The Varied Functions of Mechanisms for Causal Explanation in Political Science

• In Event: Causality, Processes, and Mechanisms

Abstract

What functions do mechanisms perform for causal explanations in political science research? The question is provoked by longstanding discussions in the qualitative and multi-methods political science field where mechanisms are widely identified as being of key importance for causal explanations. This is particularly the case in the extensive literature on process tracing often recognized as set of methods especially well-suited to generate mechanism-based explanations. More broadly, approaches from otherwise vastly different epistemological corners of political science and the social sciences seemingly find common ground in the assertion that mechanisms are important. Indeed, an extraordinary aspect of these discussions is the broad appeal of this type of explanation. This paper takes this observation as a point of departure for a discussion on the varied functions that this metaphor plays for causal explanations.
The paper grounds its discussion in a focused exploration of the concrete deployment of mechanisms in a broad set of empirical studies published over the last 20 years in leading political science and international relations journals. On the basis of this exploration the paper provides a typology including three different functions of mechanisms in causal explanation: mechanism as theory, mechanism as observed causal patterns, and mechanism as abductively derived analytical tool. The typology, it is argued, offers a way to clarify differences and similarities in understandings of mechanisms and causal explanations in different research traditions. This helps highlight how some of the fault lines on which discussions on mechanisms in political science rely are uninformative. It further provides the ground for a clearer understanding of the limits and possibilities of mechanism-based explanations and helps outline the methods that can be used to identify them.

Author

• Ludvig Norman, Stockholm University