. In particular, the son achieved what eluded the father. For many opponents of the Iraq War, the victory of George W. Bush nonetheless was not only maddening but incomprehensible. It was as if the country had learned nothing from the unhappy Vietnam experience. With John Kerry, the Democrats had nominated a candidate who was able to frame the war issue in terms that evoked the Vietnam parallel: a colossal blunder. For supporters of the President, on the other hand, the more compelling parallel was Pearl Harbor 1941. The war in Iraq was part of a war on terror unleashed by the attacks of 9/11. The American public was sharply divided over this claim, just as it was divided over the looming parallel of a Vietnam-type quagmire in Iraq. Given these conflicting opinions, it should not be taken for granted that Bush won because of the Iraq war, but perhaps in spite of it. He was reelected with one of the narrowest margins for a sitting president, struggling in polls against his opponent until the very end of the campaign and coming up short against John Kerry in the televised debates.

The 2004 election raises several questions about wartime elections. One set of questions is historical. How does the outcome of the 2004 case fit into the annals of American elections? Does the White House party win more elections in war time than it loses, and does it fare better or worse than might have been expected? Another set of questions deals with the popularity of a wartime president? In which ways did the war on terror and the war in Iraq shape the approval ratings of Bush as President? Which general hypotheses about the effect of war on presidential popularity does the Bush case confirm? The third set focuses on Election Day. Was the Iraq war a critical issue in the vote choice between Bush and Kerry in 2004? Which version of issue-voting does the Iraq case exemplify? In the end, did the war in Iraq help or hurt the electoral prospect of the Commander-in-Chief? Answers to these questions help resolve some long-standing controversies about the impact of war on voting and public opinion.

The Historical Record

Since 1789, a number of American elections have been held in wartime or right after the conclusion of a war. What lessons, if any, does history teach us about the effect of war on election outcomes? The Vietnam War has etched itself in the annals of American electoral politics for ending the political career of President Johnson in 1968. Yet World War II recalls the opposite experience, as FDR and the Democrats easily won reelection in 1944. More recently, one George Bush suffered electoral defeat in the wake of war while another George Bush secured victory. Which of these outcomes is more common in American history?

To get a handle on this question, the first thing that must be done is to settle the definition of “war.” What is indisputable is that the U.S. has fought five “declared” wars, based on a congressional Declaration of War as provided by the Constitution (the War of 1812, the Mexican War of 1846–1848, the Spanish-American War of 1898, and the two World Wars). Aside from this small set of cases, common usage refers to many other U.S. military actions as “wars.” The current military operations in Iraq carry the “war” label as did those in Vietnam in the 1960s and 1970s, or in Korea in the 1950s. Beyond that, the U.S. has engaged in many “small wars” and countless military expeditions, from the battle with Barbary pirates on the shores of Tripoli in 1801–1805 to yet another deployment in Haiti recently. As once compiled by Senator Goldwater, the list of “U.S. military hostilities abroad without a declaration of war” had reached 199, roughly one every year in American history (Congressional Record 1973, http://www.history.navy.mil/library). Including all of these military engagements would mean that war is a constant, not a variable, in U.S. elections. The question then would be which, if any, theoretically salient features of military action matter for the outcome of elections. This is an important question, to be sure, but one that goes beyond the scope of this analysis. Our focus is on the effect of wars like the one the U.S. launched in Afghanistan and Iraq during the first term of President George W. Bush, which were on-going during the 2004 election and still continue as of 2007.

To set the bar high enough, our definition of “war” includes all military interventions that meet one or more of the following criteria: (1) a specific authorization by Congress; or (2) the commitment of a sizable force (over 100,000); or (3) significant casualties (over 1,000 killed in action). The list of such wars is presented in Table 1. For each of these wars, we have identified the presidential election that was held while the war was going on as well as the one right after the completion of the war. In all, the yield is 11 wars, casting a shadow over 18 elections (out of a total of 54 since 1789). That should be large enough to let us draw some lessons, yet not so large as to make war a constant in elections.
Table 1

Wartime elections in American history



White House Party



War of 1812






War of 1812






Mexican War





Civil War






Civil War






Spanish-American War/Philippine Insurrection






Philippine Insurrection






World War I





World War II






World War II






Korean War





Korean War






Vietnam War






Vietnam War





Vietnam War






Vietnam War






Gulf War






Afghanistan/Iraq War





Note: The definition of “war” includes all military interventions that meet one or more of the following criteria: (1) a specific authorization by Congress; or (2) the commitment of a sizable force (over 100,000); or (3) significant casualties (over 1,000 killed in action). Wars in italics were completed during the presidential term prior to the election for which they are listed. “Democrat” includes the historical antecedent Democratic-Republicans in the elections of 1812 and 1816

The tabulation indicates that the White House party has won most of the presidential elections in war time (12 of the 18). The Bush victory in the 2004 election thus was the rule whereas the Bush defeat in 1992 was the exception. Electoral victory is nearly as common when the war is on-going as when it is finished. There is also not much of a partisan difference, although Republicans have done slightly better than Democrats in wartime elections. So war appears to be paying electoral dividends for the party of the Commander in Chief in the presidential race, not costing the party the White House. Yet before jumping to conclusions, we have to compare this record with what happens in peacetime elections. As it turns out, the White House party has won most of those elections, too, compiling nearly the same won-lost record (winning 21 of 36). The fact is, the incumbent party wins about two of every three presidential elections, regardless of whether the country is at war or not. Having led the country into war does not appreciably improve the odds of reelection, but it does not jeopardize the prospect either, everything else being equal.

