The Policy World Meets Academia: Designing U.S. Policy toward Russia

Chapter 7: Putin's Peers

Back to table of contents
Timothy J. Colton, Timothy Frye, and Robert Legvold
U.S. Policy Toward Russia

H. E. Goemans

After the disintegration of the Soviet Union, few institutions remained to allocate and reallocate political authority in the successor states. Many were led by local representatives or functionaries held over from the Communist Soviet Union. Over time, the majority of successor republics in the West adopted institutions to regulate the transfer of political authority and to guarantee leaders a safe retirement should they lose power. In the East, however, most successor states failed to develop such institutions, and thus evolved into personalist dictatorships.

As a result, Russia—which, arguably, is also to some degree a personalist dictatorship—is geographically close to seven such regimes, sharing a border with four. The fact that personalist dictatorships are significantly more likely to engage in international conflict than any other type of dictatorship makes Russia (surrounded by more personalist dictatorships than any country in recent history) especially vulnerable to future disputes with some of its nondemocratic neighbors. Moreover, most personalist dictatorships on or near Russia’s borders contain a significant minority of ethnic Russians. The combination of personalist dictatorships and significant ethnic minorities considerably increases the prospects for international conflict.

It would be a mistake to reflexively attribute a conflict between Russia and a neighboring state to Russia’s renewed determination to control the former Soviet Empire. Rather, such a regional conflict would more likely result directly from domestic constraints and interregional politics.

The United States can play a powerful and positive role to promote peace and stability in the region. First, the United States should continue its policy of refusing to recognize leaders who come to power through the threat or use of force. Second, preferably in consultation with Russia, the United States should vigorously oppose any leader in the Soviet successor states who resorts to ethnic repression in order to maintain power. Third, the United States might consider promoting, in cooperation with Russia, the establishment of a regional institution through which the former Soviet republics could settle territorial or other issues. However, the United States should not openly pressure the personalist dictators in Central Asia to change their domestic political institutions. Coercion would likely be met with ferocious opposition from the current leadership and perhaps from Russia as well. Abandoning a policy that promotes democratization, though not an attractive option, is decidedly better than the alternative. Any policy for democratization will have to take a long-term view.

In this paper, I examine the incentives and constraints of personalist dictators; offer a brief analysis of a comparable region and era of personalist dictators—Central America in the nineteenth century—and trace how the prevalence of conflict there was finally resolved; and draw conclusions that might inform U.S. policy toward Russia.


The groundbreaking work of political scientist Barbara Geddes opened new avenues of research into the behavioral patterns of different types of dictatorships. Geddes distinguished three major categories of dictatorships: military, single party, and personalist dictatorships, as well as several hybrids or combinations of these types. She defines military regimes as those in which “professional military . . . rules as an institution, for example, Argentina 1976–83.” Single-party regimes (including Leninist) are defined as those in which “the party . . . penetrates society to the village level and . . . officials and leaders must come up through the party, for example, the CCM in Tanzania and the KMT in Taiwan.” Finally, personalist regimes are those “in which despite possibly wearing a uniform and creating a support party, policies and personnel are chosen and disposed of at the whim of the ruler.”1

