FIRC is also distinct from state creation. Newly liberated states or states created by secession do not qualify as instances of regime change because there is no sovereign regime to depose. I also exclude attempts to eliminate nonstate actors, such as the so-called Islamic State, because these are not sovereign states ruled by internationally recognized governments. In sum, whether it succeeds or fails, regime change constitutes an attempt by a state to replace the leadership of a sovereign state to which it intends to restore sovereignty. Though the foreign power’s aim may be to produce policy change, its purpose is to install leaders more willing to embrace that change.
Method and Case Selection
I use several methods to establish the logic of my argument and test its empirical validity. The argument itself is based on a game theoretic model. To keep the argument accessible to readers unfamiliar with game theory, I lay out the informal version of the argument in Chapter 1 and present the model in Appendix 2. I also derive hypotheses from my argument, which I test with both quantitative and qualitative evidence alongside alternative arguments. For the statistical analysis, I test my main hypothesis—that domestic opposition in a target state increases the likelihood of FIRC—on a data set that includes 133 cases of attempted FIRC (see Appendix 1). These tests help establish the generalizability of my argument, showing its ability to explain a large number of cases, even when controlling for alternative explanations.
Statistical tests, however, are less effective in testing an argument’s causal logic. For this, I rely on a series of case studies. If my argument is correct, I should find evidence that when leaders faced significant domestic opposition, they resisted complying with foreign demands that could put their political power at risk. I should also find that, as long as the opposition was not more opposed to the foreign power than the leader, the foreign power attempted to use the opposition to overthrow the leader. When the leader did not face significant domestic opposition, I should find that the foreign power agreed to some form of settlement. This settlement may have taken a variety of forms, ranging from one that involved concessions from the targeted leader to one that resembled the status quo or even entailed concessions from the foreign power. Findings that would challenge my theory include evidence that leaders facing strong domestic opposition acquiesced to politically costly demands without an equally strong incentive from the foreign power to do so. Evidence that the foreign power rejected regime change and pursued negotiations in such instances would also challenge my theory. Additionally, the theory would also be falsified if a foreign power attempted to overthrow a targeted leader without the existence of a strong, friendly domestic opposition.
I use a most-similar case design in which I analyze similar events that share a number of features but differ in their outcome—one ending in a settlement, the other in regime change.39 First, I compare the response of President Dwight Eisenhower’s administration to liberal governments in Guatemala and Bolivia in 1954. I next look at the Soviet Union’s response to nationalist governments in Poland and Hungary in 1956. Finally, I examine the US decisions to impose regime change on Iraq in 2003 but negotiate with Libya, only to pursue regime change in Libya eight years later.
By choosing cases that share numerous traits, I can test my argument’s causal logic while controlling for alternative hypotheses. For example, in each case study, the targeted leaders’ regime types and ideologies are similar, which suggests that neither factor can explain the decision to pursue FIRC in these cases.40 The target state’s geostrategic location (i.e., within or outside the imposing state’s sphere of influence) is also similar in each case, which suggests that stronger powers might be just as likely to negotiate as to overthrow leaders within (or outside) these regions. Finally, the same individual(s) made policy in the stronger state during roughly the same time period. Because it is unlikely that policymakers’ personalities, views, biases, or domestic pressures varied much during the time frames involved, it is also unlikely that these influenced the divergent outcomes.
Variation across the cases also allows me to introduce additional controls. I have paired states with different regime types to control for arguments suggesting that certain combinations are more likely to lead to FIRC. In the studies of American involvement in Bolivia and Guatemala, a democracy squared off against two democratically elected leaders. In the cases of the Soviet responses to Poland and Hungary, the more powerful state was a nondemocracy, whereas its targets were popularly supported communist leaders. In the study of US responses to Libya and Iraq, the stronger nation was democratic and its targets were not. The cases also vary in terms of the structure of the international system. The first two cases occurred during the Cold War, a period of bipolarity, whereas the last took place in the post–Cold War era, a period of unipolarity in which the United States faced no peer competitors. The cases vary as to the timing of the decision to pursue regime change too. In the first and last cases, the decision to pursue FIRC in one instance preceded the decision to negotiate in the other. In the second case, the decision to pursue negotiations preceded the decision to pursue FIRC. Finally, to maximize the range of cases, I have chosen ones that varied in the amount of force used and the type of regime change pursued. The first case involves a covert operation aimed at partial regime change. The second involves a military invasion aimed at partial regime change, and the third set includes two cases aimed at full regime change, one involving a military invasion and the other a foreign-backed insurgency. Table 1 shows the three case studies and the control variables.
Another element dictating case selection is the availability of primary source materials. Access to meeting minutes, memos, and various other governmental documents allows for better insight into the policies as they were conceived at the time. The CIA and State Department have released a wide variety of materials related to the deliberations of the most central players in Guatemala and Bolivia.41 The minutes of the Central Committee of the Communist Party of the Soviet Union (CPSU CC) Presidium are available for the discussions about Poland and Hungary.42 They shed light on how Soviet decision makers weighed their policy options during the two uprisings. Finally, in the aftermath of the US invasion of Iraq, researchers acquired access to translated and transcribed recordings of high-level meetings between Saddam Hussein and his top advisers.43 Access to all these materials enhances our ability to interpret the behaviors observed.
Outline of the Book
Chapter 1 presents my argument. I explain why strong states opt to remove foreign leaders or regimes in weak states rather than negotiate settlements that would allow them to retain power. I also explain how targeted leaders respond when they know another nation seeks to depose them. I end with a discussion of alternative arguments, from which I derive testable hypotheses. In Chapter 2, I expand my argument to explain how stronger powers choose between overthrowing the target state’s leaders and overhauling its domestic institutions. I also address the costs and risks associated with these different forms of regime change and why decision makers sometimes anticipate regime change to be cheaper than it actually is.
In the chapters that follow, I subject my argument to a series of empirical tests. Chapter 3 presents the statistical tests of my hypothesis that the probability of FIRC increases with the strength of the target’s domestic opposition. I also discuss the quantitative measures that I use as proxies for the opposition’s strength and the variables for alternative arguments.
Table 1. Foreign-Imposed Regime Change Case Studies: Within-Case and Cross-Case Controls