STATISTICS OF DEMOCIDE
War, Rebellion, and Democide*
By R.J. Rummel
The political, cultural, socio-economic, and geographic measures analyzed previously were largely structural. They reflect interrelationships among a state's institutions and physical characteristics and the values, meanings, and norms of its people. What is left out of this is the behavioral occasion for democide. That is the changes in the context of a regime that give it an excuse for democide, appear to necessitate democide, or challenge power such that democide seems an appropriate defense. Such is the breakout of international war or military action, domestic or foreign rebellion, revolution, anti-regime guerrilla warfare and terrorism, or a coup-d'etat.
Such warfare is theoretically related to democide in several ways. First, democide may be part of a strategy for achieving victory. Bombing and shelling cities indiscriminately, for example, may be thought necessary to terrorize an enemy people into pressing for an end to a war and to demoralize the base of the enemy regime's power. To deter sabotage and attacks on one's soldiers in occupied territory, reprisals involving the random shooting of the inhabitants may be believed effective. Blockading any food deliveries to an enemy region can be seen as hastening victory, even though thousands of civilians could starve to death. Kidnapping and killing leading sympathizers with enemy guerrillas may be considered unavoidable if the regime is to survive.
But separately from all this, involvement in an intense and passionately fought war enables a regime to further implement its ideological, racial, nationalist, or theological imperatives through outright democide or its intensification. Once fully engaged in a war the Nazis could further their central program of making Germany Jew-free by instituting at the highest government level the "final solution," and extending it to all occupied nations in Europe. During the Civil War in China, Nationalist Chiang Kai-shek tried to further his fascist-militarist program by eliminating critics and political opponents even though they had no connection with the communist enemy. Given all of Stalin's massive democide during the 1930s, intensification is perhaps not the word for his democide during the Second World War, but it is true that he did use the war to uncover more supposed counterrevolutionaries, bourgeois thinkers, economic wreckers, and untrustworthy communists and military men, murdering some 10,000,000 of them and other such unworthies during the war.1
Moreover, in the fog of war or rebellion democide may not only become practical, but war may eliminate the foreign protection for minorities or political movements, or fear of a foreign reaction were democide unleashed. Thus once Turkey was allied with Germany in World War I, which effectively removed the protection that the Christian Powers gave the Christian Armenian minority in Turkey, the Young Turk rulers could undertake their program to completely Turkify Turkey; and the post-war Atatürk nationalists used their wars against the Constantinople regime and Armenia and Greece to purge the country of Greeks and the remaining Armenians. Similarly, the Indonesian military in 1965 and 1966 used the ostensive communist coup against the Sukarno regime to rid the country of its communists and their sympathizers, murdering perhaps some 500,000 in the process.2 And then there is the lesser known democide in Burundi in 1971 and 1972 when after a minor rebellion of the majority Hutu, the ruling Tutsi minority had their excuse to lessen the ever present Hutu threat to their rule by murdering Hutu leaders, professionals, intellectuals, and students. Including others caught up in the terror, the Tutsi regime eventually killed some 150,000 in all.3
At the most general and taking into account these various ways warfare can cause, stimulate, or excuse democide, this is part of a general process within the social field I call the conflict helix.4 In short, a society at any one time is a balance of diverse powers, a simultaneous solution to the diverse equations of interests, capabilities, and will among contending groups and elites. And this balance supports a structure of expectations--laws, agreements, contracts, etc.--governing the society and its regime as to who gets what, when, and how in benefits or sanctions. However, in time one or the other group's or regime's interest, capabilities, or will may change, creating a gap between the structure of expectations and underlying balance of powers. This gap is a tension towards a change in the structure of expectations waiting for an excuse to explode behavior in one direction or another. Specifically, changes in the domestic structure of power between groups and individuals and in the interests and desired policies of the governing elite may not be transformed into actual policy because of the thick crust of existing expectations (rules, laws, contracts, implicit agreements, and the like) and routine behavior. And there may still be sufficient countervailing powers to be a threat to success. However, war or rebellion breaks this crust and routine. It invokes extraordinary laws and policies. It favors the regime conducting the war. It weakens countervailing powers, as of potential foreign opposition. It disrupts the existing structure of expectations and may further alter the balance of powers toward the regime. Thereby it often permits the implementation of policies that would have been impossible or inconceivable before the war. This was the case with the Young Turks, Hitler, the Indonesian Generals, and Burundi's Tutsi regime in 1971.
