10 WORLD POLITICS the relative power of the United States(in its capabilities if not in its will to control or shape events).These theoretically based distinctions require us to consider the effects of the Kantian variables within differ- ent international structures,though evaluation of the post-cold war era necessarily remains tentative. We omit from our analyses all but the first year of both World War I and World War II,because bilateral trade data for those years are frag- mentary,as they are for the immediate postwar years,1919-20 and 1946-49.Omitting all but the first year of the world wars,which con- sisted of conficts between democracies and autocracies or between two autocracies,biases our results against finding evidence of the demo- cratic peace,but it also provides assurance that our results are not de- termined by these dramatic but atypical events.23 Most of our variables and data are discussed in previous publications.Here we concentrate on what is new. DEPENDENT VARIABLE:INVOLVEMENT IN MILITARIZED DISPUTES We use the Correlates of War(cow)data on interstate disputes(MIDs). We code each year that a dyad was involved in a dispute in which one or both states threatened to use force,made a demonstration of force, or actually used military force against the other.The variable DISPUTE equals 1 if a dispute was ongoing and 0 if not.Some researchers urge that only the initial year of a dispute be noted since a dispute in one year increases the chances that the dyad will experience a dispute in subsequent years.24This procedure eases some problems but raises oth- ers.If leaders are rational,as all our theories assume,they will fre- quently reevaluate their positions,whether to escalate,deescalate,or maintain the existing strategy.We agree with Blainey:"The beginning of wars,the prolonging of wars and the prolonging or shortening of pe- effects of nuclear weapons would also imply a break between 1945 and all previous years of modern history.Dating the end of the bipolar cold war system is more problematic.Waltz's definition would argue for a break at the end of 1991,when the Soviet Union was dissolved.But William Dixon and Stephen Gaarder show a decisive shift in the pattern of Soviet-American conflict in 1988;see Dixon and Gaarder,"Presidential Succession and the Cold War:An Analysis of Soviet-American Relations, 1948-1992,"Journal of Politics 54,no.1(1992). 2 Farber and Gowa (fn.3)express this concern. 24 Stuart A.Bremer,"Dangerous Dyads:Conditions Affecting the Likelihood of Interstate War," Journal of Conflict Resolution 36,no.1(1993);Katherine Barbieri,"International Trade and Conflict: The Debatable Relationship"(Paper presented at the annual meeting of the International Studies As- sociation,Minneapolis,Minn.,February 1998);Nathaniel Beck,Jonathan Katz,and Richard Tucker, "Taking Time Seriously in Binary Time-Series-Cross-Section Analysis,"American Journal of Political Science 42,no.4(1998).See,however,our comment in fn.49 below
10 WORLD POLITICS the relative power of the United States (in its capabilities if not in its will to control or shape events). These theoretically based distinctions require us to consider the effects of the Kantian variables within differ ent international structures, though evaluation of the post-cold war era necessarily remains tentative. We omit from our analyses all but the first year of both World War I and World War II, because bilateral trade data for those years are frag mentary, as they are for the immediate postwar years, 1919-20 and 1946-49. Omitting all but the first year of the world wars, which con sisted of conflicts between democracies and autocracies or between two autocracies, biases our results against finding evidence of the demo cratic peace, but it also provides assurance that our results are not de termined by these dramatic but atypical events.23 Most of our variables and data are discussed in previous publications. Here we concentrate on what is new. Dependent Variable: Involvement in Militarized Disputes We use the Correlates of War (COW) data on interstate disputes (MIDs). We code each year that a dyad was involved in a dispute in which one or both states threatened to use force, made a demonstration of force, or actually used military force against the other. The variable DISPUTE equals 1 if a dispute was ongoing and 0 if not. Some researchers urge that only the initial year of a dispute be noted since a dispute in one year increases the chances that the dyad will experience a dispute in subsequent years.24 This procedure eases some