HomeDocuments on SitePersonalDemocratic PeaceDemocide20th C. DemocideMegamurderersLesser MurderersWhy DemocideDimensionsConflictMethodsTheoryPolicyLinks PHOTOS OF DEMOCIDEGalleries

Other Democide Related Documents On This Site


"Democide vs genocide. Which is what?"

"War isn't this century's biggest killer"

"How many did communist regimes murder?"


"Democide in totalitarian states: mortacracies and megamurderers"

"The Holocaust in comparative and historical perspective"

Graduate Syllabus on Repression and Democide


"Power kills: genocide and mass murder"


Lethal Politics

China's Bloody Century

Democide: Nazi Genocide and Mass Murder

Death By Government

Statistics of Democide (entire)


By R.J. Rummel


From 1900 to 1987 state, quasi-state, and stateless groups have killed in democide (genocide, massacres, extrajudicial executions, and the like) near 170,000,000 people. Case studies and quantitative analysis show that ethnic, racial, and religious diversity, economic development, levels of education, and cultural differences do not account for this killing. Rather, democide is best explained by the degree to which a regime is empowered along a democratic to totalitarian dimension and secondarily the extent to which it is characteristically involved in war or rebellion. Combining these results with those that show that democracies don't make war on each other, the more democratic two nations are the less foreign violence between them, and that the more democratic a regime the less internal violence, strongly suggests that democracy is a general method of nonviolence.

Political regimes--governments--have probably murdered near 170,000,000 of their own citizens and foreigners in this century, about four times the number killed in all international and domestic wars and revolutions.1 Why? I will offer both a theory and empirical results on this question and then sketch the variety of tests of the theory that were conducted. But first, I will have to define what I mean by government murder and in doing this propose an appropriate concept.

A concept that has done yeoman service in denoting government murder is genocide. But this concept hardly covers the variety and extent of ruthless murder carried out by governments. To be more specific, in international conventions and the general literature, genocide has been defined in part as the intentional killing by government of people because of their race, religion, ethnicity, or other indelible group membership. Cold-blooded government killing, however, extends beyond genocide so defined, as of starving civilians to death by a blockade, assassinating supposed sympathizers of anti-government guerrillas, purposely creating a famine, executing prisoners of war, shooting political opponents, or murder by quota (as carried out by the Soviets, Chinese communists, and North Vietnamese).

To cover all such murder as well as genocide and politicide I use the concept of democide.2 This is the intentional killing of people by government. It excludes the killing of those with weapons in their hands or those indirectly killed as a result of military action and excludes judicial executions for what are normally considered capital crimes, such as murder and treason (unless such are clearly excuses for the executions, as of Stalin's show trials in the 1930s).

Democide is meant to define the killing by states as the concept of murder does individual killing in domestic society. Here intentionality (premeditation) is critical. This also includes practical intentionality. If a government causes deaths through a reckless and depraved indifference to human life, the deaths were as though intended, as in the deadly Soviet forced labor camps.

It is democide that I will try account for here.


The theoretical hypothesis is that the more democratic freedom a nation has, the less likely its government will commit foreign or domestic democide. In brief summary,3 the theory is that through democratic institutions social conflicts that might become violent are resolved by voting, negotiation, compromise, and mediation. The success of these procedures is enhanced and supported by the restraints on decision makers of competitive elections, the cross-pressures resulting from the natural pluralism of democratic--spontaneous--societies, and the development of a democratic culture and norms that emphasizes rational debate, toleration, negotiation of differences, conciliation, and conflict resolution. Moreover, democratic leaders see others, even political opponents, as within the same moral universe, as equally nonviolent, as disposed to negotiate differences peacefully.4

On the other side are totalitarian political regimes. Rather than being a means for resolving differences in views, they try to impose a particular ideology, religion, or solution to social problems on society, regardless of the opposition. For this reason such regimes try to control all aspects of society and deal with conflict by force, coercion, and fear, that is, by power. Moreover, such power breeds political paranoia by the dictator or within a narrow ruling group. This is the fear that others are always plotting to take over rule and would execute those now in power. Finally, there is one hierarchical pyramid of power rather than a multitude of such pyramids as in a democracy, one single coercive organization. This turns all socio-political and economic issues and problems into a matter of us versus them, of those with power versus those without. We should therefore find that the less democratic a regime, the more unchecked and unbalanced power at the center, the more it should commit democide. Democide becomes a device of rule, as in eliminating possible opponents, or a means for achieving one's ideological goal, as in the purification of one's country of an alien race or the reconstruction of society.

There is thus a scale of political regimes from the most democratic to the most totalitarian, from freedom in terms of political and civil rights and liberties to an absolute power under which such rights and liberties do not exist. And we should find empirically that the more democratic the less violence in foreign and domestic affairs, the more totalitarian the more violence. So far this equation between the scales of power and violence has been empirically supported. We find that democracies do not (or rarely) make war on each other, that the more democratic and less totalitarian two regimes the less foreign violence between them,5 and democracies have the least domestic violence (Rummel, "Libertarianism, Violence Within States, and the Polarity Principle", "Libertarian Propositions on Violence Within and Between Nations: A Test Against Published Research Results", 1995a). By this theory power also should be directly predictive of democide such that the less democratic a regime along the democratic to totalitarian scale of power, the more likely it will commit democide.