All this means that we should find a close relationship with war and rebellion over and above that of Power. Do we? The answer to this is given in part by the simple averages of the war, rebellion, and violence scores in Table 16A.1. These statistics are only for state regimes that have committed democide and are reproduced in Table 21.1 for different types of these regimes. Given that the scaling of the regimes for this violence was from 0 to 2, and that 2 in each case means that war, rebellion, or violence of some kind was concomitant to democide or occurred frequently enough to overlap or be related to democide, the higher the average the closer to a time connection between such violence and democide. If the average is over 1.0, then this violence is usually coincident with or just immediately preceding the occurrence of democide.
We can thus see from the table that for none of these types of regimes is democide generally coincident with war or rebellion. But when we consider both war and rebellion and look directly only at the coincidence between violence and democide in the last (stimulus) column, then we find that for both democratic and authoritarian regimes, democide usually occurs during or just after war or rebellion. Actually there is a perfect inverse correlation here with Power. (For the meaning of an inverse correlation and correlation in general, see Understanding Correlation.) The greater the Power of a regime, the more likely that genocide will be in time separate from war or rebellion, as for Stalin in the 1930s and Mao Tse-tung in the late 1950s; the less the Power the more likely a regime will commit democide, if it does so, during or right after a major war or rebellion, as for the American indiscriminate urban bombing in World War II and the largely British blockade of the Central Powers during and after World War I.
The averages of Table 21.1 reflect the coincidence of war or rebellion and democide, but what about correlations between democide and violence when those regimes apparently free of democide are included. Table 16A.1 gives the number of war and rebellion-dead for the 114 state regimes committing democide. These dead exclude democide during war. I have added to these statistics similar data from Small and Singer (1982) on battle-dead for the seventy-three regimes that when combined with those committing democide have comprised our 214 state-regime sample analyzed in the previous chapters.5 Because of the very large number of killed in war or rebellion for a few of the regimes these data were log transformed.6
Now, are the democide patterns we have already identified related to warfare or, in fact, does taking into account violence alter the patterns? Table 21.2 gives an immediate answer--yes. But before unpacking these results I again must be clear about what the correlations (loadings) with the patterns in the upper part of Table 21.2 mean. It is the democide and war-dead across regimes for their whole duration that is intercorrelated in the component analysis. Thus a war or rebellion may occur during an early period in the life of a regime and the democide in the final period. For example, the duration of the Soviet regime is from 1917 to past the final data collection year of 1987. There was a deadly civil war from 1918 to 1922 and the huge death toll during World War II. But unrelated to both of these was the millions that were killed in the collectivization campaign of the 1930s, the intentional Ukrainian famine, and the Great Terror. These deaths are a large part of the final 1917-87 democide toll for the Soviet Union and thus contributes to the correlation with the overall 1917-87 war-dead for the regime.
The results in Table 21.2 thus show the relationship between a regime's disposition to commit various kinds of democide--disposition measured by the occurrence of democide during the life of the regime--and its characteristic involvement in war. Belligerence may seem a better term for this involvement, but many nations fight wars not of their making, such as for Belgium where its involvement in both World Wars was forced upon it by the invasion of Germany. It would be misleading to say these wars index Belgium's belligerence. Disposition to democide and characteristic war or rebellion can be treated as traits. That is, they may not be manifest at any particular time nor together, but nonetheless describe behavior during the life of a regime. Moreover, by henceforth describing the results of Table 21.2 in terms of regime dispositions, or characteristics we keep in mind that the patterns shown do not necessarily reflect the temporal coincidence of behavior.
