How Many Casualties Will Americans Tolerate?
Misdiagnosis
CHRISTOPHER GELPI
In "The Iraq Syndrome" (November/December 2005), John Mueller argues that public support for the American wars in Korea, Vietnam, and Iraq can be explained with "a simple association: as casualties mount, support decreases." He goes on to say that support for the Iraq war has dropped so fast that it makes sense to talk about an "Iraq syndrome," a casualty-induced aversion to the future use of force by the United States.
Mueller's landmark contributions to the study of public opinion and war have rightly earned him much respect and inspired a large portion of the scholarly agenda taken up by me and my research partner, Peter Feaver. In this essay, however, he is only partially correct. The public is, as he notes, sensitive to casualties: casualties are the cost of war, and the public would prefer the same outcome (victory) at lower cost (fewer casualties). But when he steps from there to casualty-phobia -- a sensitivity to casualties so acute that it paralyzes policymakers -- Mueller goes far beyond his limited data and must ignore extensive evidence that runs counter to his theory. Even the limited data Mueller cites disprove his own central claim: that support drops steadily and inexorably with mounting casualties regardless of context. Along the way, Mueller grossly misstates the findings from my co-authored research and largely overlooks research by Richard Eichenberg, Bruce Jentleson, Steven Kull, Eric Larson, and others that adds more nuance to this picture.
MODERN SENSITIVITIES
Casualty sensitivity may be thought of as price sensitivity to the human cost of war. As with other forms of price sensitivity, some members of the public are more sensitive to casualties than others, and one person's sensitivity to casualties may vary over time.
In fact, the public's willingness to bear the human cost of war has varied substantially during different phases of the wars in Korea, Vietnam, and Iraq. The key variable is the perceived likelihood that the mission will succeed. Variations in perceptions of success over time have substantially altered the extent to which casualties have eroded aggregate public support for each war. Varying perceptions of success have also played a major role in determining which individuals have decided to oppose the war in Iraq, specifically, and when. It is this breadth of evidence that led me and my co-author to characterize the American public as "defeat-phobic" rather than "casualty-phobic."
Mueller cursorily dismisses our research by citing out of context a survey statistic we report in a 1999 op-ed: the mean response to a question about the number of acceptable casualties in a hypothetical U.S. military intervention in the Democratic Republic of the Congo. Mueller clings to a confused reading of that statistic, despite our extensive discussion in a subsequent book, Choosing Your Battles. We invite readers to read pages 95-148 to decide for themselves whether Mueller has accurately reported our argument. For the record, our interpretation of the (limited) survey results we discuss in that book is that a majority of the American public claims it would tolerate something on the order of a hundred casualties to bring democracy to Congo. Our main point, however, is that the focus of policymakers and pundits on casualties, without reference to context, is misguided, because public support for a military operation will erode sharply in the face of mounting casualties when the public believes the war is failing but will remain relatively robust when the public believes the war is succeeding. This argument is based on extensive analyses of scores of surveys -- including much of the survey data on which Mueller relies.
Mueller's claim that public support for the Korean War dropped as casualties mounted, for example, does not withstand scrutiny. The assertion of such a relationship is based entirely on a single observation: the drop in public support between August and December 1950. Mueller attributes this drop to the 10,000 U.S. battle deaths during this period, but he fails to note that the United States was also suffering some of its most lopsided battlefield defeats at the time, as Chinese forces swept down the peninsula. Furthermore, he fails to note that public support for the war increased by eight percentage points between December 1950 and August 1951 despite the fact that nearly 8,000 American soldiers were killed during that period. Why did public support rally despite the mounting casualties? Because U.S. forces managed to recapture South Korean territory and stabilize the front at the 38th parallel. Public support only began to wane again when casualties mounted -- though not precipitously -- after the war settled into a stalemate.
Similarly, Mueller's depiction of the drop in support for the Vietnam War gives the impression of a gradual decline in response to U.S. casualties. What he fails to note in his chart is the numbers of casualties that generated drops in support. The rate of U.S. casualties began to decline after President Richard Nixon instituted his policy of "Vietnamization" in 1969, but public support for the war declined more quickly than it had from 1965 through 1967, despite the fewer casualties. The critical turning point appears to have been the Tet offensive, in February 1968: the media's negative portrayal of Tet seems to have solidified Vietnam as a failure in the public mind, and our analyses indicate that the negative impact of casualties on support for the war tripled in Tet's wake. In fact, prior to Tet, U.S. casualties had a statistically significant impact only because of the ten-percentage-point drop in public support from March to May of 1966. Between May 1966 and December 1967, more than 12,000 U.S. soldiers died, but public support for the war dropped by only three percent. After Tet, on the other hand, the public lost hope and casualties led to a steady decline in support.
