Counterfactuals and Foreign Policy Analysis
Summary and Keywords
Counterfactuals seek to alter some feature or event of the pass and by means of a chain of causal logic show how the present might, or would, be different. Counterfactual inquiry—or control of counterfactual situations—is essential to any causal claim. More importantly, counterfactual thought experiments are essential, to the construction of analytical frameworks. Policymakers routinely use then by to identify problems, work their way through problems, and select responses. Good foreign-policy analysis must accordingly engage and employ counterfactuals.
There are two generic types of counterfactuals: minimal-rewrite counterfactuals and miracle counterfactuals. They have relevance when formulating propositions and probing contingency and causation. There is also a set of protocols for using both kinds of counterfactuals toward these ends, and it illustrates the uses and protocols with historical examples. Policymakers invoke counterfactuals frequently, especially with regard to foreign policy, to both choose policies and defend them to key constituencies. They use counterfactuals in a haphazard and unscientific manner, and it is important to learn more about how they think about and employ counterfactuals to understand foreign policy.
Counterfactual means contrary to facts. A counterfactual describes something that did not exist or occur. In everyday language, counterfactuals are characterized as “what if” statements. They alter some feature of the past, and by means of a chain of causal logic, show, from antecedent to consequent, how the present might be different. Ordinary people, policymakers, and scholars use counterfactuals for diverse ends. They help us to frame and work our way through decisions, justify choices reduce anxiety generated by unsuccessful outcomes, and buttress causal claims.
Counterfactuals are—or should be—essential tools of foreign-policy analysis (Lebow, 2010). The study of foreign policy can be factored into two successive stages: (a) the reasons actors had for behaving as they did and (b) the consequences of their behavior. The former requires reconstructing actor goals and expectations, their understandings of relevant features of their political and personal worlds, and the process by which they reached and implemented decisions. The latter involves the study of aggregation, as outcomes are the product of the interactions of multiple actors and can bear little relationship to what any of them wanted or expected. Counterfactuals feature prominently in the first stage of analysis because policymakers frequently turn to history for “lessons” to help frame or justify their policies. These lessons are often based on counterfactuals and applied to current problems in a counterfactual manner.
Various analytical approaches and tools are germane to the study of aggregation. They are most frequently employed with the goal of making causal statements about foreign-policy outcomes. Persuasive causal claims must consider counterfactuals. They must ask if the outcome in question would have occurred in the absence of its putative causes.
Foreign-policy analysis also has a normative element. Practitioners evaluate policies and their implementation. Here, too, counterfactuals are critical because evaluation is a form of comparison, and in the political, as in the personal world, these are often with imagined worlds.
Weber (1988) believed that ethical social science requires researchers to query their personal beliefs and positions in society because those largely determine the questions they ask, the methods of finding answers they choose, and even the answers they find. Social scientists cannot readily extract themselves from their lifeworlds, but they can use counterfactual thought experiments to probe their situations and beliefs, and, especially, the frameworks they use for analysis.
This article defines a counterfactual and describe two generic kinds: minimal-rewrite and miracle counterfactuals. It shows their relevance to formulating propositions and probing contingency and causation, offers a set of protocols for using both kinds of counterfactuals toward these ends, and illustrates the uses and protocols with historical examples. Policymakers invoke counterfactuals frequently, especially with regard to foreign policy, to both choose policies and defend them to key constituencies. They use counterfactuals in a haphazard and unscientific manner, and it is important to learn more about how they think about and employ counterfactuals to understand foreign policy.
Policymakers and Counterfactuals
Foreign policy often rests on counterfactual foundations. During the Cuban missile crisis. President Kennedy and his advisers used eight counterfactuals, and Nikita Khrushchev and his advisers two, to generate and evaluate their policy options (Lebow & Stein, 1996). Most of these counterfactuals were implausible because they rested on what are now known to have been demonstrably false assumptions about the motives and risk-taking propensity of the other side. They are nevertheless a powerful analytical tool because of the insight they offer into the thinking of leaders and advisers on both sides.
Policymakers, their advisers, and the media approach foreign policy in terms of what they consider to be the lessons of the past. These lessons almost always rest on or involve counterfactuals and are often applied to present situations in the form of counterfactuals. The postwar economic order is based in part on the assumption that the Great Depression of the 1930s could have been avoided, or its effects muted, if international institutions had been in place to coordinate an effective collaborative response and discourage short-term and ultimately counterproductive efforts by states to protect themselves at the expense of others.
Postwar American national security was based on another lesson of the 1930s: that appeasement invites aggression, and its corollary deterrence prevents it. When apprised by in a telephone call of the news of North Korea’s invasion of South Korea in June 1950, US president Harry Truman immediately authorized the dispatch of American forces from Japan to defend South Korea. He then hurriedly returned to Washington from his home in Missouri:
I had time to think aboard the plane. In my generation, this was not the first occasion when the strong attacked the weak. I recalled some earlier instances: Manchuria, Ethiopia, Austria. I remembered how each time the democracies failed to act, it encouraged the aggressors to keep going ahead. Communism was acting in Korea just as Hitler, Mussolini, and the Japanese, had acted, ten, fifteen, twenty years earlier. I felt certain that if South Korea were allowed to fall, Communist leaders would be emboldened to override nations closer to our shores … If this was allowed to go unchallenged it would mean a third world war, just as similar incidents had brought on the second world war.
(Truman, 1956, p. 332)
Margaret Thatcher framed the Falklands-Malvinas crisis in similar terms. She equated the General Galtieri and the Argentine Junta with Hitler and the Nazis and was convinced that any concession on her part would lead to new demands (Lebow, 1983). These and other counterfactual policy lessons take the following form: Strategy A failed miserably, and Strategy B would have worked, so apply Strategy B when this situation next arises. There are also factual policy lessons: Strategy A worked, so use it again when facing a similar situation.