While there is no way to control for all the other factors that have a bearing on electoral outcomes going back as far as 1789, it is instructive nonetheless to do so for some predictable features of presidential elections over time. What vote gain or loss accrues to the White House party in wartime elections above and beyond its predictable vote share? The two-party vote division in American presidential elections has been shown to follow a remarkable cyclical dynamic formalized by Norpoth (1995) and Midlarsky (1984).1 In a nutshell, the party taking over the White House from the other major party in a presidential election in nearly all cases holds on to the office in the immediately following election, with the sitting president seeking reelection. At the same time, the presidential party must reckon with a reversal after holding the White House for two or more terms. The two-term limit in presidential elections, whether by tradition or law, seems to be the root of this dynamic (Norpoth, 2002).

Whatever the process may be, whenever a party has controlled the White House for two or more terms, history shows that the party’s hold on that office has slipped. Seen against this background, George W. Bush was in a favored position to win reelection in 2004; he had come to office in 2000 by ousting the White House party in 2000. In contrast, his father was not in a favored position in 1992; by that time the same party had held the White House for three terms. It is extremely rare for a party to win four terms in a row in presidential elections. So perhaps neither the father’s defeat in 1992 nor the son’s victory in 2004 had much to do with war.

Vote gains and losses of the White House party, as adjusted for the cyclical baseline, are posted for wartime elections in Fig. 1.2 To begin with the most recent case, the 2004 election does not record a gain, but a loss. Bush fell about 2 points short of his predicted share. So did his father, albeit in a losing effort, in 1992. Several wars post both gains and losses. The Vietnam War did not hurt Nixon in his bid for reelection in 1972, nor Johnson in 1964. But Vietnam did hurt the Democrats in 1968 and also the Republicans in 1976. The Korean War cost the Democrats in 1952 and boosted the Republicans in 1956. World War II helped FDR win an unprecedented fourth term in 1944, when the cyclical model predicts defeat. One of the oddest cases is the 1920 election where victory in war brought no electoral comfort to the White House. Even with the adjustment for a return to normalcy (in terms of the electoral dynamic), the Democrats took a beating. Overall, there are as many wartime elections that register, in the adjusted sense, a vote gain for the White House party as do a vote loss (8 each). Why is that? Why do some wars help the electoral prospects of the war party while others hurt them? And how can the same war have one effect in one election, and the opposite effect in another?
Fig. 1

Vote gain/loss of White House Party in wartime elections (adjusted for cyclical dynamic). Note: Black bars represent ongoing wars; gray bars represent wars completed during the presidential term prior to the election. For the Election of 1900, the Spanish American War had ended, but the Philippine Insurrection was ongoing

Theoretical Perspectives

Whether war helps or harms the popular standing of a government in office is a matter of intense debate. What is remarkable, perhaps unique, about this debate is that both sides owe a profound debt to one and the same scholar. John Mueller’s groundbreaking study, War, Presidents and Public Opinion (1973), has given theoretical meaning and operational shape to two dominating ideas: the “rally” phenomenon and the “casualty” effect. It is plainly impossible to imagine an analysis of public support for war or presidential popularity (in polls or votes) that does not pay attention to one or both of these ideas. That is not to say that Mueller’s conclusions have gone unchallenged. Yet, even attempts to point to other explanations usually start with his ideas.

To begin with the “rally” phenomenon, any government entering into a war counts on the folk wisdom that “politics stops at the water’s edge” and that “you don’t change horses in midstream.” Military action abroad is expected to generate a “rally ‘round the flag” effect at home, which unites an otherwise contentious public. Faced with an external enemy, the adversaries of domestic politics suspend their disputes and close ranks behind the country’s leader, designated by the U.S. Constitution as Commander in Chief of the armed forces. It would seem unpatriotic for citizens safely out of harm’s way not to do so. Military intervention certainly meets all three criteria of Mueller’s definition of a “rally” event: (1) it is international and (2) involves the United States and particularly the president directly; and it must be (3) specific, dramatic, and sharply focused” (1973, p. 209). Indeed, the top two categories in Mueller’s scheme of rally points are sudden American military interventions (e.g., in Korea, June 1950) and major military actions in ongoing wars (e.g., Gulf of Tonkin incident, August 1964 in Vietnam).

Numerous studies have documented the surge in presidential approval in the wake of rally events (Kernell, 1978; MacKuen, 1983; Ostrom & Simon, 1985; Brody, 1991; Nickelsburg & Norpoth, 2000). At the same time, it is in the nature of a rally that it does not last forever. Rally-induced boosts of leader approval evaporate sooner or later. Do they have enough staying power to secure victory in the next election? One certified instance where this happened was in the British election of 1983 held in the wake of the Falklands War of 1982. The rally ‘round Prime Minister Thatcher remained strong enough to assure a Conservative victory (Norpoth, 1992). How common are rallies of this kind?

Many observers are skeptical. Whatever the seductive lure of patriotism, war is a destructive enterprise. For the public back home, the costs of war register quite palpably in the human toll of soldiers killed and wounded. Seeing one’s country go to war may, for a moment, make one’s throat feel lumpy and arouse feelings of pride and patriotism. But then, as the casualties mount and sacrifices grow more onerous, war can be expected to foster doubts, dissent, opposition, and protest. Using a rational-choice approach, some see war as subjecting political leaders to a “domestic hazard that threatens the very essence of the office-holding homo politicus—the retention of political power” (Bueno de Mesquita & Siverson, 1995, p. 852). In more fragile societies, the discontent unleashed by war may be directed not only against the political leaders but against the political regime itself. War, as the saying goes, is the “midwife of revolution.”