Finally, with regard to Iraq, Mueller once again overstates the evidence of a consistent relationship between casualties and public support. As his charts illustrate, public opinion data on whether the wars in Korea, Vietnam, and Iraq were a "mistake" were available only on an intermittent basis, making reliable statistical analyses difficult. Presidential-approval data, on the other hand, are available on a regular basis, and the wording of questions in presidential-approval surveys remains consistent, both of which allow for more reliable analyses. Considering President George W. Bush's approval rating over the course of operations in Iraq is instructive. During the "major combat" phase of the war, presidential approval did not decline in the face of casualties suffered as the U.S. quickly swept Saddam Hussein from power. One might dismiss this finding as a brief "rally round the flag" effect, because from June 2003 to June 2004 -- during what we label the "insurgency" phase of the war -- presidential approval appears to have declined in response to casualties (with one brief reprieve in response to the capture of Saddam). But in June 2004, the Bush administration restored sovereignty to Iraq and began preparations for elections, bolstering public optimism about success, or at least progress, in Iraq. Consequently, during the "post-sovereignty" period, from June to November 2004, U.S. casualties had no impact on presidential approval. For example, from July to November 2003, there were 200 U.S. deaths in Iraq and the president's approval rating dropped by more than eight percentage points. On the other hand, from July to November 2004, there were 300 deaths, but Bush's approval rating remained unchanged. More detailed analyses indicate that these changes were statistically significant even when accounting for other factors that influence presidential approval.
In 2005, according to the data, U.S. casualties continued to have no impact on approval ratings until March or April. Public optimism about progress in Iraq appears to have carried the Bush administration through the costly assault on Fallujah, and the president received a noticeable bounce in approval in the wake of the successful Iraqi elections at the end of January. Since that time, however, the public has had few benchmarks by which it can measure progress in Iraq. Thus, since March, we have seen the human toll begin to erode presidential approval once again.
WEIGHING THE COSTS
Public support for U.S. military operations, then, does not inexorably decrease like sand flowing through an hourglass. Instead, the American public regularly makes judgments about the potential costs and benefits of a military operation. As the likelihood of obtaining any benefits diminishes, the human cost of war becomes less tolerable, and casualties reduce support for the operation. On the other hand, if and when the public is optimistic about a successful outcome, it is far more willing to bear the human cost of war.
I would note, however, that Mueller makes at least one important observation: the same kind of drop in support that resulted from thousands of deaths in Vietnam has resulted from just hundreds of deaths in Iraq. Mueller posits a superficially plausible hypothesis for this difference -- that the public views Iraq as less important than Vietnam -- but he does not cite survey data to support that claim. It seems highly unlikely that the stakes in Vietnam were considered to be ten times as great as those in Iraq. Instead, my co-author and I put forth a more promising hypothesis: that in the years since Vietnam, the advance of technology has changed the level of "necessary" casualties that the public believes success requires. In other words, the public will tolerate the evil of casualties provided that they are truly necessary for victory -- but the technological prowess of the American military has changed the public's expectations of how many casualties are in fact needed to achieve victory. We find it more plausible that the public has seen an order-of-magnitude change in technology over the past 30 years than that it has come to believe in an order-of-magnitude change in stakes. But we recognize that these remain competing hypotheses that warrant continued research.
CHRISTOPHER GELPI is Associate Professor of Political Science at Duke University and the co-author, with Peter D. Feaver, of Choosing Your Battles: American Civil-Military Relations and the Use of Force.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Mueller Replies
I thank Christopher Gelpi for his thoughtful and extensive comments on my article. Space is limited, but let me reply to several of his key objections.
First, support tends to drop faster in the earlier phases of a war than in the later ones because weak supporters are rather quickly alienated, as has been found in all three wars I examined. But regardless, the overall pattern is one of erosion of support as casualties mount, and, given the quality of public-opinion data and the complexity of events, it seems unlikely that the data can really be pushed to obtain the precise kind of direct casualty-to-support relationship Gelpi seeks. In part this is because, as I note in the article, positive events in a war can temporarily increase support, and negative ones temporarily reduce it, even as the overall trend does not change.
It is not at all clear, however, that the Tet offensive was "the crucial turning point," as Gelpi maintains. Tet did inspire public alarm about the prospects for success in the Vietnam War, but a glance at the figure supplied with my original article will demonstrate that support simply continued to erode, not drop precipitously, during that episode. Moreover, support for the Vietnam War declined by 21 percentage points in the two and a half years before Tet but by only 14 percentage points in the three years after it.
Whatever the effect of the Tet offensive, it is important to stress, as I do in the article, that any increases or decreases in support tend to be temporary: support begins to erode again if and when further American casualties are registered. This interpretation fits the Iraq war data supplied in the article quite well. (In dealing with Iraq, Gelpi unnecessarily complicates things by abandoning measures of support for war for ones of presidential approval, which depends on many factors other than war.)