The counterfactual lesson has good evidence that a particular strategy failed—appeasement in this case—but no evidence that an alternative strategy would have succeeded—or that appeasement might not prove appropriate in some other situation. Interestingly, appeasement was practiced in the 1930s in part because one of the key revisionist policy lessons of World War I was that deterrence—although it was not called that at the time—had helped to provoke that conflict (Lebow, 1981, 2010). More interesting still, is that those espousing deterrence in the postwar era made little to no effort to discover if Hitler could have been restrained if France and Britain had demonstrated willingness to go to war to defend the European territorial status quo. German documents made this an eminently researchable question, and historians ultimately used them to try to determine at what point Hitler could no longer be deterred. Deterrence advocates were largely unfamiliar with this literature, which suggests that at least some of them advocated military buildups, forward deployments, and the use of threatening rhetoric for different reasons. Other American policymakers have accepted deterrence as conventional wisdom, even though the counterfactual was never seriously investigated, and used it to frame their responses to the Soviet Union, China, and, more recently, Iraq, Iran, and North Korea.
The factual lesson rests on the belief that strategy A was responsible for a positive outcome. It involves a hidden counterfactual: that there would have been no good outcome in the absence of strategy A. This claim is rarely investigated even when there is good historical evidence that could enhance or diminish belief in the efficacy of strategy A. It may be that the strategy worked, or alternatively, that the outcome was attributable to something else. To stay with our example of deterrence, it was confirmed on the basis of expectations instead of evidence in the Cold War. When the Soviet Union did not invade Berlin in 1948–1949, American leaders and scholars concluded that deterrence had worked, not that Stalin had never had any intention of occupying the western sectors of that city by force. They drew the same conclusion from the two subsequent Berlin crises of 1958–1959 and the Cuban missile crisis of 1962 (Lebow, 2017). In these instances, tautological confirmation of the causes of success were taken as confirmation of the prior counterfactual on which the strategy was based. In the missile crisis, Kennedy believed, incorrectly, that Khrushchev had sent missiles to Cuba because he doubted the president’s resolve and would not have done so if Kennedy had taken a stronger stand at the Bay of Pigs or in Berlin. In fact, Khrushchev never doubted Kennedy’s resolve; he thought he was brash and risk prone and likely to order the American navy to intercept the ships carrying missiles to Cuba if they were deployed openly (Lebow & Stein, 1994, pp. 72–78). Ironically, his concern for Kennedy’s resolve made him deploy the missiles secretly, which provoked the crisis. Kennedy and his advisers had already decided that there was little they could do in the face of a public missile deployment because it was so analogous to the American one in Turkey (Lebow & Stein, 1994, pp. 80–81).
Factual and counterfactual lessons alike take it for granted that policymakers can make comparisons between the past and the present in the sense of identifying similar situations. Research in cognitive psychology suggests that this assumption very problematic because ordinary people and policymakers frequently mistake the most salient features of a situation for the most important (Jervis, 1979, pp. 217–270). This leads to misleading comparisons and application of a strategy, even if demonstrably successful in the past, may not be appropriate to the present. Once again, deterrence offers an example.
Analysts and Counterfactuals
Foreign-policy analysts must be sensitive to the ways in which policymakers use counterfactuals. The examples cited here suggest they use them frequently and badly. Policymakers, their advisers, and the media are victims of entrenched, but not necessarily valid, historical lessons and cognitive biases that lead them to make superficial comparisons between past and present situations. Analysts need to take these counterfactuals and limitations into account when attempting to reconstruct foreign-policy decisions. They should also use them as indirect but compelling evidence about the assumptions used by leaders to frame and address problems. This is most compelling for private counterfactuals, those, that is, used by policymakers in their deliberations, not just publicly to sell their policy choices.
Foreign-policy analysts need to make their own counterfactuals. In the first instance, this is necessary to free themselves from the determinism that pervades the study of foreign policy and international relations. They also require counterfactuals to establish frameworks of analysis and proposition and to evaluate their causal claims.
There are two kinds of counterfactuals relevant to foreign-policy analysis: minimal-rewrite counterfactuals and miracle-world counterfactuals (Tetlock & Belkin, 1996). Minimal-rewrite counterfactuals are intended to be as realistic as possible. They make small, credible, changes in the fabric of history with the goal of probing causation and contingency. In a recent book Lebow (2014) imagined a world in which Archduke Franz Ferdinand and his wife, Countess Sophie, returned alive from their visit to Sarajevo. This counterfactual is eminently plausible because their assassination was such a near thing and never would have happened if the archduke and those responsible for his security had acted sensibly, either before the first unsuccessful attempt on his life or in its immediate aftermath. A small, credible rewrite of history has the potential to bring about a different world.
A recent but equally realistic example concerns the survival of a young Cuban boy, Elián Gonzalez (Lebow, 2014, p. 11). In November 1999, Elián fled Cuba with his mother and twelve others in a small boat with a faulty engine; Elián’s mother and ten other passengers died in the crossing. Floating in an inner tube, Elián was rescued at sea by two fishermen, who handed him over to the U.S. Coast Guard. The subsequent decision by attorney general Janet Reno to return Elián to his father in Cuba rather than let him stay with his paternal great uncle in Florida infuriated many Cuban Americans. As a result, many fewer Americans of Cuban descent voted Democratic in the 2000 presidential election. If Elián had drowned, Al Gore would have carried Florida and have become president of the United States. His election would not have prevented 9/11, but it almost certainly would have prevented the invasion of Iraq. Harvey (2011) argues against this counterfactual with one of his own. He contends that President Gore would also have invaded Iraq. Neither of these opposing counterfactuals is definitive, but both have the virtue of sharpening the debate about the causes of the Iraq War and raise new questions amenable to historical research.
Miracle-world counterfactuals make implausible changes in reality. They violate our understanding of what is “realistic,” or even conceivable, but they are valuable when they allow us to reason our way to the causes and contingency of real events or the dynamics that govern them. Consider the hypothesis that Europe achieved its global military advantage in the nineteenth century because it was the only region of the world where no long-standing hegemony had been established. Prolonged competition among its leading political units made them lean and mean, better armed, and more efficient in the use of large-scale violence. To advance this hypothesis it is necessary to posit a counterfactual world in which some state—perhaps a better-organized and better-led Spain in the sixteenth century—achieved something approaching continental hegemony. It is also necessary to reason through what a hegemonic Europe would have been like and how it would have differed from historical Europe in the dimensions relevant to the argument. Miracle counterfactuals are essential tools in developing and evaluating competing historical and political interpretations and theories.
Mueller’s (1989) world without nuclear weapons involved a miracle rewrite of a century of scientific and political history to “uninvent” nuclear weapons. He then explored how the Cuban missile crisis would have played out in this world. In contrast to the conventional wisdom that attributes the long peace between the superpowers to nuclear deterrence, he argues that Moscow and Washington were restrained by their general satisfaction with the status quo, memories of the human and economic costs of World War II, and belief in the utter destructiveness of large-scale, conventional warfare. Mueller contended that nuclear weapons were redundant and possibly counterproductive to efforts to maintain the peace.
Both kinds of counterfactuals are relevant to the three tasks relevant to foreign-policy analysts to consider.
Receptivity to Contingency
Fischoff (1975) demonstrated that outcome knowledge affects our understanding of the past by making it difficult for us to recall that we were once unsure about what was going to happen. Events deemed improbable by experts (e.g., peace between Egypt and Israel, the end of the Cold War) are often considered overdetermined and all but inevitable after they have occurred (Hawkins & Hastie, 1990). By tracing the path that appears to have led to a known outcome, we diminish our sensitivity to alternative paths and outcomes. We may fail to recognize the uncertainty under which actors were operating and the possibility that they could have made different choices that might have led to different outcomes.
Many psychologists believe the certainty of hindsight effect is deeply rooted and difficult to eliminate. But the experimental literature suggests that counterfactual intervention can assist people in retrieving and making explicit their massive but largely latent uncertainty about historical junctures—that is, to recognize that they once thought, perhaps correctly, that events could easily have taken a different turn. The proposed correctives use one cognitive bias to reduce the effect of another. Ross, Lepper, Strack, and Steinmetz (1977) exploited the tendency of people to inflate the perceived likelihood of vivid scenarios to make them more responsive to contingency. People they presented with scenarios describing possible life histories of post-therapy patients evaluated these possibilities as more likely than did members of the control group who were not given the scenarios. This effect persisted even when all the participants in the experiment were told that the post-therapy scenarios were entirely hypothetical. Tetlock and Lebow (2010) used four scenarios about the outcome of the Cuban missile crisis to demonstrate the utility of counterfactuals in overcoming determinism. Each scenario was more elaborate and vivid than the previous one and was seen as more vivid, and hence more credible, by a sample of policymakers and international-relations scholars.
Policymakers rarely act in response to explicit theories but commonly rely on more informal understandings of how the world works. They display the same belief in the retrospective near-inevitability of important historical outcomes as their academic counterparts. In interviews with numerous American, Soviet, and European politicians, diplomats, and military officers who played prominent roles in the end of the Cold War, Herrmann and Lebow (2001) found that almost all of them believed that the Cold War had to end when and how it did. At the same time, these policymakers insisted on the contingency of developments critical to this outcome in which they played a major role. They explained how easily such developments (e.g., arms control, the unification of Germany) could have been forestalled or worked out differently if it had not been for their skill, relationships with their opposite numbers, or ability to collaborate with them behind the backs of their respective governments. They seemed unaware of the contradiction between these two positions and struggled to reconcile them when pushed to do so.
There is something wrong with this story. If major historical developments are so inevitable, the pattern of events leading to them should not be so contingent. If events are overdetermined, the underlying conditions responsible for those events should have been apparent at the time to scholars and policymakers alike, making them—although not their timing and specific expression—to some degree predictable. None of the events in question were self-evident at the time. In the decade prior to 1914, there was a general expectation among many military authorities and some, but by no means all, political leaders that a European war was likely. There was nevertheless remarkable optimism within the diplomatic and business community that mutual trade and investment had made war increasingly irrational and less likely. On the eve of the war books advancing both arguments were run-away bestsellers (Angell, 1910; Bernhardi, 1912). European opinion was also divided on World War II, with many of those in power in France and Britain and the Soviet Union convinced that Hitler had limited aims or could be bought off with territorial and other concessions. For quite different reasons, Churchill and Roosevelt expected to be able to do business with the Soviet Union after World War II. Writing in the late fifties, Herz (1959), reminded us that the advent of bipolarity was as unexpected as the atomic age, in part responsible for it. Hardly anybody predicted the onset of the Cold War or its demise, let alone the collapse of the Soviet Union. In the early decades of the Cold War, American foreign-policy experts worried that the Soviet model would be more appealing to the so-called Third World countries than liberal capitalism. In its latter decades, members of the U.S. national security community thought it possible, if not likely, that the Soviet Union would pull ahead militarily and act more aggressively. Both expectations were wide of the mark.
