Back to Listing

Longitudinal Data Analysis 2-Day Workshop with Donald Hedeker

Event Type
The Center for Biostatistical Development
May 13, 2013 - May 14, 2013   All Day
Donald Hedeker

Topics covered:

This workshop will focus on analysis of longitudinal data using mixed models. Models for continuous normally distributed outcomes will be presented and described, and then focus will turn to analyses of non-normal outcome types:

Binary outcomes, focusing on application of the mixed logistic regression model;

Nominal outcomes (unordered categorical), focusing on mixed multinomial logistic regression;

Ordinal outcomes (ordered categorical), focusing on mixed proportional odds and non-proportional odds models;

Count outcomes, including examples of mixed Poisson regression;

Time-to-event or survival outcomes using grouped and/or discrete time survival models;

Throughout the workshop, use of the SuperMix software program will be described and illustrated: SuperMix is especially useful for mixed modeling of non-normal outcomes, as it performs full-likelihood analysis for up to three-level models in a computationally efficient manner.  Participants will perform analyses using the SuperMix program.  

Reference for SuperMix: Hedeker D, Gibbons RD, du Toit M & Cheng Y (2008). SuperMix: Mixed Effects Models. Scientific Software International, Lincolnwood, IL.

Much of this workshop will be based on the book: Hedeker D & Gibbons RD (2006). Longitudinal Data Analysis. Wiley, New York.


A good knowledge of multiple regression analysis and logistic regression. Some prior knowledge of mixed models is beneficial.  Participants should bring a laptop computer to the course, and the free student version of SuperMix needs to be downloaded and installed on your laptop, see http://www.ssicentral.com/supermix/downloads.html.  

Sponsored by:

The Center for Biostatistical Development

The Center for Biostatistical Development (CBD) provides a biostatistical research environment aimed at fostering and developing innovative methods to address health-related scientific problems.  The CBD also provides education and training, via lectures and workshops, to increase the accessibility and use of the novel statistical methods.  The overall goal of the CBD is to provide and disseminate biostatistical research that will address and advance questions in the health sciences.  For more information please visit http://publichealth.uic.edu/departments/epidemiologyandbiostatistics/centerforbiostatisticaldevelopment/

link for robots only