To hypothesize that democracy is inversely related to democide leaves open the question as to how this democide is measured. I argue theoretically that democracy will be most related to the total magnitude of domestic democide logged. First, the more democracy the less the number of its own people the regime will kill for the reasons given. However, these restraints do not well operate in times of hot or cold war, when the military or intelligence services operate in secret, and in their foreign operations are in effect totalitarian enclaves within a democratic structure. This is because in wartime the democracies largely give the military their head, secrecy prevails, wartime controls over the nation are instituted, and it is an open question whether democracies in wartime retain their full democratic character (consider the internment of Japanese-Americans in concentration camps during World War II, for example), particularly in their foreign operations. For this reason I argue that the primary inverse relationship between democracy and democide is to domestic killing. This is not to say that democracies will murder as many noncombatants in wartime as will nondemocracies, which they clearly have not, but that the correlation between democracy and foreign democide will be much lower than for the domestic or total amount of democide [on the nature of correlation and the correlation coefficient, see Understanding Correlation].

Second, the closer one gets to democracy on the hypothetical democracy-totalitarianism scale, the more the restraints on democide should kick in. This is because even a moderate liberalization of a totalitarian regime, as after the death of Stalin or Mao tse-tung, creates countervailing forces that makes democide difficult to carry out or less the approved means to achieve policies. When power at the center is limited by tradition or other power groups, even in authoritarian regimes such as those of Saudi Arabia, Iraq, or Iran, the ability to kill unlimited numbers of people is sharply limited. We should expect, therefore, that as regime types vary from democratic to somewhat democratic to authoritarian to somewhat totalitarian to totalitarian, that there should be a virtual logarithmic increase in the number of people a regime kills.

And finally, this should not only be an absolute relationship between the democratic-totalitarian scale and domestic democide, but also one with the number killed as a proportion of the population per year of the regime--the rate of democide. Democratic restraints should operate not only on the sheer number killed over the life of a regime, but also on the relative number killed. That is, the inverse relationship between democracy and democide should hold regardless of the duration of the regime and its population.

Finally, we should expect that the relationship between democracy and democide should be greatest with domestic democide (on which the democratic restraints will have their greatest effect), secondly with total democide, which includes foreign, and there should be little if any relationship to foreign democide by itself.


I have gone through five research stages to test the hypothesis that democracy is causally and inversely related to democide. Specifically for this test, I collected data on all democide for all regimes, 1900-1987, for which estimates in English were available in the literature.6

Second, I delineated the dimensionality of these data through factor analysis [for a description of factor analysis, see "Understanding Factor Analysis"].

Third, attending now to the hypothetical independent variable, I determined various ways of measuring democracy over the same years for different regimes. I then used factor analysis to define the prime indicator of the theoretical democracy-totalitarianism continuum.

Fourth, I collected data on a number of a control variables, particularly those defining cross-national socio-cultural diversity, culture, war and rebellion, wealth, and power. I also separately factor analyzed these data to uncover their major indicators and reduce the number of variables and their multicolinearity in the tests.

Finally, I then applied factor analysis, interactive multiple regression analyses, canonical analyses, and time series regression to test whether of all indicators the democracy-totalitarian one best accounted for democide, as it should. It did, regardless of the controls or type of tests.

Because of the sheer magnitude of the analyses--just one of the factor analyses or canonical analyses by itself could have constituted an article, I will have to be very brief in presenting the most important results. I will try to be as explicit as possible where it is most important to be so, which is in the actual tests.


Turning now to the data on democide, the operational question is whether total, domestic, and foreign democide, and the rate of democide (as defined above) are empirically different patterns in the democide data. This is already a kind of test of the above theory, since if these types of democide are highly intercorrelated, then democracy cannot be both highly related to domestic democide and largely unrelated to foreign democide, as the theory suggests.

To determine this I need to define different foreign and domestic democide types--variables. In doing this, three criteria are important. One is that these types are conceptually and empirically meaningful. The second is that they can be identified among the flow of events and especially in the fog of war and violence. And the third is that there are data that can be so defined. The types consistent with these requirements are listed in Table 1.

In total 218 regimes (141 state regimes and seventy-seven quasi-state and group regimes) committed some sort of democide in this century for which I could find estimates, no matter how small. How many state regimes did not commit democide? This is a difficult question, simply because it requires that all regimes existing during this century be identified. Now, as used here a regime is a government that is identified by certain political characteristics that exist for a specifiable period. These characteristics define the nature and distribution of a regime's coercive and authoritative power and the manner in which this power is exercised and power-holders changed. For example, the change of regime from the rule of the Czar over Russia to the Kerensky government, and then within the same year to the Bolsheviks gives us three regimes. The change from the Kaiser monarchy to the Weimar Republic to Hitler's rule also gives us three different German regimes. Mainly but not completely relying on Ted Robert Gurr's (1990) political characterization of regimes (polities) from 1800 to 1986, I count 432 distinct state regimes during 1900-1987.7 And 141 of these, or about one-third, committed some form of democide.

