Abstract # 100:

Scheduled for Friday, June 22, 2007 09:30 AM-12:00 PM: Session 10 (North Main Hall E) Workshop


Statistics for Primatologists Workshop: Analyses of Longitudinal Data Using Multilevel Modeling

M. L. Van Horn
University of South Carolina, Department of Psychology, Columbia, SC, USA
line
     Regardless of the type of primatology you practice, properly disentangling between and within subject variance is a must. Research studying developing organisms typically uses repeated measures designs. Recently, multilevel modeling statistical tools have greatly improved the analysis of developmental data, and new statistical software makes these models accessible to most researchers. Growth curve analyses, a subset of the multilevel approach, focus on change as a subject level process with individual trajectories rather than group means. These models also relax many assumptions of other methods, which may increase statistical power. This workshop will concern the application of multilevel models to the analysis of longitudinal research. The techniques will be illustrated with data from a randomized experiment studying the effects of anti-psychotic drugs on weight gain in juvenile pigtail monkeys (Macaca nemestrina). The workshop's goal is to provide an introduction to growth curve analyses, including the research questions answered and the assumptions made by these models. Handout materials will demonstrate the application of these techniques using a SAS program, with the results contrasted to those obtained using alternative methods. Resources for further information on the application of growth curve models will also be provided.