Applications of Mixed Models to the Analysis of Longitudinal Multiple Source Data

• Activity Number: 347
• Type: Topic Contributed
• Date/Time: Wednesday, August 14, 2002 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology*
• Author(s): Garrett Fitzmaurice*+
• Affiliation(s): Harvard University
• Keywords: Longitudinal analysis; Multivariate data

Abstract

We consider regression-based methods for analyzing multiple sources data arising from a longitudinal study. We use the term "multiple source" data to encompass all cases where data are simultaneously obtained from multiple informants or raters (e.g., self-reports, family members, health care providers, administrators) or via different/parallel instruments or methods (e.g., symptom rating scales, standardized diagnostic interviews, or clinical diagnoses). However, we restrict our use of this term to data that are commensurate. In this talk we outline some of the potential advantages of jointly analyzing the data from multiple sources. However, one of the key challenges in the analysis of longitudinal multiple source data is the proliferation of covariance parameters. Linear mixed models provide a very flexible, yet parsimonious, structure for the covariance among longitudinal multiple source outcomes. The main ideas are illustrated using data on family functioning from a longitudinal study comparing two forms of cognitive, psycho-educational, preventive intervention targeted at families in which one or both parents had experienced serious affective disorders.