In their article “Just War and Unjust Soldiers,” Scott Sagan and Benjamin Valentino provide evidence about how the American public views issues of just and unjust actions in wartime. According to their survey data, the public puts much more emphasis on the justice or injustice of leaders’ actions initiating warfare, and less on soldiers’ actions during war, than do moral philosophers of both the traditionalist and the revisionist schools of just war thought. A majority of respondents in a 2014 survey experiment conducted by the authors declared that victors of a war would be justified in punishing, with prison terms, soldiers who had participated in an unjust attack (figure 6, condition A), and almost half of the respondents regarded prison terms as justified even if the soldiers were conscripts (figure 6, condition C).Footnote 1 Even when the experimental prompt included no mention of war crimes, fewer than 20 percent of respondents regarded conscripts who participated in an unjust war as behaving ethically (figure 5, condition C). It seems that the American public condemns soldiers for following wrong commands by their leaders, and that roughly half of the public would mete out severe legal punishments to these soldiers.
Conversely, over one-third of respondents in the Sagan-Valentino experiment would exempt, even from moral criticism, soldiers acting in a war of just cause who committed blatant war crimes (figures 3 and 5, condition E). That figure is halved when subjects are told that the war in which those soldiers participated was unjust (figures 3 and 5, condition D). The authors conclude both that the American public blames soldiers for the unjust actions of their leaders (figure 5, condition A) and that the same public offers an “an intuitive moral license . . . to soldiers believed to be fighting for a just cause” (p. 435).
These are potentially important findings. If substantiated by further work, as discussed below, they will be disturbing to everyone—including, presumably, virtually all moral philosophers—who wishes to hold soldiers responsible for crimes committed during wartime.
Implications for Philosophical Positions
Sagan and Valentino pose two questions of moral responsibility: First, must soldiers share moral responsibility with their leaders for participating in an unjust war, regardless of the quality of their own wartime actions? Second, must soldiers accept moral responsibility for their own actions in war, regardless of the justice or injustice of the war? The results from the authors’ survey show that with regard to the first question, the American public agrees with the revisionists: soldiers share responsibility with their leaders for participating in an unjust war. Sagan and Valentino do not explicitly discuss the views of traditionalist and revisionist philosophers on the second question, but it seems clear that both sets of scholars believe that soldiers are responsible for their own actions in wartime, and certainly must not commit war crimes. The American public, however, is divided on the second question: only two-thirds agree that soldiers are responsible for not committing war crimes, even if they are fighting for a just cause. Table 1 summarizes the answers to these questions by group.
Table 1. Responsibility in Warfare
A substantial proportion of the Americans surveyed by Sagan and Valentino display what I will call a distinctive “condemnation-absolution syndrome”: condemning soldiers for participating (willingly or not) in unjust warfare, and absolving them for crimes committed in the course of a war launched for just cause. For anyone who wishes to ensure that American soldiers behave justly in war and are held accountable when they do not, the willingness of much of the American public to absolve soldiers from responsibility for war crimes in an otherwise just war should be a source of concern.
Sagan and Valentino also seek to connect their findings to philosophical debates in the just war tradition. Traditional just war doctrine, in their interpretation, “delineates a division of moral responsibility in which political leaders are responsible for the initiation of a war, while soldiers are responsible only for their own conduct during the war” (p. 411). Revisionist theorists, they claim, hold that “soldiers who fight for an unjust cause bear some responsibility for their part in advancing the unjust war, even if they conduct themselves according to jus in bello rules” (p. 412). Sagan and Valentino not only challenge assumptions about public attitudes that they attribute to both sets of moral philosophers; they conclude that their findings “suggest that revisionism, if put into practice, could undermine the protection of noncombatants during war” (p. 435).
In evaluating this statement, much depends on what “put into practice” means. It could mean simply the acceptance of revisionist arguments by the American public, or another public. Under this interpretation, the Sagan-Valentino claim does not hold up, since revisionists do not absolve soldiers fighting in a just war from jus in bello responsibility. Accepting their position therefore would not undermine protection of noncombatants.
