Multilevel Methods

Abstract

The study of multilevel phenomena in organizations involves a complex interplay between methods and statistics on one hand and theory development on the other. In this introduction, the authors provide a short summary of the five articles in this feature topic and use them as a platform to discuss the broad need for work in the two areas of (a) multilevel construct validation and measurement and (b) statistical advances in variance decomposition. Within these two broad frameworks, the authors specifically discuss, first, the need to continue moving beyond notions of isomorphism in developing and testing aggregate-level constructs. Second, they discuss the potential value of using discontinuous growth models to understand transitions in longitudinal studies. Finally, they discuss some of the issues surrounding the ability to decompose variance in multilevel modeling of dichotomous and other nonnormal outcome data.