BIOMETRICS SECTION NEWS, MAY 2006

Edited by Brian Caffo, Department of Biostatistics, Johns Hopkins University

Continuing Education Courses at JSM 2006

Misrak Gezmu, 2006 CE Chair.

The Biometrics Section is proud to co-sponsor two short courses during the annual meeting in Seattle.

Generalized linear latent and mixed models

Anders Skrondal and Sophia Rabe-Hesketh will present a one-day course, “Generalized linear latent and mixed models”. Generalized linear mixed (or multilevel) models (GLMMs) are useful for longitudinal data, cluster-randomized trials, surveys with cluster-sampling, genetic studies, meta-analysis, etc. The random effects in GLMMs are latent variables representing between-cluster variability and inducing within-cluster dependence. Latent variables are also often used to represent true values of variables measured with error, e.g. diet (continuous) or diagnosis (categorical). Measurement models specifying the relationship between measured and latent variables (factor, item response or latent class models) can form part of regression models, giving structural equation models (SEMs), such as covariate measurement error models. SEMs can also be used to model dependence between different processes, for instance the response of a clinical trial and drop-out. Skrondal is Professor in Statistics and Chair in Social Statistics at the London School of Economics and Rabe-Hesketh is a professor of educational statistics at the University of California, Berkeley. Skrondal and Rabe-Hesketh are co-authors of the book “Generalized Latent Variable Modeling: Multilevel, Longitudinal and Structural Equation Models” (Chapman& Hall, 2004). The book will be used as a text book for this course. The presenters are experienced statistical consultants and have published 6 books and 100 peer-reviewed papers. Their course at JSM 2004 was one of the most highly attended and received very positive evaluations.

The intended audience is statisticians and graduate students in statistics interested in problems involving longitudinal or clustered data or fallible measurements. Familiarity with generalized linear models is essential and some experience with random effects models would be beneficial.

Applied Longitudinal Analysis

Garrett Fitzmaurice will present a one day course entitled “Applied Longitudinal Analysis”. Garrett Fitzmaurice is Associate Professor of Biostatistics at the Harvard School of Public Health and Associate Professor of Medicine (Biostatistics) at Harvard Medical School. His research and teaching interests are in methods for analyzing longitudinal and repeated measures data. The goal of this course is to provide an introduction to statistical methods for analyzing longitudinal data. The main emphasis is on the practical rather than the theoretical aspects of longitudinal analysis. The course begins with a review of established methods for analyzing longitudinal data when the response of interest is continuous. A general introduction to linear mixed effects models for continuous responses is presented. When the response of interest is categorical (e.g., binary or count data), a number of extensions of generalized linear models to longitudinal data have been proposed. A broad overview of two main types of models: "marginal models" and "generalized linear mixed models" will be presented. While both classes of models account for the within-subject correlation among the repeated measures, they differ in approach. Moreover, these two classes of models have regression coefficients with quite distinct interpretations and address somewhat different questions regarding longitudinal change in the response. In this course the highlight will be on the main distinctions between these two types of models and to discuss the types of scientific questions addressed by each. Fitzmaurice is the co-author the book “Applied Longitudinal Analysis” (Wiley, 2004), this book will be used as a text book for this class. Dr. Fitzmaurice has taught several courses and workshops on this topic in universities as well as industry, both in the US and abroad.

Attendees should have a strong background in linear regression and some minimal exposure to generalized linear models (e.g., logistic regression).

JSM 2007

It’s already time to start thinking about invited sessions for next year's Joint Statistical Meetings, which will be held July 29- August 2 in Salt Lake City, Utah. Anyone who is interested in organizing an invited session or who has ideas for one, please contact our 2007 Program Chair, Runze Li, at rli@stat.psu.edu.

A typical invited session consists of three 30-minute talks followed by a 10-minute invited discussion and 10 minutes of floor discussion. However, other formats are possible. The 2006 program, which is online at the ASA’s website (www.amstat.org), is a good source for examples.

Remember, the most mature ideas will have an advantage in competing for the limited number of slots, so it's best to have your ideas in final form by the middle of June. The Biometrics Section will have at least four invited sessions, but if we generate enough good ideas we will be able to compete for additional slots as well.

Decisions will be made in July, so don't delay!