Of the fourteen countries that share a border with Russia, four are led by personalist dictators.2 In Kazakhstan, President Nursultan Nazarbayev has been the effective leader for almost two decades. In Belarus, President Alyaksandr Lukashenka has ruled since July 1994. After independence, Heydar Aliyev first ruled Azerbaijan; since his death in October 2003, his son Ilham Aliyev has ruled. Finally, North Korea’s Kim Jong Il similarly inherited the mantle of his father in 1997. The noncontiguous successor states of Uzbekistan, Kyrgyzstan, and Turkmenistan also suffered (and in the case of Uzbekistan, continue to suffer) under personalist dictators. In Uzbekistan, President Islam Karimov has ruled for two decades. Askar Akayev led Kyrgyzstan until he was deposed in March 2005. Kurmanbek Bakiyev, who led the revolution that overthrew Akayev, has since ruled the country and overwhelmingly won the elections of July 2009. (The elections were subsequently deemed “a disappointment” by the Organisation for Economic Co-operation and Development.)3 Although perhaps not a full-fledged personalist dictator yet, Bakiyev appears to be determined to hold on to power, if necessary by repression. Perhaps the most extravagant and flamboyant of the personalist dictators of Central Asia was Saparmurat Niyazov, who promoted an elaborate cult of personality and ruled Turkmenistan until his death in 2006. As in the case of Bakiyev, Niyazov’s successor rose to power in an irregular manner, which, we shall see below, also significantly affects the international behavior of leaders. “Owezgeldi Atayev, who, according to the . . . constitution [of Turkmenistan], should have succeeded Niyazov, was instead relieved of his post as speaker of the 50-member Majlis (the unicameral legislature) after an unspecified criminal case was brought against him by the prosecutor-general’s office.” In his place, Deputy Prime Minister Gurbanguly Berdimuhammedov became Turkmenistan’s new ruler.4 A year after he came to power, Berdimuhammedov began a purge of officials from the Niyazov regime. Although perhaps not as extreme as his predecessor, Berdimuhammedov is well on his way to establishing another personalist dictatorship in Turkmenistan.5

Barbara Geddes and, subsequently, political scientist Erica Franz6 postulated that a fundamental factor in these regimes’ behavior, in domestic as well as international affairs, is their significantly different numbers of veto players—that is, actors who can single-handedly block policy—and thus the significantly different constraints on their policy choices. (This argument is echoed by political scientists Dan Reiter and Allan Stam,7 but offers only a shallow explanation of why personalist dictatorships are prone to war.) However valid the argument, because the concept of “veto players” is empirically difficult to operationalize, I work from a different set of principles, one that is well established in the literature on comparative politics. I argue that regimes can be usefully distinguished by the degree to which they allow their leaders a safe and prosperous retirement. In this vein, scholars of comparative politics have offered two ideal types to distinguish regimes. As political philosopher Karl Popper explains,

[W]e may distinguish two main types of government. The first type consists of governments of which we can get rid without bloodshed—for example, by way of general elections; that is to say, the social institutions provide means by which the rulers may be dismissed by the ruled, and the social traditions ensure that these institutions will not easily be destroyed by those who are in power. The second type consists of governments which the ruled cannot get rid of except by way of a successful revolution—that is to say, in most cases, not at all.8

The fundamental difference between these two ideal types, according to political scientist William H. Riker, stems from the degree to which institutions protect politicians and leaders after they lose office:

Almost everything . . . that we think of as civil liberties (the rights of a speedy trial, habeas corpus, and security against unreasonable search and seizure, for example) originated to protect politicians who feared prosecution if and when they lost office. Thus the historic purpose of these fundamental democratic liberties has been not to provide freedom as an end in itself, but to render effective both political participation and the process of choice in voting.9

Leaders of countries that do not rely on such well-established norms, rules, and procedures lack institutional protections to shield them against ensuing, sometimes severe, punishment after they lose office. Riker’s argument thus suggests a close institutional link between the manner and consequences of losing office. Leaders lose office in a regular manner (that is, following the existing rules, norms, and procedures, including elections, term limits, and voluntary retirement) because they can afford to. Leaders lose office in an irregular manner (by the threat or use of force) because holding on to power provides their only protection against potential punishment. Since the voluntary retirement of such leaders exposes them to potential punishment, they cling to power. The only way to remove them from office is by the use or threat of force.