problems but raises oth ers. If leaders are rational, as all our theories assume, they will fre quently reevaluate their positions, whether to escalate, deescalate, or maintain the existing strategy. We agree with Blainey: "The beginning of wars, the prolonging of wars and the prolonging or shortening of pe effects of nuclear weapons would also imply a break between 1945 and all previous years of modern history. Dating the end of the bipolar cold war system is more problematic. Waltz's definition would argue for a break at the end of 1991, when the Soviet Union was dissolved. But William Dixon and Stephen Gaarder show a decisive shift in the pattern of Soviet-American conflict in 1988; see Dixon and Gaarder, "Presidential Succession and the Cold War: An Analysis of Soviet-American Relations, 194$-1992,n Journal ofPolitics 54, no. 1 (1992). 23 F?rber and Gowa (fn. 3) express this concern. 24 Stuart A. Bremer, "Dangerous Dyads: Conditions Affecting the Likelihood of Interstate War," Journal of Conflict Resolution 36, no. 1 (1993); Katherine Barbieri, "International Trade and Conflict: The Debatable Relationship" (Paper presented at the annual meeting of the International Studies As sociation, Minneapolis, Minn., February 1998); Nathaniel Beck, Jonathan Katz, and Richard Tucker, "Taking Time Seriously in Binary Time-Series-Cross-Section Analysis," American Journal of 'Political Science 42, no. 4 (1998). See, however, our comment in fn. 49 below
KANTIAN PEACE 11 riods of peace all share the same causal framework.The same explana- tory framework and the same factors are vital in understanding each state in the sequel of war and peace."25 Moreover,we investigated 166 multiyear disputes during the post-World War II era and found that more than half involved a change in the level of force employed over the course of the dispute or that a new dispute arose as the first was concluding.Thus we report analyses of states'involvement in disputes rather than of just their onset;but as in earlier studies of the cold war era,26 we reestimated key analyses using only the first year of disputes without finding material differences from those reported below. DYADIC INDEPENDENT VARIABLES We lag all independent variables by one year to ensure that they were not affected by a dispute to be explained.For some explanatory variables this precaution is clearly important;for example,conflict may limit trade just as trade may constrain conflict.A similar reciprocal relationship can be imagined for international organizations and conflict,as many IGOs-though hardly all-are formed among states that maintain peaceful relations.For other variables such considerations are irrelevant.Geographically proximate countries are prone to conflict, but the frequency of their disputes does not affect their proximity.To be consistent,however,we lag all the independent variables.This precaution does not put to rest all questions about the direction of causality,but it is a reasonable step at this time.27 All the variables are listed by their acronyms in the appendix. DEMOCRACY We use the Polity III data to compute a summary measure of the political character of regimes,subtracting every country's score on the autocracy scale from its score on the democracy scale.The resulting variable(DEM)ranges from-10 for an extreme autocracy to +10 for the most democratic states.Because a dispute can result from the actions of a single state,the likelihood of conflict is primarily a function of the degree of constraint experienced by the less constrained state in each dyad.As that state is the weak link in the chain of peaceful 2s Geoffrey Blainey,The Causes of War,3d ed.(New York:Free Press,1988). 26 Oneal and Russett(fn.6,1999). 27Kim(fn.6),using a simultaneous equation model,finds that the effect of trade on conflict is much stronger than the reciprocal one.Russett,Oneal,and Davis(fn.6)construct a model for predicting IGO membership that includes,among other factors,the absence of confict.There is an effect,but it is weaker than the influence of IGOs on conflict
