The perils of quasi‐likelihood information criteria
Abstract
In this paper, we consider some potential pitfalls of the growing use of quasi‐likelihood‐based information criteria for longitudinal data to select a working correlation structure in a generalized estimating equation framework. In particular, we examine settings where the fully conditional mean does not equal the marginal mean as well as hypothesis testing following selection of the working correlation matrix. Our results suggest that the use of any information criterion for selection of the working correlation matrix is inappropriate when the conditional mean model assumption is violated. We also find that type I error differs from the nominal level in moderate sample sizes following selection of the form of the working correlation but improves as sample size is increased as the selection is then concentrated on a single correlation structure. Our results serve to underline the potential dangers that can arise when using information criteria to select correlation structure in routine data analysis. Copyright © 2015 John Wiley & Sons, Ltd.Keywords
Funding Information
- Fonds de recherche de Québec en santé
- Natural Sciences in Engineering Research Council of Canada
- Fonds de recherche de Québec en nature et technologies
This publication has 15 references indexed in Scilit:
- Functional group diversity increases with modularity in complex food websNature Communications, 2015
- Serum Lipids, Lipoproteins, and Risk of Breast Cancer: A Nested Case-Control Study Using Multiple Time PointsJNCI: Journal of the National Cancer Institute, 2015
- Working‐correlation‐structure identification in generalized estimating equationsStatistics in Medicine, 2008
- QIC Program and Model Selection in GEE AnalysesThe Stata Journal: Promoting communications on statistics and Stata, 2007
- Akaike's Information Criterion in Generalized Estimating EquationsBiometrics, 2001
- Improving generalised estimating equations using quadratic inference functionsBiometrika, 2000
- A cautionary note on inference for marginal regression models with longitudinal data and general correlated response dataCommunications in Statistics - Simulation and Computation, 1994
- The Impact of Model Selection on Inference in Linear RegressionThe American Statistician, 1990
- Generalized Linear Models1989
- Longitudinal data analysis using generalized linear modelsBiometrika, 1986