Using Multivariate Data to Structure Developmental Change
This chapter examines the ideas in detail, using a contemporary modeling approach. It discusses some issues centered on the application of selected multivariate techniques to the study of change and process — the sine qua non of development. The chapter presents standard multivariate results based on exploratory-factor analysis, canonical correlation, and multivariate analysis of variance. It also presents details on a confirmatory version of the traditional common-factor model and the choice between alternative structural models in both quantitative and qualitative detail. Multivariate investigations usually begin with some preliminary examination of the univariate data and bivariate relations. The individual differences in growth can be found by either the traditional multivariate methods or by the newer structural equation methods. The chapter suggests that initially plotting the raw data and using preliminary multivariate techniques, such as factor analysis, canonical correlation, and MANOVA to obtain a broad picture of the developmental changes within a longitudinal, multivariate, data set.