Generalized Linear Models With Random Effects; A Gibbs Sampling Approach

Abstract

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.. American Statistical Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of the American Statistical Association.Generalized linear models have unified the approach to regression for a wide variety of discrete, continuous, and censored response variables that can be assumed to be independent across experimental units. In applications such as longitudinal studies, genetic studies of families, and survey sampling, observations may be obtained in clusters. Responses from the same cluster cannot be assumed to be independent. With linear models, correlation has been effectively modeled by assuming there are clusterspecific random effects that derive from an underlying mixing distribution. Extensions of generalized linear models to include random effects has, thus far, been hampered by the need for numerical integration to evaluate likelihoods. In this article, we cast the generalized linear random effects model in a Bayesian framework and use a Monte Carlo method, the Gibbs sampler, to overcome the current computational limitations. The resulting algorithm is flexible to easily accommodate changes in the number of random effects and in their assumed distribution when warranted. The methodology is illustrated through a simulation study and an analysis of infectious disease data.

The use of hierarchical models for estimating relative risks of individual genetic variants: An application to a study of melanoma

“…In the following, we present the joint posterior distribution of the unknown parameters arising for a GLMM for canonical one-parameter exponential families [ 26,27 ]. Assume that conditional on a random effect δ, the responses y i are independent and follow an exponential family with conditional mean related to the linear predictor through g(μ) = Uα + Xδ = η, with δ ~ N(0, G).…”

“…Details on the use of Gibbs sampler and on the full conditional distributions in the case of generalized linear mixed models can be found in Zeger and Karim [ 26 ], Gamerman [ 29 ], and Natarajan and Kass [ 27 ]. WinBUGS [ 30 ] software was used to generate the samples and derive inferences on the parameters of interest and the WinBUGS code is supplied in the Appendix.…”

Bayesian analysis for the bivariate Poisson regression model: Applications to road safety countermeasures

“…The data augmentation algorithm can be used to estimate the parameters with sampling based approaches like the Gibbs sampler. The Gibbs sampler is proposed by Geman and Geman (1984), and is widely used as a general method in Bayesian computation (Gelfand and Smith, 1990;Zeger and Karim, 1991).…”

Ascertainment adjustment in complex diseases

“…A Monte Carlo approach would use an approach like Metropolis Hastings. The model can also be fit using the Gibbs sampling approach [Zeger and Karim, 1991], as done by Burton et al [2000].…”