The Logic of Comparison

Abstract

We have already argued in brief that the logic of comparative analysis is different from the logic of other forms of social science research, especially from the statistical method that tends to dominate social research. This difference is evident in terms of the tendency of the unit of analysis for statistical research to be the individual, as opposed to the collectivity more commonly encountered in comparative research. It is seen more importantly in terms of the manner in which the two methods deal with the crucial question of controlling sources of extraneous variance. In the complex social world, there are any number of factors associated with the variance observed in any number of other variables. Finding significant statistical correlations among those factors is rarely a problem for social researchers. What is more of a problem is determining whether those correlations are empirically and theoretically meaningful; that is, can we say confidently that X causes Y, or whether they are merely the product of other, unmeasured, variables affecting the variables actually measured. In other words, the real difficulty for the social sciences is making convincing statements about the causation of political phenomena, given the complexity of interactions among the whole range of social phenomena and the number of external sources of variance.