Use and misuse of levels of analysis in leadership research: An illustrative review of leader–member exchange

Prior reviews have focused on if leadership literature pays homage to levels of analysis; our purpose is to take stock of how well and in what ways levels issues are implemented in theory and hypotheses formulation, construct conceptualization, measurement, and data analysis. To illustrate these notions, we provide an in-depth review of 163 multi-level empirical leader–member exchange (LMX) and vertical dyad linkage (VDL) studies published between 1972 and 2012. Our findings indicate that the volume of multi-level studies in LMX has drastically increased recently and the majority of these studies cast theory and hypotheses at the same level. With regard to alignment of levels of theory/hypotheses with measurement and with data analyses, however, roughly one out of two published studies suffers from misalignment of levels. Most of this misalignment occurs in studies where the dyad level of analysis, a key component of LMX, is an explicit or implicit focus. We conclude with a discussion of the implications for the LMX literature, and present recommendations to guide multi-level leadership research.