Let us now look at how this democide is empirically patterned across the 432 regimes. By a pattern is meant the intercorrelation of certain types of democide such that when a regime kills people in one kind of democide there is a high probability that it also will have committed or will commit the other kinds of democide. Ideally, this intercorrelation--pattern--should be so defined that the influence of other patterns is statistically partialled out.

I must make clear that the various democide types are totaled over the life of a regime. For regimes surviving for only a couple of years, the different types of democide are probably simultaneous. For very long lived regimes, such as the Soviet Union or United States, different types of democide and even the different occurrences of democide for a particular type, may have been committed in years separated by decades or even half a century or more. A high correlation, then, between two democide types, such as terror and genocide, should be interpreted to mean that a regime characteristically committed both types of democide or that both are characteristic behavior of the regime, not that both types were simultaneously committed. A pattern of interrelated democide types then means that these are interrelated behavioral characteristics of regimes.

Moreover, since a few regimes may have very high democide compared to others (e.g., the Soviet Union and China), all data were log transformed (base 10), as indicated by the suffix "L" added to the type's name. To be sure this is kept in mind, "L" is added to the type's name.

Using component analysis I identified the patterns of democide just for the 218 regimes with democide and also only for the 141 state regimes.8 For both analyses (not shown) the patterns were much the same. Since the state regimes will be the focus of subsequent analysis and tests, I will concentrate on these patterns.

The fourteen democide types reduces to five empirical patterns (or dimensions) of democide and their five indicators (italicized).9 The first and most important pattern centrally involves domestic democide, with which is correlated terror, massacres, the domestic democide rate, and total democide. The second and statistically independent pattern is one of foreign democide, and also including forced labor deaths, camp deaths, and POWs killed. Two other independent patterns comprise the annual domestic democide rate and democidal bombing deaths. The final pattern is that of mainly genocide and secondarily massacres.

The italicized indicators are now our fundamental measures of democide. They will be the basis of all subsequent analyses of democide and should be looked at as fundamental causal foci. That is, each empirical pattern reflects underlying first order causes and conditions that differ from those related to other patterns.10 Note therefore that since genocide is a statistically independent causal pattern in the democide data, the causes that underlay this form of mass murder are separate and distinct from those causing the other type of government killing. This is not to deny that there is an overall explanation for democide in general, but that within this general explanation there are particular patterns of democide, such as genocide, explained by more specific causes and conditions.

As mentioned, this component analysis is also an initial test of one aspect of the hypothesis. The expectation was that democracy would have its greatest inverse relationship to domestic democide and little to the foreign. For this to be true, domestic and foreign democide must be near independent empirical patterns in the data. This is indeed the case, where the component analysis and orthogonal and oblique rotations showed domestic and foreign democide to be different, uncorrelated, dimensions of democide.11 While this does not substantiate that it is democracy that is the cause, this separation of the patterns is a necessary condition for the hypothesis, as elaborated, to be true.


To keep the data collection manageable for all subsequent analyses the sample will have to be limited to: (a) all 141 state regimes committing democide, and (b) those state regimes not committing democide that (c) involve a large shift in power from previous or succeeding regimes. For example, although the communist Afghanistan regime (1978-) is included because it committed democide, a previous non-communist regime (1965-73) is also included. Austria committed no democide, as best I can determine, but it had two very different regimes, one autocratic regime in the pre-Hitler takeover period, 1934-38, and the other the post-Second World War democratic regime, 1946-. Both Austrian regimes are included. I also tried to pick regimes such that all major cultures, national characteristics, socio-economic attributes, and regime variation would be represented. This selection procedure gives me a sample of 214 state regimes, including the 141 with democide, for the years 1900-1987. Hereafter, this is the basic sample for all analyses.

In Table 2 I list a variety of political measures that in one way or another define regime types.12 The question now is which of these measures centrally define the theoretical democratic-totalitarian scale that is supposed to predict to democide, if indeed, there is such an independent empirical political pattern. These measures were selected to span the variety of regime types, whether liberal democratic, absolute monarchies, communist, noncommunist totalitarian, military dictatorships, oligarchic republics, or personalist autocracies.

Table 3 shows the five statistically independent political patterns that emerge from a component analysis for these 214 regimes. As shown, there are five major patterns.13 The substantive nature of these patterns is identified by the coefficients (loadings) in the matrix, which give the correlation between the democide types and the pattern. Squaring these correlations then defines the amount of variation in the democide related to the pattern. I have outlined in the table each of these correlations for which there is 25 percent or more covariation between political type and pattern. As can be seen (similar types are set off by horizontal lines), the first and most important of these involves a democratic to totalitarian continuum, or looking at all the measures correlated with the pattern, a continuum measuring the degree to which coercive regime power penetrates and controls political and socio-economic institutions, functions, and individual behavior. To keep this idea foremost, I have named this the Totalitarian Power pattern. Both the democratic and totalitarian scales are among those measures most highly correlated with this pattern. Its indicator will be constructed by adding the totalitarian scale to the inverse of the democratic scale, with the result that totalitarian regimes will be at the high end.14 The resulting indicator I will simply call TotalPower.

