Abstract
Using a unique data set of causal usage drawn from research articles published between 2006-2008 in the American Journal of Sociology and American Sociological Review, this article offers an empirical assessment of causality in American sociology. Testing various aspects of what we consider the conventional wisdom on causality in the discipline, we find that (1) variablistic or covering law models are not the dominant way of making causal claims, (2) research methods affect but do not determine causal usage, and (3) the use of explicit causal language and the concept of mechanisms to make causal claims is limited. Instead, we find that metaphors and metaphoric reasoning are fundamental for causal claims-making in the discipline. On this basis, we define three dominant causal types used in sociology today, which we label the Probabilistic, Initiating and Conditioning types. We theorize this outcome as demonstrating the primary role that cognitive models play in providing inference-rich metaphors that allow sociologists to map causal relationships on to empirical processes.