Generalized linear longitudinal mixed models with linear covariance structure and multiplicative random effects

René Holst, Bent Jørgensen

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

The paper proposes a versatile class of multiplicative generalized linear longitudinal mixed models (GLLMM) with additive dispersion components, based on explicit modelling of the covariance structure. The class incorporates a longitudinal structure into the random effects models and retains a marginal as well as a conditional interpretation. The estimation procedure is based on a computationally efficient quasi-score method for the regression parameters combined with a REML-like bias-corrected Pearson estimating function for the dispersion and correlation parameters. This avoids the multidimensional integral of the conventional GLMM likelihood and allows an extension of the robust empirical sandwich estimator for use with both association and regression parameters. The method is applied to a set of otholit data, used for age determination of fish.
Original languageEnglish
JournalChilean Journal of Statistics
Volume6
Issue number1
Pages (from-to)15-36
Number of pages22
ISSN0718-7912
Publication statusPublished - 2015

Cite this