Models of causal inference: Imperfect but applicable is better than perfect but inapplicable

Abstract

We assess a recent paper by Durand and Vaara (2009) that advances causal graph modeling as a tool for inferring causes in strategy research. We focus on the Markov condition, a key assumption on which causal graph modeling is based, and show why this condition is invariably violated in strategic management in general and the resource‐based view of the firm in particular. We then introduce vector space modeling as a quantitative alternative to causal graph modeling, and consider how improved methods of causal inference might enhance our ability to test some of the central propositions of the resource‐based view. Copyright © 2013 John Wiley & Sons, Ltd.