Case Studies and Theory Development in the Social Sciences - Alexander L. George [75]
Thus, causal mechanisms provide more detailed and in a sense more fundamental explanations than general laws do. The difference between a law and a mechanism is that between a static correlation (“if X, then Y”) and a “process” (“X leads to Y through steps A, B, C”). As Jon Elster notes:
The scientific practice is to seek explanation at a lower level than the explanandum. If we want to understand the pathology of the liver, we look to cellular biology for explanation… . To explain is to provide a causal mechanism, to open up the black box and show the nuts and bolts… . The role of mechanisms is two-fold. First, they enable us to go from the larger to the smaller: from molecules to atoms, from societies to individuals. Secondly, and more fundamentally, they reduce the time lag between the explanans and explanandum. A mechanism provides a continuous and contiguous chain of causal or intentional links; a black box is a gap in the chain… . The success of the reduction is constrained by the extent to which macro-variables are simultaneously replaced by micro-variables… . The search for micro-foundations … is in reality a pervasive and omnipresent feature of science… .278
In our view, this commitment in principle to consistency between a theory and what is known at the lowest observable level of space and time does not rule out positing and testing theories on the macrolevel. The commitment to consistency with the microlevel also does not mean that the explanatory weight or meaningful variation behind any particular phenomenon occurs at this level. If all individuals behave the same in the same social structure, then the interesting causal and explanatory action is at the level of the social structure, even if it must operate through the perceptions and calculations of individuals.
The acceptable level of generality of causal mechanisms will vary depending on the particular research question and research objectives under investigation.279 As we note in Chapter 10 on process-tracing, social science research never delves into the finest level of detail observable. Macrosocial mechanisms can be posited and tested at the macrolevel, as is common in the field of economics, and this is often the most cost-effective way to test such mechanisms. All that the commitment to microlevel consistency entails is that individuals must have been capable of behaving, and motivated to behave as the macrolevel theory states, and that they did in fact behave the way they did because of the explicit or implicit microlevel assumptions embedded in the macrolevel theory.
Some simplification of the microfoundations of macrotheories is tolerable for the purposes of parsimony or pedagogy. At the frontiers of research, however, social scientists need to discard stylized simplifying assumptions and build upon the most accurate microlevel mechanisms that can be discerned. David Dessler gives a good example of this process from physics:
In the ideal gas model, the gas is said to behave as if the molecules occupy no volume and have no interactions. These are idealizations. They are useful because they lay bare the essential workings of a gas … the idealizations also restrict the model’s range of applicability … the theory’s explanatory power increases as its false assumptions are “relaxed”—that is, as the assumptions distorting, idealizing, or simplifying effects are removed. At each step in the process, it is the assumptions that are true that carry the explanatory burden. To the extent the theory remains false, its range and power are restricted.280
Thus, while our theories rely on simplifying assumptions, advancing beyond the boundaries of our knowledge requires that we make our assumptions as accurate as possible.
The commitment to consistency with the microfoundational level raises another question: