From Process to Mechanism: Varieties of Disaggregation

McAdam, Tarrow and Tilly (henceforth MTT) persuasively argue that causal mechanisms can be measured, and they show us how this can be done. The authors argue that causal mechanisms may or may not be observable, but that their presence or absence can be assessed, if not directly, then at least through indirect measurement. The authors also argue convincingly that social processes can generally be disaggregated into component causal mechanisms, and that by doing so, we can begin to measure (and study!) processes. We concur with both claims. In fact, we see this article as a significant contribution to the literature on causal mechanisms because it anchors the debate beyond the definitional issue of what constitutes a causal mechanism (for an extensive list of definitions, see Mahoney 2001) in the larger epistemological project of understanding and measuring causation. However, and precisely because the authors’ main goal is to “identify and measure the ‘causal chain and casual mechanism’ in a process rather than assume we can best understand its as an ‘x follows y story’,” we highlight in our response the conditions under which disaggregation of processes into causal mechanisms is, and is not, likely to produce compelling causal explanations. In their article, MTT elide quite different ways of disaggregating processes (and concepts). We believe that the differences among these disaggregation strategies have consequences for measurement validity and for causal explanation. We identify five different levels of mechanismic concepts that in the original article are labeled simply as either “mechanisms” or “processes.” Each of these levels demands particular strategies for measurement and evidentiary standards, and implies particular types of causal claims. Qual Sociol (2008) 31:333–339 DOI 10.1007/s11133-008-9102-4