While that is a remote prospect for the U.S., the casualties incurred through war threaten public support for it at home. This is a thesis forcefully argued by Mueller (1973), his work on the rally-effect notwithstanding. His analysis of public opinion on the wars in Korea and Vietnam shows that the cumulative count of U.S. casualties (killed, wounded, and missing) closely predicts the level of popular support for these wars over time (Table 3.4, Mueller, 1973, p. 61). The climbing toll made the American public turn against those wars and the commanders-in-chief. Both Harry Truman and Lyndon Johnson suffered unrelenting drops of their job approval while the wars in Korea and Vietnam piled up mounting casualties during their terms of office (Kernell, 1978; Ostrom & Simon, 1985).3 And both presidents ended up declining to seek their party’s re-nomination for another term after faring poorly in early primaries. Even in World War II, as some studies now suggest, American casualties may have taken a toll on presidential approval (Baum & Kernell, 2001; Kriner, 2006), although FDR’s standing remained high throughout the war and he was comfortably reelected in the wartime election of 1944. So why do casualties take such a different toll on presidential approval in some wars than in others?

One compelling alternative is the expectation of success. When the public believes a war is winnable it may be expected to support the war as well as the war party, regardless of casualties, but when the public believes the war is unwinnable it will oppose the war and turn against the president. How the perceptions of victory and defeat are shaped may depend not only on material gains and losses but also on psychological factors and media coverage (Johnson & Tierney, 2006). Victory or expectation of victory has proved to be a powerful determinant of public support for military operations (Feaver & Gelpi, 2004; Eichenberg, 2005). Contrary to widespread belief, these studies suggest the Vietnam experience has not made the American gun-shy about the use of force abroad. Even during the Vietnam War, as well as the one in Korea, there were strong signs that victory was a major concern for the general public. Gen. MacArthur’s boast, “There is no substitute for victory,” struck a popular chord in a public where demand for victory vastly outnumbered demand for withdrawal. This was true, according to the lengthy poll trend assembled by Mueller (1973, pp. 75–97) for both the Korean and the Vietnam Wars. Truman as well as Johnson paid dearly in popularity for failing to satisfy the hawkish side of public opinion. Truman’s firing of Gen. MacArthur was highly unpopular and may have doomed his political prospect more than anything else related to the Korean War. And for all the dovish anti-war protest against the Vietnam War, President Johnson’s challenger, Eugene McCarthy, got far more votes in the New Hampshire Primary in 1968 from frustrated hawks than from doves (Converse, Miller, Rusk, & Wolfe, 1969, p. 1092). Even in the general election that year, the demand for stronger action in Vietnam topped the demand for withdrawal by about a 2–1 ratio. Unhappiness with prospects for success was a major reason why the American public lost faith in the war and their commander in chief.

Yet history also teaches us that victory on the battlefield may not be enough to guarantee victory at the polls. The defeat of Saddam Hussein in the first Gulf War did not assure George H. W. Bush’s reelection, nor did victory in World War I or in the Mexican War keep the White House in the hands of the war party in the next presidential contest. How ungrateful and forgetful can the public get! Aside from rival factors (such as the economy) that are capable of blunting the electoral punch of military success, even a victorious war raises questions that may undermine popular support. Is (was) the cause of the war worthy enough? Some objectives strike the general public as more worthy than others and hence elicit more support. Public reactions to military interventions appear to be quite discriminating, based on the objective of the mission rather than knee-jerk one way or the other (Eichenberg, 2005). Rarely is the answer as simple as in the case of an undeniable attack on one’s own country by a clearly identified enemy using conventional military means (Pearl Harbor 1941). When conditions are murky, many Americans may need some guidance to figure out the pros and cons of a military intervention in places few would be able to find on a map.

On matters of foreign policy, in general, the public is more dependent on elite cues for its own opinions. When the opinion leadership of the country, including both major parties in Congress, agree and support the president on a particular military action abroad, the general public tends to accept that decision as being right and just. In such cases, the public is willing to rally to the president’s side in approval ratings (Brody, 1991). But when Congress is divided and the media give ample voice to dissent over a war, ordinary Americans may follow suit and question the worthiness of the cause. While Congress was near-unanimous in its Gulf of Tonkin Resolution in 1964 (only two dissenting votes, one more than on the declaration of war against Japan in 1941), the elite consensus collapsed a few years later, a development not seen during World War II. On the merits of taking action against Saddam Hussein, there was sharp disagreement in Congress from the very start in both 1990 and 2002, just as there was in the general public. Even independent of the elite response, some argue, it is the tenor of media coverage that determines whether or not the general public rallies to the president at moments of international crisis (Baum & Groeling, 2004).

Whatever the specific explanation for the rally phenomenon, there is wide agreement nowadays that foreign policy exerts a significant influence on presidential popularity and voting that may rival the influence of the economy, which has received more attention (e.g., MacKuen, Erikson, & Stimson, 1989, 1992, 2000; Norpoth, 1996; Nadeau et al., 1999; Marra, Ostrom, & Simon, 1990; James & Rioux, 1998). Some studies, which have examined the relationship between economic and foreign policy evaluations, have found that these forces have a roughly equal influence on overall approval (Nickelsburg & Norpoth, 2000; Cohen, 2002). Given the events of and subsequent to September 11th, we would expect this balance to tilt, if anything, in favor of foreign policy for the current administration.

