Correcting for Self-selection Based Endogeneity in Management Research: Review, Recommendations and Simulations

Foundational to management is the idea that organizational decisions are a function of expected outcomes; hence, the customary empirical approach to employ multivariate techniques that regress performance outcome variables on discrete measures of organizational choices (e.g., investments, trainings, strategies and other managerial decision variables) potentially suffer from self-selection based endogeneity bias. Selection-effects represent an internal validity threat as they can lead to biased parameters that render erroneous empirical results and incorrect conclusions with regard to the veracity of theoretical assertions. Our review of the empirical literature suggests that selection-effects have received increasing attention in both micro- and macro-based research in recent years. Yet even when researchers acknowledge the issue, the techniques to correct for selection-effects have not always been employed in the proper manner; thus, estimations often suffer from shortcomings that potentially render flawed empirical findings. We explain the nature of self-selection based endogeneity bias and review the techniques available to researchers in management to correct for selection-effects when organizational decisions are discrete in nature. Furthermore, we engage in Monte Carlo simulations that demonstrate the tradeoffs involved with alternative techniques.