Mixed Models, Longitudinal and Incomplete Data

Based on Verbeke and Molenberghs (Springer, 1997, 2000, 2004), a general introduction to longitudinal data and the linear mixed model for continuous responses will be presented. The topic will be approached from the modeler's and practitioner’s points of view. Emphasis will be on model formulation, parameter estimation, and hypothesis testing, as well as on the distinction between the random-effects (hierarchical) model and the implied marginal model. Illustrations will be given based on the SAS procedure MIXED.

When the response of interest is categorical, the linear mixed model concepts can be extended towards generalized linear mixed models. An alternative approach is the use of generalized estimating equations (GEE). A lot of emphasis will be put on the fact that the regression parameters in both types of models have different interpretations. Advantages and disadvantages of both procedures will be discussed and compared in detail, and illustrations will be based on the SAS procedures GENMOD and NLMIXED.

Finally, when analyzing longitudinal data, one is often confronted with missing observations, i.e., scheduled measurements have not been made, due to a variety of (known or unknown) reasons. It will be shown that, if no appropriate measures are taken, missing data can cause seriously biased results, and interpretational difficulties.

Instructors:

Geert Molenberghs is Professor in Biostatistics at the Limburgs Universitair Centrum in Belgium. He published methodological work on repeated categorical data and on the analysis of nonresponse in clinical and epidemiological studies.

Gert Verbeke is Associate Professor in Biostatistics at the Biostatistical Centre of the Katholieke Universiteit Leuven in Belgium. He wrote his dissertation as well as a number of methodological papers, on various aspects of linear mixed models for longitudinal data.

Both presenters are editor and author of three books on the use of linear mixed models for the analysis of longitudinal data (Springer Lecture Notes 1997, Springer Series in Statistics 2000, Springer Series in Statistics 2004-to appear), and they have taught several (short) courses on the topic in universities as well as industry.

Audience:

Any interested scholar.

Prerequisite:

Throughout the course, it will be assumed that the participants are familiar with basic statistical modeling, including linear models (regression and analysis of variance), as well as generalized linear models (logistic and Poisson regression). Moreover, pre-requisite knowledge should also include general estimation and testing theory (maximum likelihood, likelihood ratio).

Provisions:

Enrollees will receive lecture notes, continental breakfast; break time for lunch (lunch not provided).

Date:

Friday, March 25, 2005


Time:

8:30 a.m. to 5:00 p.m.

Location:

Alumni Center, 200 Fletcher Street, University Main Campus.

Fee:

$150 for University affiliated faculty and staff, and for ASA members;
$ 75 for students; $325 for others.
Fees can be paid by check or will be billed to a University of Michigan project/grant.

Registration:

Call CSCAR at 734-764-7828 (ext 0). Enrollment is limited. Make check payable to CSCAR--University of Michigan, or give the University of Michigan Account to be billed. Send check to CSCAR, University of Michigan, 3550 Rackham Bldg., 915 E. Washington St., Ann Arbor, MI 48109-1070.