Research questions arise when events strike us as interesting or anomalous. To conceive of an event as anomalous we need a benchmark that establishes what is normal. In the hard sciences, benchmarks can sometimes be derived from a well-established laws or statistical generalizations; cold fusion would have been contrary to several of these laws and thus a strikingly anomalous event. There are few, if any, laws or universally valid statistical generalizations applicable to the social world, but we all hold to theories, or at least to strong, if informal views, about how that world works. These theories and views give rise to expectations, and when these expectations are unfulfilled, to counterfactual worlds. These alternative worlds may appear more probable than the actual state of affairs. During the Cold War, the preeminent question in the security field was “the long peace” between the superpowers. In international political economy, a key question in the late 1970s and 1980s was whether the survival of the postwar international economic order could survive in the face of America’s seeming decline as a hegemon. Some security specialists (Waltz, 1976; Gaddis, 1987) considered it remarkable—that is, contrary to their theories and expectations—that the superpowers had avoided war, unlike rival hegemons of the past. Some international political economists were equally surprised that neither Germany nor Japan had sought to restructure international economic relations to their advantage in response to the relative U.S. decline (Gilpin, 1981, pp. 222–223; Keohane, 1984). Both research agendas assumed that the status quo was an extraordinary anomaly that required an equally extraordinary explanation. For researchers who assumed that neither superpower was so unhappy with the existing state of affairs that it was willing to risk war to change it, or that Germany and Japan were not dissatisfied enough with the institutions for managing the international economy that the United States had established at the end of World War II to risk the disruption entailed in trying to change them, the seeming robustness of the postwar political and economic orders posed no intellectual puzzle.
These examples indicate how ideology and culture provide the unspoken assumptions that shape inquires of the social world. These understandings generate expectations, which if they are not met, in the future or historically, give rise to seeming anomalies and interesting research questions. Ideological and cultural biases, like their political counterparts, readily assume the guise of science and lend themselves to tautological confirmation. This tightens the hold of the biases on the minds of scholars, policymakers, and public opinion. Counterfactuals are a useful means of exposing the subjective foundations of beliefs and providing a possible means for scholars to step outside their own cultures and belief structures. Suppose we ask whether there would have been a Cold War if the Soviet Union had been a liberal democracy in 1945 or if there had been no nuclear weapons? What if China had accessible coal reserves and iron deposits close to major commercial centers? Would it have begun its own industrial revolution and become a rival to Europe? These counterfactuals are utterly unrealistic, but they compel us to examine the unspoken assumptions that guide our expectations about the real world.
Testing and Evaluation
All counterfactuals are causal assertions. Conversely, all causal assertions entail counterfactuals, making them fundamental to testing or evaluating theories and interpretations. If we hypothesize that x caused y, we assume that y would not have happened, ceteris paribus, in the absence of x. Quantitative research attempts to get around this problem, and the contrapositive form of the fallacy of affirmation, by constructing a sample of comparable cases large enough to contain adequate variation on dependent (what is to be explained) and independent (what allegedly explains it) variables (Dawes, 1996). This strategy is only effective if there are no causes beyond those considered that vary systematically with the error term. To rule out this possibility researchers need to pose the counterfactual of what would have happened if variables in the error term were altered? In actual experiments, this problem can only partially be solved by random assignment. In case studies and historical narratives this is not possible.
Case study researchers typically attempt to establish causation by means of process tracing (Eckstein, 1975; George & Bennett, 2005). They document the links between a stated cause and a given outcome in lieu of establishing a statistical correlation. This works best at the individual level of analysis, but only when there is enough evidence to document the calculations and motives of the actors—and this is generally not the case, for the reasons discussed here. Even on the rare occasions when such evidence is available, it still may not be possible to determine the relative weight of the several hypothesized causes and which, if any, might have produced the outcome in the absence of others or in combination with other causes not at work in the case. Like statistical analysis, process tracing is based on a prior set of beliefs about what is important, what makes for a case, and working assumptions about the causal nexus in the cases in question. Historians should, but rarely do, make these assumptions explicit. To sustain causal inference, it is generally necessary to engage in comparative analysis.
Comparative analysis is also possible within a single case and can take two forms: intracase comparison and counterfactual analysis. Intracase comparison breaks a case down into a series of similar interactions that are treated as separate and independent cases for the purposes of analysis. Numerous studies of arms control and superpower crises make use of this technique (George & Smoke, 1974; George & Simmons, 1994; Hopf, 1994). Like any form of comparative analysis, intracase comparisons try to obtain as much variation as possible on dependent (explanandum) and independent (explanans) variables. This is sometimes more difficult to do than in cross-case comparisons. The independence of cases is also more problematic, as the process and outcome of past decisions are likely to have considerable influence on subsequent decisions about similar issues. But intracase comparison confers a singular benefit: it builds variation within a fundamentally similar political and cultural context, controlling better than intercase comparison for many factors that may be important but otherwise unrecognized. Unfortunately, not every case can be broken down into multiple decision points for purposes of comparison. This works only for those where actors have made multiple decisions about the same problem and considered roughly similar options on these occasions.
When intracase comparison is impossible, variation can be generated within a case by counterfactual experimentation. This latter strategy lies at the core of many simulations where variables are given a wide range of counterfactual values to determine the sensitivity of the outcome to changes in one or more of them. Counterfactual simulation can identity key variables and the range of values in which they will be have the most impact on the outcome. Information obtained this way, especially if it has counterintuitive implications, can guide subsequent empirical work intended to test the model or generate information necessary to make it a better representation of reality. Counterfactual simulation can evaluate theories more directly. Lars-Eric Cederman (1996) used this method to test the realist assumption that balancing is the inevitable consequence of international anarchy. It follows, Cederman reasoned, that global hegemons should rarely emerge in real or counterfactual worlds. He nevertheless found that they appear with regularity in counterfactual simulations, especially under conditions of defense-dominance, the best case for neorealism.
Counterfactual experiments and simulations can tease out the assumptions on which theories and historical interpretations rest. Apologists for the Soviet system insist that communism would have evolved differently if Lenin had lived longer or been succeeded by someone other than Stalin (Breslauer, 1996). Attempts to address this question have not resolved the controversy but have compelled historians to be more explicit about the underlying assumptions that guide and underpin contending interpretations of Stalin, and the nature of the Communist party and the Soviet state. Those assumptions have now become the focus of controversy, and scholars have looked for evidence with which to evaluate them.
