Hypothesis generation in latent growth curve modeling using principal components

While confirmatory latent growth curve analyses provide procedures for testing hypotheses about latent growth curves underlying data, one must first derive hypotheses to be tested. It is argued that such hypotheses should be generated from a combination of theory and exploratory data analyses. An exploratory components analysis is described and illustrated. The analysis can be used to derive hypotheses about the number and form of curves underlying growth phenomena. Alternatives to the components analysis are discussed along with challenges to the implementation of the analysis.