Longitudinal data analysis: Applications of random coefficient modeling to leadership research

Understanding how leaders develop, adapt, and perform over time is central to many theories of leadership. However, for a variety of conceptual and methodological reasons, such longitudinal research remains uncommon in the leadership domain. The purpose of this paper is to introduce the leadership scholar to conceptual issues involved in longitudinal design and analysis, and then demonstrate the application of random coefficient modeling (RCM) as a framework capable of modeling longitudinal leadership data. The RCM framework is an extension of the traditional regression model, so many readers will already have the fundamental knowledge required to use RCM. However, RCM has the additional capability of analyzing the kinds of data commonly found in longitudinal studies, including correlated observations, missing data, and heterogeneity over time. Further, the RCM allows for testing predictors of change over time. Thus, we introduce conceptual issues related to longitudinal research, discuss RCM within the context of regression, and conclude with an application of the RCM approach. We use a common substantive example throughout the paper to facilitate our discussion of the RCM.