Approval for the “War President”

To test propositions about war and presidential popularity, we assembled time series of presidential approval ratings from several polling organizations (Gallup, CBS News/New York Times, and NBC News/Wall Street Journal polls). In addition to overall job approval,4 we also obtained approval for two specific areas, the economy5 and foreign policy.6 For each type of approval we aggregated the polling data from the three organizations, using an algorithm developed by Stimson (1999) and covering the time period from January 1978 to December 2005.7 Was foreign policy, indeed, more influential in shaping the approval of George W. Bush than it was for previous presidents, and was the economy less of a factor? Separately obtained estimates for the presidency of George W. Bush and pre-Bush administrations are presented in Table 2.
Table 2

Foreign policy versus the economy as determinants of presidential approval


Pre-Bush April 1978 to December 2000

George W. Bush January 2001 to November 2005





Foreign Policy Approval





Economic Approval























Number of Observations





Ljung-Box Q





Test: βForeign Policy = βEconomic

t = −0.194


t = 3.255*


p < .05, ** p < .01, *** p < .001

Note: Each of the three presidential approval measures is an aggregate based on polls conducted by the Gallup organization, the CBS News/New York Times Poll, and the NBC News/Wall Street Journal Poll, January 1978 to November 2005. The estimates in the table are for the first differences of approval

For the 1978–2000 period, foreign policy and economics have close to equal effects on overall approval, which are not statistically different from each other.8 In contrast, for the Bush presidency (2001 to 2005), a one-point increase in foreign policy approval leads to an increase in overall approval that is four times larger than one associated with economic approval, a highly significant difference.9 With foreign policy having the greater weight in the case of President Bush, we now turn to the two sides of the war effect on approval—rallies and casualties—as specified in the following model10:

$$ {\hbox{Approval}}_t = \beta _0 + \frac{{\omega _{10} {\hbox{Sept}}11_t + \omega _{11} {\hbox{Sept}}11_{t - 1} }} {{1 - \delta _1 L}} + \frac{{\omega _{20} {\hbox{Iraq}}_{t - 1} }} {{1 - \delta _2 L}} + \beta _1 {\hbox{Casualties}}_t + \beta _2 ({\hbox{Casualties}}_t \cdot {\hbox{War}}\,{\hbox{Opposition}}_t ) + \beta _3 {\hbox{ICS}}_t + \varepsilon _t $$

The rally phenomenon is modeled as a dynamic intervention with immediate impact parameters (omega) and a decay parameter (delta) that captures the staying power of the rally: the higher the delta, the longer the rally. We include the two historic rally events of the Bush Presidency, the terrorist attacks of 9/11 and the invasion of Iraq, commenced in March of 2003. Given the timing of approval polling, it was decided to capture the immediate impact of the 9/11 event with two omega-parameters (September and October of 2001), and that of the Iraq invasion with one (April of 2003, 1 month after the start of the war). Turning to the casualty-hypothesis, we follow the reasoning, as outlined in the previous section, that American casualties suffered in war action affect presidential approval in conjunction with public feelings about the war. In other words, casualties will only take a toll on the president’s standing if the public opposes the war. Hence we constructed a weighted measure of casualties (Ostrom & Simon, 1985), though not using the cumulative count of casualties. Our measure weights the monthly casualty figure by the proportion that does not support the decision to go into Iraq (“war opposition”).11

While war and peace is the dominant issue of the Bush Administration, the economy without much doubt demands attention, too. To capture its influence on Bush approval, we relied on a commonly used measure, the Index of Consumer Sentiment from the Survey of Consumers (Survey Research Center, University of Michigan). We also recognize that the media play an important role in shaping opinions on foreign policy issues. These days, as has been shown, many people get their foreign policy information from entertainment-driven, soft-media sources, which in the case of the Iraq war may have lead to a highly opinionated public (Baum, 2006). Our theoretical framework does recognize, however indirectly, such a media effect by including public opinion about the war. War opposition, regardless of its creation, is seen as conditioning the effect of casualties on presidential approval. Opinion about the war in Iraq is an indispensable ingredient in our model, which attempts to relate casualties to approval of President Bush.

Figure 2 shows monthly time series for Bush approval, Iraq war opposition, and U.S. casualties. The left vertical axis marks the scale, in percentages, for presidential approval and Iraq war opposition while the right vertical axis scales the monthly casualties figure (number of U.S. military deaths). The first note to make is that months with high casualties do not necessarily correspond to large dips in approval. In fact, casualties only appear to have a relationship with approval towards the end of the series. After April 2005, casualties appear to trend upwards and approval downwards. Note that this is after opposition to the Iraq war has passed a certain threshold—a figure slightly greater than 50% disapproval. The model estimates presented in Table 3 express this relationship in statistical terms.
Fig. 2

Bush approval, opposition to Iraq war, and U.S. casualties

Table 3

Rally events, war casualties and Bush approval




September 11th Rally

    Impact in September 2001



    Impact in October 2001






Iraq war rally

    Impact in April 2003






Iraq war casualties



Casualties × war-opposition



Consumer sentiment index












Number of observations



Ljung-Box Q (12)



p < .05, ** p < .01, *** p < .001

Note: The presidential approval measure is an aggregate based on polls conducted by the Gallup organization, the CBS News/New York Times Poll, and the NBC News/Wall Street Journal Poll, January 2001 to November 2005. The estimates in the table are for the level of approval; casualties, war opposition, and the consumer sentiment index were all entered in level form, too

The results in Table 3 confirm the immense power of the September 11th effect on presidential approval.12 Our estimate gives President Bush a bump in approval of 38.5 points, which survives at the astounding rate of almost 95% a month. The decay is so slow that a significant residue of the 9/11 rally, amounting to nearly 5% approval, could still be felt at election time in 2004.13 The Iraq War rally, in contrast, registers with a relatively small impact that was short-lived (roughly 5 months). So it would appear that the Iraq war paid no discernible dividends for Bush’s standing with the American public and may have ended up hurting it if U.S. casualties prove costly.

