Generalized linear mixed models (GLMM) are generalized linear models with normally distributed random effects in the linear predictor. Penalized quasi-likelihood (PQL), an approximate method of inference in GLMMs, involves repeated fitting of linear mixed models with “working” dependent variables and iterative weights that depend on parameter estimates from the previous cycle of iteration. The generality of PQL, and its implementation in commercially available software, has encouraged the application of GLMMs in many scientific fields. Caution is needed, however, since PQL may sometimes yield badly biased estimates of variance components, especially with binary outcomes.
Recent developments in numerical integration, including adaptive Gaussian quadrature, higher order Laplace expansions, stochastic integration and Markov chain Monte Oarlo (MOMO) algorithms, provide attractive alternatives to PQL for approximate likelihood inference in GLMMs. Analyses of some well known datasets, and simulations based on these analyses, suggest that PQL still performs remarkably well in comparison with more elaborate procedures in many practical situations. Adaptive Gaussian quadrature is a viable alternative for nested designs where the numerical integration is limited to a small number of dimensions. Higher order Laplace approximations hold the promise of accurate inference more generally. MOMO is likely the method of choice for the most complex problems that involve high dimensional integrals
KeywordsCovariance Convolution Stein Verse Estima
Unable to display preview. Download preview PDF.
- D. G. Clayton. Generalized linear mixed models. In W. R. Gilks, S. Richardson, and D. J. Spiegelhalter, editors, Markov Chain Monte Carlo in Practice, chapter 16, pages 275–301. Chapman and Hall, London, 1999.Google Scholar
- CYTEL Software Corporation. EGRET for Windows. CYTEL Software Corporation, Cambridge, MA, 1999.Google Scholar
- [l1]P. J. Davis and I. Polonsky. Numerical interpolation, differentiation and integration. In M. Abramowitz and I. A. Stegun, editors, Handbook of Mathematical Functions, chapter 25, pages 875–924. U.S. Government Printing Office, Washington, D.C., 1964.Google Scholar
- H. Goldstein. Multilevel Statistical Models. Edward Arnold, London, 1995.Google Scholar
- R. C. Littell, G. A. Milliken, W. W. Stroup, and R. D. Wolfinger. SAS System for Mixed Models. SAS Institute Inc., Cary, N.C., 1996.Google Scholar
- K. J. McKonway, M. C. Jones, and P. C. Taylor. Statistical Modelling using GENSTAT. Arnold, London, 1999.Google Scholar
- J. Myles and D. Clayton. GLMMGibbs: An R Package for Estimating Bayesian Generalised Linear Mixed Models by Gibbs Sampling. Imperial Cancer Research Fund, London, 2001.Google Scholar
- J. Pinheiro and D. M. Bates. Approximations to the log-likelihood function in the nonlinear mixed-effects model. Journal of Computational and Graphical Statistics, 4:12–35, 1995.Google Scholar
- J. Rasbash, W. Browne, H. Goldstein, M. Yang, I. Plewis, M. Healy, G. Woodhouse, D. Draper, I. Langford, and T. Lewis. A User’s Guide to MLwiN. Institute of Education, London, 2000.Google Scholar
- S. W. Raudenbush, A. S. Byrke, Y. F. Cheong, and R Congdon. HLM 5: Hierarchical Linear and Nonlinear Modeling. Scientific Software International, Lincolnwood, IL, 2000.Google Scholar
- D. J. Spiegelhalter, A. Thomas, N. G. Best, and W. R. Gilks. BUGS: Bayesian Inference using Gibbs Sampling, Version 0.30. Medical Research Council Biostatistics Unit, Cambridge, 1994.Google Scholar
- SAS Institute Inc. Staff. The NLMIXED procedure. In SAS/STAT User’s Guide Version 8, chapter 46, pages 2421–2504. SAS Publishing, Cary, NC, 2000.Google Scholar