There is also a Political Power pattern (Factor 5) which should not be confused with Totalitarian Power and is largely statistically (and conceptually) independent of it. The political power measure defines this pattern and indexes the degree to which political power is centralized, politically autocratic, or dictatorial, without any elector system, legislature (rubber stamp or not), or other representative body. TotalPower well reflects this centralization of political power in totalitarian systems, of course, but also the regime's penetration of and control over the nonpolitical aspects of society as well, such as religion, the economy, and culture, which is not measured by the Political Power pattern alone.

There is much confusion in the literature between totalitarian and political power that must be clarified here. Because of the lack of any electoral system and even nominal representative body, authoritarian regimes like that of Saudi Arabia may have an higher score on political power than the Soviet Union. And because of the lack of any meaningful legislature or other control over the executive, a regime like that in Kuwait with an absolute monarch is often coded with greater political power than many communist countries where a legislature exists, all be it largely a rubber stamp, and where a politburo may provide some executive restraints as in the Soviet Union of the 1970s. For this reason many scales of democracy will position communist countries closer to the democratic end then the absolute monarchies or dictatorships without any legislature or electoral system. The political power scores used here are primarily based on the work of Ted Robert Gurr (1990). He codes the political power of each regime as a combination of its regulation of participation and executive recruitment, the competitiveness of executive recruitment, the constraints on the chief executive, whether the executive is monocratic or not, and the centralization of the state. Note that all these are political characteristics and do not define, for example, the degree to which there is a command economy, or regime control over the media, religion, or other non-political institutions.

There is also an authoritarian versus totalitarian pattern (see Factor 2 of Table 3), fundamentally the opposition between the two.15 Both types of regimes are nondemocratic, but they differ sharply in the degree to which power regulates and controls all of society. We have here the same distinction between totalitarian and political power, but now largely limited to nondemocratic regimes.

Of the remaining two patterns listed in Table 3, Factor 3 defines absolute monarchies and Factor 4 reflects the power of a society's traditional elite (clan or church leaders, historic economic elite, chiefs and tribal leaders, aristocrats, etc.).


With these indicators defined, what are we to expect of the relationship between democide and the five political patterns? Foremost, as noted, TotalPower should have the highest positive relationship to the domestic democide pattern, which also includes, secondarily total democide. The more TotalPower, the more democide. Second, TotalPower should also have a positive but moderate relationship to the annual democide rate and genocide patterns. Genocide is a more specific democidal behavior and thus more effected by idiosyncratic causes and conditions, while the annual democide rate is partially dependent on a regime's population and duration, neither being characteristics much influenced by totalitarianism. For domestic democide, the annual rate, and genocide, the political power of a regime should be second in relationship. It reflects an important aspect of power, but not the absolute totalitarian power that is most democidal.

To test these expectations, we might simply consider the product moment correlations between the democide and political indicators. In fact by far the highest correlation between the political and democide measures is .55 for TotalPower and domestic democide. But because this correlation and the others are influenced by the interrelations among all the indicators, they can be only suggestive. The best way of untangling (partialling out) the interrelationships among the correlations and defining the independent lines of causation is through component analysis.

Table 4 shows the unrotated and orthogonal (statistically independent) components (factors).16 Each reflects an independent causal pattern or nexus, where the largest and most pervasive one is that of the first unrotated component. But the unrotated component often obscures lesser patterns that might be more theoretically important. For this reason the components should be rotated to see if these other patterns are present in the data and to best define the tightest interrelationships among them. Note also that whether they are rotated or unrotated, each of these rotated factors will delineate a causal nexus such that the influences involved in the other factor patterns are partialled out.

It is therefore important that we find that the most general causal pattern, the first unrotated factor, most centrally involves domestic democide, and secondarily the annual rate and genocide. And the only political indicator included is for TotalPower. On rotation, this causal nexus is more clearly defined, with political power now playing a secondary role. [stuff excluded]

Aside from this cluster and looking again at Table 4, we find that foreign democide, including bombing, forms a pattern by itself, as also does the three authoritarian type indicators.

The causal weight of TotalPower in accounting for domestic democide can be visually displayed by disaggregating it into the democratic and totalitarian scales of which it is composed, and graphing domestic democide against both of them. The resulting three-dimensional surface is drawn in Figure 1.17

There are several things to note about this surface. At the democratic corner it shows virtually no domestic democide for both scales. Then as we move away from the democratic corner toward either opposing end, democide increases.18 Moreover, the mid-surface--the joint effect of the democracy and totalitarian scales, or TotalPower-- is almost uniformly slanted upward until it approaches the diagonal corner from democracy and then curves upward even more. This means that TotalPower squared rather than TotalPower alone should be more predictive of domestic democide in regression analysis, which in fact is the case as we will see.

