Modeling Growth Data Using Multidimensional Scaling Profile Analysis

摘要

This paper discusses how Multidimensional Scaling (MDS) can provide an exploratory technique for identifying major growth profiles, which may be indicative of growth associated with subgroups. It briefly overviews the conventional growth models and growth mixture models, examines the assumptions related to these models, and indicates some limitations associated with these models. It then proposes an exploratory growth profile analysis using the MDS model as a complement to more specification-oriented techniques. It describes the Profile Analysis via Multidimensional Scaling model (PAMS) and extends the model for longitudinal data. The MDS profile model can solve for the growth parameters such that each MDS dimension corresponds to a major growth profile. It is argued that the MDS model provides an exploratory tool for identifying growth trends and studying individual differences with respect to those growth trends. Since MDS has not traditionally been used for longitudinal studies, the MDS growth analysis can serve as the basis for studies of the kind discussed in the paper.