As noted by Popper, the distinction is often drawn to clearly separate “democracies” from “dictatorships.”10 The logic of Popper and Riker can be extended further to distinguish different types of dictatorships and explain their distinctive behavior in international politics. Specifically, personalist dictators face a significantly higher probability of personal punishment after they lose office. This, in turn, makes such personalist dictators extremely reluctant to give up power and even likely to resort to international conflicts in gambles for survival.11 Thus, removing personalist dictators often requires the use of force, either by domestic military forces or by domestic rebels with foreign military support. This argument is corroborated by two sets of simple cross-tabs, which examine the manner as well as the consequences of losing office for the different types of dictatorships. Table 1 compares whether leaders were removed in a regular manner according to institutional norms, rules, and procedures, as a result of ill health or a natural death, or as the result of an irregular removal from office. Irregular removals from office occur when leaders are removed 1) contrary to well-established norms, rules, and procedures or 2) when no such norms, rules, or procedures have been established. Table 1 shows a striking result: personalist dictators are significantly more likely to be removed in an irregular manner than other dictators.

Table 1: How Dictators Lose Office

Image of Table 1

Regime type as provided by Barbara Geddes in “Minimum-Winning Coalitions and Personalization in Authoritarian Regimes,” paper presented at the annual meeting of the American Political Science Association, Chicago, September 2–5, 2004. Source: Author’s calculations based on data from Geddes, “Minimum-Winning Coalitions and Personalization in Authoritarian Regimes” and from H. E. Goemans, Kristian Skrede Gleditsch, and Giacomo Chiozza, “Introducing Archigos: A Dataset of Political Leaders,” Journal of Peace Research 46 (2) (2009): 269–283.

A more in-depth examination of how leaders are removed—omitted here for brevity’s sake—delivers two striking patterns that explain why personalist dictators tend to go to war more frequently than other dictators. First, 12 percent of personalist dictators were ousted by domestic rebel forces with foreign support. Except for one military/personalist-hybrid leader (6 percent) and two single-party hybrids (5 percent), none of the other types of dictators left office under such conditions. Second, personalist leaders were twice as likely as military leaders and four times as likely as single-party leaders to be removed through a coup. A striking 40 percent of the personalist dictators who lost office were removed by a coup. (If military-against-military coups are included, about 32 percent of the military dictators lost office as a result of a coup.) Therefore, it is not surprising that the post-tenure fate of personalist dictators, as demonstrated in Table 2, is particularly bleak.

Table 2: The Post-Tenure Fate of Dictators

Image of Table 2

Regime type as provided by Barbara Geddes in “Minimum-Winning Coalitions and Personalization in Authoritarian Regimes,” paper presented at the annual meeting of the American Political Science Association, Chicago, September 2–5, 2004. Source: Author’s calculations based on data from Geddes, “Minimum-Winning Coalitions and Personalization in Authoritarian Regimes” and from H. E. Goemans, Kristian Skrede Gleditsch, and Giacomo Chiozza, “Introducing Archigos: A Dataset of Political Leaders,” Journal of Peace Research 46 (2) (2009): 269–283.


Personalist dictators are particularly likely to become involved in international conflict because it can significantly reduce the leaders’ risk of losing office. First, when personalist dictators face a domestic rebellion supported from abroad, they have strong incentives to interdict such military support by attacking rebel bases beyond their countries’ borders. Second, escalating threats to deter foreign support of domestic rebels can easily trigger an international crisis. The foreign supporter may decide to increase its military support or even become directly involved in the fight with its own forces, especially when ethnic brethren form a substantial part of the rebel forces or suffer repression.

Third, international conflict allows the leader to send potentially troublesome and ambitious officers out of the capital and to the front. Following the ancient example of King David and Uriah the Hittite, Idi Amin, the military dictator and president of Uganda during the 1970s, used this strategy to eliminate opposition from within the armed forces. In 1977, Great Britain broke off diplomatic relations and, together with the United States, imposed harsh economic sanctions on Idi Amin’s Uganda. The sanctions exacerbated the deterioration of an already faltering economy and, by diminishing Amin’s ability to buy off his core supporters in the military, created unrest among those supporters.12 Determined to maintain control, Amin began to purge his inner circle, including his longtime second in command, Vice President and Commander of the Armed Forces General Idris Mustafa Adrisi.13 After Adrisi suffered a highly suspicious car accident, his supporters in the army, particularly the crack Simba (Lion) Regiment and the Chui (Leopard) Regiment, openly revolted. While the revolt was brutally suppressed, survivors fled across the border into Tanzania.14 The 1978 war between Uganda and Tanzania began when Amin sent his soldiers in pursuit of the rebels. Contemporaries agree that the primary goal of Amin’s invasion was to deal with a threat from his own military forces. Milton Obote, the former president of Uganda who was in exile in Tanzania at that time, put it bluntly: the invasion “was a desperate measure to extricate Amin from the consequences of the failure of his own plots against his own army.”15 By turning on some of his remaining core supporters, Amin risked antagonizing the very forces that underpinned his brutal regime. Thus, he tried to blame the Tanzanian forces for the executions of rebels from the Simba Regiment. After the Tanzanian forces recaptured the Kagera salient, they found “[s]cattered in the bush . . . the bodies of 120 Ugandan soldiers. There had been no Tanzanian troops in the area before, and there was no sign that Tanzanian artillery had landed there.”16 The truth was inescapable: “The Tanzanian commanders deduced the corpses had been dumped to look as if they were battle fatalities, although they were actually executed mutineers.”17

While I cannot offer a direct statistical test of the different mechanisms whereby personalist dictators are likely to become involved in conflict, I can establish that personalist dictators are indeed more likely to become involved in international conflict. Improving on the work of Mark Peceny, Caroline Beer, and Shannon Sanchez-Terry,18 Dan Reiter and Allan Stam19 were the first to show that personalist dictators are particularly likely to challenge democracies but are unlikely to have democracies challenge them. Nonetheless, particular dyads may be conflict prone because of country-specific factors rather than regime type; therefore I run a fixed-effect logit model, grouping observations by country, to control for any country-specific effects on the overall probability of conflict involvement.

Table 3 shows that personalist dictators are significantly more likely to become involved in international conflict than hybrid regimes.20 This likelihood is more pronounced early in the tenures of those who entered office in an irregular manner and dissipates over time, though these effects do not appear to be very robust. Notably, leaders are less likely to become involved in conflict as they grow older, a trend that is quite robust.21

Table 3: Dictatorial Regime Types and Conflict Involvement

Image of Table 3

Fixed-effect logit; observations grouped by country. Coefficients of each regime type must be interpreted relative to the excluded category, hybrid regimes. Source: Author’s calculations based on data from Geddes, “Minimum-Winning Coalitions and Personalization in Authoritarian Regimes,” paper presented at the annual meeting of the American Political Science Association, Chicago, September 2–5, 2004 and from H. E. Goemans, Kristian Skrede Gleditsch, and Giacomo Chiozza, “Introducing Archigos: A Dataset of Political Leaders,” Journal of Peace Research 46 (2) (2009): 269–283.

To probe more deeply into the factors that drive personalist dictators toward war, in the next section I analyze a region that experienced a concentration of personalist dictators similar to that which currently prevails in Central Asia: that is, Central America from 1840 to the 1920s. This brief history strongly suggests some disturbing parallels to the current situation in Central Asia. At the same time, the history of Central America also highlights steps the United States can take to mitigate the dangers of international conflict.


In Central America between 1840 and 1919, the caudillo, a personalist dictator par excellence, decided when to go to war. During this period, the leaders of Costa Rica, Nicaragua, Honduras, El Salvador, and Guatemala went to war much more often than is generally known. No fewer than seventeen wars were fought between 1840 and 1919,22 and a striking pattern emerged in eleven of them. Leaders in Central America became increasingly likely to lose office through the use of force when the ideological balance of power between so-called liberals and conservatives23 in the region changed, either because a leader in a neighboring country was replaced or because a neighboring leader changed his allegiance.

When, for example, a Conservative leader was overthrown by a Liberal in Guatemala, Conservative leaders elsewhere had reason to worry that their own Liberal exiles would obtain support from Guatemala’s new Liberal regime and that as a result, the domestic balance of power between Conservatives and Liberals could shift against the leader. However, a Conservative leader could not bargain for a power-sharing deal with his political opponent because there was no guarantee the agreement would hold if circumstances changed in his favor. Instead, violent conflict between the leader and the opposition became more likely. The opposition would gather abroad, where Guatemala’s Liberal opponent would give them shelter, arms, and a safe place to organize and prepare an invasion.