There another way to test and better understand the hypothetical relationship between democide and TotalPower, which is shown in Figure 2. For this the 18 point TotPower indicator of was divided into five groups, such that the low and high groups comprised the lowest and highest scale values, the mid-group the five mid-scale values, and the rest distributed between the low-mid and high-mid groups. The resulting plot of group means is almost perfect. It curves upward continuously to absolute totalitarian power.


Consider now 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 the best 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'état.

Such warfare is theoretically related to democide in several ways. First, democide can become part of the strategy for achieving victory. Bombing and shelling cities indiscriminately, for example, is believed to terrorize the enemy people into pressing for an end to war and to demoralize the base of the enemy regime's power.

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 domestic democide or its intensification. For example, once fully engaged in a war the Nazis could further their central program of making Germany Jew-free by instituting the "final solution," and extending it to all occupied nations in Europe.

Blending with this rationale for democide is the excuse that war and rebellion give to initiate large scale democide. Minorities or opposing political movements or parties may have been perceived by the power elite as a long-standing threat to the regime. In the fog of war or rebellion democide may become practical, or war may eliminate the foreign protection for such groups 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 Armenian minority in Turkey, the Young Turk rulers could undertake their program to completely Turkify Turkey through genocide.

We therefore should find a close relationship over and above that of TotalPower with war and rebellion. Do we? For each of the 141 regimes committing democide I determined from a variety of sources the number of their war and rebellion dead, excluding their democide dead during these wars.19 I have added to these statistics similar data20 on battle-dead for those regimes without democide that comprised my 214 state-regime sample. Because of the very large number of killed in war or rebellion for a few of the regimes these data were log10 transformed.

Before continuing I should clarify what the correlations between war-dead, rebellion-dead, and democide will mean empirically. These correlations are for the democide and war-dead or rebellion-dead at any time in the life of a regime. Thus a war or rebellion may occur during a regime's early period 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 1941-1945 participation in 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 contribute to the correlations with the overall 1917-87 war and rebellion-dead for the regime.

The correlation between war or rebellion-dead and democide thus define 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 or rebellion. 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.

Now, are the democide patterns we have already identified related to characteristic warfare? A component analysis (not shown) gives an immediate answer--yes.21 War-dead is correlated .86 with a foreign democide and bombing pattern and rebellion-dead is correlated .93 with an independent pattern of domestic democide. But war and rebellion causes the domestic democide pattern to break up into two, one being the magnitude of domestic democide related to rebellion-dead 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 in general 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 generally 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 considerable genocide is carried out even by regimes that have had or will have relatively few or no war-dead, as the current genocide of perhaps 500,000 people in Rwanda within a month attests.

The question now is whether a regime's characteristic war and rebellion affect the relationship between its TotalPower, our measure of the democracy-totalitarian scale, and democide. A component analysis (not shown) says no. There is 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 TotPower and the other characteristic rebellion-dead.


It may well be that the relationships between TotalPower, war-dead, rebellion-dead and domestic and foreign democide may be due to or moderated by a regime's socio-economic, cultural, and physical environment. To test this I followed four steps. First I collected data on eight measures of a nation's social diversity (religious, racial, ethnic, linguistic, etc.), fourteen measures of a nation's culture (Catholic society, Moslem society, English influence, cultural/geographic region, etc.), and twenty-one measures of a nation's socio-economic (GNP/capita, energy consumption per capita, educational level, etc.), demographic (population, density, etc.), and geographic (size, percent arable land, etc.) basis. Second, I did separate component analyses (not shown) on the democide types and each of these three kinds of measures. The only meaningful relationship found was between an indicator of national power (measured by a regime's population) and foreign democide. This makes sense, since powerful nations have the capability and greater opportunity to make war, and as we have seen, war is itself related to foreign democide. Third, I redid these component analyses now including the political indicators, with the result that the fundamental relationship between TotalPower and domestic democide remained unaffected. And fourth, a full scale component analysis (not shown) was done with all the democide types, political indicators, war and rebellion-dead, and the indicators of the diversity, culture, socio-economic, and geographic environment of regime behavior. The result was that the correlation between TotalPower, war-dead, rebellion-dead, and the domestic democide and foreign democide patterns remained largely the same.


The result so far is that for state regimes in general there is virtually no relationship of democide to cross-national diversity, culture, religion, regional variation, economics, education, health, transportation, demography, and geography. Overall eighty-three measures22 were analyzed, twenty-four independent patterns delineated, their best indicators selected, and all these indicators used to define the relationships between the democide committed by regimes and their attributes and context. As a result we can say that the dominant pattern of democide, that centrally involving domestic democide, is exclusively related to patterns of power and the likelihood of rebellion against a regime. This so far has been a positive test of the hypothesis that democracy is inversely related to democide.

The aim now is to more specifically test this employing interactive multiple regression analysis and using the above defined indicators as controls.

Table 5 presents the results of a regression analysis of the distinct patterns of democide previously identified, excluding bombing, and also including overall democide. In addition, because the effect of TotalPower appeared nonlinear, as shown in Figure 1, and the same was possible true for war-dead and rebellion-dead, these were also squared in the regression. Moreover, since foreign democide may well be impacted by the interaction between national power and war-dead, I also included national power times war-dead.23

I did a forward interactive regression, which began with the twenty-four independent variables listed in the table, their partial correlations, F-ratio, and significance level for each of them, with none entered into the regression. I successively selected those to enter or subsequently to remove from the regression, calculating the regression results for each entry or subsequent removal, until I had the best multiple R, regression coefficient t-tests, and intuitive and theoretical substantive fit, and no significant measures left to enter. The regression results shown in the table are the end result of this, where the multiple R and independent variable t-tests listed are only for the final regression. If there are no t-tests given for an independent variable, then it was not included in the final regression.24

Now looking at the results of the table, overall democide (logged) is best accounted for by the power and violence indicators and their interaction terms, and secondarily by refugees and the family basis of the social structure (i.e., social modernization). Six indicators in all account for almost three-quarters of the variation (R2) in democide across all 214 regimes, a remarkable result. And among all independent variables, as theoretically expected, the indicator most significantly accounting for this is TotalPower squared. Again, as for the results described previously, but now treating all the other variables as controls, it is not only that the greater TotalPower the more democide, but the greater the TotalPower the more its effect is multiplied. This effect also holds true for domestic democide alone (the second column of results shown in the table) and, less so, for genocide and the annual rate. The tightest relationship is between TotalPower squared and domestic democide, secondarily with total democide, thirdly with the democide rate, as predicted. Moreover, for each of these TotalPower squared is the best predictor. All the results so far supported the hypothesis, but these regression results are the most direct and persuasive.

As also predicted, and consistent with the previous findings, neither TotalPower or TotalPower squared significantly explain foreign democide. Why this should be so has already been explained, but to be sure this is understood some elaboration might be helpful. Totalitarian power and the other measures for regimes have been defined as central government characteristics. What I have not measured is the islands of near absolute power that can exist at one time or another even within democratic regimes. This is most notable in time of war, when for democracies the military are given considerable if not near absolute power within a restricted domain, and absolute secrecy and even deception of elected representatives by military and political leaders is practiced. Although the regime would still be characterized as democratic, in the pursuit of victory in war totalitarian-military power can flourish in defense related areas. It is thus not inconsistent with my other findings that foreign democide, generally occurring for democratic and many authoritarian regimes in time of war, should have little relationship whether a regime is, centrally, democratic or totalitarian.

Note the very close relationship of national power times war-dead to overall democide, almost the same as that for TotalPower. This is in part due to democide including the foreign component, to which this interaction term has a significant relationship; it has none to domestic democide. The higher the national power of a regime measured by its population (and which also reflects its size and energy consumption per capita) times the greater the characteristic severity of the wars a regime is likely to be involved in, the greater its foreign democide. The greater national power and characteristic war-dead, the more the effect is multiplied.

A final and the broadest test possible of the hypothesis was also carried out using canonical analysis (not shown). The set of dependent variables comprised all the fourteen types of democide listed in Table 2; the independent set was made up of the twenty-four variables employed in the above regressions. The result was seven significant pairs of canonical variates, with the first pair of canonical variates having a canonical correlation of .92. The independent variables loading this were much the same as those shown for democide in the regression of Table 5. I also did a discriminate analysis of democide (not shown), with regimes grouped by democide magnitude. The first canonical correlation also was .92 and again nearly the same independent variables were responsible for this excellent predictability. In all cases, TotalPower was a central predictor (along with the war, rebellion, and national power variables), again supporting the hypothesis.


All tests of the hypothesis were positive. Empirically, at least for our century, democracy is inversely related to democide.

Among a variety of social diversity (e.g., race, ethnicity, religion, language), socio-economic, cultural, geographic, and other indicators, the best way of accounting for and predicting democide is by the degree to which a regime is totalitarian along a democratic-totalitarian scale. That is, the extent to which a regime controls absolutely all social, economic, and cultural groups and institutions, the degree to which its elite can rule arbitrarily, largely accounts for the magnitude and intensity of genocide and mass murder. The best assurance against democide is democratic openness, political competition, leaders responsible to their people, and limited government. In other words, Power kills, absolute Power kills absolutely.

That Power kills is the primary and for domestic democide singular general explanation of democide. This is true even when we consider how regimes differ in their underlying ethnic, religious, and racial diversity. It is also true in general when we consider whether they are Christian, Moslem, European, or their cultural region. It is true when taking into account different levels of education or economic development. It is true for differences in sheer size. And it is true even for the trend of overall democide through time (not shown).

However, the tendency of regimes to fight severe domestic rebellions or foreign wars also predicts to democide. But for both Power is a causal agent. The more totalitarian a regime's power, the more total their wars or rebellions are likely to be, and the more totalitarian power and bloody their wars and rebellions, the more it probably will commit democide.

As mentioned in the beginning, we now have solid empirical evidence that democracies don't (or rarely) war on each other, the more democratic two regimes the less violence between them, and the more democratic the less domestic collective violence. Now we find also that as a regime is less democratic its democide increases exponentially.