“Put into practice” could also mean that publics, not being well versed in philosophical distinctions, become confused by the acceptance of revisionist views on soldiers’ responsibility for the justice of the wars in which they fight. Accepting revisionist views on this subject would in this interpretation somehow lead publics to absolve soldiers on the just side from the obligation to fight justly in bello. Thus, accepting the stronger revisionist claim about the responsibility of soldiers for the justice of a war's cause would confuse publics into accepting a weaker moral claim about soldiers’ responsibility to avoid committing war crimes. Perhaps this could occur, but it makes no logical sense and a convincing psychological explanation would need to be put forward to make this odd confusion plausible.
A third meaning of “put into practice,” as suggested by Sagan and Valentino's abstract, would be that revisionist principles become incorporated into the laws of war. If the resulting legal language clearly differentiated between the responsibility to fight only in just wars and the responsibility to fight justly, whatever the nature of the war, we would once again have little reason to worry about a slippery slope toward accepting war crimes. Only if the legal text implied the absolution for war crimes that much of the American public seems to offer would there be cause for concern. The problem identified by the Sagan-Valentino experiment is not that the American public agrees with the revisionist philosophers on soldiers’ responsibility for the wars in which they fight, but that much of the public disagrees with both sets of philosophers on soldiers’ responsibility not to commit war crimes.
Validity and Generality: Unpacking the Sagan-Valentino Inferences
I devote the rest of my commentary to a different question: How seriously should we take the attitudes expressed by a representative sample of 750 members of the American public in one survey experiment conducted in 2014? Do we really know that public views of wartime behavior differ from those of moral philosophers, as table 1 suggests? Before discussing the implications of the condemnation-absolution syndrome, it would be wise for researchers to ascertain both the validity and generality of this finding. To do so, we need to know whether the Sagan-Valentino experiment passes tests of descriptive inference and replicability.
The relevance of an experimental finding such as this one depends on our belief that we can construct valid descriptive inferences from it. Drawing a descriptive inference is “the process of understanding an unobserved phenomenon on the basis of a set of observations.”Footnote 2 The unobserved phenomenon for Sagan and Valentino is the attitudes of the American public toward justice and injustice in warfare in 2014. Like any social science findings, the findings of their particular experiment are subject to nonsystematic variation, which could include random variations in the composition of the survey sample, the weather when the survey was taken, recent events reported by the media, or individuals’ mood fluctuations. An optimal measure of American public attitudes—clearly infeasible within the budgets of social scientists—would involve conducting a large number of such surveys and administering each to a different representative sample using the same methodology. The first question of valid descriptive inference therefore involves whether it is correct to infer that the findings of Sagan and Valentino are actually representative of the American public's attitudes toward justice and injustice in warfare in 2014.
In this sense, the issue of the validity of descriptive inferences is similar to the issue of replicability in psychology or social psychology: Under similar circumstances, would a survey of a different representative sample of the same population generate essentially similar conclusions? An active controversy about replicability is currently raging in fields such as psychology, where scientists such as those on the Open Science Collaboration team have claimed that many experimental results turn out not to be replicable.Footnote 3
Sagan and Valentino do not merely want to ascertain, as historians, what American public attitudes were, in 2014, toward just behavior in wartime; they also want to generalize about contemporary American attitudes on this subject. The significance of their argument is therefore premised on the validity of a second descriptive inference: that their findings, assumed for the sake of argument to be representative of public views in 2014, are also valid today. Testing this inference is feasible. I therefore suggest, as the next research step for these authors, replicating their experiment in 2020. If the study reported here were successfully replicated, it would suggest that the Sagan-Valentino findings are stable over time, and not an artifact of specific attributes of the American public in 2014. Their research could become the basis for a well-founded descriptive inference about contemporary American public attitudes toward justice in war.
If the Sagan-Valentino findings are valid for the contemporary United States, they would suggest comparative questions. Is the condemnation-absolution syndrome distinctive to American publics at various times, or is it more general? If more general, does it apply to democratic publics in different cultural contexts, Western and non-Western? Does it apply to nondemocratic publics? Each of these questions asks about a further descriptive inference. These descriptive inferences could then provide the basis for eventual causal theory. Valid descriptive inferences about public attitudes toward justice in war constitute crucial intermediate steps toward the development of an explanatory theory of public attitudes toward justice in war.
Consider, for instance, the plausible causal hypothesis that, due to their status as voting citizens, citizens of democracies will be more likely than subjects of autocracies to hold soldiers responsible for the justice of the wars in which they engage. If democratic citizens considered soldiers responsible for the justice of wars in which they participate, their views would more closely resemble those of revisionist just war theorists than would the views of autocratic subjects. The logic behind this hypothesis turns on the belief that citizens of democracies are more accustomed than subjects of autocracies to take responsibility for the policies adopted by their governments. The following statement by a soldier in Shakespeare's Henry V, quoted in Walzer's Just and Unjust Wars,Footnote 4 and also by Sagan and Valentino (p. 416), is classically authoritarian: “We know enough if we know we are the king's men. If his cause be wrong, our obedience to the king wipes the crime of it out of us.” These words could not have come from a democratically minded citizen-soldier. Since democratic citizens are more likely than autocratic subjects to accept responsibility for their governments’ policies, we might hypothesize that they would be more likely to hold soldiers responsible for their governments’ actions in initiating war.
Testing this hypothesis would require, first, running the Sagan-Valentino experiment in a number of countries that are at various levels on a democracy-autocracy scale. Such experiments would yield descriptions of the attitudes of representative samples of the publics of different countries at specific points in time. These descriptions would in turn provide the basis for a series of specific descriptive inferences: that a specific public's attitudes toward soldiers’ conduct in warfare were of a certain type. The specific inferences could then be pooled to make descriptive inferences about the attitudes of “democratic publics” as compared to those of subjects of more autocratic regimes. Only if these differences were significant in the expected direction would investigators proceed to the next step: developing a strategy to explain these differences by identifying the causal mechanisms involved. If the proposition to be tested is that citizenship status is crucial, experiments in democracies comparing the attitudes of citizens with those of noncitizens could constitute an appropriate next step.
Quite apart from such a causal analysis, comparative descriptive inferences would provide some evidence about the generality of the Sagan-Valentino findings. Have they simply identified specific socially and historically constructed attitudes that vary by society or even by time period? Or have they discovered a general human set of intuitions that could somehow reflect more fundamental general human attributes deeply embedded in our neurological systems?Footnote 5 Human attitudes toward issues involving cooperation vary enormously by society.Footnote 6 My expectation therefore is that attitudes toward justice in warfare will also exhibit a large degree of variation. One source of variation is likely to be, as suggested above, the nature of the polity's political system. Other sources of variation could be the society's prevailing religious beliefs or social norms regarding conflict. The presence of substantial intersocietal variation would indicate that Sagan and Valentino have not discovered a fundamental property of human nature. In any event, the Sagan-Valentino findings are sufficiently important that it would be worthwhile to determine how general they are.
Sagan and Valentino have discovered an intriguing pattern of response to hypothetical actions by soldiers during wartime, which I have dubbed the condemnation-absolution syndrome. If propositions about the existence of this syndrome in the United States withstand tests of descriptive inference, the syndrome would be disturbing for moral philosophers and for anyone who seeks to ensure that U.S. soldiers are held accountable for unjust wartime behavior. If the syndrome seems to be a general one—existing across societies or some subset of societies—proponents of justice in bello would have even more cause for concern. However, before embarking on soul-searching explorations of these issues, social scientists should determine whether the Sagan-Valentino results pass multiple tests of valid descriptive inference.
