Not logged in.
Quick Search - Contribution
Contribution Details
Type | Conference Presentation |
Scope | Discipline-based scholarship |
Title | The Endogeneity Problem in Random Intercept Models: Are Most Published Results Likely False? |
Organization Unit | |
Authors |
|
Presentation Type | paper |
Item Subtype | Original Work |
Refereed | Yes |
Status | Published in final form |
Language |
|
Publisher | Academy of Management |
ISSN | 2151-6561 |
Series Name | Academy of Management Proceedings |
Number | 1 |
Page Range | 18927 |
Event Title | 79th Annual Meeting of the Academy of Management |
Event Type | conference |
Event Location | Boston, Massachusetts |
Event Start Date | August 9 - 2019 |
Event End Date | August 13 - 2019 |
Abstract Text | Entities such as individuals, teams, or organizations can vary systematically from one another. Researchers typically model such data using random effects, multilevel models, which assume that the random effects are uncorrelated with the regressors. Violation of this assumption creates an endogeneity problem. We review the various modeling approaches in the presence of endogeneity and show with a series of Monte Carlo simulations that popular “go-to” solutions can produce biased and inconsistent estimates depending on the model estimated. Our results show that researchers should instead use cluster means of the Level 1 explanatory variables as controls (i.e., the correlated random effects or Mundlak approach). To examine the state of the science, we reviewed 150 randomly drawn articles from organizational science journals, finding that only 70 articles properly deal with the random- effects assumption. Alarmingly, most models also failed on the usual exogeneity requirement of the regressors, leaving only 13 articles (8.67%) that potentially reported trustworthy multilevel estimates. We offer a set of practical recommendations for researchers to model multilevel data appropriately. |
Digital Object Identifier | 10.5465/AMBPP.2019.18927abstract |
Export | BibTeX |