Tying all these results together, then, the final conclusion is that democracy is a general method of nonviolence. 


* From the pre-publisher edited R.J. Rummel, "Democracy, power, genocide, and mass murder," The Journal of Conflict Resolution 39 (March 1995): 3-26.

1. See Rummel (1994, Chapter 1).

2. For a precise definition and elaboration of this concept, see Rummel (1994, Chapter 2).

3. The theory is conceptually developed in Rummel (1995a)

4. This is in effect the same theoretical explanation that others have given for democracies not making way on each other. See in particular Russett (1993), Ray (1995), and Weart (1995). I see democracy as a general method of nonviolence with the same explanation applying across the board for why it should eliminate or minimize violence, including democide.

5. See Russett (1993) and Rummel (1985, 1995a) for supporting evidence and studies on this and the previous proposition.

6. These data are based on almost 8,200 estimates of war, domestic violence, genocide, mass murder, and other relevant data, that I recorded from over a thousand sources, which include general works, specialized studies, human rights reports, journal articles, and news sources. For the tables of estimates on the Soviet Union, China, and Nazi Germany, see Rummel (1990, 1991, 1992). For all other estimates, see Rummel (1995b). For totals and a statistical overview of the data, see Rummel (1994, Chapter 1).

7. I also consulted the lists of regimes in Calvert (1970) and Russett (1993).

8. I did both varimax orthogonal and oblique rotation of the components was done. Some of the democide types are arithmetically related. Thus, total democide is the sum of domestic and foreign democide and camp dead is related to foreign labor dead. This is not a problem as long as it is recognized and taken into account in the interpretation of the results (a composite variable and its two parts can only be two dimensional at most).

Also, except for the variance for which they account, there would be no change in the patterns were the democide of all regimes 1900-1987 component analyzed.

9. The full results of this analysis are given in Rummel (1995b).

10. Some deny that factor analysis can be used in causal analysis. But consider. If two uncorrelated empirical patterns are found by factor analysis, they both cannot be the result of the same causes. There must be two different sets of causes at work. This is all that is being said here: that factor analysis uncovers different causal foci, not that it identifies what these causes are or proves causation. Similarly, correlation does not prove causation, but if two variables are perfectly uncorrelated (even after third variables are partialled out), then one cannot be the cause of the other. Lack of correlation does disprove causation.

11. This lack of correlation holds even through oblique rotation.

12. There are a variety of published scales of democracy that could not be used in here in toto, since a first requirement of any data set was that it be available for or applicable to all state regimes from 1900 to 1987, a requirement met by only a few data sets. For some political measures, however, these scales were useful for data for particular regimes that fell within their time span.

13. I determined the number of components to rotate by the substantive meaning of the factor and the eigenvalue-one criterion. The fifth unrotated component was slightly below this criterion at .92. Oblique rotation does not produce substantively different patterns than those shown.

14. Since each scale is 0-9 and with totalitarianism and democracy at the high end of each and their correlation is negative, then TotalPower = (TotalScale) + (9 - DemocrScale).

15. This is consistent with previous component analysis results for all nations, 1955-1963. See Rummel (1979).

16. I rotated various numbers of factors and the three factor solution gives the cleanest and most theoretically satisfying solution.

17. The inverse squared distance technique used to draw the surface shown in the figure is not based on regression, but interpolates domestic democide logged (the Z height of the surface at a XY point) as the weighted average of the totalitarian and democratic (X and Y) scales. The squared Euclidean distances across the totalitarian and democratic scales comprise the weights.

18. Based on the joint data points, the surface shown in the figure has been extrapolated to the full range of the two political scales. Thus, even though there were no regimes that were scaled both democratic and totalitarian (scale values of 0 and 1 for the democratic scale and 8 and 9 for the totalitarian--not theoretically impossible), the surface was extrapolated to the region in the right-hand corner. Moreover, although there is a strong negative correlation between democracy and totalitarianism, this correlation is not perfect (some democracies are more statist than others). That the democratic and totalitarian scales in the figure are shown at right angles does not imply lack of correlation, therefore, anymore than would a standard two-dimensional scatter plot of these two scales, where the scales are shown at right angles.

19. Data are given in Table 16A.1 of (Rummel, 1995b)

20. From Small and Singer (1982).

21. The specific results for this and the other analyses described below are given in Rummel (1995b).

22. This includes nine measures of war and rebellion, where only the best indicators of the resulting two patterns have been presented here.

23. In cross-national component analyses, one of the well-defined dimensions is that of national power. It is indexed by energy consumption and population. See, for example, Rummel (1972) and the studies referenced therein.

24. This is important to understand, for were all twenty-four included in the final regression, with only the significant independent variables shown, the multiple R would capitalize on the small but non-zero covariance between the dependent variables and the many non-significant independent variables, thus making R misleadingly high. But also, this would distort the t-tests for those independent variables that have high correlations, such as including both TotalPower and TotalPower squared.


Adelman, Irma and Cynthia Taft Morris (1973) ECONOMIC GROWTH AND SOCIAL EQUITY IN DEVELOPING COUNTRIES. Stanford, California: Stanford University Press.