Because every causal argument has its associated counterfactual, critics of a particular theory or interpretation have two generic strategies open to them. They can offer a different and more compelling theory or interpretation—far and away the most common strategy—or show that the outcomes associated with the theory or interpretation would have happened in the absence of their hypothesized causes. Mueller’s study of the Cold War is a nice example of the second approach.
Counterfactuals are a key component of evaluation in personal and political issues. Psychological studies suggest that people invent counterfactuals to evaluate their choices and their outcomes, especially when they are negative (Kahneman, Slovic, & Tversky, 1982, pp. 201–208; Roese & Olson, 1995). Upward counterfactuals—which lead to better outcomes—improve affect—while downward counterfactuals—can serve as wake-up calls that prompt or lead to calls for preparatory responses (Landman, 1992; Baron, 1992; Mandel, 2003).
Was the development of nuclear weapons a blessing or a curse for human kind? What about affirmative action, free trade, or the economic and political integration of Europe? Serious and thoughtful people can be found on all sides of these controversies. Their arguments share one thing in common: they use counterfactual benchmarks—most often, implicitly—to assess the merits of real-world policies, outcomes, or trends. John Mueller’s assessment of the role of nuclear weapons in the Cold War and the controversy surrounding Lenin’s survival or alternate successors for the development of the Soviet Union is one example that has been mentioned. More recently, the Iraq War has spawned counterfactuals on both sides of the debate. Those opposed to the Anglo-American invasion of Iraq insist that it has provoked, rather than minimized terrorism against the United Kingdom and the United States. Their argument rests on an imaginary benchmark of what the level of terrorist attacks would have been like in the absence the war. Conversely, supporters of the George W. Bush administration contend that the president reduced the incidence of terrorism against the West by taking the war to the Islamic heartland “at low cost.”
Assessment can be significantly influenced, even determined, by the choice of counterfactual. The conventional wisdom holds that the allied victory in World War I was a positive outcome: it prevented an expansionist Germany from achieving hegemony in continental Europe. This assessment makes sense from the vantage point of the corporate boardrooms and corridors of power in Paris, London, New York, and Washington. From the perspective of European Jewry, the outcome of the First World War was a disaster. If Germany had won, there almost certainly would have been no Hitler and no Holocaust. This counterfactual reveals the strikingly different interests of the groups in question and how groups are drawn to different counterfactuals or evaluate the same ones differently. As with historical analogies, the interesting, and eminently researchable question becomes the extent to which counterfactuals guide evaluation or are chosen to justify positions that people have reached for quite different reasons (Khong, 1992).
Protocols for Thought Experiments
If the past is highly contingent, so is the counterfactual past. We can never assert with any authority that a change in the past would have led to a particular counterfactual world. The most we can do is make good counterfactual arguments, arguments that follow appropriate protocols and are as rich in evidence as possible.
The impossibility of making valid, as opposed to good, counterfactuals is in no way an obstacle to good counterfactual analysis. The goal is not to make the case for any particular future. Rather, the goal is to explore the feasibility and contingency more generally of types of alternative worlds and to use them to assess the causes and contingency of historical outcomes. Toward this end, here are eight protocols for historical counterfactual thought experiments germane to foreign policy analysis.
1. Realism. Blase Pascal famously observed that the history of the Roman Empire would have been different if Cleopatra had had an ugly nose, because Marc Anthony would not have fallen for her. This counterfactual lacks realism. There is no particular reason why Cleopatra should have been less attractive. She might have broken and disfigured her nose in childhood accident, but it is not self-evident that this would have weakened the ardor of Anthony, who sought power as much as beauty and seems to have been taken in as much by Cleopatra’s persona as her looks. Good counterfactuals ought to arise from the context. There should be compelling mechanisms to bring them into being that require only plausible rewrites of history. Moreover, these rewrites must be consistent with the pattern of decisions or behavior that follow from the counterfactuals themselves. The Franz Ferdinand and Elián Gonzalez examples are cases in point.
2. Clarity: All causal arguments should define as unambiguously as possible what is to be explained (the consequent in counterfactual arguments), what accounts for this outcome (the antecedent) and the chain of logic linking the two. Good counterfactuals should also specify the conditions that would have to be present for the antecedent to occur. Some historians have argued that timely public health measures could have significantly reduced the mortality in Europe associated with the Black Death pandemic of the 14th century. For European communities to have implemented these measures they would have had to recognize that human intervention could affect the spread of disease and had the authority and will to impose draconian measures on travel and trade over the likely objections of the wealthy and merchant classes (Hawthorn, 1991, pp. 31–60). These conditions, Hawthorn (1991, pp. 54–60) argues, are unrealistic given the values, knowledge, and political structure of the age; large-scale quarantines would not be implemented to combat infectious diseases until the 18th century. Plausible world counterfactuals do not only require antecedents that are understood to be realistic; the antecedents themselves must not require other implausible conditions or counterfactuals.
3. Logical consistency or cotenability: Every counterfactual is a shorthand statement of a more complex argument that almost invariably requires a set of connecting conditions or principles. The antecedent should not undercut any of the principles linking it to the consequent. A case in point is Fogel’s (1964) famous claim that if the railroads had not existed, the American economy in the 18th century would have grown only slightly more slowly than it did, because there would have been a much stronger incentive to invent the internal combustion engine sooner. Elster (1978) rightly objects to this counterfactual on the grounds that if the technology had been available to invent and produce automobiles, it would almost certainly have also led to the prior invention and development of railroads. Consider another example: the outcome of the Cuban missile crisis if Richard Nixon had won—as he nearly did—the 1960 presidential election. Nixon would have been more responsive to the demands of the military for an air strike and invasion, and this would have led to a military, possibly nuclear, conflict with the Soviet Union. The problem is that for the same reason Nixon would have been responsive to demands of the CIA and the Pentagon for committing American forces to save the faltering Bay of Pigs invasion. If so, there would have been no Castro-led Cuba a year later and no possibility of the Soviet Union secretly deploying missiles there.
