New Ways of Seeing: Pitfalls and Opportunities in Multilevel Research

Abstract

Researchers once lamented the paucity of multilevel theory, models, and research in the literature (e.g., O’Reilly, 1990; Staw, 1984), but now management journals are replete with such studies. Around a decade ago, Hitt, Beamish, Jackson, and Mathieu (2007) noted that about a quarter of recent management publications were multilevel—undoubtedly, the trajectory remains positive. The proliferation may provide support for the adage that “the squeaky wheel gets the grease,” but it likely also reflects the field’s desire to develop more comprehensive, context-rich theory and findings. Moreover, the availability of “how to” volumes for developing multilevel theory and analyzing the associated data (e.g., Johns, 2001, 2006; Kozlowski & Klein, 2000), as well as the widespread availability of accessible statistical packages, contributes to the movement. The shift is both symbolic and substantive. The multilevel context—once treated as an unknown or messy source of error variance that needed to be controlled—is frequently at the heart of theorizing on a variety of topics. This is perhaps most evident in the teams literature (see Mathieu, Maynard, Rapp, & Gilson, 2008, for a review) where multilevel studies examine direct cross-level effects as well as contextual moderators that influence lower-level processes and outcomes (e.g., Yu & Zellmer-Bruhn, 2018).1But, the influence is apparent in other streams as well, including strategic human resource management (Ployhart,Weekley, & Ramsey, 2009), emotions (Scott, Barnes, & Wagner, 2012), social networks (Brass, Galaskiewicz, Greve, & Tsai, 2004), and many others.