Exiles across the border, organizing to overthrow the leader, were often the cause of interstate war. As war correspondent Frederick Palmer24 put it, “A favorite means of warfare of one President on another was to support the organization of a revolutionary army within his borders to invade his neighbor’s territory when it was ready.” Leaders, who often led their army, would try to preempt such invasions by invading their hostile neighbor first, provoking international warfare.

We can detect a disturbing analogy to this pattern in Central Asia when we consider the treatment of or foreign support for Russian ethnic minorities living in successor republics. Azerbaijan has the lowest percentage of ethnic Russians living within its borders, with 1.8 percent. Turkmenistan (4 percent) and Uzbekistan (5.5 percent) have slightly higher fractions of ethnic Russians. In Belarus and Kyrgyzstan, the percentage rises to 11.4 percent and 12.5 percent, respectively, and is highest in Kazakhstan, where 30 percent of the population is ethnic Russian. As different leaders in these countries rise to power and repress or support Russian minorities (which occurs for domestic political reasons, as well as foreign) the danger of international conflict rises dramatically. Recall that the statistical analysis (not shown) confirmed that personalist dictators were particularly likely to lose power as a result of domestic rebellion supported from abroad—with the obvious foreign supporter, in this case, being Russia.

The United States played an effective role in significantly decreasing the risk of war in Central America, a role it should assume in Central Asia today. As I noted above, the support of exiles time and again led to war in Central America. When, for example, Honduras and Nicaragua were ruled by a Conservative and a Liberal, respectively, the Conservative leader of Honduras would support the Conservative exiles from Nicaragua, while the Liberal leader of Nicaragua would support the Liberal exiles from Honduras. Why would leaders make the same mistake over and over again, especially when it often led to war and their forcible removal from office? The answer is simple: to maintain support among Liberals (or Conservatives) at home, the Liberal (or Conservative) leader had to support Liberal (or Conservative) exiles. As a result, the dominant strategy for maintaining power—supporting exiles with a similar ideological bent—created a classic Prisoner’s Dilemma. Even though both sides would have been better off if neither supported exiles, the logic of their strategic interaction ensured that both sides supported exiles, which increased the probability of war. To escape this dilemma, the sparring Liberals and Conservatives needed a credible enforcer, a party that would hold leaders accountable for supporting exiles.

The Washington Treaty of December 1907 would finally provide such an enforcer. U.S. Secretary of War Elihu Root determined that wars in Central America were against the interests of the United States. He organized a conference of all five states in December 1907 in Washington, under the auspices of Mexico and the United States.25 On December 20, the parties attempted to deal with the fundamental causes of recurrent warfare on the isthmus. Articles XVI and XVII declared:

Article XVI–Desiring to prevent one of the most frequent causes of disturbances in the Republics, the contracting Governments shall not permit the leaders or principal chiefs of political refugees, nor their agents, to reside in the departments bordering on the countries whose peace they might disturb. Those who may have established their permanent residence in a frontier department may remain in the place of their residence under the immediate surveillance of the Government affording them an asylum, but from the moment when they become a menace to public order they shall be included in the rule of the preceding paragraph.

Article XVII–Every person, no matter what nationality, who, within the territory of one of the contracting Parties, shall initiate or foster revolutionary movements against any of the others, shall be immediately brought to the capital of the Republic, where he shall be submitted to trial according to law.26

An Additional Convention to the General Treaty contained three further clauses intended to do away with the threat of a forcible removal from office. It stated:

Article I–The Governments of the High Contracting Parties shall not recognize any other Government which may come into power in any of the five Republics as a consequence of a coup d’état, or of a revolution against the recognized Government, so long as the freely elected representatives of the people thereof, have not constitutionally reorganized the country.

Article II–No Government of Central America shall in case of civil war intervene in favor of or against the Government of the country where the struggle may take place.

Article III–The Governments of Central America, in the first place, are recommended to endeavor to bring about, by the means at their command, a constitutional reform in the sense of prohibiting the re-election of the President of a Republic, where such prohibition does not exist, secondly to adopt all measures necessary to effect a complete guarantee of the principle of alternation in power.27

To enforce these terms, the Treaty established a Central American Court of Justice, in Cartago, Costa Rica. In subsequent years, with some relatively minor exceptions, the United States showed that it intended to hold the Central American states to their promises. Thus, “[t]he United States became the enforcer of the 1907 treaty.”28 The old pattern was broken. Even though a Conservative leader replaced a Liberal leader in Nicaragua in 1911, and the Conservative Manuel Bonilla regained power in Honduras in 1913, these exogenous shocks did not trigger international crises or wars as they had so many times before. Thus, the 1907 Washington Treaty solved the commitment problem by cushioning the effects of any temporary shock to a leader’s capabilities and legitimacy. By stipulating that any leader who came to power through a coup would not be recognized, the Treaty fundamentally altered the costs and benefits of any coup or revolt.

To minimize the risks of international conflict in Central Asia, the United States should seek to emulate the position it took more than a century ago. The United States should vigorously support its current policy of refusing to recognize any leader who comes to power through the use or threat of force. It should, as is current practice, withhold foreign aid (including military aid) until credible elections have been held. It should oppose any attempts at ethnic repression, both for moral as well as for practical reasons. Finally, the United States should perhaps consider proposing to Russia a joint establishment of a tribunal to settle any disputes among the Soviet successor states.


Vladimir Putin, Dmitry Medvedev, and the Russian leaders who will succeed them face a precarious situation in Central Asia, where Russia is neighbor to between four and seven personalist dictators and confronts an elevated risk of international conflict. Adversaries of the personalist dictators know that there exists only one way to obtain power: through the threat or use of force. Domestic political opponents of personalist dictators, particularly if they are ethnically organized and have ethnic brethren in neighboring countries, are likely to seek support from neighboring leaders. If they succeed, the probability of international conflict, unsurprisingly, drastically increases.

Basic statistical patterns in regime type and the consequences of losing office show that personalist dictators have strong reasons to cling to power (Table 2). As a result of these pressures, personalist dictators are significantly more likely to become involved in international conflict than other regime leaders (with the potential exception of single-party dictators; see Table 3).

The history of Central America between 1840 and 1919 shows that the presence of exiles abroad (or ethnic brethren, as is the case today in Central Asia), and an exogenous regional shock in favor of those exiles, can explain the prevalence of conflict in that period. In 1907, by declaring its unwillingness to recognize leaders who came to power through the use of force, and by instituting a Court of Arbitration to address disputes among the Central American states, the United States successfully defanged the most important factors that had led to recurrent war. The United States should pursue a similar policy in Central Asia, preferably in close cooperation and consultation with Russia.


1. Geddes collected data on 170 authoritarian regimes between 1945 and 1996; Barbara Geddes, “Minimum Winning Coalitions and Personalization in Authoritarian Regimes,” paper presented at the annual meeting of the American Political Science Association, Chicago, September 2–5, 2004. It is important to note that Geddes collected the data largely by country. For the empirical analyses below I merged Geddes’s data with Archigos, taking care to attribute the right regime type to the appropriate leader; Henk E. Goemans, Kristian Skrede Gleditsch, and Giacomo Chiozza, “Introducing Archigos: A Dataset of Political Leaders,” Journal of Peace Research 46 (2) (2009): 269–283. This merged data and the DO-files are available on request. Please note that CCM refers to Taiwan’s Chama Cha Mapinduzi (Party of the Revolution); KMT refers to Kuomintang (the Chinese Nationalist Party).

2. These fourteen countries (with the mileage shared along the borders) are Kazakhstan (4,254 miles), China (2,265 miles), Mongolia (2,170 miles), Ukraine (9 miles), Finland (816 miles), Belarus (596 miles), Georgia (449 miles), Poland (268 miles), Latvia (181 miles), Estonia (180 miles), Azerbaijan (177 miles), Lithuania (141 miles), Norway (122 miles), and North Korea (11 miles).

3. Keesing’s Record of World Events 55 (July 2009): Kyrgyzstan, 49323.

4. Keesing’s Record of World Events 52 (December 2006): Turkmenistan, 47641.


6. Erica Emily Franz, “Breaking Down the Residual Category: Policy Stability among Dictatorships from a Veto Players Perspective” (n.p., 2006),

7. Dan Reiter and Allan C. Stam, “Identifying the Culprit: Democracy, Dictatorship, and Dispute Initiation,” American Political Science Review 97 (2) (2003): 333–337.

8. Karl R. Popper, The Open Society and its Enemies (New York: Routledge, 1963), 124.

9. William H. Riker, Liberalism Against Populism (Prospect Heights, Ill.: Waveland Press, 1982), 6–7; emphasis added.

10. Popper, The Open Society and its Enemies, 124.

11. Giacomo Chiozza and H. E. Goemans, “Leaders and International Conflict” (unpublished book manuscript, University of Rochester and Vanderbilt University, 2010).

12. Amji Omara-Otunnu, Politics and the Military in Uganda, 1890–1985 (New York: St. Martin’s Press, 1987), 139–141.

13. Tonay Avirgan and Martha Honey, War in Uganda, The Legacy of Idi Amin (Westport, Conn.: Lawrence Hill & Company, 1982), 48–51; George Ivan Smith, Ghosts of Kampala (New York: St. Martin’s Press, 1980), 176–178.

14. Ibid., 178.

15. Quoted in Avirgan and Honey, War in Uganda, 52; emphasis original.

16. Ibid., 69.

17. Joseph Kamau and Andrew Cameron, Lust to Kill: The Rise and Fall of Idi Amin (London: Corgi Books, 1979), 306.

18. Mark Peceny, Caroline C. Beer, and Shannon Sanchez-Terry, “Dictatorial Peace?” American Political Science Review 96 (1) (2002): 15–26.

19. Reiter and Stam, “Identifying the Culprit,” 333.

20. Using a one-tailed test, the difference between personalist dictators and military dictators is just barely significant at 10 percent. The differences between personalist and single-party dictators, however, fail to reach conventional measures of significance. Nevertheless, the predicted probability of war involvement is about 0.040 for personalist dictators and only 0.025 for leaders of single-party dictatorships.

21. Michael Horowitz, Rose McDermott, and Allan C. Stam, “Leader Age, Regime Type, and Violent International Relations,” Journal of Conflict Resolution 49 (5) (2005): 661–685.

22. See Chiozza and Goemans, “Leaders and International Conflict”; Lorenzo Montúfar, Reseña Histórica. Centro-America, vol. Tomo Sexto (Guatemala: Tipografia “La Union,” 1887); Hubert Howe Bankcroft, The Works of Hubert Howe Bancroft: History of Central America, vol. III, 1801–1887 (San Francisco: The History Company Publishers, 1887); Gregorio Bustamante Mace, Historia Militar de El Salvador, 2nd ed. (San Salvador, El Salvador: Publicaciones del Ministerio del Interior, 1951); Robert L. Scheina, Latin America’s Wars: Volume I (Washington, D.C.: Brassey’s, 2003); Mario Rodriguez, Central America (Englewood Cliffs, N.J.: Prentice Hall, 1965).

23. These labels do not correspond to our current usage of the terms. Liberals were largely anticlerical and pro-trade.

24. Frederick Palmer, Central America and Its Problems (New York: Moffat, Yard & Company, 1910), 293.

25. Palmer, Central America and Its Problems, 292–294.

26. Reproduced in ibid., 307–317, Appendix A.

27. Reproduced in ibid, 316–317.

28. Scheina, Latin America’s Wars, 261.