4. Enabling counterfactuals should not undercut the antecedent: Counterfactuals often require other counterfactuals to make them possible. For Germany to win World War II its armies would have had to defeat the Soviet Union and the United Kingdom or, at least, have compelled both nations to sue for peace. In the aftermath of victory, Hitler would have had to rein in his aggressive impulses so as not to provoke war with the United States and make Germany the target of American atomic bombs. For Hitler to have shown restraint he would have had to become a very different person. If restraint and caution had been part of his character all along, it is unlikely that he would have come to power and even more unlikely that he would have gambled on occupying the Rhineland, Anschluss with Austria, challenging the British and French over Czechoslovakia and invading the rump Czech state after gaining territorial concessions at Munich—the very successes that paved the way for a German invasion of Poland and subsequent challenge of the Soviet Union. Herwig (2006) argues persuasively that to win World War II, Germany would have required jet fighter planes and nuclear weapons, and here, too, it has been well documented, Hitler’s personality stood in the way.
5. Historical consistency: Weber (1988) insisted that plausible counterfactuals should make as few historical changes as possible on the grounds that the more we disturb the values, goals, and contexts in which actors operate, the less predictable their behavior becomes. Counterfactual arguments that make a credible case for a dramatically different future on the basis of one small change in reality are very powerful, and the plausible rewrite rule should be followed whenever possible. The nature of the changes made by the experiment are nevertheless more important than number of changes. A plausible rewrite that makes only one alteration in reality may not qualify as a plausible world counterfactual if the counterfactual is unrealistic or if numerous subsequent counterfactual steps are necessary to reach the hypothesized consequent.
A counterfactual based on several small changes, all of them appearing plausible, may be more plausible, especially if they lead more directly to the consequent. Lebow’s (1981, 2010) investigation of the contingency of World War I used plausible rewrites to break, or at least alter the timing of, the causal chains that led to gestalt shifts in Germany and Russia, making them more prone to risk. This dramatically increases the probability of preventing that conflict. Lebow (2014) also remade the political context for Austria by forestalling the Italian landing at Tripoli in 1911, which effectively forestalls the First and Second Balkan Wars, leaving Serbia a smaller and less threatening state and Franz Ferdinand alive. Those who employ plausible rewrites of history should be explicit about the choices and why they fulfill this and other conditions better than alternative counterfactuals.
6. Theoretical consistency: There are few, if any, generally accepted theories in the social sciences and none in international relations, comparative politics, or history. For the purposes of counterfactual analysis, it is nevertheless useful to reference any theories, empirical findings, or historical interpretations on which the causal principles or connecting arguments are based. Because assumptions are so critical for the plausibility of factual and counterfactual arguments, efforts to do this allow the researcher, to some degree, to prime readers about the assumptions in which the arguments in question are anchored.
7. Avoid the conjunction fallacy: There are good statistical grounds for the plausible-rewrite rule, since the probability of a consequent is the multiple of the probability of each counterfactual step linking it to the antecedent. The laws of statistics suggest that the probability of any compound counterfactual is exceedingly low. This does not mean that our world is overdetermined, only that it is highly unlikely that hypothesized antecedents will produce specific consequences at any temporal remove. Social and political developments are highly contingent, and the future is undetermined—as was the past when it was still the future. The long-term consequences of change are almost always unpredictable. If Mozart had lived to the age of 65, today’s world could well have turned out to be strikingly different from the world we know. But many alternative worlds are possible, and the probability of any one of them coming to pass is exceedingly low. Counterfactuals may have changed the world, but in ways that become exponentially more difficult to track over time because of the additional branching points that enter the picture. The probabilities associated with these outcomes will vary enormously, so researchers need to specify whether their counterfactuals are intended to produce a specific world, a set of worlds with particular characteristics, or any world (on a specific dimension) other than the one that actually came to pass.
8. Recognize the interconnectedness of causes and outcomes: Surgical counterfactuals are unrealistic because causes are interdependent and have important interaction effects. Changes we make in the past may require other changes to make them possible, and those other changes, in turn, are almost certain to produce changes in addition to the ones we expect to produce the consequent. History is like a spring mattress. If one of the springs is cut, or simply subjected to extra pressure, the other springs will, to varying degrees, shift their location and tension. Good counterfactuals must specify, within reason, what else is likely to change as a result of a hypothesized antecedent and consider how the change that appears to be the most important (a choice that also requires elaboration) might influence the probability of the consequent. In Herwig’s (2006) counterfactual probe of World War II, he reasoned that not even victory in the east would have led to a German victory in World War II because it would have made Germany, not Japan, the target of the first American atomic weapons. In the real world, German successes were the catalyst for the American crash effort to build nuclear weapons and now make it the preferred target for those weapons for the same reason. German victory is thus impossible unless some other plausible counterfactual can be devised to untrack the Manhattan Project.
9. Consider second order counterfactuals: Even when there is good to reason to believe that the antecedent will produce the desired consequent, the possibility remains that subsequent developments will return history, more or less, to the course from which it was initially diverted by the antecedent. This might be the long-term result of enabling counterfactuals necessary to bring about the antecedent, of follow-on counterfactuals produced by the antecedent, of counterfactuals arising from the consequent, or of interaction among any combination of these counterfactuals. Interaction effects among second-order counterfactuals might be considered “third order” counterfactuals, and they, too, can have profound consequences for the subsequent course of developments.
Attempts to identify and analyze all the counterfactuals arising from the antecedent and consequent would quickly lead to an infinite regress. Researchers should nevertheless try to identify what, in their judgment, is the most likely course or courses of events that could unravel their consequent or negate its value as an outcome. The last point entails the recognition that we choose a consequent because of some larger effect it is intended to produce. If other developments make it unlikely that the consequent will have the desired effect, it loses its attractiveness. No counterfactual argument is complete without some argument about alternative, alternative futures, and some assessment of their likelihood and implications for both the consequent and its utility as a consequent.