Arat, Zehra F. (1991) DEMOCRACY AND HUMAN RIGHTS IN DEVELOPING COUNTRIES. Boulder, Colorado: Lynne Rienner Publishers.

Banks, Arthur S. (1971) CROSS-POLITY TIME SERIES DATA. Cambridge: The MIT Press.

Bollen, Kenneth A. (1980) "Issues in the comparative measurement of political democracy." AMERICAN SOCIOLOGICAL REVIEW 45 (June): 370-390.

Calvert, Peter (1970) A STUDY OF REVOLUTION. Oxford, Great Britain: Clarendon Press.

Cole, Timothy Michael (1987) UNITED STATES LEADERSHIP AND THE LIBERAL COMMUNITY OF STATES. Ph.D. Dissertation, University of Washington.

Coppedge, Michael and Wolfgang H. Reinicke (1990) "Measuring polyarchy." STUDIES IN COMPARATIVE INTERNATIONAL DEVELOPMENT 25 (Spring): 51-72.

Coulter, Phillip (1975) "Framework for analysis: theory and research design," in Phillip Coulter, SOCIAL MOBILIZATION AND LIBERAL DEMOCRACY: A MACROQUANTITATIVE ANALYSIS OF GLOBAL AND REGIONAL MODELS. Lexington, Massachusetts: D.C. Heath.

Cutright, Phillips and James A. Wiley (1969) "Modernization and political representation: 1927-1966." STUDIES IN COMPARATIVE INTERNATIONAL DEVELOPMENT 5: 23-44.

Dahl, Robert A. (1971) POLYARCHY. New Haven, Yale University Press.

Gurr, Ted Robert (1990) POLITY II: POLITICAL STRUCTURES AND REGIME CHANGE, 1800-1986. Ann Arbor, Michigan: Inter-university Consortium for Political and Social Research [Polity II data code book].

Lake, David A. (1992) "Powerful pacifists: democratic states and war." AMERICAN POLITICAL SCIENCE REVIEW 86 (March): 24-37.

Ray, James Lee (1995) DEMOCRACY AND INTERNATIONAL POLITICS: AN EVALUATION OF THE DEMOCRATIC PEACE PROPOSITION. Columbia, South Carolina: University of South Carolina Press.

Rummel, R.J. (1972) DIMENSIONS OF NATIONS. Beverly Hills: Sage Publications.

___________ (1979) NATIONAL ATTRIBUTES AND BEHAVIOR. Beverly Hills: Sage Publications.

___________ (1984) "Libertarianism, Violence Within States, and the Polarity Principle." COMPARATIVE POLITICS 16 (July): 443-462.

___________ (1985) "Libertarian Propositions on Violence Within and Between Nations: A Test Against Published Research Results" THE JOURNAL OF CONFLICT RESOLUTION, 29 (September): 419-55.

___________ (1990) LETHAL POLITICS: SOVIET GENOCIDE AND MASS MURDER. New Brunswick, New Jersey: Transaction Publishers.

___________ (1991) CHINA'S BLOODY CENTURY: GENOCIDE AND MASS MURDER SINCE 1900. New Brunswick, New Jersey: Transaction Publishers.

___________ (1992) DEMOCIDE: NAZI GENOCIDE AND MASS MURDER. New Brunswick, New Jersey: Transaction Publishers.

___________ (1994) DEATH BY GOVERNMENT: GENOCIDE AND MASS MURDER SINCE 1900. New Brunswick: Transaction Publishers.

___________ (1995a [published in 1997a) POWER KILLS: DEMOCRACY AS A METHOD OF NONVIOLENCE, New Brunswick, New Jersey: Transaction Publishers.

___________ (1995b [published in 1997b) STATISTICS OF DEMOCIDE. New Brunswick, New Jersey: Transaction Publishers.

Russett, Bruce (1993) GRASPING THE DEMOCRATIC PEACE: PRINCIPLES FOR A POST-COLD WAR WORLD. Princeton, New Jersey: Princeton University Press.

Sachs, Moshe Y. [ed.] (1971) WORLDMARK ENCYCLOPEDIA OF THE NATIONS, Volumes 1-5. New York: Harper & Row.

Small, Melvin and J. David Singer (1982) RESORT TO ARMS: INTERNATIONAL AND CIVIL WARS 1816-1980. Beverly Hills, California: Sage Publications.

Smith, Arthur K. Jr. (1969) "Socio-economic development and political democracy: a causal analysis." MIDWEST JOURNAL OF POLITICAL SCIENCE 13 (February): 95-125.

Vanhanen, Tatu (1990) THE PROCESS OF DEMOCRATIZATION: A COMPARATIVE STUDY OF 147 STATES, 1980-88. New York: Crane Russak, 1990.

Weart, Spencer (1995 [published in 1998]) NEVER AT WAR: WHY DEMOCRACIES WILL NOT FIGHT ONE ANOTHER. New Haven: Yale University Press.

You are the visitor since 11/27/02

Go to top of document