The effects of leader member exchange (LMX) have been widely studied, yet the mechanisms and its valence as a contingency factor are muddled. When modeled as a moderator variable, LMX has been shown to moderate relationships both positively and negatively. Beyond some values of LMX as a moderator, detrimental inversions can occur wherein LMX changes the main effect of a typically beneficial factor (e.g., a positive leadership style) on an outcome factor (e.g., follower performance) from positive to negative. Given the potential counter-intuitive effects of too much or too little LMX, we review different possible cases of such inversions and propose theoretical mechanisms and patterns of effects to classify and integrate them. We analyze a sample of 37 moderated relationships from published studies wherein LMX served as a moderator. We observe that inversions may occur both in the high and the low ranges of LMX. Using an innovative methodology, we also demonstrate that inversions probably occur in 36 of the 37 primary studies in our sample, but yet the authors did not test for those potential inversions, and probably lacked the range of data to do so. Future leadership research should theorize and systematically collect data to assess the entire range of LMX as well as other contingent leadership factors which may trigger similar inversions. This approach allows researchers to avoid the false perceptions that certain leadership factors create inherently either beneficial or detrimental effects, since inversions can occur at certain levels of the moderator as demonstrated here with LMX.