These criteria will not allow researchers to validate plausible-world counterfactuals, but they will help them weed out poor counterfactuals on the basis of clarity and logical and substantive completeness. Counterfactuals that pass this rigorous set of tests are also likely to appear more plausible to readers. Most of these criteria are not applicable to miracle-world counterfactuals, which, by definition, are not required to meet any real-world tests. The value of such counterfactuals is based entirely on their ability to provoke, inspire, and compel researchers to think about issues and problems they would not otherwise recognize or address or to look at them in a new light. For a field in which careful, technical work is increasingly valued over imagination, miracle-world counterfactuals have the potential to refocus our attention on a deeper set of important questions.
Many foreign-policy analysts are concerned with the future. The honest ones are skeptical about their ability to predict anything but the most short-term outcomes. We cannot, however, throw our hands up in the face of uncertainty, contingency, and unpredictability. In a complex society, individuals, organizations, and states require a high degree of confidence—even if it is misplaced—in the short-term future and a reasonable degree of confidence about the longer term. In its absence, they would be unable to commit themselves to decisions, investments, and policies. Like nudging the frame of a pinball machine to influence the path of the ball, we cope with the dilemma of uncertainty by doing what we can to make our expectations of the future self-fulfilling. We seek to control the social and physical worlds, not only to make them more predictable, but also to reduce the likelihood of disruptive and damaging shocks (e.g., floods, epidemics, stock-market crashes, or foreign attacks).
Thinking about the future involves the construction of what analysts believe to be the most likely possible worlds. It requires extrapolations from the present. Like counterfactuals, forecasts consist of causal chains. They go from the present via a chain of logic to a future envisaged consequent, not from the past in the form of an imagined antecedent. Past and future are logically equivalent, as the future in due course becomes the past. Logical equivalences and methodological similarities create link future worlds and counterfactual presents.
The construction of counterfactuals requires us to consider not just one, but many alternate worlds. First, there is the world we create by means of our minimal rewrite of history. The instant we intervene in history, it is the only alternative world. As time goes by, it has the potential to evolve into many possible worlds. Consider the analogy to a chess game. We have made a move in the form of a counterfactual. History, our opponent, must respond. But as there is no history sitting across from us, we, in effect, play both boards. We must consider a variety of moves and work through the consequences of the most likely ones to ascertain, in the first instance, if it is reasonable to assume that events are likely to diverge from the actual path of history or somehow come back to something similar. Second, and of equal importance, we want to see if any of the possible moves and countermoves will undercut our chain of logic or produce “second order” counterfactuals that will bring history more or less back on course even if our consequent is achieved. Our initial counterfactual world has the potential of creating an infinity of counterfactual worlds, and the more of them we consider and the possible worlds to which they give rise, the more convincing our arguments are likely to be about the desired effects of the initial intervention in history.
Looking at the future also involves the consideration of multiple possible worlds. In this connection, we must distinguish between two ways of addressing the future: prediction and scenario generation. The former is theory driven and uses a theory or propositions based on seeming regularities to predict a future development. The prediction can be hedged as either-or or be stated probabilistically, but either way it is an either-or proposition in the sense that it comes to pass (more or less) or does not. Scenario generation, also called forecasting, is a more labile and open-ended process (Bernstein, Lebow, Stein, & Weber, 2000). It, too, starts from a set of theoretical assumptions, but it understands them as the starting place for analysis, not the basis for making predictions. This difference reflects a deeper difference in epistemology. Forecasting, like evolutionary biology, acknowledges the all-important nature of context and, further, that most of the important features of context must lie outside of any theory. Scenarios recognize the hazardous nature of forecasting so consider multiple future worlds, where they branch off from one another and the kind of information would need at different times upstream to increase or decrease confidence in any of these worlds and the story lines leading to them. In effect, the protocols developed for counterfactuals apply equally to forecasting.
Angell, N. (1910). The great illusion. London: Macmillan.Find this resource:
Baron, J. (1992). The effect of normative beliefs on anticipated emotions. Journal of Personality and Social Psychology, 63, 320–330.Find this resource:
Bernhardi, Friedrich von (1912). Deutschland und der nächste Krieg (2 Vols.). Stuttgart, GE: Cotta.Find this resource:
Bernstein, S., Lebow, R. L., Stein, J. L., & Weber, S. (2000). Physics got all the easy problems: adapting social science to an unpredictable world. European Journal of International Relations, 6, 43–76.Find this resource:
Breslauer, G. W. (1996). Counterfactuals reasoning in western studies of Soviet politics and foreign relations. In P. E. Tetlock & A. Belkin (Eds.), Counterfactual thought experiments in world politics (pp. 69–94). Princeton, NJ: Princeton University Press.Find this resource:
Cederman, L.‑E. (1996). Rerunning history: Counterfactuals simulation in world politics. In P. E. Tetlock & A. Belkin (Eds.), Counterfactual thought experiments in world politics (pp. 247–267). Princeton, NJ: Princeton University Press.Find this resource:
Dawes, R. M. (1996). Counterfactual inferences as instances of statistical inferences. In P. E. Tetlock & A. Belkin (Eds.), Counterfactual experiments in world politics (pp. 301–308). Princeton, NJ: Princeton University Press.Find this resource:
Eckstein, H. (1975). Case study and theory in political science. In F. I. Greenstein & N. W. Polsby (Eds.), Handbook of political science: Vol. 7. Political science: Scope and theory (pp. 94–137). Reading, MA: Addison-Wesley.Find this resource:
Elster, J. (1978). Logic and society: Contradictions and possible worlds. New York: Wiley.Find this resource:
Fischoff, B. (1975). Hindsight is not equal to foresight: The effect of outcome knowledge on judgment under uncertainty. Journal of Experimental Psychology: Human Perception and Performance, 1, 288–299.Find this resource:
Fogel, R. (1964). Railroads and American economic growth: Essays in econometric history. Baltimore, MD: Johns Hopkins University Press.Find this resource:
Gaddis, J. L. (1987). The long peace: Inquiries into the history of the Cold War era. New York: Oxford University Press.Find this resource:
George, A. L., & Bennett, A. (2005). Case studies and theory development. Cambridge, MA: M.I.T. Press.Find this resource:
George, A. L., & Simmons, W. E. (Eds.). (1994). The limits of coercive diplomacy. Boulder, CO: Westview.Find this resource:
George, A. L., & Smoke R. (1974). Deterrence in American foreign policy: Theory and practice. New York: Columbia University Press.Find this resource:
Gilpin, R. (1981). War and change in world politics. New York: Cambridge University Press.Find this resource:
Harvey, F. (2011). Explaining the Iraq war: Counterfactual theory, logic and evidence. Cambridge, U.K.: Cambridge University Press.Find this resource:
Hawkins, S. A., & Hastie, R. (1990). Hindsight: Biased judgments of past events after the outcomes are known. Psychological Bulletin, 107, 311–327.Find this resource:
Hawthorn, G. (1991). Possible worlds: Possibility and understanding in history and the social sciences. Cambridge, U.K.: Cambridge University Press.Find this resource:
Herrmann, R. K., & Lebow, R. L. (2001). Policymakers and the Cold War’s end: Micro and macro assessments of contingency. Cold War History Bulletin, 12–13, 337–340.Find this resource:
Herwig, H. H. (2006). Hitler wins in the east but still loses World War II. In P. E. Tetlock, R. L. Lebow, & G. Parker, Unmaking the West: “What-if” scenarios that rewrite world history (pp. 323–362). Ann Arbor: University of Michigan Press.Find this resource:
Herz, J. (1959). International politics in the atomic age. New York: Columbia University Press.Find this resource:
Hopf, T. (1994). Peripheral visions: Deterrence theory and American foreign policy, 1965–1990. Ann Arbor: University of Michigan Press.Find this resource:
Jervis, R. (1979). Perception and misperception in international relations. Princeton, NJ: Princeton University Press.Find this resource:
Kahneman, D., Slovic P., & Tversky, A. (1982). Judgment under uncertainty: Heuristics and biases. New York: Cambridge University Press.Find this resource:
Keohane, R. O. (1984). After hegemony. Princeton, NJ: Princeton University Press.Find this resource:
Khong, Y. F. (1992). Analogies at war: Korea, Munich, Dien Bien Phu, and the Vietnam decisions of 1965. Princeton, NJ: Princeton University Press.Find this resource:
Landman, J. (1992). Regret and elation following action and inaction: Affective responses to positive versus negative outcomes. Personality and Social Psychology Bulletin, 13, 524–536.Find this resource:
Lebow, R. N. (1981). Between peace and war: The nature of international crisis. Baltimore, MD: Johns Hopkins University Press.Find this resource:
Lebow, R. N. (1983). Miscalculation in the South Atlantic: British and Argentine intelligence failures in the Falkland crisis. Journal of Strategic Studies, 6, 1–29.Find this resource:
Lebow, R. N., & Stein, J. G. (1994). We all lost the Cold War. Princeton, NJ: Princeton University Press.Find this resource:
Lebow, R. N. (2010). Forbidden fruit: Counterfactuals and international relations. Princeton, NJ: Princeton University Press.Find this resource:
Lebow, R. N. (2014). Archduke Franz Ferdinand Lives! A world without world war I. London: Palgrave Macmillan.Find this resource:
Lebow, R. N. (2017). Back from the brink: Conflict management and resolution. London: Palgrave Macmillan.Find this resource:
Lebow, R. N., & Stein, J. L. (1996). Back to the past: Counterfactuals and the Cuban missile crisis. In P. E. Tetlock & A. Belkin (Eds.), Counterfactual thought experiments in world politics (pp. 119–148). Princeton, NJ: Princeton University Press.Find this resource:
Mandel, D. R. (2003). Counterfactuals, emotions and context. Cognition and Emotion, 17, 139–159.Find this resource:
Mueller, J. (1989). Retreat from doomsday: The obsolescence of major war. New York: Basic.Find this resource:
Roese, N. J., & Olson, J. M. (1995). Counterfactual thinking: A critical overview. In N. J. Roese & J. M. Olson (Eds.), What might have been: The social psychology of counterfactual thinking (pp. 1–56). Mahwah, NJ: Erlbaum.Find this resource:
Ross, L. M., Lepper, R., Strack, F., & Steinmetz, J. (1977). Social explanation and social expectation: Effects of real and hypothetical explanations on subjective likelihood. Journal of Personality and Social Psychology, 35, 817–829.Find this resource:
Tetlock, P. E., & Belkin, A. (1996). Counterfactual thought experiments in world politics. In P. E. Tetlock & A. Belkin (Eds.), Counterfactual thought experiments in world politics (pp. 1–38). Princeton, NJ: Princeton University Press.Find this resource:
Tetlock, P. E., & Belkin, A. (Eds.). (1996). Counterfactual thought experiments in world politics: Logical, methodological and psychological perspectives. Princeton, NJ: Princeton University Press.Find this resource:
Tetlock, P. E., & Lebow, R. L. (2002). Poking counterfactual holes in covering laws: Cognitive styles and political learning. American Political Science Review, 95, 829–843.Find this resource:
Truman, H. S. (1956). Memoirs: Vol. 2. 1946–1952: Years of trial and hope (p. 332). New York: Doubleday.Find this resource:
Waltz, K. N. (1976). Theory of international politics. Reading, MA: Addison-Wesley.Find this resource:
Weber, Max (1988). Die ‘Objektivität’ sozialwissenschaftlicher und sozial politischer Erkenntnis. In Gesammelte Aufsätze zur Wissenschaftslehre (pp. 146–214). Tübingen, GE: Mohr Siebeck. Original work published in 1904.Find this resource: