AN INTRODUCTION TO MIXED MODEL ANALYSIS
sponsored by
THE ANN ARBOR CHAPTER OF THE AMERICAN STATISTICAL ASSOCIATION (ASA)
and
THE UNIVERSITY OF MICHIGAN CENTER FOR STATISTICAL CONSULTATION AND RESEARCH (CSCAR)

Monday, March 4, 9:00 a.m. -- 4:00 p.m.
Tuesday, March 5, 9:00 a.m. -- 4:00 p.m.
Wednesday, March 6, 9:00 a.m. -- 12:00 noon

3001 SPH I, 109 South Observatory, Ann Arbor

Description: Most traditional statistical methods, such as linear and logistic regression, assume independent observations. Thus, they are unsuited to analyzing data from many fields, including longitudinal measurements from health and business research, panel data from social science research, complex survey data requiring small area estimation, achievement scores from multi-level designs in education, degradation data from engineering, and spatial data from environmental and agricultural research. Mixed model methodology, on the other hand, provides a framework for analyzing data with dependent observations. Recent advances in statistical software have made mixed model methods available to practitioners and have led to a greater appreciation of their value. This course will include a General Introduction to Mixed Models and the Use of SAS PROC MIXED, followed by modules on Applications to Longitudinal Continuous Data, Generalized Linear Mixed Models for Discrete Data, and other special topics.

Instructor: Walter Stroup, Ph.D., Professor of Biometry at the University of Nebraska. Dr. Stroup is the co-author of a forthcoming book on mixed models in SAS and has published many articles on the theory and applications of mixed models. He has taught short courses for statisticians in pharmaceutical companies, agricultural research facilities, and health institutes in the United States, Morocco, Niger, and Burundi, and he developed a course on mixed models at the University of Nebraska.

Audience: Practicing statisticians and others who wish to learn how to use mixed models in their work. This short course will emphasize concepts and applications over technical details, but will presume familiarity with basic statistical concepts and methods, including multiple linear regression, logistic regression, and maximum likelihood.

Supplies: All participants will receive a booklet containing copies of transparencies, data sets, and SAS examples. Refreshments will be provided during morning and afternoon breaks. Software and data sets will be available through School of Public Health computer networks.

Further information: contact Jonathan Raz (313-936-1009,jonraz@umich.edu).