KANTIAN PEACE 11 riods of peace all share the same causal framework. The same explana tory framework and the same factors are vital in understanding each state in the sequel of war and peace."25 Moreover, we investigated 166 multiyear disputes during the post-World War II era and found that more than half involved a change in the level of force employed over the course of the dispute or that a new dispute arose as the first was concluding. Thus we report analyses of states' involvement in disputes rather than of just their onset; but as in earlier studies of the cold war era,26 we reestimated key analyses using only the first year of disputes without finding material differences from those reported below. Dyadic Independent Variables We lag all independent variables by one year to ensure that they were not affected by a dispute to be explained. For some explanatory variables this precaution is clearly important; for example, conflict may limit trade just as trade may constrain conflict. A similar reciprocal relationship can be imagined for international organizations and conflict, as many IGOs?though hardly all?are formed among states that maintain peaceful relations. For other variables such considerations are irrelevant. Geographically proximate countries are prone to conflict, but the frequency of their disputes does not affect their proximity. To be consistent, however, we lag all the independent variables. This precaution does not put to rest all questions about the direction of causality, but it is a reasonable step at this time.27 All the variables are listed by their acronyms in the appendix. DEMOCRACY We use the Polity III data to compute a summary measure of the political character of regimes, subtracting every country's score on the autocracy scale from its score on the democracy scale. The resulting variable (DEM.) ranges from -10 for an extreme autocracy to +10 for the most democratic states. Because a dispute can result from the actions of a single state, the likelihood of conflict is primarily a function of the degree of constraint experienced by the less constrained state in each dyad. As that state is the weak link in the chain of peaceful 25 Geoffrey Blainey, The Causes of War, 3d ed. (New York: Free Press, 1988). 26 Oneal and Russett (fn. 6,1999). 27 Kim (fn. 6), using a simultaneous equation model, finds that the effect of trade on conflict is much stronger than the reciprocal one. Russett, Oneal, and Davis (fn. 6) construct a model for predicting IGO membership that includes, among other factors, the absence of conflict. There is an effect, but it is weaker than the influence of IGOs on conflict
12 WORLD POLITICS relations,28 we expect that the less democratic state(DEM)in a dyad is the stronger determinant of interstate violence.Conversely,the more democratic that state,the more constrained it will be from engaging in a dispute and the more peaceful the dyad.In previous analyses we found,as Kant had expected,that the difference between states'political regimes also affects the likelihood of conflict.Democratic-autocratic dyads were the most conflict-prone in the cold war era;two autocracies were less likely to fight,and two democracies were the most peaceful. We reconsider these findings below. The Polity III regime scores exhibit some problems of comparability over time.Until 1918 about 40 percent of British males(dispropor- tionately working class)were disfranchised by residence requirements; female suffrage was granted partially in 1918 and fully only in 1928.29 In the United States women obtained the vote only in 1920,and blacks were systematically excluded until the 1960s.Swiss women achieved the franchise only in 1971.Some of these changes are reflected in the Polity data and hence in rising levels of democracy in the international system.For example,the United Kingdom goes from 6 to 7 on the de- mocracy scale in 1880,to 8 in 1902,and jumps to 10 only in 1922.But Switzerland is coded at 10 from 1848,as is the United States from 1871.The consequences of these restrictions on political participation for foreign policy may not be trivial.In the contemporary United States,for example,women are significantly more averse to the use of military force than are men and vote in part on this basis.30 Thus the exclusion of women from the franchise in earlier periods could have profoundly reduced the tendency of even the most"democratic"states to avoid conflict. 28 William J.Dixon,"Democracy and the Peaceful Settlement of International Conflict,"American Political Science Review 88,no.1(1994). 2 Trevor Wilson,The Myriad Faces of War:Britain and the Great War,1914-1918 (Cambridge,En- gland:Polity Press,1986),660-61;Kenneth MacKenzie,The English Parliament(Harmondsworth: Penguin,1950),106. 