Flexible Models of Change: Using Structural Equations to Match Statistical and Theoretical Models of Multiple Change Processes

Abstract

To introduce and illustrate recent advances in statistical approaches to simultaneous modeling of multiple change processes. Provide a general overview of how to use structural equations to simultaneously model multiple change processes and illustrate the use of a theoretical model of change to guide selection of an appropriate specification from competing alternatives. The selected latent change score model is then fit to data collected during an adolescent weight-control treatment trial. A latent change score model is built starting with the foundation of repeated-measures analysis of variance and illustrated using graphical notation. The assumptions behind using structural equations to model change are discussed as well as limitations of the approach. Practical guidance is provided on matching the statistical model to the theory underlying the observed change processes and the research question(s) being answered by the analyses.