Now looking at Table 21.2 in detail, we still have the foreign democide pattern, but including war and rebellion in the analysis causes the domestic democide pattern to break up into two, one being the magnitude of domestic democide and the other an independent pattern of genocide and the domestic democide annual rate. The reason for this separation into two patterns is the very high correlation of characteristic rebellion with the magnitude of democide and virtually no correlation with genocide and the annual rate. That is, the amount of overall democide is highly related to the number of people characteristically killed in rebellions, but this in turn has little to do with a regime's disposition to commit genocide or its annual rate of domestic democide.
Nor is the characteristic intensity of war related to genocide either. Apparently regimes in general plan and implement their genocidal policies independent of the characteristic occurrence and intensity of their wars. This finding requires careful digestion. It does not mean that all genocides are independent of a regime's tendency to be involved in other forms of violence. This is patently false, as the Nazi and Young Turk cases, among others, would attest. But it is to say that genocide is carried out even by regimes that have had or will have few or no war-dead
Finally, the foreign democide pattern not only retains its identity, but is also very highly correlated with a regime's war-dead. That is, in general regimes who murder foreigners are those that also have been involved in war, the more war-dead they have characteristically suffered the greater the number of foreigners they generally have killed.
The question now is whether a regime's characteristic war and rebellion effect the relationship between its Power and democide. Table 21.3 gives the answer, which is no. We see virtually the same structure of relationship between democide and the political measures. But in addition we find that domestic democide splits into two uncorrelated patterns, one involving TotalPower and the other characteristic rebellion (this independence even holds for oblique rotation, where as given in Table 23.1 the correlation between the two parts of the domestic democide pattern--factors 1 and 4--is .23, or about 5 percent variation in common).
Because of the strong relationship between democide, politics, war and rebellion, we should expect that characteristic war and rebellion might influence the relationship with the other indicators we have considered. Table 21.4 displays a component analysis of all these indicators together, and the usual four-factor orthogonal solution. Although the factor order is changed (which is not of substantive importance here), we still have the two democide patterns, with totalitarian power related to domestic democide. And characteristic rebellion is, as one would expect from Table 21.3, also related to this pattern. (To understand component analysis and its patterns, see "Understanding Factor Analysis".)
Even within the context of a regime's overall war and rebellion, diversity has virtually no relationship with any kind of democide. Nor does culture, except for a moderate relationship between whether regimes are or are not in Central and South American. These Latin regimes tend to have less foreign democide, have less national power and, especially, less characteristic war-dead. This pattern strongly involves foreign democide and characteristic war-dead, the same pattern we saw in Table 21.3, and includes the indicator of national power as well, the relationship we saw in the last Chapter.
One important consideration remains. How does Power enter into the relationship between the characteristic severity of war, rebellion, and the commission of democide. Figure 21.1 illustrates this. I have argued elsewhere and I believe the empirical evidence supports that democracies do not make war on each other and that the less democracy in a totalitarian direction the more likely foreign and collective domestic violence (as measured by those killed).7 That is, Power not only causes democide directly, but also causes war and rebellion. And thus through them, Power is also an indirect cause of democide.
So far we have found that democide is dependent on Power, war-dead, and rebellion-dead. Let us now test for the dependence of war and rebellion-dead on Power. To do this adequately we must recognize that this dependence is probabilistic in this sense. The more the Power, the greater the disposition to fight very bloody wars or rebellions. This does not mean that where a regime has near absolute Power it will necessarily have such violence. This is a disposition. It may be inhibited by contextual factors (such as the military power of a potential victim or enemy). The flip side of this is that when such near total wars or rebellions do occur they probably were caused by and centrally involved regimes with near absolute Power. Operationally, this means that I cannot rely on correlations alone to test this, but must also study the actual distribution of war and rebellion-dead for different levels of Power.
