Democracy, Transparency, and Secrecy in Crisis

Democracy, Transparency, and Secrecy in Crisis Abstract Why is it that some democracies are able to effectively engage in secrecy and mobilize their forces privately, while others are unable to keep enough information secret to effectively carry out such an attack? Recent work by Colaresi (2014) suggests that a democracy's ability to keep information secret depends on its retrospective oversight institutions, which allow democracies to keep information secret in the present with the promise of revealing it to the public in the future. Though Colaresi (2014) finds that states with retrospective oversight have greater military spending, more support for military action, and more success in crises, to our knowledge there has yet to be empirical support for the argument that democracies with greater retrospective oversight can actually conduct foreign policy with greater opacity than those without retrospective oversight. In this research note we empirically explore this question, using data on private mobilizations by democracies with varying levels of retrospective oversight. This gives us greater insight into strategic military choices and shows how military effectiveness can be affected by political institutions. I would think that Mr. Khrushchev operating a totalitarian system, which has many advantages as far as being able to move in secret, and all the rest—there is a terrific disadvantage not having the abrasive quality of the press applied to you daily, to an administration, even though we never like it, and even though we wish they didn't write it, and even though we disapprove, there isn't any doubt that we could not do the job at all in a free society without a very, very active press. —John F. Kennedy, December 1962 interview with Sander Vanocur, NBC Despite conventional wisdom that autocracies have an inherent advantage in the use of secrecy in foreign policy, previous work on the relationship between transparency and secrecy, specifically Colaresi (2012, 2014), finds that democracy and secrecy can co-exist.1 Democracies’ retrospective oversight institutions (ROI) allow the public to give the democratic leader leeway to conduct policy in secret, with the expectation that this information will be made public in the future.2 This leads to some democracies, those with institutions such as freedom of information laws, being able to better engage in secrecy in foreign policy than others. Though Colaresi (2014) finds that states with retrospective oversight have greater military spending, more support for military action, and more success in crises, to our knowledge there has yet to be empirical support for the argument that democracies with greater retrospective oversight can actually be less transparent in conducting foreign policy than those with less retrospective oversight. In this research note, we empirically explore this question by studying the relationship between a democracy's level of retrospective oversight and its use of private mobilization in crisis.3 If Colaresi's (2012, 2014) argument is correct, we should find that not only is there variation in the willingness and ability of democracies to use secrecy but that this variation will depend on their institutional set-up. An example of an effective use of secrecy by a democracy is that of Israel during the 1956 Suez Crisis. Israel, with the support of the British and French governments, launched an attack against Egypt on October 29, beginning with IDF paratroopers seizing the Mitla Pass and continuing with two other battalions joining them within twenty-four hours to seize a variety of key Egyptian points along the Gulf of Suez. While the Egyptians were not surprised by the fact that Israel had chosen to attack, both the location and timing of the attack were indeed a surprise to them. Based on intentional deception by the Israelis, both the Egyptians and the Americans were expecting Israel to strike at the Israeli-Jordanian border (Knorr and Morgan 1983). In fact, Israeli intelligence intentionally bolstered this perception, with Israeli reservists being told that they were being mobilized as a response to Iraqi mobilizations into Jordan and stories being planted in the press about a potential incursion into Jordan (Betts 1982). This deception was achieved largely by involving only the smallest number of top Israeli leaders (as well as those of the British and French) in the planning of the attack (Bar-On 1989). Though in the 1950s FOI laws were uncommon in the international system (the United States’ Freedom of Information Act of 1966 was one of the first of its kind), Israel did indeed have some oversight institutions in place at the time (Berliner 2014). Israel's formal Freedom of Information Act did not pass until 1998, but its legislature, the Knesset, had some oversight powers in 1956. The Knesset was able to demand that the executive provide it with additional information through its committee system or in the full session of the body. As stated by Israel's penal code (Part II, Ch. 7, Article 4, Section 113), it is the Knesset's Foreign Affairs Committee that approves the designation of information as “secret information” and thus allows the executive to keep it from the public. This approval must be then made public through the Knesset's official legal journal, Reshumot (Laws of the State of Israel 1983). Though its committee system has had some inefficiencies, it is still a way in which the legislature has oversight power over the executive (Friedberg 2008, 2011, 2013). In contrast, there are cases in which democratic leaders could have used a private mobilization, and might have benefitted from doing so, but did not mobilize privately. Colaresi (2014), for example, points to the case of France in the interwar years of 1934 to 1939, which, faced with the German threat of the militarization of the Rhineland, failed to mobilize against the German army and adequately react to intelligence that showed increased German bellicosity. As Colaresi (2014) argues, the French government, which did not pass freedom of information laws until 1978, was facing a great deal of distrust from its civilian population when it came to military matters, and thus would have been unable to obtain the resources necessary to mobilize privately against the Germans. At the time, even a decision to extend military service from one to two years was controversial, and thus France found itself without enough forces to complement fortifications (the Maginot line) and counter potential German surprise attacks (Hughes 2006). In the remainder of this research note, we empirically test the theoretical relationship between retrospective oversight and transparency and the use of secrecy in conflict by democracies, as originally proposed by Colaresi (2012, 2012). In addition, this work contributes to the body of literature that identifies the institutional characteristics that affect the combat effectiveness of democracies (Reiter and Stam 2002, Avant 2007, among others). It thus provides further support for the study of conflict outcome determinants from an institutional approach. Democracy, Retrospective Oversight, and Secrecy Despite a significant body of literature showing that democracies are more likely to win the conflicts they fight (Lake 1992; Bennett and Stam 1996; Bennett and Stam 1998; Reiter and Stam 2002; Bueno de Mesquita et al. 2005), democracies have traditionally been thought to be at a military disadvantage in the effective use of secrecy (Almond 1956). Democracies are believed to have an increased level of transparency that can make it difficult for them to keep secrets from their population, the media, and eventually other states.4 One expectation is that when democratic states attempt to carry out secret military operations, they are not effective at them (Almond 1956).5 An alternate expectation is that given these transparency costs, democracies do not actually engage in transparency when it comes to national security matters (Desch 2002; Downes 2009; Schuessler 2010). There are a variety of reasons for why a state would choose to keep information secret, many of them relating to the military advantage that it can allow them. As argued by Colaresi (2014), secrecy can allow states to deceive their opponents and thus suppress their capabilities. Slantchev (2010) also provides a theoretical explanation for why states in a crisis have an incentive to “feign weakness” and not engage in costly signaling in order to maintain the strategic advantage that comes from mobilizing privately in the event of actual military force. He goes as far as to state that this places democratic regimes, which cannot hide their resolve, at a disadvantage in bargaining. Lindsey (2015) shows in a crisis-bargaining model that a consideration of military strategy helps in understanding why both strong and weak states have an interest in keeping information about military strategy and capabilities secret. Secrecy, as a force multiplier, can increase the combat capability of the side using it, as well as prevent the opponent from using its capabilities efficiently (von Clausewitz, cited in Betts 1982; Pilisuk 1984; Colaresi 2014). Further, secrecy allows leaders to preserve their ability to act during conflict without incurring audience costs from their populations, which allows them a broader range of options (Sagan and Suri 2003; Carson 2016). This is particularly true of democracies, which can face pressures from their publics to escalate a conflict (Carson 2016). Democracies can use secret policy actions in a way that conveys a signal to local allies and/or strategic adversaries without committing themselves. When states are transparent to their publics in sharing information about, for example, their militaries, the state's enemies in the international system are also able to acquire this information (Ben-Zvi 1976; Lai 2004). Thus, states incur what Colaresi (2014) refers to as transparency costs. Transparency costs refer to the fact that once states reveal information about a policy, that policy becomes less useful to them.6 Democracies have traditionally been found to be more transparent than autocracies (Hollyer, Rosendorff, and Vreeland 2011). If this is indeed true, and we accept that transparency costs exist, it is difficult to understand how it is that democracies are more effective at fighting conflicts than autocracies are (Reiter and Stam 2002). Colaresi (2014) resolves this problem by suggesting that executives in democracies do indeed classify information and keep it secret from their publics but that they can do so without losing accountability to the population. He argues that they do so through security accountability institutions, which allow the public to scrutinize the actions of the government in the future, even if in the present these actions are kept secret (some examples are freedom of information acts, investigative committees, and freedom of the press). Rosato (2003) similarly states that even though democracies are generally more transparent than autocracies, anecdotal evidence suggests that this does not prevent the use of secrecy. He argues that democratic politics still allow for secrecy during times of crisis and that expectations