Part of the dispute between analytical sociology and critical realists comes down to a complicated interplay between ontology and methodology. Both groups have strong (and conflicting) ideas about social ontology, and both think that these ideas are important to the conduct of social-science research. Analytical sociologists tend towards an enlightened version of methodological individualism: social entities derive from the actions and nature of the individuals who constitute them. Critical realists tend toward some version or another of emergentism: social entities possess properties that are emergent with respect to the individual activities that constitute them.
Both groups tend to design social science methodologies to correspond to the ontological theories that they advance. So they tacitly agree about what I regard as a questionable premise -- that ontology dictates methodology.
I want to argue for a greater degree of independence between ontology and methodology than either group would probably be willing to countenance. With the analytical sociologists I believe that social facts depend on the availability of microfoundations at the level of ensembles of individuals. This is an ontological fact. But with the critical realists I believe that it is entirely appropriate for social scientists to examine the causal and structural properties of social entities without being forced to attempt to provide the microfoundations of these properties. This is an observation about the locus and nature of explanation. There are stable structural and causal properties at the social level, and it is entirely legitimate to investigate these properties in full empirical detail. Sociologists, organizational theorists, and institutional researchers should be encouraged to investigate in detail the workings, arrangements, and causal properties of the regimes that they study. And this is precisely the kind of investigation that holds together researchers as diverse as Michael Mann, Kathleen Thelen, Charles Perrow, Howard Kimmeldorf, and Frank Dobbin. (Use the search box to find discussions of their work in earlier posts.)
What this implies is that sociologists can legitimately pursue meso- and macro-level inquiries into the nature of the social entities that most interest them. Organizational theory is an especially good example. We can approach the study of organizations from a number of points of view. But one perfectly legitimate approach is to attempt to discover some of the dynamic and causal properties of organizations with specified features. This takes the form of trying to discover what propensities a given organizational form has when embedded into a given institutional or social context. And this is a form of causal inquiry that is analogous to metallurgy or materials science: what are the properties of conductivity, thermal expansion, ductility, etc., of metals or ceramics of a given structural composition?
This means in turn that the ontology of individualism does not imply very much about methodology and research strategies. Ontology is not irrelevant to methodology; but it provides only weak constraints on the nature of the methodologies social scientists may choose in their pursuit of better understanding of the social world.
Can we say more about how ontological confidence about the nature of social entities is consistent with methodological pluralism? One part of the answer derives from the idea of the relative autonomy of various levels of the natural and social world. This is the argument that Jerry Fodor put forward with respect to the "special sciences" like psychology. Fodor argued persuasively that psychologists are entitled to investigate psychological properties without being obliged to reduce these properties to facts about the central nervous system. The rationality of science does not force us to be reductionist. Instead, it is legitimate to examine the properties of the system-level structures of a domain without attempting to say how these properties derive from more fundamental features of the stuff. And this has equally compelling implications for sociology as it does for other special sciences.
The upshot of this set of considerations is important for the conduct of social science. The ontological truism that social phenomena are constituted by individual-level activities does not imply that social science methodology must proceed along the lines of reductionism or aggregative explanation (vertical explanations, reproducing the struts of Coleman's boat). We can be individualist in ontology and macro-ist in our methodology.
The constraint of the ontological truism of microfoundationalism has really only two significant implications:
But generally speaking, all of this suggests that we should be methodological pluralists in the social sciences -- make use of the research strategies that seem most promising for understanding a specific range of phenomena without a lot of concern for how the method aligns with our most refined ontological thinking.