30 Carole Kennedy Chaney,R.Michael Alvarez,and Jonathan Nagler,"Explaining the Gender Gap in U.S.Presidential Elections,"Political Researcb Quarterly 51,no.2(1998).To take such changes into account,Zeev Maoz uses an adjusted threshold of democracy for all countries that shifts upward in 1870 (for general male suffrage)and 1920 (female suffrage);see Maoz,Domestic Sources of Global Change(Ann Arbor:University of Michigan Press,1996),54.Our use of unadjusted democracy scores thus leans against our hypothesis of democratic peace before World War I.Kristian Gleditsch and Michael Ward note that our continuous measure,Democracy minus Autocracy score,has the virtues of being symmetric and transitive;but the relative importance of its components is unstable over time; see Gleditsch and Ward,"Double Take:A Re-examination of Democracy and Autocracy in Modern Polities,"Journal ofConfict Resolution 41,no.3(1997).For the period 1880-1969 this aggregated mea- sure is largely infuenced by the degree of competition for executive recruitment;subsequently con- straints on the executive are the main determinant.Fortunately the relatively stable earlier period covers all the pre-cold war years we add here.As no analysis of the democratic peace after World War II has yet addressed the 1969 break,we too leave that for later investigation
12 WORLD POLITICS relations,28 we expect that the less democratic state (DEML) in a dyad is the stronger determinant of interstate violence. Conversely, the more democratic that state, the more constrained it will be from engaging in a dispute and the more peaceful the dyad. In previous analyses we found, as Kant had expected, that the difference between states' political regimes also affects the likelihood of conflict. Democratic-autocratic dyads were the most conflict-prone in the cold war era; two autocracies were less likely to fight, and two democracies were the most peaceful. We reconsider these findings below. The Polity III regime scores exhibit some problems of comparability over time. Until 1918 about 40 percent of British males (dispropor tionately working class) were disfranchised by residence requirements; female suffrage was granted partially in 1918 and fully only in 1928.29 In the United States women obtained the vote only in 1920, and blacks were systematically excluded until the 1960s. Swiss women achieved the franchise only in 1971. Some of these changes are reflected in the Polity data and hence in rising levels of democracy in the international system. For example, the United Kingdom goes from 6 to 7 on the de mocracy scale in 1880, to 8 in 1902, and jumps to 10 only in 1922. But Switzerland is coded at 10 from 1848, as is the United States from 1871. The consequences of these restrictions on political participation for foreign policy may not be trivial. In the contemporary United States, for example, women are significantly more averse to the use of military force than are men and vote in part on this basis.30 Thus the exclusion of women from the franchise in earlier periods could have profoundly reduced the tendency of even the most "democratic" states to avoid conflict. 28 William J. Dixon, "Democracy and the Peaceful Settlement of International Conflict," American Political Science Review 88, no. 1 (1994). 29 Trevor Wilson, The Myriad Faces of War: Britain and the Great War, 1914-1918 (Cambridge, En gland: Polity Press, 1986), 660-61; Kenneth MacKenzie, The English Parliament (Harmondsworth: Penguin, 1950), 106. 30 Carole Kennedy Chaney, R. Michael Alvarez, and Jonathan Nagler, "Explaining the Gender Gap in U.S. Presidential Elections," Political Research Quarterly 51, no. 2 (1998). To take such changes into account, Zeev Maoz uses an adjusted threshold of democracy for all countries that shifts upward in 1870 (for general male suffrage) and 1920 (female suffrage); see Maoz, Domestic Sources of Global Change (Ann Arbor: University of Michigan Press, 1996), 54. Our use of unadjusted democracy scores thus leans against our hypothesis of democratic peace before World War I. Kristian Gleditsch and Michael Ward note that our continuous measure, Democracy minus Autocracy score, has the virtues of being symmetric and transitive; but the relative importance of its components is unstable over time; see Gleditsch and Ward, "Double Take: A Re-examination of Democracy and Autocracy in Modern Polities," Journal of Conflict Resolution 41, no. 3 (1997). For the period 1880-1969 this aggregated mea sure is largely influenced by the degree of competition for executive recruitment; subsequently con straints on the executive are the main determinant. Fortunately the relatively stable earlier period covers all the pre-cold war years we add here. As no analysis of the democratic peace after World War II has yet addressed the 1969 break, we too leave that for later investigation