about transparency are dropped during wartime. This allows the government to mobilize their resources in secret while also giving the public the opportunity to later verify whether resources were indeed used for the intended purpose (Rosato 2003; Colaresi 2014; Fenster 2015; Birchall 2015). Beyond the first-mover advantage secrecy creates, it can thus allow the government to send a credible signal (through a mobilization) to certain actors (such as local allies or even adversaries with advanced intelligence capabilities) while also keeping it private from others, such as domestic or foreign publics (Sagan and Suri 2003; Hansen, Christensen, and Flyverbom 2015; Carson 2016; Carson and Yahri-Milo 2017).7 Though democracy has often been used as a proxy for transparency (Leeds and Davis 1997; Miller 1999; Schultz 1999; Mitchell and Prins 2004), previous work on transparency has shown that the two are not perfectly correlated and that even within democracies there can be variation in levels of external transparency (Bell 2013; Bell et al. 2014). There is also variation across democracies with regards to the amount of retrospective oversight institutions they have (Colaresi 2014). Thus, those states that do not have security oversight institutions will have to pay transparency costs in order to get their publics’ support for their foreign policies. Without a mechanism to hold the government accountable in the future and ensure that funds were not used in a way that violates the public interest, the public will not be willing to give the executive the resources needed to carry out a policy without a prior justification. In a democracy with retrospective oversight, the public is willing to give the executive the freedom to use secrecy to maneuver in foreign policy, knowing that if resources are misused, the public will find out about it in the future and punish the executive (Colaresi 2014). This leads to the conclusion that democracies that commit to reveal information to their publics in the future are able to conduct foreign policy with greater secrecy in the present. Thus, we expect that democratic states that have more retrospective oversight will be more able to use secrecy in foreign policy. Existing research has assumed this to be the case and drawn out implications of retrospective oversight on war outcomes, that is, success (Colaresi 2012, 2014). It is also important to consider the intermediate step of whether states with retrospective oversight have a greater ability to actually conduct their foreign policy actions in private. There are a number of foreign policy actions that leaders may see some advantage in conducting in private. We focus on the mobilization of forces. Private mobilization requires an effective use of secrecy, and we expect that democracies with more retrospective oversight will be more likely to mobilize privately. While mobilization for war is a rare event, it potentially has higher stakes than any other foreign policy action a state takes and is harder to hide than many other foreign policy actions. If retrospective oversight institutions provide the ability to mobilize in private, we should have some confidence that these institutions allow greater secrecy across foreign policy making. We thus derive the following hypothesis: H1: The more retrospective oversight a democracy has, the more likely it is to privately mobilize during a crisis. In the section that follows, we will empirically test this hypothesis using data on both the retrospective oversight of democracies and on private mobilizations during crises. Research Design Our sample consists of all crisis-directed dyads between the years 1970 and 1994 (these temporal restrictions are due to data availability of both our dependent and independent variables). We use data from the International Crisis Behavior (ICB) dataset to identify when a crisis occurs (Brecher and Wilkenfeld 1997). The ICB project defines a crisis as an event that produces “a heightened probability of military hostilities.” Because we use Lai's measure of private mobilization, we consider only crises that are not intrastate or intrawar.8 Even though we are modeling the choice of a single state to privately mobilize, we construct the dataset with a crisis-directed dyad unit of analysis (Hewitt 2003). This setup allows us to accommodate the directional aspect of private mobilization during crises. The crisis-directed dyad unit of analysis also allows us to take into account characteristics of not only the (potential) mobilizing state but also of the state that it is involved in a crisis with. The Dependent Variable, Private Mobilization Our dependent variable is the private mobilization of troops in crises, an indicator of an effective use of secrecy by the state. To measure private mobilizations, we use Lai's (2004) data on the actions taken by states during crises. He conceptualizes private mobilizations as the military preparations that a state takes for conflict and not for diplomatic purposes. Lai identifies whether private mobilization has occurred by looking for two specific behaviors in the crisis.9 First, a state is coded as privately mobilizing if it “does not make a verbal threat or claim and that same state attempts to conceal its mobilization or that state's crisis action is to attempt a first strike (221).” Second, a state is also coded as privately mobilizing if the “state's only crisis activity is to launch a military strike against the opposing state (221).” To code these actions Lai (2004) uses both the ICB data's case summaries, which were the ones used to create the full ICB dataset, as well as the secondary sources that are referenced in the ICB case summaries (Brecher and Wilkenfeld 1997). The variable is coded dichotomously, taking on a value of 1 in cases that experience private mobilization and 0 otherwise. Given our dichotomous dependent variable, we estimate a series of probit models, with the error terms clustered by country.10 Measuring Retrospective Oversight To measure retrospective oversight, we rely on the retrospective security policy information institutions (r-SPII) index created by Colaresi (2012). The dataset includes information on democracies (considered as states with a Polity score greater than 6). The index is a composite measure of the existence and implementation of freedom of information laws, legislative oversight (the existence and efficacy of legislative committees that investigate national security), and the Freedom House civil liberties measure (rescaled to 0–1, with higher values indicating more liberties). The index ranges from 0 to 1, with higher values indicating greater retrospective oversight (Colaresi 2012, 2014). Control Variables We include in our analysis a series of control variables that we expect to have an effect on a state's probability of engaging in private mobilization during crisis. Given that our theory is about separating the effects of retrospective oversight from those of democracy we begin by including a measure of regime type in our models: the Polity IV combined democracy-autocracy score (Marshall, Jaggers, and Gurr 2004). We also control for contiguity, as we should expect that privately mobilizing troops for longer distances will be more difficult (Starr and Most 1976). This is generated from the Correlates of War Direct Contiguity Data, version 3.1 and is operationalized as a dichotomous variable equal to 1 if two states are separated by 150 miles of water or less or by a land or river border (Stinnett et al. 2002). Given that there can be strategic reasons to mobilize in private given the relative strength of the states involved in a crisis, we control for state capabilities. Using both states’ Comparative Index of National Capabilities (CINC) scores, we also include a measure of relative capabilities (Singer, Bremer, and Stuckey 1972). This is operationalized as State A's capabilities over the capabilities of both states in the dyad. This can vary from 0 to 1, with low values indicating that State A is weak, and high values indicating that State B is weak. Finally, with the understanding that allied states will be less likely to engage in private mobilizations against each other, we include the COW alliance variable, indicating whether the two states in a dyad share any alliances (Gibler 2009). Results and Discussion Table 1 includes estimates from four models in which private mobilization is the dependent variable. Model 1 includes only the retrospective oversight measure as an independent variable, while in Model 2 we include all of the control variables. Table 1. Probit models: Retroactive oversight in democracies and private mobilization, 1970–94.   (1)  (2)  (3)  (4)  Retrospective Oversight  0.277  0.895**        (0.412)  (0.420)      Polity 2 Score, State A    −0.0169    0.00490      (0.0247)    (0.0316)  Contiguous Dyad    0.551***    0.558***      (0.163)    (0.182)  Alliance    0.0907    0.117      (0.0875)    (0.0956)  Power Ratio    0.174    0.280      (0.305)    (0.321)  Freedom of Information Index      −0.721  −0.585        (0.814)  (0.755)  Legislative Oversight Index      1.036**  1.153**        (0.490)  (0.467)  Civil Liberties Index      0.0258  0.0463        (0.0724)  (0.114)  Constant  −1.721***  −2.587***  −1.910***  −2.946***    (0.157)  (0.469)  (0.414)  (0.723)  Observations  504  498  459  453  pr2            (1)  (2)  (3)  (4)  Retrospective Oversight  0.277  0.895**        (0.412)  (0.420)      Polity 2 Score, State A    −0.0169    0.00490      (0.0247)    (0.0316)  Contiguous Dyad    0.551***    0.558***      (0.163)    (0.182)  Alliance    0.0907    0.117      (0.0875)    (0.0956)  Power Ratio    0.174    0.280      (0.305)    (0.321)  Freedom of Information Index      −0.721  −0.585        (0.814)  (0.755)  Legislative Oversight Index      1.036**  1.153**        (0.490)  (0.467)  Civil Liberties Index      0.0258  0.0463        (0.0724)  (0.114)  Constant  −1.721***  −2.587***  −1.910***  −2.946***    (0.157)  (0.469)  (0.414)  (0.723)  Observations  504  498  459  453  pr2          Standard errors in parentheses. Two-tailed significance tests used: * p < .10, ** p < .05, *** p < .01. View Large Table 1. Probit models: Retroactive oversight in democracies and private mobilization, 1970–94.   (1)  (2)  (3)  (4)  Retrospective Oversight  0.277  0.895**        (0.412)  (0.420)      Polity 2 Score, State A    −0.0169    0.00490      (0.0247)    (0.0316)  Contiguous Dyad    0.551***    0.558***      (0.163)    (0.182)  Alliance    0.0907    0.117      (0.0875)    (0.0956)  Power Ratio    0.174    0.280      (0.305)    (0.321)  Freedom of Information Index      −0.721  −0.585        (0.814)  (0.755)  Legislative Oversight Index      1.036**  1.153**        (0.490)  (0.467)  Civil Liberties Index      0.0258  0.0463        (0.0724)  (0.114)  Constant  −1.721***  −2.587***  −1.910***  −2.946***    (0.157)  (0.469)  (0.414)  (0.723)  Observations  504  498  459  453  pr2            (1)  (2)  (3)  (4)  Retrospective Oversight  0.277  0.895**        (0.412)  (0.420)      Polity 2 Score, State A    −0.0169    0.00490      (0.0247)    (0.0316)  Contiguous Dyad    0.551***    0.558***      (0.163)    (0.182)  Alliance    0.0907    0.117      (0.0875)    (0.0956)  Power Ratio    0.174    0.280      (0.305)    (0.321)  Freedom of Information Index      −0.721  −0.585        (0.814)  (0.755)  Legislative Oversight Index      1.036**  1.153**        (0.490)  (0.467)  Civil Liberties Index      0.0258  0.0463        (0.0724)  (0.114)  Constant  −1.721***  −2.587***  −1.910***  −2.946***    (0.157)  (0.469)  (0.414)  (0.723)  Observations  504  498  459  453  pr2          Standard errors in parentheses. Two-tailed significance tests used: * p < .10, ** p < .05, *** p < .01. View Large Once we control for all of the other factors that we expect to influence the probability of a private mobilization during crisis, the coefficient on the retrospective oversight coefficient is positive and significant at the .05 level. This finding is supportive of Hypothesis 1. Although, to our knowledge, there is no previous published research examining private mobilization as a dependent variable, the results on the control variables are somewhat surprising. First, it appears that the level of democracy is not statistically related to private mobilization. Second, neither military capabilities nor the existence of an alliance are statistically related to the decision to privately mobilize, once states find themselves in a crisis. Finally, dyads of contiguous states are more likely to experience a private mobilization. As described above, the r-SPII index is composed of three different elements (Freedom of Information Laws, Legislative Oversight, and Civil Liberties). We also try decomposing the index into its various components as a way to further understand what it is about retrospective oversight in democracies that allows them to effectively engage in secrecy in their foreign policy. In Models 3 and 4 of Table 1, we present the results of two more probit models. In Model 3, we include only the three different components of the r-SPII index as independent variables. In Model 4, we again include all of the control variables that are expected to have an effect on the probability of a private mobilization occurring. In both Models 3 and 4, the only component of the r-SPII index that achieves statistical significance is the Legislative Oversight Index. The coefficient for this variable is positive and significant at the .05 level. This is not surprising, as legislative institutions are likely in a more credible position to demand information from the executive at a later date compared to the ability of the public through FOI laws. Figures 1 and 2 show the predicted probabilities generated from the models estimated above, with 95 percent confidence intervals. Figure 1 shows the relationship between the full retrospective oversight scale and private mobilization (from Model 2), and Figure 2 shows the relationship between legislative oversight index and private mobilization (from Model 4). Both figures illustrate the expected positive relationship. In Figure 2, with legislative oversight set to its mean (.08), the probability of a private mobilization is 5.7 percent.11 A change from the mean level of legislative oversight to one standard deviation above the mean (.28) leads to an increase in the probability of private mobilization to 8.9 percent, an increase of 56 percent. An increase to two standard deviations above the mean (.48) of legislative oversight, increases the probability of private mobilization to 13.2 percent, more than doubling the probability compared to the mean level case.12 Figure 1. View largeDownload slide Retrospective oversight and private mobilization Figure 1. View largeDownload slide Retrospective oversight and private mobilization Figure 2. View largeDownload slide Legislative oversight and private mobilization Figure 2. View largeDownload slide Legislative oversight and private mobilization One potential issue to address regarding these results is the possibility that retrospective oversight institutions may be biasing our sample. In other words, it is possible that we are more likely to have information on secret actions carried out by democracies with ROI than by those without, while democracies without ROI may have carried out secret actions that are never known.13 While this is certainly possible, there are three factors that alleviate this concern for this particular study. The first is that our sample data end in 1994, which is more than twenty years from the writing of this manuscript. This means that even in democracies without ROI, it is most likely that enough time has gone by that most of these actions no longer represent a national security concern and have either been declassified by the government or investigated by the media. The second is that we restrict our analysis to private mobilizations of the state's military during crises, that is, the concentration of military forces in one area (Lai 2004). These are the types of actions that will certainly give a state a first-mover advantage in conflict but that are difficult to keep indefinitely concealed. As Carson and Yahri-Milo (2017) note, very few military mobilizations can actually be kept completely secret from all audiences (for example, local allies might observe the military mobilization). Private military mobilizations can be contrasted with other covert actions that could indeed be kept secret by democracies without ROI, such as covert special forces operations, assassination attempts, or secret alliances. Finally, it is also important to keep in mind that the implication of Colaresi's argument about ROI is not that democracies without ROI are immune to being transparent. Instead, it is argued that these governments are going to be expected by their populations to provide information about security activities as they happen. The democracies without ROI are expected to reveal information earlier than the democracies with ROI. ROI allows the democratic leader to hide information in the present with the promise to reveal information down the road. If we were analyzing a sample that included autocracies, this would be a bigger issue, as autocrats without ROI would be unlikely to ever have the information revealed. By focusing on the differences across consolidated democratic regimes, we believe this concern is diminished. Conclusion In our analysis we find that among democracies, states that have retrospective oversight institutions are more likely to use private mobilizations during a crisis. This is consistent with Colaresi's (2012, 2014) argument that retrospective oversight institutions allow states to avoid incurring transparency costs, and thus enables them to use secrecy effectively while engaging in foreign policy actions. Colaresi's (2012, 2014) previous work shows that democracies with retrospective oversight are indeed more effective at obtaining their desired goals in their international interactions. Yet his theoretical argument for why this occurs, that democracies with ROI are better able to use secrecy in their foreign policy, remains untested. In theory, retrospective oversight could lead to more successful avenues, for example, by making leaders more risk averse and less willing to embark on risky conflicts if they have to be held accountable to them in the future.14 This project, by showing a positive correlation between ROI and private mobilizations, provides support for Colaresi's (2012, 2014) theoretical argument and for the idea that some democracies are better able to perform in conflict due to their ability to use secrecy effectively. This research also has important implications for our understanding of the role that democratic politics play in crisis behavior. Transparency and democracy are concepts that are routinely conflated; with the assumption made that democratic states are uniformly transparent. The models estimated in this paper show that retrospective oversight institutions lead to transparency having its own independent effect on crisis behavior apart from democratic regime type. Another finding to note is that when we decompose the r-SPII index, the only component that has a positive and statistically significant effect on private mobilization is legislative oversight. Future work on the institutional characteristics that allow states to mobilize their troops in private can look more deeply into this relationship. Why is it that giving the legislature greater oversight power allows the executive to mobilize the military in private? Why is this effect greater than that of FOI laws? The evidence in this study consistently demonstrates that democracies with retrospective oversight institutions are more likely to engage in private mobilization than those democracies without ROI. This has important implications for studies of bargaining and crisis escalation. Lai (2004) tells us that states are likely to mobilize publicly when they are actually seeking to signal information and come to a bargained settlement, while private mobilization occurs when states give up on bargaining and seek an advantage in actual war fighting. Recent work by Carson (2016) and Carson and Yahri-Milo (2017) (as well as by Sagan and Suri 2003) suggests that, beyond the military first-mover advantage they provide, private mobilizations may actually send a costly signal to select actors (such as local allies and/or adversaries with advanced intelligence agencies), while also allowing leaders the option of avoiding audience costs and bargaining with an opponent without being forced into escalation by their domestic audiences. Democracies with retrospective oversight institutions may thus be better able to avoid the escalation of conflict by being able to bargain with an opponent free of domestic pressures stemming from audience costs. Retrospective oversight institutions thus serve to widen the scope of available options a democracy has during conflict. This work also points to several future research questions. If retrospective oversight indeed allows for greater secrecy in conducting foreign policy, we should observe democratic states with retrospective oversight institutions also engaging in other forms of secret behavior, such as secret alliances and secret basing negotiations. Further work can explore how it is that retrospective oversight affects these other forms of foreign policy behavior. In addition, future work may use a case study approach to study states before and after the implementation of retrospective oversight institutions and see if their use of secrecy in their foreign policy behavior changes. Carla Martinez Machain is an associate professor of political science at Kansas State University. She received her PhD from Rice University in 2012. Her research specializations include conflict outcomes, security policy, and military cooperation. She was the recipient of a 2017 Minerva Grant from the US Department of Defense. Sam Bell is an associate professor of political science at Kansas State University. He received his PhD from Binghamton University in 2009. His research examines the causes of political violence. His work has previously appeared in the Journal of Politics, International Studies Quarterly, and the Journal of Conflict Resolution, among other outlets. Footnotes 1 We define transparency following Bell (2013) as “the inability of government leaders to hide government processes, decision-making and events taking place within a state from external audiences.” 2 Following Colaresi (2014), we define retrospective oversight institutions as those that “empower actors outside of the executive with tools to investigate national security issues and publicize information” and “allow for investigations into previously secret information and to reveal executive abuses of secrecy powers” (2014, 9). 3 Following Lai (2004), we define private mobilizations as the concentration of military forces in one area without the knowledge of the opposing state. 4 There is not complete uniformity of thought on the effect that democracy and transparency have on secrecy. See Kam (1988) and Finel and Lord (1999) for alternate views. 5 One example that often is used to illustrate this point is that of the Associated Press reporting on the Bay of Pigs invasion even before it had taken place. 6 We note that in some cases states will have an incentive to keep information from their publics in order to avoid audience costs but will count on local allies and/or strategic rivals with “sophisticated intelligence capabilities” to observe these actions and thus rely on them as a costly signal of their resolve (Carson and Yahri-Milo 2017, 125). 7 Carson and Yahri-Milo (2017, 125) argue that “covert action is rarely completely secret,” as parts of an action such as a private mobilization can be observed by the intelligence apparatus of local allies and/or even rivals. States are thus able to use a private mobilization as a costly signal of their resolve while also not incurring audience costs, which would be particularly costly for democracies (Carson 2016). 8 These are the only cases that can take on a value of 1 for our dependent variable, because they are the only cases coded by Lai (2004). 9 In the analysis presented in Lai (2004), the unit of analysis is the crisis. We obtained the state-level data from Lai. This allows us to identify which state in a crisis engaged in private mobilization. 10 The results remain substantively the same when we cluster on the dyad. 11 All other variables were set to means and modes. All predictions were generated with margins command in Stata 15. 12 Figure 1 illustrates a similar set of substantive predictions. 13 We thank an anonymous reviewer for pointing this out. 14 We thank an anonymous reviewer for pointing out this observation to us. References Almond G. A. 1956. “ Public Opinion and National Security Policy.” Public Opinion Quarterly  20 ( 2): 371– 78. Google Scholar CrossRef Search ADS   Avant Deborah. 2007. “ Political Institutions and Military Effectiveness: Contemporary United States and United Kingdom.” Creating Military Power: The Sources of Military Effectiveness , 80– 105. Bar-On Mordechai. 1989. “ David Ben-Gurion and the Sevres Collusion.” In Suez 1956: The Crisis and Its Consequences , edited by Louis William Roger, Owen Roger, 145– 60. Oxford: Clarendon. Google Scholar CrossRef Search ADS   Bell Sam R. 2013. “ What You Don't Know Can Hurt You: Information, External Transparency, and Interstate Conflict, 1982–1999.” Conflict Management and Peace Science  30 ( 5): 452– 68. Google Scholar CrossRef Search ADS   Bell Sam R., Chad Clay K., Murdie Amanda, Piazza James. 2014. “ Opening Yourself Up: The Role of External and Internal Transparency in Terrorism Attacks.” Political Research Quarterly  67 ( 3): 603– 14. Google Scholar CrossRef Search ADS   Bennett Scott D., Stam III Allan C.. 1998. “ The Declining Advantages of Democracy: A Combined Model of War Outcomes and Duration.” Journal of Conflict Resolution  42 ( 3): 344– 66. Google Scholar CrossRef Search ADS   Bennett D. Scott, Stam Allan C.. 1996. “ The Duration of Interstate Wars, 1816–1985.” American Political Science Review  90 ( 2): 239– 57. Google Scholar CrossRef Search ADS   Ben-Zvi Abraham. 1976. “ Hindsight and Foresight: A Conceptual Framework for the Analysis of Surprise Attacks.” World Politics  28 ( 3): 381– 95. Google Scholar CrossRef Search ADS   Berliner D., 2014. “ The Political Origins of Transparency.” The Journal of Politics  76 ( 02): 479– 91. Google Scholar CrossRef Search ADS   Betts Richard K. 1982. Surprise Attack: Lessons for Defense Planning . Brookings Institution Press. Birchall Clare. 2015. “ ‘Data. Gov-in-a-Box’ Delimiting Transparency.” European Journal of Social Theory  18 ( 2): 185– 202. Google Scholar CrossRef Search ADS   Brecher Michael, Wilkenfeld Jonathon. 1997. A Study of Crisis . Ann Arbor: University of Michigan Press. Google Scholar CrossRef Search ADS   Bueno de Mesquita Bruce, Smith Alastair, Siverson Randolph M., Morrow James D.. 2005. The Logic of Political Survival . The MIT Press. Carson Austin. 2016. “ Facing Off and Saving Face: Covert Intervention and Escalation Management in the Korean War.” International Organization  70 ( 1): 103– 31. Google Scholar CrossRef Search ADS   Carson Austin, Yarhi-Milo Keren. 2017. “ Covert Communication: The Intelligibility and Credibility of Signaling in Secret.” Security Studies  26 ( 1): 124– 56. Google Scholar CrossRef Search ADS   Colaresi M. 2012. “ A Boom with Review: How Retrospective Oversight Increases the Foreign Policy Ability of Democracies.” American Journal of Political Science  56 ( 3): 671– 89. Google Scholar CrossRef Search ADS   Colaresi M. 2014. Democracy Declassified: The Secrecy Dilemma in National Security . Oxford: Oxford University Press. Google Scholar CrossRef Search ADS   Desch Michael C. 2002. “ Democracy and Victory: Why Regime Type Hardly Matters.” International Security  27 ( 2): 5– 47. Google Scholar CrossRef Search ADS   Downes Alexander B. 2009. “ How Smart and Tough Are Democracies? Reassessing Theories of Democratic Victory in War.” International Security  33 ( 4): 9– 51. Google Scholar CrossRef Search ADS   Fenster Mark. 2015. “ Transparency in Search of a Theory.” European Journal of Social Theory  18 ( 2): 150– 67. Google Scholar CrossRef Search ADS   Finel Bernard I., Lord Kristin M.. 1999. “ The Surprising Logic of Transparency.” International Studies Quarterly  43: 315– 39. Google Scholar CrossRef Search ADS   Friedberg Chen. 2008. “ Legislative Oversight and the Israeli Committee System: Problems and Solutions.” Oversight  217. Friedberg Chen. 2011. “ From a Top-Down to a Bottom-Up Approach to Legislative Oversight.” The Journal of Legislative Studies  17 ( 4): 525– 44. Google Scholar CrossRef Search ADS   Friedberg Chen. 2013. “ Oversight by the State Control Committee in the Israeli Parliament: Form of Accountability Under Stress.” Israel Affairs  19 ( 2): 306– 20. Google Scholar CrossRef Search ADS   Gibler Douglas M. 2009. International Military Alliances, 1648–2008 . Washington, DC: CQ Press. Google Scholar CrossRef Search ADS   Hansen Hans Krause, Christensen Lars Thøger, Flyverbom Mikkel. 2015. “ Introduction: Logics of Transparency in Late Modernity Paradoxes, Mediation and Governance.” European Journal of Social Theory  18 ( 2): 117– 31 Google Scholar CrossRef Search ADS   Hewitt J. Joseph. 2003. “ Dyadic Processes and International Crises.” Journal of Conflict Resolution  47 ( 5): 669– 92. Google Scholar CrossRef Search ADS   Hollyer James R., Rosendorff B. Peter, Vreeland James Raymond. 2011. “ Democracy and Transparency.” Journal of Politics  73 ( 4): 1191– 205. Google Scholar CrossRef Search ADS   Hughes J. M., 2006. To the Maginot Line: The Politics of French Military Preparation in the 1920’s , Vol. 64. Cambridge, MA: Harvard University Press. Kam Ephraim. 1988. Surprise Attack: The Victim's Perspective . Cambridge, MA: Harvard University Press. Google Scholar CrossRef Search ADS   Kennedy J. F. 1962. “ Interview by S. Vanocur, Transcript.” NBC , https://www.jfklibrary.org/JFK/JFK-in-History/John-F-Kennedy-and-the-Press.aspx Knorr Klaus Eugen, Morgan Patrick M. 1983. Strategic Military Surprise: Initiatives and Opportunities . New Brunswick, NJ: Transaction Books. Lai Brian. 2004. “ The Effects of Different Types of Military Mobilization on the Outcome of International Crises.” Journal of Conflict Resolution  48 ( 2): 211– 29. Google Scholar CrossRef Search ADS   Lake David A. 1992. “ Powerful Pacifists: Democratic States and War.” American Political Science Review  86 ( 1): 24– 37. Google Scholar CrossRef Search ADS   Laws of the State of Israel (1948–), LC Call No. KMK 13.I874. 1983. Jerusalem: The Government Printer. Leeds Brett Ashley, Davis David R.. 1997. “ Domestic Political Vulnerability and International Disputes.” Journal of Conflict Resolution  41 ( 6): 814– 34. Google Scholar CrossRef Search ADS   Lindsey David. 2015. “ Military Strategy, Private Information, and War.” International Studies Quarterly  59 ( 4): 629– 40. Marshall Monty G., Jaggers Keith, Gurr Ted Robert. 2004. “Polity IV.”  College Park: University of Maryland. Miller Ross A. 1999. “ Regime Type, Strategic Interaction, and the Diversionary Use of Force.” Journal of Conflict Resolution  43 ( 3): 388– 402. Google Scholar CrossRef Search ADS   Mitchell Sara McLaughlin, Prins Brandon C.. 2004. “ Rivalry and Diversionary Uses of Force.” Journal of Conflict Resolution  48 ( 6): 937– 61. Google Scholar CrossRef Search ADS   Pilisuk Marc. 1984. “ Experimenting with the Arms Race.” Journal of Conflict Resolution  28 ( 2): 296– 315. Google Scholar CrossRef Search ADS   Reiter Dan, Stam Allan C.. Democracies at War . Princeton, NJ: Princeton University Press, 2002. Google Scholar CrossRef Search ADS   Rosato Sebastion. 2003. “ The Flawed Logic of the Democratic Peace Theory.” American Political Science Review  97 ( 4): 585– 602. Google Scholar CrossRef Search ADS   Sagan Scott D., Suri Jeremi. 2003. “ The Madman Nuclear Alert: Secrecy, Signaling, and Safety in October 1969.” International Security  27 ( 4): 150– 83. Google Scholar CrossRef Search ADS   Schuessler John M. 2010. “ The Deception Dividend: FDR's Undeclared War.” International Security  34 ( 4): 133– 65. Google Scholar CrossRef Search ADS   Schultz Kenneth A. 1999. “ Do Democratic Institutions Constrain or inform? Contrasting Two Institutional Perspectives on Democracy and War.” International Organization  53 ( 2): 233– 66. Google Scholar CrossRef Search ADS   Singer J. David, Bremer Stuart, Stuckey John. 1972. “ Capability Distribution, Uncertainty, and Major Power War, 1820–1965.” Peace, War, and Numbers  19– 48. Slantchev Branislav L. 2010. “ Feigning Weakness.” International Organization  64 ( 3): 357– 88. Google Scholar CrossRef Search ADS   Starr Harvey, Most Benjamin A.. 1976. “ The Substance and Study of Borders in International Relations Research.” International Studies Quarterly  581– 620. Stinnett Douglas M., Tir Jaroslav, Diehl Paul F., Schafer Philip, Gochman Charles. 2002. “ The Correlates of War (COW) Project Direct Contiguity Data, Version 3.0.” Conflict Management and Peace Science  19 ( 2): 59– 67. Google Scholar CrossRef Search ADS   © The Author(s) (2018). Published by Oxford University Press on behalf of the International Studies Association. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Foreign Policy Analysis Oxford University Press

Democracy, Transparency, and Secrecy in Crisis

Loading next page...
 
/lp/ou_press/democracy-transparency-and-secrecy-in-crisis-NlNaeRjMpK
Publisher
Wiley-Blackwell
Copyright
© The Author(s) (2018). Published by Oxford University Press on behalf of the International Studies Association.
ISSN
1743-8586
eISSN
1743-8594
D.O.I.
10.1093/fpa/ory004
Publisher site
See Article on Publisher Site

Abstract

Abstract Why is it that some democracies are able to effectively engage in secrecy and mobilize their forces privately, while others are unable to keep enough information secret to effectively carry out such an attack? Recent work by Colaresi (2014) suggests that a democracy's ability to keep information secret depends on its retrospective oversight institutions, which allow democracies to keep information secret in the present with the promise of revealing it to the public in the future. Though Colaresi (2014) finds that states with retrospective oversight have greater military spending, more support for military action, and more success in crises, to our knowledge there has yet to be empirical support for the argument that democracies with greater retrospective oversight can actually conduct foreign policy with greater opacity than those without retrospective oversight. In this research note we empirically explore this question, using data on private mobilizations by democracies with varying levels of retrospective oversight. This gives us greater insight into strategic military choices and shows how military effectiveness can be affected by political institutions. I would think that Mr. Khrushchev operating a totalitarian system, which has many advantages as far as being able to move in secret, and all the rest—there is a terrific disadvantage not having the abrasive quality of the press applied to you daily, to an administration, even though we never like it, and even though we wish they didn't write it, and even though we disapprove, there isn't any doubt that we could not do the job at all in a free society without a very, very active press. —John F. Kennedy, December 1962 interview with Sander Vanocur, NBC Despite conventional wisdom that autocracies have an inherent advantage in the use of secrecy in foreign policy, previous work on the relationship between transparency and secrecy, specifically Colaresi (2012, 2014), finds that democracy and secrecy can co-exist.1 Democracies’ retrospective oversight institutions (ROI) allow the public to give the democratic leader leeway to conduct policy in secret, with the expectation that this information will be made public in the future.2 This leads to some democracies, those with institutions such as freedom of information laws, being able to better engage in secrecy in foreign policy than others. Though Colaresi (2014) finds that states with retrospective oversight have greater military spending, more support for military action, and more success in crises, to our knowledge there has yet to be empirical support for the argument that democracies with greater retrospective oversight can actually be less transparent in conducting foreign policy than those with less retrospective oversight. In this research note, we empirically explore this question by studying the relationship between a democracy's level of retrospective oversight and its use of private mobilization in crisis.3 If Colaresi's (2012, 2014) argument is correct, we should find that not only is there variation in the willingness and ability of democracies to use secrecy but that this variation will depend on their institutional set-up. An example of an effective use of secrecy by a democracy is that of Israel during the 1956 Suez Crisis. Israel, with the support of the British and French governments, launched an attack against Egypt on October 29, beginning with IDF paratroopers seizing the Mitla Pass and continuing with two other battalions joining them within twenty-four hours to seize a variety of key Egyptian points along the Gulf of Suez. While the Egyptians were not surprised by the fact that Israel had chosen to attack, both the location and timing of the attack were indeed a surprise to them. Based on intentional deception by the Israelis, both the Egyptians and the Americans were expecting Israel to strike at the Israeli-Jordanian border (Knorr and Morgan 1983). In fact, Israeli intelligence intentionally bolstered this perception, with Israeli reservists being told that they were being mobilized as a response to Iraqi mobilizations into Jordan and stories being planted in the press about a potential incursion into Jordan (Betts 1982). This deception was achieved largely by involving only the smallest number of top Israeli leaders (as well as those of the British and French) in the planning of the attack (Bar-On 1989). Though in the 1950s FOI laws were uncommon in the international system (the United States’ Freedom of Information Act of 1966 was one of the first of its kind), Israel did indeed have some oversight institutions in place at the time (Berliner 2014). Israel's formal Freedom of Information Act did not pass until 1998, but its legislature, the Knesset, had some oversight powers in 1956. The Knesset was able to demand that the executive provide it with additional information through its committee system or in the full session of the body. As stated by Israel's penal code (Part II, Ch. 7, Article 4, Section 113), it is the Knesset's Foreign Affairs Committee that approves the designation of information as “secret information” and thus allows the executive to keep it from the public. This approval must be then made public through the Knesset's official legal journal, Reshumot (Laws of the State of Israel 1983). Though its committee system has had some inefficiencies, it is still a way in which the legislature has oversight power over the executive (Friedberg 2008, 2011, 2013). In contrast, there are cases in which democratic leaders could have used a private mobilization, and might have benefitted from doing so, but did not mobilize privately. Colaresi (2014), for example, points to the case of France in the interwar years of 1934 to 1939, which, faced with the German threat of the militarization of the Rhineland, failed to mobilize against the German army and adequately react to intelligence that showed increased German bellicosity. As Colaresi (2014) argues, the French government, which did not pass freedom of information laws until 1978, was facing a great deal of distrust from its civilian population when it came to military matters, and thus would have been unable to obtain the resources necessary to mobilize privately against the Germans. At the time, even a decision to extend military service from one to two years was controversial, and thus France found itself without enough forces to complement fortifications (the Maginot line) and counter potential German surprise attacks (Hughes 2006). In the remainder of this research note, we empirically test the theoretical relationship between retrospective oversight and transparency and the use of secrecy in conflict by democracies, as originally proposed by Colaresi (2012, 2012). In addition, this work contributes to the body of literature that identifies the institutional characteristics that affect the combat effectiveness of democracies (Reiter and Stam 2002, Avant 2007, among others). It thus provides further support for the study of conflict outcome determinants from an institutional approach. Democracy, Retrospective Oversight, and Secrecy Despite a significant body of literature showing that democracies are more likely to win the conflicts they fight (Lake 1992; Bennett and Stam 1996; Bennett and Stam 1998; Reiter and Stam 2002; Bueno de Mesquita et al. 2005), democracies have traditionally been thought to be at a military disadvantage in the effective use of secrecy (Almond 1956). Democracies are believed to have an increased level of transparency that can make it difficult for them to keep secrets from their population, the media, and eventually other states.4 One expectation is that when democratic states attempt to carry out secret military operations, they are not effective at them (Almond 1956).5 An alternate expectation is that given these transparency costs, democracies do not actually engage in transparency when it comes to national security matters (Desch 2002; Downes 2009; Schuessler 2010). There are a variety of reasons for why a state would choose to keep information secret, many of them relating to the military advantage that it can allow them. As argued by Colaresi (2014), secrecy can allow states to deceive their opponents and thus suppress their capabilities. Slantchev (2010) also provides a theoretical explanation for why states in a crisis have an incentive to “feign weakness” and not engage in costly signaling in order to maintain the strategic advantage that comes from mobilizing privately in the event of actual military force. He goes as far as to state that this places democratic regimes, which cannot hide their resolve, at a disadvantage in bargaining. Lindsey (2015) shows in a crisis-bargaining model that a consideration of military strategy helps in understanding why both strong and weak states have an interest in keeping information about military strategy and capabilities secret. Secrecy, as a force multiplier, can increase the combat capability of the side using it, as well as prevent the opponent from using its capabilities efficiently (von Clausewitz, cited in Betts 1982; Pilisuk 1984; Colaresi 2014). Further, secrecy allows leaders to preserve their ability to act during conflict without incurring audience costs from their populations, which allows them a broader range of options (Sagan and Suri 2003; Carson 2016). This is particularly true of democracies, which can face pressures from their publics to escalate a conflict (Carson 2016). Democracies can use secret policy actions in a way that conveys a signal to local allies and/or strategic adversaries without committing themselves. When states are transparent to their publics in sharing information about, for example, their militaries, the state's enemies in the international system are also able to acquire this information (Ben-Zvi 1976; Lai 2004). Thus, states incur what Colaresi (2014) refers to as transparency costs. Transparency costs refer to the fact that once states reveal information about a policy, that policy becomes less useful to them.6 Democracies have traditionally been found to be more transparent than autocracies (Hollyer, Rosendorff, and Vreeland 2011). If this is indeed true, and we accept that transparency costs exist, it is difficult to understand how it is that democracies are more effective at fighting conflicts than autocracies are (Reiter and Stam 2002). Colaresi (2014) resolves this problem by suggesting that executives in democracies do indeed classify information and keep it secret from their publics but that they can do so without losing accountability to the population. He argues that they do so through security accountability institutions, which allow the public to scrutinize the actions of the government in the future, even if in the present these actions are kept secret (some examples are freedom of information acts, investigative committees, and freedom of the press). Rosato (2003) similarly states that even though democracies are generally more transparent than autocracies, anecdotal evidence suggests that this does not prevent the use of secrecy. He argues that democratic politics still allow for secrecy during times of crisis and that expectations about transparency are dropped during wartime. This allows the government to mobilize their resources in secret while also giving the public the opportunity to later verify whether resources were indeed used for the intended purpose (Rosato 2003; Colaresi 2014; Fenster 2015; Birchall 2015). Beyond the first-mover advantage secrecy creates, it can thus allow the government to send a credible signal (through a mobilization) to certain actors (such as local allies or even adversaries with advanced intelligence capabilities) while also keeping it private from others, such as domestic or foreign publics (Sagan and Suri 2003; Hansen, Christensen, and Flyverbom 2015; Carson 2016; Carson and Yahri-Milo 2017).7 Though democracy has often been used as a proxy for transparency (Leeds and Davis 1997; Miller 1999; Schultz 1999; Mitchell and Prins 2004), previous work on transparency has shown that the two are not perfectly correlated and that even within democracies there can be variation in levels of external transparency (Bell 2013; Bell et al. 2014). There is also variation across democracies with regards to the amount of retrospective oversight institutions they have (Colaresi 2014). Thus, those states that do not have security oversight institutions will have to pay transparency costs in order to get their publics’ support for their foreign policies. Without a mechanism to hold the government accountable in the future and ensure that funds were not used in a way that violates the public interest, the public will not be willing to give the executive the resources needed to carry out a policy without a prior justification. In a democracy with retrospective oversight, the public is willing to give the executive the freedom to use secrecy to maneuver in foreign policy, knowing that if resources are misused, the public will find out about it in the future and punish the executive (Colaresi 2014). This leads to the conclusion that democracies that commit to reveal information to their publics in the future are able to conduct foreign policy with greater secrecy in the present. Thus, we expect that democratic states that have more retrospective oversight will be more able to use secrecy in foreign policy. Existing research has assumed this to be the case and drawn out implications of retrospective oversight on war outcomes, that is, success (Colaresi 2012, 2014). It is also important to consider the intermediate step of whether states with retrospective oversight have a greater ability to actually conduct their foreign policy actions in private. There are a number of foreign policy actions that leaders may see some advantage in conducting in private. We focus on the mobilization of forces. Private mobilization requires an effective use of secrecy, and we expect that democracies with more retrospective oversight will be more likely to mobilize privately. While mobilization for war is a rare event, it potentially has higher stakes than any other foreign policy action a state takes and is harder to hide than many other foreign policy actions. If retrospective oversight institutions provide the ability to mobilize in private, we should have some confidence that these institutions allow greater secrecy across foreign policy making. We thus derive the following hypothesis: H1: The more retrospective oversight a democracy has, the more likely it is to privately mobilize during a crisis. In the section that follows, we will empirically test this hypothesis using data on both the retrospective oversight of democracies and on private mobilizations during crises. Research Design Our sample consists of all crisis-directed dyads between the years 1970 and 1994 (these temporal restrictions are due to data availability of both our dependent and independent variables). We use data from the International Crisis Behavior (ICB) dataset to identify when a crisis occurs (Brecher and Wilkenfeld 1997). The ICB project defines a crisis as an event that produces “a heightened probability of military hostilities.” Because we use Lai's measure of private mobilization, we consider only crises that are not intrastate or intrawar.8 Even though we are modeling the choice of a single state to privately mobilize, we construct the dataset with a crisis-directed dyad unit of analysis (Hewitt 2003). This setup allows us to accommodate the directional aspect of private mobilization during crises. The crisis-directed dyad unit of analysis also allows us to take into account characteristics of not only the (potential) mobilizing state but also of the state that it is involved in a crisis with. The Dependent Variable, Private Mobilization Our dependent variable is the private mobilization of troops in crises, an indicator of an effective use of secrecy by the state. To measure private mobilizations, we use Lai's (2004) data on the actions taken by states during crises. He conceptualizes private mobilizations as the military preparations that a state takes for conflict and not for diplomatic purposes. Lai identifies whether private mobilization has occurred by looking for two specific behaviors in the crisis.9 First, a state is coded as privately mobilizing if it “does not make a verbal threat or claim and that same state attempts to conceal its mobilization or that state's crisis action is to attempt a first strike (221).” Second, a state is also coded as privately mobilizing if the “state's only crisis activity is to launch a military strike against the opposing state (221).” To code these actions Lai (2004) uses both the ICB data's case summaries, which were the ones used to create the full ICB dataset, as well as the secondary sources that are referenced in the ICB case summaries (Brecher and Wilkenfeld 1997). The variable is coded dichotomously, taking on a value of 1 in cases that experience private mobilization and 0 otherwise. Given our dichotomous dependent variable, we estimate a series of probit models, with the error terms clustered by country.10 Measuring Retrospective Oversight To measure retrospective oversight, we rely on the retrospective security policy information institutions (r-SPII) index created by Colaresi (2012). The dataset includes information on democracies (considered as states with a Polity score greater than 6). The index is a composite measure of the existence and implementation of freedom of information laws, legislative oversight (the existence and efficacy of legislative committees that investigate national security), and the Freedom House civil liberties measure (rescaled to 0–1, with higher values indicating more liberties). The index ranges from 0 to 1, with higher values indicating greater retrospective oversight (Colaresi 2012, 2014). Control Variables We include in our analysis a series of control variables that we expect to have an effect on a state's probability of engaging in private mobilization during crisis. Given that our theory is about separating the effects of retrospective oversight from those of democracy we begin by including a measure of regime type in our models: the Polity IV combined democracy-autocracy score (Marshall, Jaggers, and Gurr 2004). We also control for contiguity, as we should expect that privately mobilizing troops for longer distances will be more difficult (Starr and Most 1976). This is generated from the Correlates of War Direct Contiguity Data, version 3.1 and is operationalized as a dichotomous variable equal to 1 if two states are separated by 150 miles of water or less or by a land or river border (Stinnett et al. 2002). Given that there can be strategic reasons to mobilize in private given the relative strength of the states involved in a crisis, we control for state capabilities. Using both states’ Comparative Index of National Capabilities (CINC) scores, we also include a measure of relative capabilities (Singer, Bremer, and Stuckey 1972). This is operationalized as State A's capabilities over the capabilities of both states in the dyad. This can vary from 0 to 1, with low values indicating that State A is weak, and high values indicating that State B is weak. Finally, with the understanding that allied states will be less likely to engage in private mobilizations against each other, we include the COW alliance variable, indicating whether the two states in a dyad share any alliances (Gibler 2009). Results and Discussion Table 1 includes estimates from four models in which private mobilization is the dependent variable. Model 1 includes only the retrospective oversight measure as an independent variable, while in Model 2 we include all of the control variables. Table 1. Probit models: Retroactive oversight in democracies and private mobilization, 1970–94.   (1)  (2)  (3)  (4)  Retrospective Oversight  0.277  0.895**        (0.412)  (0.420)      Polity 2 Score, State A    −0.0169    0.00490      (0.0247)    (0.0316)  Contiguous Dyad    0.551***    0.558***      (0.163)    (0.182)  Alliance    0.0907    0.117      (0.0875)    (0.0956)  Power Ratio    0.174    0.280      (0.305)    (0.321)  Freedom of Information Index      −0.721  −0.585        (0.814)  (0.755)  Legislative Oversight Index      1.036**  1.153**        (0.490)  (0.467)  Civil Liberties Index      0.0258  0.0463        (0.0724)  (0.114)  Constant  −1.721***  −2.587***  −1.910***  −2.946***    (0.157)  (0.469)  (0.414)  (0.723)  Observations  504  498  459  453  pr2            (1)  (2)  (3)  (4)  Retrospective Oversight  0.277  0.895**        (0.412)  (0.420)      Polity 2 Score, State A    −0.0169    0.00490      (0.0247)    (0.0316)  Contiguous Dyad    0.551***    0.558***      (0.163)    (0.182)  Alliance    0.0907    0.117      (0.0875)    (0.0956)  Power Ratio    0.174    0.280      (0.305)    (0.321)  Freedom of Information Index      −0.721  −0.585        (0.814)  (0.755)  Legislative Oversight Index      1.036**  1.153**        (0.490)  (0.467)  Civil Liberties Index      0.0258  0.0463        (0.0724)  (0.114)  Constant  −1.721***  −2.587***  −1.910***  −2.946***    (0.157)  (0.469)  (0.414)  (0.723)  Observations  504  498  459  453  pr2          Standard errors in parentheses. Two-tailed significance tests used: * p < .10, ** p < .05, *** p < .01. View Large Table 1. Probit models: Retroactive oversight in democracies and private mobilization, 1970–94.   (1)  (2)  (3)  (4)  Retrospective Oversight  0.277  0.895**        (0.412)  (0.420)      Polity 2 Score, State A    −0.0169    0.00490      (0.0247)    (0.0316)  Contiguous Dyad    0.551***    0.558***      (0.163)    (0.182)  Alliance    0.0907    0.117      (0.0875)    (0.0956)  Power Ratio    0.174    0.280      (0.305)    (0.321)  Freedom of Information Index      −0.721  −0.585        (0.814)  (0.755)  Legislative Oversight Index      1.036**  1.153**        (0.490)  (0.467)  Civil Liberties Index      0.0258  0.0463        (0.0724)  (0.114)  Constant  −1.721***  −2.587***  −1.910***  −2.946***    (0.157)  (0.469)  (0.414)  (0.723)  Observations  504  498  459  453  pr2            (1)  (2)  (3)  (4)  Retrospective Oversight  0.277  0.895**        (0.412)  (0.420)      Polity 2 Score, State A    −0.0169    0.00490      (0.0247)    (0.0316)  Contiguous Dyad    0.551***    0.558***      (0.163)    (0.182)  Alliance    0.0907    0.117      (0.0875)    (0.0956)  Power Ratio    0.174    0.280      (0.305)    (0.321)  Freedom of Information Index      −0.721  −0.585        (0.814)  (0.755)  Legislative Oversight Index      1.036**  1.153**        (0.490)  (0.467)  Civil Liberties Index      0.0258  0.0463        (0.0724)  (0.114)  Constant  −1.721***  −2.587***  −1.910***  −2.946***    (0.157)  (0.469)  (0.414)  (0.723)  Observations  504  498  459  453  pr2          Standard errors in parentheses. Two-tailed significance tests used: * p < .10, ** p < .05, *** p < .01. View Large Once we control for all of the other factors that we expect to influence the probability of a private mobilization during crisis, the coefficient on the retrospective oversight coefficient is positive and significant at the .05 level. This finding is supportive of Hypothesis 1. Although, to our knowledge, there is no previous published research examining private mobilization as a dependent variable, the results on the control variables are somewhat surprising. First, it appears that the level of democracy is not statistically related to private mobilization. Second, neither military capabilities nor the existence of an alliance are statistically related to the decision to privately mobilize, once states find themselves in a crisis. Finally, dyads of contiguous states are more likely to experience a private mobilization. As described above, the r-SPII index is composed of three different elements (Freedom of Information Laws, Legislative Oversight, and Civil Liberties). We also try decomposing the index into its various components as a way to further understand what it is about retrospective oversight in democracies that allows them to effectively engage in secrecy in their foreign policy. In Models 3 and 4 of Table 1, we present the results of two more probit models. In Model 3, we include only the three different components of the r-SPII index as independent variables. In Model 4, we again include all of the control variables that are expected to have an effect on the probability of a private mobilization occurring. In both Models 3 and 4, the only component of the r-SPII index that achieves statistical significance is the Legislative Oversight Index. The coefficient for this variable is positive and significant at the .05 level. This is not surprising, as legislative institutions are likely in a more credible position to demand information from the executive at a later date compared to the ability of the public through FOI laws. Figures 1 and 2 show the predicted probabilities generated from the models estimated above, with 95 percent confidence intervals. Figure 1 shows the relationship between the full retrospective oversight scale and private mobilization (from Model 2), and Figure 2 shows the relationship between legislative oversight index and private mobilization (from Model 4). Both figures illustrate the expected positive relationship. In Figure 2, with legislative oversight set to its mean (.08), the probability of a private mobilization is 5.7 percent.11 A change from the mean level of legislative oversight to one standard deviation above the mean (.28) leads to an increase in the probability of private mobilization to 8.9 percent, an increase of 56 percent. An increase to two standard deviations above the mean (.48) of legislative oversight, increases the probability of private mobilization to 13.2 percent, more than doubling the probability compared to the mean level case.12 Figure 1. View largeDownload slide Retrospective oversight and private mobilization Figure 1. View largeDownload slide Retrospective oversight and private mobilization Figure 2. View largeDownload slide Legislative oversight and private mobilization Figure 2. View largeDownload slide Legislative oversight and private mobilization One potential issue to address regarding these results is the possibility that retrospective oversight institutions may be biasing our sample. In other words, it is possible that we are more likely to have information on secret actions carried out by democracies with ROI than by those without, while democracies without ROI may have carried out secret actions that are never known.13 While this is certainly possible, there are three factors that alleviate this concern for this particular study. The first is that our sample data end in 1994, which is more than twenty years from the writing of this manuscript. This means that even in democracies without ROI, it is most likely that enough time has gone by that most of these actions no longer represent a national security concern and have either been declassified by the government or investigated by the media. The second is that we restrict our analysis to private mobilizations of the state's military during crises, that is, the concentration of military forces in one area (Lai 2004). These are the types of actions that will certainly give a state a first-mover advantage in conflict but that are difficult to keep indefinitely concealed. As Carson and Yahri-Milo (2017) note, very few military mobilizations can actually be kept completely secret from all audiences (for example, local allies might observe the military mobilization). Private military mobilizations can be contrasted with other covert actions that could indeed be kept secret by democracies without ROI, such as covert special forces operations, assassination attempts, or secret alliances. Finally, it is also important to keep in mind that the implication of Colaresi's argument about ROI is not that democracies without ROI are immune to being transparent. Instead, it is argued that these governments are going to be expected by their populations to provide information about security activities as they happen. The democracies without ROI are expected to reveal information earlier than the democracies with ROI. ROI allows the democratic leader to hide information in the present with the promise to reveal information down the road. If we were analyzing a sample that included autocracies, this would be a bigger issue, as autocrats without ROI would be unlikely to ever have the information revealed. By focusing on the differences across consolidated democratic regimes, we believe this concern is diminished. Conclusion In our analysis we find that among democracies, states that have retrospective oversight institutions are more likely to use private mobilizations during a crisis. This is consistent with Colaresi's (2012, 2014) argument that retrospective oversight institutions allow states to avoid incurring transparency costs, and thus enables them to use secrecy effectively while engaging in foreign policy actions. Colaresi's (2012, 2014) previous work shows that democracies with retrospective oversight are indeed more effective at obtaining their desired goals in their international interactions. Yet his theoretical argument for why this occurs, that democracies with ROI are better able to use secrecy in their foreign policy, remains untested. In theory, retrospective oversight could lead to more successful avenues, for example, by making leaders more risk averse and less willing to embark on risky conflicts if they have to be held accountable to them in the future.14 This project, by showing a positive correlation between ROI and private mobilizations, provides support for Colaresi's (2012, 2014) theoretical argument and for the idea that some democracies are better able to perform in conflict due to their ability to use secrecy effectively. This research also has important implications for our understanding of the role that democratic politics play in crisis behavior. Transparency and democracy are concepts that are routinely conflated; with the assumption made that democratic states are uniformly transparent. The models estimated in this paper show that retrospective oversight institutions lead to transparency having its own independent effect on crisis behavior apart from democratic regime type. Another finding to note is that when we decompose the r-SPII index, the only component that has a positive and statistically significant effect on private mobilization is legislative oversight. Future work on the institutional characteristics that allow states to mobilize their troops in private can look more deeply into this relationship. Why is it that giving the legislature greater oversight power allows the executive to mobilize the military in private? Why is this effect greater than that of FOI laws? The evidence in this study consistently demonstrates that democracies with retrospective oversight institutions are more likely to engage in private mobilization than those democracies without ROI. This has important implications for studies of bargaining and crisis escalation. Lai (2004) tells us that states are likely to mobilize publicly when they are actually seeking to signal information and come to a bargained settlement, while private mobilization occurs when states give up on bargaining and seek an advantage in actual war fighting. Recent work by Carson (2016) and Carson and Yahri-Milo (2017) (as well as by Sagan and Suri 2003) suggests that, beyond the military first-mover advantage they provide, private mobilizations may actually send a costly signal to select actors (such as local allies and/or adversaries with advanced intelligence agencies), while also allowing leaders the option of avoiding audience costs and bargaining with an opponent without being forced into escalation by their domestic audiences. Democracies with retrospective oversight institutions may thus be better able to avoid the escalation of conflict by being able to bargain with an opponent free of domestic pressures stemming from audience costs. Retrospective oversight institutions thus serve to widen the scope of available options a democracy has during conflict. This work also points to several future research questions. If retrospective oversight indeed allows for greater secrecy in conducting foreign policy, we should observe democratic states with retrospective oversight institutions also engaging in other forms of secret behavior, such as secret alliances and secret basing negotiations. Further work can explore how it is that retrospective oversight affects these other forms of foreign policy behavior. In addition, future work may use a case study approach to study states before and after the implementation of retrospective oversight institutions and see if their use of secrecy in their foreign policy behavior changes. Carla Martinez Machain is an associate professor of political science at Kansas State University. She received her PhD from Rice University in 2012. Her research specializations include conflict outcomes, security policy, and military cooperation. She was the recipient of a 2017 Minerva Grant from the US Department of Defense. Sam Bell is an associate professor of political science at Kansas State University. He received his PhD from Binghamton University in 2009. His research examines the causes of political violence. His work has previously appeared in the Journal of Politics, International Studies Quarterly, and the Journal of Conflict Resolution, among other outlets. Footnotes 1 We define transparency following Bell (2013) as “the inability of government leaders to hide government processes, decision-making and events taking place within a state from external audiences.” 2 Following Colaresi (2014), we define retrospective oversight institutions as those that “empower actors outside of the executive with tools to investigate national security issues and publicize information” and “allow for investigations into previously secret information and to reveal executive abuses of secrecy powers” (2014, 9). 3 Following Lai (2004), we define private mobilizations as the concentration of military forces in one area without the knowledge of the opposing state. 4 There is not complete uniformity of thought on the effect that democracy and transparency have on secrecy. See Kam (1988) and Finel and Lord (1999) for alternate views. 5 One example that often is used to illustrate this point is that of the Associated Press reporting on the Bay of Pigs invasion even before it had taken place. 6 We note that in some cases states will have an incentive to keep information from their publics in order to avoid audience costs but will count on local allies and/or strategic rivals with “sophisticated intelligence capabilities” to observe these actions and thus rely on them as a costly signal of their resolve (Carson and Yahri-Milo 2017, 125). 7 Carson and Yahri-Milo (2017, 125) argue that “covert action is rarely completely secret,” as parts of an action such as a private mobilization can be observed by the intelligence apparatus of local allies and/or even rivals. States are thus able to use a private mobilization as a costly signal of their resolve while also not incurring audience costs, which would be particularly costly for democracies (Carson 2016). 8 These are the only cases that can take on a value of 1 for our dependent variable, because they are the only cases coded by Lai (2004). 9 In the analysis presented in Lai (2004), the unit of analysis is the crisis. We obtained the state-level data from Lai. This allows us to identify which state in a crisis engaged in private mobilization. 10 The results remain substantively the same when we cluster on the dyad. 11 All other variables were set to means and modes. All predictions were generated with margins command in Stata 15. 12 Figure 1 illustrates a similar set of substantive predictions. 13 We thank an anonymous reviewer for pointing this out. 14 We thank an anonymous reviewer for pointing out this observation to us. References Almond G. A. 1956. “ Public Opinion and National Security Policy.” Public Opinion Quarterly  20 ( 2): 371– 78. Google Scholar CrossRef Search ADS   Avant Deborah. 2007. “ Political Institutions and Military Effectiveness: Contemporary United States and United Kingdom.” Creating Military Power: The Sources of Military Effectiveness , 80– 105. Bar-On Mordechai. 1989. “ David Ben-Gurion and the Sevres Collusion.” In Suez 1956: The Crisis and Its Consequences , edited by Louis William Roger, Owen Roger, 145– 60. Oxford: Clarendon. Google Scholar CrossRef Search ADS   Bell Sam R. 2013. “ What You Don't Know Can Hurt You: Information, External Transparency, and Interstate Conflict, 1982–1999.” Conflict Management and Peace Science  30 ( 5): 452– 68. Google Scholar CrossRef Search ADS   Bell Sam R., Chad Clay K., Murdie Amanda, Piazza James. 2014. “ Opening Yourself Up: The Role of External and Internal Transparency in Terrorism Attacks.” Political Research Quarterly  67 ( 3): 603– 14. Google Scholar CrossRef Search ADS   Bennett Scott D., Stam III Allan C.. 1998. “ The Declining Advantages of Democracy: A Combined Model of War Outcomes and Duration.” Journal of Conflict Resolution  42 ( 3): 344– 66. Google Scholar CrossRef Search ADS   Bennett D. Scott, Stam Allan C.. 1996. “ The Duration of Interstate Wars, 1816–1985.” American Political Science Review  90 ( 2): 239– 57. Google Scholar CrossRef Search ADS   Ben-Zvi Abraham. 1976. “ Hindsight and Foresight: A Conceptual Framework for the Analysis of Surprise Attacks.” World Politics  28 ( 3): 381– 95. Google Scholar CrossRef Search ADS   Berliner D., 2014. “ The Political Origins of Transparency.” The Journal of Politics  76 ( 02): 479– 91. Google Scholar CrossRef Search ADS   Betts Richard K. 1982. Surprise Attack: Lessons for Defense Planning . Brookings Institution Press. Birchall Clare. 2015. “ ‘Data. Gov-in-a-Box’ Delimiting Transparency.” European Journal of Social Theory  18 ( 2): 185– 202. Google Scholar CrossRef Search ADS   Brecher Michael, Wilkenfeld Jonathon. 1997. A Study of Crisis . Ann Arbor: University of Michigan Press. Google Scholar CrossRef Search ADS   Bueno de Mesquita Bruce, Smith Alastair, Siverson Randolph M., Morrow James D.. 2005. The Logic of Political Survival . The MIT Press. Carson Austin. 2016. “ Facing Off and Saving Face: Covert Intervention and Escalation Management in the Korean War.” International Organization  70 ( 1): 103– 31. Google Scholar CrossRef Search ADS   Carson Austin, Yarhi-Milo Keren. 2017. “ Covert Communication: The Intelligibility and Credibility of Signaling in Secret.” Security Studies  26 ( 1): 124– 56. Google Scholar CrossRef Search ADS   Colaresi M. 2012. “ A Boom with Review: How Retrospective Oversight Increases the Foreign Policy Ability of Democracies.” American Journal of Political Science  56 ( 3): 671– 89. Google Scholar CrossRef Search ADS   Colaresi M. 2014. Democracy Declassified: The Secrecy Dilemma in National Security . Oxford: Oxford University Press. Google Scholar CrossRef Search ADS   Desch Michael C. 2002. “ Democracy and Victory: Why Regime Type Hardly Matters.” International Security  27 ( 2): 5– 47. Google Scholar CrossRef Search ADS   Downes Alexander B. 2009. “ How Smart and Tough Are Democracies? Reassessing Theories of Democratic Victory in War.” International Security  33 ( 4): 9– 51. Google Scholar CrossRef Search ADS   Fenster Mark. 2015. “ Transparency in Search of a Theory.” European Journal of Social Theory  18 ( 2): 150– 67. Google Scholar CrossRef Search ADS   Finel Bernard I., Lord Kristin M.. 1999. “ The Surprising Logic of Transparency.” International Studies Quarterly  43: 315– 39. Google Scholar CrossRef Search ADS   Friedberg Chen. 2008. “ Legislative Oversight and the Israeli Committee System: Problems and Solutions.” Oversight  217. Friedberg Chen. 2011. “ From a Top-Down to a Bottom-Up Approach to Legislative Oversight.” The Journal of Legislative Studies  17 ( 4): 525– 44. Google Scholar CrossRef Search ADS   Friedberg Chen. 2013. “ Oversight by the State Control Committee in the Israeli Parliament: Form of Accountability Under Stress.” Israel Affairs  19 ( 2): 306– 20. Google Scholar CrossRef Search ADS   Gibler Douglas M. 2009. International Military Alliances, 1648–2008 . Washington, DC: CQ Press. Google Scholar CrossRef Search ADS   Hansen Hans Krause, Christensen Lars Thøger, Flyverbom Mikkel. 2015. “ Introduction: Logics of Transparency in Late Modernity Paradoxes, Mediation and Governance.” European Journal of Social Theory  18 ( 2): 117– 31 Google Scholar CrossRef Search ADS   Hewitt J. Joseph. 2003. “ Dyadic Processes and International Crises.” Journal of Conflict Resolution  47 ( 5): 669– 92. Google Scholar CrossRef Search ADS   Hollyer James R., Rosendorff B. Peter, Vreeland James Raymond. 2011. “ Democracy and Transparency.” Journal of Politics  73 ( 4): 1191– 205. Google Scholar CrossRef Search ADS   Hughes J. M., 2006. To the Maginot Line: The Politics of French Military Preparation in the 1920’s , Vol. 64. Cambridge, MA: Harvard University Press. Kam Ephraim. 1988. Surprise Attack: The Victim's Perspective . Cambridge, MA: Harvard University Press. Google Scholar CrossRef Search ADS   Kennedy J. F. 1962. “ Interview by S. Vanocur, Transcript.” NBC , https://www.jfklibrary.org/JFK/JFK-in-History/John-F-Kennedy-and-the-Press.aspx Knorr Klaus Eugen, Morgan Patrick M. 1983. Strategic Military Surprise: Initiatives and Opportunities . New Brunswick, NJ: Transaction Books. Lai Brian. 2004. “ The Effects of Different Types of Military Mobilization on the Outcome of International Crises.” Journal of Conflict Resolution  48 ( 2): 211– 29. Google Scholar CrossRef Search ADS   Lake David A. 1992. “ Powerful Pacifists: Democratic States and War.” American Political Science Review  86 ( 1): 24– 37. Google Scholar CrossRef Search ADS   Laws of the State of Israel (1948–), LC Call No. KMK 13.I874. 1983. Jerusalem: The Government Printer. Leeds Brett Ashley, Davis David R.. 1997. “ Domestic Political Vulnerability and International Disputes.” Journal of Conflict Resolution  41 ( 6): 814– 34. Google Scholar CrossRef Search ADS   Lindsey David. 2015. “ Military Strategy, Private Information, and War.” International Studies Quarterly  59 ( 4): 629– 40. Marshall Monty G., Jaggers Keith, Gurr Ted Robert. 2004. “Polity IV.”  College Park: University of Maryland. Miller Ross A. 1999. “ Regime Type, Strategic Interaction, and the Diversionary Use of Force.” Journal of Conflict Resolution  43 ( 3): 388– 402. Google Scholar CrossRef Search ADS   Mitchell Sara McLaughlin, Prins Brandon C.. 2004. “ Rivalry and Diversionary Uses of Force.” Journal of Conflict Resolution  48 ( 6): 937– 61. Google Scholar CrossRef Search ADS   Pilisuk Marc. 1984. “ Experimenting with the Arms Race.” Journal of Conflict Resolution  28 ( 2): 296– 315. Google Scholar CrossRef Search ADS   Reiter Dan, Stam Allan C.. Democracies at War . Princeton, NJ: Princeton University Press, 2002. Google Scholar CrossRef Search ADS   Rosato Sebastion. 2003. “ The Flawed Logic of the Democratic Peace Theory.” American Political Science Review  97 ( 4): 585– 602. Google Scholar CrossRef Search ADS   Sagan Scott D., Suri Jeremi. 2003. “ The Madman Nuclear Alert: Secrecy, Signaling, and Safety in October 1969.” International Security  27 ( 4): 150– 83. Google Scholar CrossRef Search ADS   Schuessler John M. 2010. “ The Deception Dividend: FDR's Undeclared War.” International Security  34 ( 4): 133– 65. Google Scholar CrossRef Search ADS   Schultz Kenneth A. 1999. “ Do Democratic Institutions Constrain or inform? Contrasting Two Institutional Perspectives on Democracy and War.” International Organization  53 ( 2): 233– 66. Google Scholar CrossRef Search ADS   Singer J. David, Bremer Stuart, Stuckey John. 1972. “ Capability Distribution, Uncertainty, and Major Power War, 1820–1965.” Peace, War, and Numbers  19– 48. Slantchev Branislav L. 2010. “ Feigning Weakness.” International Organization  64 ( 3): 357– 88. Google Scholar CrossRef Search ADS   Starr Harvey, Most Benjamin A.. 1976. “ The Substance and Study of Borders in International Relations Research.” International Studies Quarterly  581– 620. Stinnett Douglas M., Tir Jaroslav, Diehl Paul F., Schafer Philip, Gochman Charles. 2002. “ The Correlates of War (COW) Project Direct Contiguity Data, Version 3.0.” Conflict Management and Peace Science  19 ( 2): 59– 67. Google Scholar CrossRef Search ADS   © The Author(s) (2018). Published by Oxford University Press on behalf of the International Studies Association. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Foreign Policy AnalysisOxford University Press

Published: Apr 30, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off