Context (archaeology)

Neuroscience

Control variable

Statistics

Causal inference

Archaeology

Variables

Variable (mathematics)

Economics

Randomized experiment

Estimation

Psychology

Endogeneity

Mathematics

Management

Geography

Causal model

Econometrics

Endogeny

Internal medicine

Mathematical analysis

Instrumental variable

Medicine

Perception

The Leadership Quarterly

Omitted variables create endogeneity and thus bias the estimation of the causal effect of measured variables on outcomes. Such measured variables are ubiquitous and include perceptions, attitudes, emotions, behaviors, and choices. Even experimental studies are not immune to the endogeneity problem. I propose a solution to this challenge: Experimentally randomized instrumental variables (ERIVs), which can correct for endogeneity bias via instrumental variable estimation. Such ERIVs can be generated in laboratory or field settings. Using perceptions as an example of a measured variable, I examine 74 recent articles from two top-tier management journals. The estimation methods commonly used exposed estimates to potential endogeneity bias; yet, authors incorrectly interpreted the estimated coefficients as causal in all cases. Then I demonstrate the mechanics of the ERIV procedure using simulated data and show how researchers can apply this methodology in a real experimental context.

Estimating the causal effect of measured endogenous variables: A tutorial on experimentally randomized instrumental variables