Discussion of the special issue on growth models for longitudinal data in educational research

In his introduction to this special issue, Roger Millsap has given an excellent summary of its contents. This eases our task considerably. In what follows, we first present some general observations about growth curve modeling of longitudinal data. Among other things, we will address the criticism that the popular latent growth curve model may yield spurious (linear, quadratic, etc.) components. Then we will proceed with a discussion of each of the 5 contributing papers. In closing, we will draw attention to new methodological developments that are of interest to analyses of developmental processes in general.