Correcting Selection Bias in Innovation and Entrepreneurship Research: A Practical Guide to Applying the Heckman Two-Stage Estimation

Abstract

Concerns about selection bias in empirical innovation and entrepreneurship research receive an increasing relevance and awareness. Mitigating a selection bias can significantly improve the validity of an empirical study and help identify causal inferences. The Heckman two-stage estimation emerged as a powerful method to detect and mitigate sample-induced endogeneity. Our meta-study finds numerous examples in the innovation and entrepreneurship literature where the Heckman two-stage estimation is applied incorrectly and incompletely. We sense considerable confusion concerning theoretical assumptions and methodological aspects in this context. For that reason, this paper offers a practical and comprehensible guide to applying Heckman’s technique. The guide supports scholars and practitioners regardless of their econometrical expertise by offering stepwise descriptions and elaborating on common econometric difficulties.

Keywords: Causal inferences; endogeneity; Heckman estimation; selection bias; two-step estimator; research methods