First, let me answer a question not asked by the title: What do we not do with CR? I said it at the end of the last chapter, and Christian Smith says I always say it. I say it again here: CR is not a theory; therefore do not treat it as such. Strictly speaking, there is no CR explanation of anything. No more than there is a positivist or post-positivist account of this or that. What is the positivist theory of social stratification? There may be multiple theories of stratification all positivist in nature, but there is no particular theory that is singularly the positivist theory of stratification or of anything. The same goes for CR.

Like positivism, postmodernism, or pragmatism, CR is a metatheory rather than a theory. As such it parameterizes – from its criteria – what good theories are. Thus, it provides grounding for what it considers tenable research and challenge to research that departs from its premises. But that is all it does.

Is that enough? Enough to explain anything? No, but it is still something. Something needed and something important.

So how do we use CR? In the first place we use it for defense and for attack. We know that from the positivist point of view, which still reigns in sociology, that ethnography and historical narrative are suspect. Not so from the standpoint of CR. From the standpoint of CR, neither of these approaches is second best. Instead, from a CR perspective, even purely descriptive work is scientifically important. It is important to know what is out there, especially if attention has not previously been drawn to what is described and especially if, in the process, what is identified is described in theoretical language that highlights the significance of what is identified.

From a CR perspective as well, neither ethnography nor historical narrative need encompass only description. Insofar as CR separates causation from generality, ethnographic description of operative mechanisms can contribute to causal explanation even if what operates does so in only one place and time. Historical narrative, on the other hand, is the only way to bring together the causal effects of conjunctures of mechanisms. As I said before, narrative, not a regression equation, is the canonical form of a CR explanation as it is of any scientific account of anything in the real world outside the laboratory.