My interpretation seems to be supported by poll results tallied since the article went to press. Support for the war dropped at the time of Hurricane Katrina, as Americans began to wonder about the nation's priorities, but this decline was more than reversed by the successful Iraqi elections of November 15, the kind of progressive benchmark lauded by Gelpi. Within days, however, support dropped again to a level slightly lower than was registered before either event took place. In all, the phenomenon suggests that a significant and lasting reversal of the erosion of support for the war in Iraq is unlikely. Those who already consider the costs of the war to be too high will probably not permanently reverse their opinion even in the event of good news.
Data from the Korean War are sparse, but they can be interpreted in the same way. A plunge in support after the Chinese entered the war -- inflicting major casualties on U.S. troops -- was, as Gelpi notes, partly (but only partly) reversed when better news came in from the battlefield. But, as he also notes, support sagged thereafter as casualties continued to accrue. There was also a bump upward in support, perhaps based on wishful thinking, when Dwight Eisenhower was elected president at the end of 1952. But other data show a decline again to the end of the war (see pages 44-53 of my War, Presidents, and Public Opinion). Most important for present purposes, the data suggest that those who abandoned their support for the war, particularly in the early months, never really came back to stay.
In the end, my interpretation may not be all that different from Gelpi's. In Choosing Your Battles, he and Peter Feaver repeatedly declare that "the public is defeat phobic, not casualty phobic." Their analysis actually concludes, however, not that the public as a whole is defeat-phobic, but only that some 20 percent of it is (see page 145). In their view, this minority group is particularly sensitive to seeming advances or setbacks. I would simply add that any shifts in opinion as a result tend to prove temporary. Gelpi and Feaver are also much too confident that presidential cheerleading can increase support for a military venture, as Bush, like Lyndon Johnson before him, seems to be finding out.
Second, Gelpi is wrong to charge that I misstate his findings. In seeking a vivid example of the notion that Americans are defeat-phobic rather than casualty-phobic, I (rather incidentally) applied information from a November 7, 1999, Washington Post op-ed in which he and Feaver proclaim it a "myth" that "Americans are casualty shy" and attempt to demonstrate that "a majority of the American people will accept combat deaths -- so long as the mission has the potential to be successful." To support their case, they cite results from a poll they conducted indicating that Americans would, on average, accept 6,861 battle deaths to "stabilize a democratic government in Congo." They also conclude that Americans would accept an average of 29,853 battle deaths "to prevent Iraq from obtaining weapons of mass destruction." (I contributed to a rebuttal to the op-ed at the time, but the Post did not publish it. It is posted at psweb.sbs.ohio-state.edu/faculty/jmueller/klarevas.doc.)
In their subsequent book, Gelpi and Feaver do acknowledge in a footnote that the numbers in the op-ed were "overly susceptible to misinterpretation," and they then rejigger their analysis of the same poll question and essentially conclude that the figure for Congo was some 6,800 percent too high. The damage, however, had already been done: this unfortunate op-ed has been widely cited, applied, and misinterpreted, particularly in military publications.
Third, Gelpi goes on to speculate that improved technology may have reduced the level of casualties the public will tolerate to achieve success. I find this to be an interesting idea, and I certainly agree with him that it deserves further research. I am somewhat skeptical, however, because of the experience of 2001. Although their casualty tolerance was never put to the test, quite a bit of poll data suggest that Americans, deeply threatened and in a rage after 9/11, were willing to suffer enormous casualties in the fight against al Qaeda in Afghanistan.
And I suspect that, whatever the changes in technology, Americans would react to another raid on Pearl Harbor in much the same way that they did in 1941. A degree of support for this proposition comes from data gathered in 1990 after Iraq invaded Kuwait. Some U.S. troops were sent to neighboring Saudi Arabia to deter an attack on that country, and poll data made it quite clear that if Iraq had attacked those troops, support in 1990 for going after Iraq would have been overwhelming. Indeed, limited data suggest that prospective support for retaliating against Iraq in the event of such an attack was higher than prospective support was in 1941 for retaliating against Japan if that country were to attack Hawaii (see page 123 of my book Policy and Opinion in the Gulf War).
If the stakes are high enough, Americans will still accept considerable casualties. In Somalia, Bosnia, Haiti, Kosovo, and Congo, the public clearly found the stakes too low to justify many casualties at all, whereas in Korea, Vietnam, and, especially, Iraq, the stakes proved insufficient as costs rose. And, as suggested in my article, if the public comes to consider a venture to be scarcely worth further American bloodshed (as it did in regard to Vietnam in 1975, Lebanon in 1983, and Somalia in 1993), it will be quite willing to cut its losses and get out -- that is, to rise above any supposed phobias and accept defeat.