Generalized Estimating Equations

Rating:
Summary: GEE explained with all the computational details
Review: GEE is an extension of the Generalized Linear Models that handles the correlation structure found in panel data and longitudinal or repeated measures analysis. The idea goes back to a paper by Zeger and Liang in 1986 and is well cover in their 1994 book with Diggle and its most recent revision. However, over the years the statistical packages have implemented GEE and variations of it in many different ways. This important new book not only provides a detailed description of GEE for theoreticians and practitioners but it also presents comparisons of how the various software products implement it. The packages include GLIM, SAS STAT/GENMOD, S-PLUS, Stata and RTI's product SUDAAN (designed specifically for survey data).

This book is well written and comprehensive and will make a great reference book. It even provides a chapter on model diagnostics. Many examples are illustrated using the various software packages.