Rejoinder to “Endogeneity bias in marketing research: Problem, causes and remedies”

Abstract There are a few significant errors in Zaefarian, Kadile, Henneberg, and Leischnig's (2017) (henceforth ZKHL) discussion of remedies to address endogeneity issues in their recent survey article in Industrial Marketing Management. Most notably, they incorrectly describe 2-stage least squares (2SLS) estimation procedure. We provide the correct methodology here, along with sample data and code, and compare estimations using ZHKL's suggested methodology with proper 2SLS and ordinary least squares. We show that the method they suggest will actually result in greater bias in coefficient estimates while proper 2SLS addresses the endogeneity problem. Also, they incorrectly describe 3-stage least squares (3SLS), both in terms of implementation procedure and appropriate setting. We address both of these issues by describing when 3SLS would be used and the benefits of using 3SLS versus other methods to estimate simultaneous equations models. We discuss further issues with their paper and provide R code and simulated data.