Statistical Models for Change

Abstract

Discusses different definitions of change and their substantive implications. Change is conceptualized in this chapter as occurring along many different dimensions, each with its own meaning. Emphasis is placed on a priori determinations of the type of change of interest within a particular research question, and the subsequent adoption of statistical tools that will allow the appropriate conclusions to be drawn. The chapter begins by presenting different ways of conceptualizing the change process at a descriptive level. Next, statistical methods are proposed to deal with these different conceptualizations of change. Structural equation models with latent variables are used as the well-suited framework to discuss these statistical developments. Hypotheses regarding both change in means (location) and in variance/covariances (scale) are discussed, as they are typically represented in terms of either growth curve or autoregressive models. Strictly intraunit-level or "time series" models are reviewed briefly. A brief discussion of event history or survival models follows. Finally, the chapter concludes with the theoretical and conceptual limitations of the proposed framework. (PsycINFO Database Record (c) 2016 APA, all rights reserved)