First, the product moment correlations. That of Power with the log of war-dead is .16 and with rebellion logged is .27 (and partialling out national power does not significantly change the correlations). Both are significant (one-tailed) in the theoretical direction at p<.01.
Next look at the plots in Figure 21.2 for both war and rebellion-dead. The line in each plot rising to the right is the regression line. As can be seen in both plots the highest number of dead for each level of Power tends to rise to the right also, as it should. But also, as is to be expected, there are a large number of regimes near or at zero dead for all ranges of Power. Not all near or actual totalitarian regimes have made or been involved in war or rebellions. The best way of illustrating the underlying effect of Power in line with the theory is to display only the highs for each level of Power, as by the area plots in Figure 21.3.
The area plots generally follow the theory: characteristic war and rebellion tend to get more extreme with increasing Power. The war-dead highs for the democracies at the left of the plot are due to the great number of dead the democracies suffered in World War II, a war thrust on them by totalitarian Powers. For this reason, no one of these charts will be fully satisfactory until we are able to code the war-dead according to whether they are for the aggressor or victim. This would show, I believe, a much greater tendency of the war-dead to be associated with near absolute Power.
Leaving this aside, I also collapsed the totalitarian power scaling into first five-groups and then two and plotted the average dead for each level of Power, as shown in Figure 21.4. Except for the aforementioned battle-dead of the democracies in World War II which causes the mean to be high for the lowest Power level in the first plot, all the other means in the four plots are in order of increasing Power.
In sum, then, we have found in this Chapter that there is a very close relationship between a regime's disposition to commit democide and its likelihood of being involved in bloody wars and rebellions. This, however, has no or little effect on the other relationships we have found between democide and the indicators of politics, diversity, culture, society, and geography. This means, as we further explored and verified, that war and rebellion are also a product of Power, rather than independent of it, and that the causal network is as drawn in Figure 21.1.
This is not the end of analysis, however, for we still have to define more precisely than component analysis allows the dependency of democide on these indicators. Moreover, the reader has a right to ask how well the theoretical and empirical results enable us to group regimes in terms of their actual democide. That is, can we predict (postdict actually) what regime did or did not commit democide and to what degree? These are the tasks of the next Chapter
* From the pre-publisher edited manuscript of Chapter 21 in R.J. Rummel, Statistics of Democide, 1997. For full reference to Statistics of Democide, the list of its contents, figures, and tables, and the text of its preface, click book.
1. Rummel (1990, Table 7A, line 433).
2. See Table 14.1C, line 1134.
3. See Table 14.1A, line 240.
4. Rummel (The Conflict Helix: Principles and Practices..., especially Chapter 13)
5. In generating the war-dead figures for regimes that have committed democide I included Small and Singer's (1982) figures along with that of others. Their battle-dead totals were usually close to what other sources defined as war-dead. Moreover, they limit their collection to wars or rebellions in which 1,000 or more were killed. This thus creates a slight bias at the very low violence end for regimes that did not commit democide. But I compensate for this by defining war and rebellion in terms of number killed, rather than simply by counting the number of wars and rebellions, and by transforming all war-dead to logarithms.
6. The Small and Singer (1982) battle dead totals stop at 1980. I have filled in subsequent years from diverse sources, as listed in the various democide tables (e.g., Table 14.1A, and Table 15.1A). I also employed in subsequent analyses war and rebellion-dead divided by population. This per capita measure of intensity usually accounted for less variance and had less significant relationships than did the log10(x+1) transformation of war and rebellion. Moreover, I also redid the subsequent analyses without logging war and rebellion, but with less meaningful results.
7. Rummel (Vol. 4: War, Power, Peace; "Libertarianism and International Violence"; "Libertarianism, Violence Within States, and the Polarity Principle"; and Power Kills).