But consider the counterfactual: what if Bush had not invaded Iraq? Would the 9/11 rally in that case have proved as long-lived as it did after Bush went to war with Iraq? To gauge the counterfactual effect of the 9/11 rally, we re-estimated the impact and staying power of the 9/11 rally for only the period up to March 2003, before the invasion. This would show how the 9/11 rally would have played out in the absence of war with Iraq. Our estimate of this counterfactual 9/11 effect produces an immediate impact nearly identical to the actual one, but a decay parameter that is considerably smaller (.904 instead of .948). Though this difference may not appear very big, over time it translates into a much faster decay for the counterfactual case. As Fig. 3 makes clear, this would have left Bush with no benefit in his approval by the time of the November 2004 election, or even the November 2002 election. So the Iraq war may have extended the life span of the 9/11 rally, not diminished it, as many believe.
Fig. 3

The 9/11 rally effect on Bush approval: with and without the Iraq war

At the same time, should one not expect the casualties of the Iraq war to depress Bush approval? The answer is not a simple yes or no. The effect of casualties, we submit, is contingent on public sentiment about the war. It is the interaction of casualties with public support that produces a significant effect of casualties on presidential approval:
$$ 0.279\,({\hbox{Casualties}}_t ) - 0.522\,({\hbox{Casualties}}_t \times {\hbox{War}}\,{\hbox{Opposition}}_t ) = (0.279 - 0.522\,{\hbox{War}}\,{\hbox{Opposition}}_t )\,{\hbox{Casualties}}_t . $$
Taken together, these estimates suggest that casualties begin to hurt once war opposition crosses a certain threshold. As can be seen in Fig. 4, the threshold turns out to be .53 (meaning 53% oppose the war).14 Above it, casualties depress presidential popularity, below it they do not.
Fig. 4

The conditional effect of casualties on Bush approval

In view of the results so far, should one claim that foreign policy was a friend or a foe to Bush in the 2004 election? Looking at November 2004, we have already demonstrated that the rally effect (September 11th and Iraq) still gave Bush an extra five points in approval. That same month, opposition to the Iraq war (against the decision to go to war) stood at 53%. Whatever the U.S. casualties that month, they would not lower Bush’s approval rating since war opposition had not crossed the threshold (also .53).15 We can conclude that foreign policy, in general, helped Bush, mustering an approval rating conducive to reelection. The Iraq war did not hurt, so it appears from the aggregate analysis of presidential approval. How does the analysis of vote choice between Bush and Kerry compare with this finding?

The War Issue and the 2004 Vote

To turn from approval of a sitting president to a choice between competing candidates on Election Day sounds like a simple act, but using an issue of public policy, whether foreign or domestic, for making a voting decision is anything but simple. For many it is an arduous chore that requires mastering a series of steps none of which can be skipped if the issue is to guide the voting decision (Campbell, Converse, Miller, & Stokes, 1960). Voting your policy preferences is not for everyone (Abramson, Aldrich, & Rohde, 2003). In one sense, war may be an easy issue to handle. It is a matter of high public concern that is bound to generate strong feelings among voters. And given the President’s role as commander-in-chief, he is closely identified with one side of the issue. Yet in another sense, voters may find war a more difficult issue to handle. Public opinion on war may be swayed by patriotism and government propaganda. Meanwhile the opposition faces a quandary: dissent from the government and risk appearing unpatriotic, support its course and provide no alternative to the voters. What is more, an on-going war asks voters not just to take a stand on the issue itself, but also to evaluate the consequences. It is one thing to support a certain policy, and another to be impressed with the results.

To begin with, policy-voting requires that voters have genuine opinions about the policy in question. How firm or fickle was American opinion on the war in Iraq? Did the public simply react to circumstance (9/11), get swept up by seeming manipulation of intelligence, and the drumbeat of war preparations? The “rally” phenomenon hints at a powerful role for presidential leadership in the formation of public opinion on matters of foreign policy. It is not farfetched to imagine a voter with a preference for Bush taking a position in favor of the war, because that is the president’s policy. On Election Day 2004, according to the NEP exit poll, just about the same majority chose George W. Bush over John Kerry as supported the U.S. decision to go to war with Iraq. Was this a case of public opinion following the leader rather than guiding the choice of leaders?

There is a rich and long history of polling the American public on its support for or against removing Saddam Hussein by force. And the pattern is unambiguous: in every poll from the end of the first Gulf War in 1991 up until the invasion of Iraq in March of 2003, a vast majority, often near 75%, favored the use of force to remove Saddam from power (Huddy, Khatib, & Capelos, 2002, pp. 448–449; Tyler & Elder, 2003). Hence popular support to take military action against Iraq was not a reaction to the 9/11 attack. Nor can it be construed in any way as a result of White House manipulation in preparation for war against Iraq after those attacks. On the contrary, the high and unwavering support for military action can be read as popular pressure on policymakers to take such action. There was no need for any distortion or fabrication of intelligence to get an otherwise reluctant public on board of the mission.

Given the long and deep roots of animosity toward Saddam Hussein in the American public, it is understandable that many came to see military action against Iraq as part of the war on terrorism. The connection between the two wars turned into a highly controversial issue during the 2004 campaign. While the Bush Administration insisted that the war in Iraq was just one battle in the broader campaign against worldwide terrorism, John Kerry disputed this connection and called the Iraq operation a “colossal blunder” that only served to undermine the real battle against terrorism. On this question, more voters in the exit poll sided with the Bush position than with Kerry’s. Without much doubt, the ability to frame issues in terms of terrorism helped Bush.

