Sample selection bias and Heckman models in strategic management research

Abstract

Research summary: The use of Heckman models by strategy scholars to resolve sample selection bias has increased by more than 700 percent over the last decade, yet significant inconsistencies exist in how they have applied and interpreted these models. In view of these differences, we explore the drivers of sample selection bias and review how Heckman models alleviate it. We demonstrate three important findings for scholars seeking to use Heckman models: First, the independent variable of interest must be a significant predictor in the first stage of a model for sample selection bias to exist. Second, the significance of lambda alone does not indicate sample selection bias. Finally, Heckman models account for sample‐induced endogeneity, but are not effective when other sources of endogeneity are present. Managerial summary: When nonrandom samples are used to test statistical relationships, sample selection bias can lead researchers to flawed conclusions that can, in turn, negatively impact managerial decision‐making. We examine the use of Heckman models, which were designed to resolve sample selection bias, in strategic management research and highlight conditions when sample selection bias is present as well as when it is not. We also distinguish sample selection bias, a form of omitted variable (OV) bias, from more traditional OV bias, emphasizing that it is possible for models to have sample selection bias, traditional OV bias, or both. Accurately identifying the type(s) of OV bias present is essential to effectively correcting it. We close with several recommendations to improve practice surrounding the use of Heckman models. Copyright © 2015 John Wiley & Sons, Ltd.

Good, better, engaged? The effect of company-initiated customer engagement behavior on shareholder value

“…The self-selection correction is not perfect, requiring more attention in future research. Self-selection may for instance not be detected due to the use of weak selection variables (Certo et al 2016). Moreover, one could argue that firms are selective in which of the implemented customer engagement initiatives they announce.…”

Geographic Co-location of Partners and Rivals: Implications for the Design of R&D Alliances

“…10 Nevertheless, the 10 Significant coefficients of inverse Mills ratio (i.e., lambda) do not necessarily imply selection concerns if the independent variable of interest is not significant in the first stage (i.e., selection) model. However, one limitation we have in employing Heckman models is that the independent variable (i.e., risk from partner-rival co-location) is available only for realized deals in our setting, so we cannot test the significance of the independent variable in the first stage model (Certo, Busenbark, Woo, & Semadeni, 2016). Furthermore, given (1) the negative association between the independent variable and task interdependence and (2) the negative correlation between error terms in the selection and outcome equations which is indicated by the negative lambda, Model 7 is subject to Type II error (i.e., failures to detect real significant relationships), which makes our main results more conservative (Certo et al, 2016), results from the Heckman probit models again provided results consistent with those from probit models.…”

International Channel Choice of Industrial Exporting Firms

“…Even though, I have followed current guidelines to account for endogeneity in strategy research, newer approaches exist 192 (e.g. Certo, Busenbark, Woo, & Semadeni, 2016) that might be superior to the current praxis in the literature.…”

“…However, there exist newer guidelines for management research to attend to this issue (Certo, Busenbark, Woo, & Semadeni, 2016). Yet, this praxis has not been adapted in the current literature as it is difficult to implement (Martin, 2013) 15 .…”

“…Even though I have followed current guidelines to account for endogeneity, newer approaches exist that might be superior to current praxis (e.g. Certo, Busenbark, Woo, & Semadeni, 2016 …”