What are mechanisms and what are they good for

Almost a century ago, Bertrand Russell called the idea of causality “a relic of a bygone era, surviving, like the monarchy, only because it is erroneously supposed to do no harm.” Into the 1980s, widely read statements of the philosophy and methodology of political science omitted any reference to causality. But in the past quarter-century, causality has made a comeback. Gary Goertz and James Mahoney begin their essay, “Causal Mechanisms and Process Tracing,” by stating that we all share an intuitive mechanism-based understanding of causation. The ubiquity of concepts like causal inference, causal effects, causal processes and pathways, and causal mechanisms in contemporary discourse supports their claim; recent research in cognitive psychology, summarized by Sloman (2005), affirms it as well. But perhaps this apparent consensus masks some fundamental disagreements. For example, Jasjeet Sekhon emphatically claims that we have over-inflated the importance of mechanisms, as “We do not need to have much or any knowledge about mechanisms in order to know that a causal relationship exists” (Sekhon 2008: 292). We may have strong grounds, Sekhon insists, for believing that an intervention has great therapeutic value without any understanding, intuitively or theoretically, about the underlying mechanisms. I believe this claim is correct; while mechanisms may be useful for causal inference (George and Bennett 2005; Waldner 2007; Hedström 2008), we can certainly make causal inferences without reference to causal mechanisms and without using process tracing. I thus disagree with the slogan Goertz and Mahoney coin for qualitative researchers, “No strong causal inference without process tracing.”