At the same time, there is no question that when Bush faced the voters in November of 2004, popular support for the war in Iraq was far lower than it had been prior to the invasion. A vast majority had shrunk to a bare one, dangerously close to threatening electoral defeat for Bush. Had the public changed its mind in view of the facts on the ground, or had its support all along not meant what it said in polls? An important clue is provided by a New York Times/CBS News poll on the eve of the invasion of Iraq that probed support for military action yet to be taken under certain conditions. Given a not implausible scenario that military action were to result in substantial U.S. casualties, the poll showed that popular support dropped about 20% from the familiar high level (Tyler & Elder, 2003). Similar declines registered for other highly plausible conditions, namely that the U.S. would be involved in a war for months or years, or that military action would result in substantial Iraqi civilian casualties. Hence about 20% of the American public fit the mold of what Feaver and Gelpi (2004) call “casualty-phobic:” they support a mission provided it is extremely low cost. Once it became clear that this was pie in the sky, actual support settled at a level very close to what the contingent measures of the Times/CBS poll had indicated.

Moving from an opinion on an issue such as the war in Iraq to a vote decision requires some sense of where the rival candidates stand. While that seems easy for the President running for reelection—after all, as Commander-in-Chief he led the country into it—the position of the challenger may be more difficult to discern. What stand did he take on entering the war? Is there any daylight that an ordinary voter can spot between President and opponent on this issue? Is the opponent offering a new policy for dealing with the war? Dissent in wartime always carries special risks, in particular, the risk of appearing unpatriotic. So the opposition party has to tread carefully, which may confuse voters about its stand on the war. John Kerry certainly did not make it easy for voters to discern his position on the Iraq war during the 2004 campaign. He was on record as having endorsed the congressional resolution authorizing the use of force against Saddam Hussein, but he claimed that this did not mean approval of the military action Bush took in early 2003. His notorious statement about first voting for the $87 billion Iraq measure before voting against it was widely exploited by the Bush campaign in efforts to depict him as a flip-flopper (“the gift that keeps on giving,” in Karl Rove’s words). In this way a policy issue turned into a personal problem for the opponent, calling into question his qualification for the Presidency. The exit poll shows that nearly half of the Democrats and eight in ten Independents who doubted Kerry’s consistency defected to George W. Bush.16

At the time of the election, the war in Iraq was not only a matter of policy, but also one of performance. Regardless of the principles involved in going to war against Saddam Hussein, there was the question of success or failure. Are we winning the war? Is the war winnable, at all? Whatever forecasts the public makes about the prospect of success derives, in part, from assessments of current conditions (Norpoth, 1996). You judge the prospect of success by how well things have been going lately, but also perhaps by some independent cues about the future (MacKuen et al., 1992). Voters on Election Day, by a small margin, were more pessimistic than optimistic about the U.S. prospects in Iraq. A majority thought that things were going badly right now for the U.S. in Iraq, and also that the war in Iraq had not improved the long-term security of the United States. These gloomy views of the future contrasted with support, albeit by a narrow margin too, for the invasion as the right thing to do. Performance clashed with policy. The resolution of this conflict would most likely determine the imprint of the Iraq war on the outcome of the 2004 election.

Our analysis controls for long-term predispositions (party identification, liberal-conservative orientation); demographics (race, sex); short-term attitudes (candidate qualities, economic conditions); and, considering the ballyhoo over moral values in this election, also religious attendance and evangelical denomination (among white voters).17 The estimates in Table 4 make clear that both Iraq war measures register highly significant and substantial effects on the vote, as do nearly all the other variables included in the model. Figure 5 depicts the probability of voting for Bush, given a voter’s opinion on the two dimensions of the Iraq war, with all the other factors held constant. There is no surprise that voters of the “right & well” type—who approved of the decision to invade Iraq and felt the situation in Iraq now was going well—were highly inclined to vote for Bush. This was also a large portion of the electorate, though no majority on Election Day. Equally unsurprising is the low Bush estimate for the “wrong & badly” type of voters—who disapproved of the decision and felt the situation in Iraq now was going badly. Roughly matching the pro-war segment of the electorate in size, the anti-war segment fell short of a majority as well. The suspense clearly lies in the behavior of the two groups with conflicting war opinions. The predicted probabilities give Bush a narrow edge in the group that believed it was the right thing to do, but sees it turning out badly. Bush also edged Kerry among those who opposed the Iraq invasion, but now see it turning out well nonetheless. So both conflicted groups give their nod to Bush, as far as predicted probabilities go.18 As long as there was something to support about the Iraq war, be it the cause or the outcome, Bush prevailed over Kerry. Had it been the other way around, Kerry most certainly would have defeated Bush, everything else being equal. In this sense, one can make the claim that the Iraq war helped rather hurt the reelection of George W. Bush.
Table 4

Determinants of presidential vote choice in 2004


Coefficient (SE)

The Iraq war issue

    U.S. decision to invade Iraq

1.75** (.14)

    Situation for the U.S. in Iraq now

1.84** (.14)

Financial situation

0.48** (.09)

Candidate Quality


1.03** (.21)


1.72** (.16)


−1.30** (.27)

Party identification

1.46** (.09)

Political ideology

0.83** (.10)


1.18** (.28)


−0.18** (.13)

Religious attendance

0.20** (.05)

White evangelical

0.55** (.16)


−14.4** (.71)

Correct prediction


Number of observations


p < .01, ** p < .001

Source: Edison/Mitofsky National Election Pool Exit Poll on Election Day, November 2, 2004. Note: Entries are logit estimates, with standard errors in parentheses. The dependent variable is the major-party presidential vote decision (Kerry vote = 0, Bush vote = 1). The independent variables for this analysis were coded as follows: U.S. decision to invade Iraq (1 = disapprove, 2 = approve); Situation for the U.S. in Iraq now going (1 = badly, 2 = well); Financial situation, compared to 4 years ago (1 = worse today, 2 = about the same, 3 = better today); Candidate quality (1 = if mentioned, 0 = if not mentioned); Party identification (1 = Democrat, 3 = Republican, 2 = Independent or something else); Political ideology (1 = liberal, 2 = moderate, 3 = conservative); Race (1 = black, 2 = non-black); Sex (1 = female, 2 = male); Religious attendance (1 = never, 2 = a few times a year, 3 = a few times a month, 4 = once a week, 5 = more than once a week); Evangelical Christian or born-again (0 = no, 1 = yes)

Fig. 5

Assessments of the Iraq war and presidential vote predictions. Note: The vote predictions come from a logit model of the voting decision that controls for party identification, ideological orientation, candidate traits (honesty, leadership, and intelligence), economic situation, sex, race, church attendance, and white evangelical denomination. See Table 4 for model estimates

Some will find this claim hard to accept. Is it not true that voters whose main issue concern was the war in Iraq chose Kerry over Bush in the 2004 election? Indeed it is, by a 3–1 margin in the exit poll. So Iraq must have been a losing issue for Bush. If war helped Bush win the 2004 election, many claim, it was not the one in Iraq but the war on terror. Indeed Bush piled up a big vote margin in the exit poll among voters whose main concern was terrorism. Did the issue of terrorism save Bush from an inevitable electoral disaster inflicted by his war in Iraq? To address this question, we have added to the controls in the vote model a measure of terrorism that asks for an evaluation of whether the country is safer from terrorism or less safe from terrorism, compared to 4 years ago. If Bush needed this issue to offset the supposedly bad effect of Iraq one would expect that the inclusion of terrorism would push the Bush probabilities into the loss region. But this is not what happens. The inclusion of terrorism in the vote model does not appreciably change the probability estimates for Iraq war support shown in Fig. 3. In particular, each of the two critical groups still favors Bush over Kerry (.55 for the “wrong & well” type, and .52 for the “right & badly” type). Bush still wins on Iraq even with terrorism factored into the vote calculus. He did not need terrorism to help avert defeat in 2004 over what some see as an unpopular war in Iraq. To the contrary, it may have been the Iraq war that kept the issue of terrorism salient in the American public. Even the most gripping issue sooner or later loses its punch without some further news. Lack of any additional terrorist attacks on the United States was bound to dim the memory of 9/11. The daily reports of carnage from Iraq after the invasion kept the topic of terrorism alive. Whatever adverse reactions the Iraq war triggered in the American public, on Election Day a majority of voters saw this war bound up with the war on terror. If there was a way to distinguish the two, for many it was a distinction without a difference.


Winning a wartime election like the one in 2004, history teaches us, is a common achievement for the White House party. But it is by no means inevitable, as the case of the 1968 election keeps reminding us. Electoral victory in war time depends on several conditions identified by students of public opinion. Key among them are the strength and longevity of the “rally” effect generated by war; the costs of war (especially casualties); the righteousness of the cause for war; the prospect of success; and the challenge from the opposition party and its leaders.

Our analysis of presidential approval confirms that George W. Bush benefited immensely from “rally” effects. The war on terror unleashed by the attacks of 9/11 sparked a surge more powerful and long-lasting than probably any other such rally in American history, including the one triggered by the attack on Pearl Harbor in 1941. The estimates from a dynamic intervention model indicate that a significant portion of the 9/11 rally survived long enough to still benefit Bush on Election Day 2004. While the invasion of Iraq registered a far less impressive rally in Bush approval, its impact appears to have extended the life of the 9/11 rally. In the absence of military action against Iraq, we estimate, the 9/11 rally would have decayed much faster. There would have been nothing left of it in the President’s approval rating by the time of November 2004. The war in Iraq kept the topic of terrorism salient, which helped Bush retain his 9/11 boost.

The casualties of the Iraq war (and subsequent occupation) hurt Bush’s approval, to be sure, but they did so only in a highly contingent way. Our findings lend support to the hypothesis that casualties affect presidential approval to the extent that the public supports or opposes the war (as being right). It is only below a certain threshold of such support that casualties lead to a drop in presidential popularity. So long as the public has enough faith in the war, a president need not worry about the fallout of casualties. By the end of 2005, support had fallen below the threshold, but not yet by the time of the 2004 election.

Our analysis of presidential choice between Bush and Kerry focused on two measures of the Iraq war issue: whether it was the right thing to go to war with Iraq, and how things were going there now for the United States. Both measures proved highly significant and about equal in strength in a vote model that controlled for key long-term and short-term forces. While one favored President Bush, the other did not. On Election Day (as well in surveys before), most Americans supported the decision to attack Iraq, but at the same time reacted negatively to the situation there now, though each time quite narrowly. The proverbial fog of war does not just cloud the perspectives on the battlefield, but also those of the general public back home. On Election Day, American voters cut through this fog of ambivalence about the war by giving the nod to the Commander-in-Chief rather than the challenger. Bush won the 2004 election because, not in spite, of the war in Iraq.


The dynamic has the form of a second-order autoregressive process, with the following parameter estimates and standard errors for elections from 1828 to 2004: \( {\hbox{VOTE}}(t){\hbox{ }} = \mathop {49.3}\limits_{(.81)} + \mathop {.48}\limits_{(.12)} \,{{\hbox{VOTE}}(t - 1)}{\hbox{ }} - \mathop {.51}\limits_{(.11)}\,{{\hbox{VOTE}}(t - 2)} .\quad N = 45\quad {\hbox{SER}} = 5.4 \)

VOTE is the Democratic percentage of the two-party vote in a presidential election. All parameter estimates are statistically significant beyond the .001 level, and the Ljung-Box Q-test indicates that no significant autocorrelation remains in the residuals.


