Implementation of a New pdMat Class of Variance Covariance Structures for Random Effects in R

• Activity Number: 309
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Computing
• Author(s): Andrzej T. Galecki*+ and Tomasz Burzykowski
• Companies: University of Michigan and Hasselt University
• Keywords: nlme package ; mixed effects models ; random effects ; variance-covariance structure ; Kronecker product

Abstract

Several pdMat classes, which represent various variance covariance structures for random effects, are available in the nlme package in R. They include, among others, pdCompSymm (compound symmetry), pdDiag (diagonal), pdSymm (a general variance covariance matrix). Following the work of Galecki (1994), we implement a new class of variance-covariance structures based on the Kronecker product of existing structures. The use of the new class is illustrated by an analysis of clustered data (repeated samples of muscle fibers from subjects) from a randomized trial.