All sociology is implicitly comparative, since social phenomena are invariably held in some way to be typical, representative, or unique, all of which imply appropriate comparison. Émile Durkheim was therefore correct to insist that ‘comparative sociology is not a particular branch of sociology; it is sociology itself, in so far as it ceases to be purely descriptive and aspires to account for facts’ (The Rules of Sociological Method, 1895). Consequently, there is no one comparative method, since all research techniques can be used to facilitate comparison.
Where a sociological analysis is explicitly held to be comparative, this usually involves the study of particular social processes across nation-states, or across different types of society (such as capitalist and state socialist). Much of what is normally referred to as comparative sociology is perhaps more accurately described as cross-national research. In his Presidential Address of 1987 to the American Sociological Association, Melvin L. Kohn delivered a manifesto for this form of research (‘Cross-National Research as an Analytical Strategy’, American Sociological Review, 1987).
Two general orientations to this type of comparative analysis are evident in the literature. First, there are those studies which seek similarity, usually starting from some well-defined a priori general theory which is then tested in different social (and possibly historical) contexts. Much functionalist inspired research, for example almost the whole corpus of modernization theory, takes this form. Similarly, structuralists in both sociology and anthropology have attempted to identify the models and general processes that underlie the apparently different orderings of experience in different societies, as for example in the work of structural Marxists. The danger of this approach is that context is ignored in the search for illustrations of allegedly universal propositions.
At the other extreme are studies which search for variance. These emphasize the specific historicity of societies, reject the search for general theories or laws, and use comparative research to shed light on the differences between cultures in order to understand better the specific arrangements that are found within each. Max Weber's comparative sociology offers a good example. Correspondingly, the problem here is that sociological explanation may be sacrificed on the altar of context, so that one arrives at the conclusion that cross-cultural or cross-national differences in particular social phenomena are entirely the result of historical contingency. The division of labour, crime-rates, organization of religion (or whatever) are different in Britain and Germany because Britain is Britain and not Germany. The units of analysis (in this case nation-states) are simply so many case studies to be interpreted.
In a stimulating analysis of this dilemma, A. Przeworski and H. Teune (The Logic of comparative Social Inquiry, 1970) have argued that the objective of comparative sociological research should be one of replacing the names of nations with the names of variables. That is, the explanans of particular dependent variables which differ cross-nationally should be rooted not simply in the different histories of the societies involved, but in specific national characteristics (such as degree of income inequality or type of political regime) that can be subsumed under variable names about which the sociologist can meaning-fully generalize.