The elections of 1812 and 1816 could not be considered for the estimation of the autoregressive vote model since popular voting was not widespread enough then.


The evidence, however, is statistically suspect. The use of a cumulative casualty count introduces a deterministic trend that is bound to produce a spurious correlation with a trend in presidential popularity. The cumulative count of any other phenomenon would do just the same.


Gallup and the NYT/CBS polls use the following wording: “Do you approve or disapprove of the way [Name] is handling his job as president?” The WSJ/NBC polls ask the following: “In general, do you approve or disapprove of the job [Name] is doing as President?


NYT/CBS polls always ask and Gallup almost always asks the following: “Do you approve or disapprove of the way [Name] is handling the economy?” In this time period (1990 through 2004), Gallup has asked two variations of this question. The first, used through March 1992, asked: “As I read off each problem, one at a time, tell me whether you approve or disapprove of the way [Name] is handling that problem... Economic conditions in this country.” From April 1992 through February 1993, Gallup asked the following: “Now please tell me whether you approve or disapprove of the way [Name] is handling some specific problems facing the country... the economy.” The WSJ/NBC polls ask: “Do you generally approve or disapprove of the job [Name] is doing in handling the economy?


The NYT/CBS polls ask the following: “Do you approve or disapprove of the way [Name] is handling foreign policy?” Gallup has used two different questions in this period. The first, used until 1992, asked: “As I read off each problem, one at a time, tell me whether you approve or disapprove of the way [Name] is handling that problem... Foreign policy.” From 1992 to 2004, Gallup asked: “Do you approve or disapprove of the way [Name] is handling foreign affairs?” The WSJ/NBC polls ask: “In general, do you approve or disapprove of the job [Name] is doing in handling our foreign policy?


The algorithm for aggregating polls drops the last observation, so our analysis covers each of the series until November of 2005.


A two-tailed t-test for the difference between the coefficients produced a test statistic of −0.194, which has a p-value of 0.846 (267 degrees of freedom).


The test of the difference between these two coefficients produces a t-statistic of 3.255, which corresponds to a p-value of 0.002 (54 degrees of freedom).


This also includes a second-order autoregressive process for the error term.:


The “war opposition” measure was constructed by an aggregation of results from three polling organizations. The CBS News/New York Times question asked: “Looking back, do you think the United States did the right thing in taking military action against Iraq, or should the U.S. have stayed out?” The Pew Center question asked: “Do you think the U.S. made the right decision or the wrong decision in using military force against Iraq?” And the Gallup polls asked: “In view of the developments since we first sent our troops to Iraq, do you think the United States made a mistake in sending troops to Iraq, or not?


All variables of the approval model have been treated in level form, not in differenced form. Diagnostic checks have confirmed this choice. What is more, model estimates for the variables in first differences do not suggest any drastic revisions of the results in level form.


The 9/11 rally, so our estimate, had a life span of 69 months, using the following equation: (Impact in 9/01 + Impact in 10/01) × (DecayX) = 0, where X equals a number of months since the final impact.


Note that, while the Bush approval series begins in January of 2001, the time series for war opposition and casualties only begin in March 2003. Setting values for these two series equal to zero for the time points prior to that might raise objections. It could be argued that this step is bound to overestimate the effect of casualties on approval by “holding” casualties at zero when in fact they do not exist. To address this objection, we tried several alternate measures of casualties. Specifically, we used (1) total War on Terror (including Iraq) casualties, which begin to be reported in October of 2001; (2) Iraq casualties with the mean casualty figure imputed for all time points prior to March 2003; and (3) Iraq casualties imputing the minimum casualty figure for those months. Regardless of casualty measurement, the results were remarkably consistent: casualties begin to have a negative effect on approval only when Iraq opposition crosses a threshold around 53%. We also estimated the model on a post-Iraq sub-sample, beginning in Mach 2003. Again we find the threshold is around 53%.


Our measure of war opinion tends to overstate the extent of opposition. It was derived by subtracting the proportion supporting the war from 1.0, and thus includes those without opinion as well.


The question in the 2004 exit poll asked, “In general, does John Kerry mostly say what he believes, or say what he thinks people want to hear?” A quarter of the Democrats and slightly over half of the Independents chose “say what he thinks people want to hear.”


The estimates for the logit model are provided in Table 4. It must be noted that the 2004 exit poll did not ask all voters the same set of questions. Most questions were put only to subsets of the sample. The large subset that was asked about the Iraq invasion and current conditions in Iraq was not given the questions about whether the Iraq war improved the long-term security of the United States, or whether the war in Iraq was part of or separate from the war on terrorism, each of which was asked in separate and smaller subsets of the exit poll.


In contrast, Gelpi, Feaver, and Riefler (2006) report a vote probability for the “wrong & well” type of voters (“Pottery Barn” in their terminology) that favors Kerry by a good margin. But in their analysis this group is also much larger than it is in the exit poll, while the anti-war segment is much smaller. What may account for the discrepancy is the possibility of movement from one group to the other since the time before the election when Gelpi et al. conducted their survey and Election Day. The Gelpi et al. vote estimate for the other conflicted group (“Noble Failure”) is consistent with ours.



A special note of thanks goes to Herbert F. Weisberg for convening the conference, “The Wartime Election of 2004,” at the Ohio State University, January 12–15, 2006, where a first draft of this paper was presented. For comments at various stages of this research, we are grateful to Jeffrey Cohen, Peter Feaver, Chris Gelpi, Holly Goerdel, Gary Jacobson, John Kessel, Matt Lebo, John Mueller, Jason Reifler, Jill Rickershauser, Jeff Segal, and Jim Stimson as well as the reviewers of this manuscript.

Copyright information

© Springer Science+Business Media, LLC 2007