Statistics of Democide
Chapter 1: Summary and Conclusions [Why Democide?...]
Other Democide Related Documents On This Site
How to test for this possible role of culture is no little problem. Every international traveler can distinguish different cultures, but quantifying this intuition is not easy. The approach I will use is the same as that applied to the political characteristics and social diversity. I will try to uncover the independent intercorrelations, the empirical patterns, among many different cultural measures. And I will select indicators of these patterns to employ in an analysis of democide, Power, and diversity.
Table 19A.1 lists the data on the fourteen cultural measures I will use. Table 19.1 gives the measures, direction of scaling, and source; Table 19.2 defines the empirical patterns among these measures for the 214 regimes. Table 19.3 names them and lists their indicators.
I found that these fourteen cultural measures divide into five empirical patterns. Helped by the six geographic variables, their interpretation is straight forward and little about them need be said, except for the correlations between the oblique versions. (On the nature of these correlation coefficients, see Understanding Correlation). Now, while oblique patterns will be distinct they can be highly correlated and those between the (primary) oblique versions of the orthogonal (uncorrelated) patterns is shown in Table 19.2. Note then that the African Culture pattern (Factor 1) is positively correlated with Moslem Culture (Factor 2) and inversely correlated with English influenced cultures (Factor 5) and, to a much lesser extent, Asian Culture (Factor 3). These are commonsensical correlations and should be kept in mind in evaluating subsequent results. The naming and selection of indictors would not differ had I focused on the oblique rather the orthogonal (uncorrelated) patterns, and the orthogonal ones are easier to interpret (because the loadings are correlations between measure and factor).
Do these cultural patterns condition or otherwise influence democide? The answer to this is given in Table 19.4. We have the by now familiar two democide patterns, virtually unchanged by the inclusion of five cultural indicators, the same in number as the democide ones. That the same democide patterns emerged in spite of this shows how little general relationship there is between variation in the cultural context of regimes and their democide. The highest product moment correlation between the cultural and democide indicators is no more than an absolute .27; the highest partial correlation is even lower at an absolute .23. The patterns in Table 19.4 underline this lack of relationship, where no cultural indicator has a correlation over an absolute .32 with any pattern involving democide.
That is, whether the culture is English in origin or influence, African, Asian, or dominated by Islam, does not predict whether the associated regimes will or will not commit more or less democide than other regimes. This notwithstanding, as with diversity, culture may be an underlying or hidden influence on the empirical relationship between democide and Power, or to democide's lack of a general relationship to diversity.
To investigate this I did a component analysis of all the indicators of democide, politics, diversity, and culture together for the 214 regimes, with the results shown in Table 19.5. (On the nature of component analysis, a form of factor analysis, see "Understanding Factor Analysis")
The analysis shows that there is practically no change in any pattern or interpattern relationship when taking into account culture. Power and diversity in their relationship to democide operate generally the same regardless of the major cultural context. Cultural differences at the most extensive global level are apparently irrelevant.
So far the necessary conclusion to this so far is that the structure of democide, and the relationship of patterns of democide to power and diversity, are statistically independent of the variation in major cultural differences and similarities. What this means is essential to understand, especially because of the results I will present in a moment. This independence is not to say that within a cultural region there will not be different patterns, as between diversity and democide. It is to say that global differences from one culture to another across the major cultures does not have an effect on the global relationships between democide, politics, and diversity.
However, any analysis across groups of the kind conducted here may differ significantly for that within groups. As an hypothetical example, public opinion poll results may show little relationship between favoring higher defense expenditures and nationalized health care, and the correlation between these may not be altered for a sample by including a variable measuring sex (say, 0 = men, 1 = women). The conclusions would be that the relationship is unaffected by a difference in sex. But if this correlation was calculated for each sex separately, we might find that for men there was a negative correlation between favoring greater defense expenditures and nationalized health care, while for women the correlation is positive. What happened statistically is that for the whole sample these positive and negative correlations canceled each other out. Both of these results (across groups or within) are true. Which one focuses on depends on the theory and practical interests (such as launching an advertising campaign to appeal to all people or particularly to women).
Now, with this in mind let us turn to component analyses of the relationships between democide, politics, and diversity within cultures. The first of these is separate analyses of Christian and non-Christian, Moslem and non-Moslem cultures. Table 19.6 gives the results, along with some descriptive statistics to help evaluate them. For simplicity I have shown only those patterns with which democide has an absolute correlation of at least .5, and I also give the global patterns from Table 19.5 for comparison.
In Table 19.7 I present similarly organized patterns for European, Central and South American, African, and Asian cultures. And Table 19.8 summarizes the results for each culture.
For these results, then, what can be said about the role of culture? Globally, it has no effect. For any program trying to predict or deal with global democide in terms of regime's political characteristics and diversity, or for any theory trying to generally account for, say, genocide, by such measures, then cultural influences can be largely ignored. Moreover, what applies globally to the relationship to democide, politics, and diversity also mainly applies to regimes within Christian and Non-Moslem cultures. This would also be largely true for European cultures, except that diversity now has a moderate relationship to domestic democide, as was the case for that in Germany, Russia, the Soviet Union, Ukraine, Poland, Yugoslavia, and Rumania.
As to the other cultures, they differ significantly in the patterns of democide, politics, and diversity. Consider, for example, Africa. Even though ethnic and tribal differences played a key role in the democide of Burundi, Rwanda, and Nigerian regimes, among others, there were African regimes, such as Ethiopia, Mozambique, and Angola, in which ideology drove the killing. Social diversity in Africa is unrelated to the patterns of democide across African regimes. But also because there is considerable genocide by some authoritarian regimes and ideological based killing by totalitarian regimes, TotalPower has only a low relationship with domestic democide.
Of all cultures, the relationships are most complex for Central and South American ones. Democide is related to diversity and politics over four patterns, one of them being that between the annual rate of domestic democide and totalitarian and political power, and another separate one between this rate and the weakness of the traditional elite. Then we find that domestic democide, including genocide, is related to authoritarianism, the standard type of regime for this culture through most of this century. Finally, there is a close relationship between the minorities at risk under a regime and the amount of foreign, not domestic, democide.
For all these cultures and results, Asia is the only other culture where this particular indicator of diversity has any meaningful relationship to democide. And in the case of Asia its relationship is what many would predict would be true for all regimes, or at least those in Africa, and that is between minorities at risk and genocide plus the annual rate of domestic democide.
Table 19.5, 19.6, 19.7, and 19.8 are fascinating in what they reveal about all these relationships globally and in these different cultures. But they do not alter our general finding. Cultural variation does not significant effect the lack of correlation between diversity and democide, nor the positive correlation between Power and domestic democide, the intensity of democide (the annual rate), or genocide. Neither does culture much effect these relationships among Christian or Non-Moslem regimes, nor just that of Power and democide for European and Asian regimes.
In any case, many will point out that culture and diversity may or may not be important, but that the educational or economic context of a regime are significant, if not indeed the determining factors in democide. This is the subject of the next Chapter.
* From the pre-publisher edited manuscript of Chapter 19 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.
Go to top of document