Omitted variable bias: A threat to estimating causal relationships

Abstract

We aim to raise awareness of the omitted variable bias (i.e., one special form of endogeneity) and highlight its severity for causal claims. Firstly, we demonstrate via analytic proof that omitting a relevant variable from a model which explains the independent and dependent variable leads to biased estimates. Secondly, we offer an easy-to-understand visualization for the problem. Finally, we discuss two remedies, diminishing the risk of the omitted variable bias, namely the instrument variable or two-stage least squares estimator and the regression discontinuity design. We hope that our review will motivate researchers to use them more often in future research.

Topics & Concepts