A Method for Analysis of Ideal Types and Configurations

The social science literature draws on three dominant modes of theorizing. First is a universalistic perspective that comprises direct effects between independent and dependent variables. Second is a contingency perspective, which extends universalistic relationships and suggests interactions between independent variables. Third is a configurational perspective, which aims at identifying unique patterns of factors. For example, a specific pattern of individuals' personality traits or organizations' human resources (HR) practices might increase performance beyond what can be explained by direct or interaction effects. However, empirical research has mostly been confined to the first two modes, arguably due to the lack of suitable methods to test configurational predictions. In this article, I present a new approach to testing configurations (e.g., ideal types, archetypes, and patterns of factors) that is based on the concept of spatial dependence. I develop a stepwise statistical approach that first tests for the existence of ideal types or configurations in the data and then allows for the identification of pre-defined types and patterns. The utility of this approach is demonstrated in empirical data and Monte Carlo simulations.