An exploration of supervisor-subordinate agreement on leader-member exchange

Date of Award




Degree Name

Doctor of Philosophy (Ph.D.)

First Committee Member

Linda L. Neider, Committee Chair


Although Leader-Member Exchange (LMX) research has acknowledged the lack of supervisor-subordinate agreement in LMX ratings (Gerstner & Day, 1997), relatively little is known about this phenomenon. The primary purpose of this dissertation was to test rival explanations for poor SLMX-LMX agreement. This project is the first attempt to both qualitatively and quantitatively study this issue. In particular, three studies were conducted to test two most probable explanations: (1) lack of measurement invariance, and (2) perceptual differences in the LMX construct.The measurement equivalence of standard LMX instruments was examined in Studies 1 and 2. With few exceptions, nonequivalent items were identified in the LMX scales. However, the level of convergence in leader-member descriptions of their relationship was not improved even after the nonequivalent items were eliminated from the initial scales. Therefore, the results suggest that the failure of observed ratings to converge between supervisors and subordinates may be less a function of measurement problems than of substantive differences between these two perspectives.Consequently, Studies 2 and 3 were conducted to examine the second explanation with respect to perceptual differences. The results provided strong support for the lack of convergence on LMX subdimension perceptions as an explanation for poor supervisor-subordinate agreement on LMX ratings. In particular, these findings suggest that leaders are more likely to pay attention to the task-oriented dimensions and attribute higher importance to them when making an overall evaluation of the LMX relationship. However, subordinates are more oriented toward the social aspects of the working relationship and perceive them as more important. Moreover, this pattern of differential perceptions of LMX subdimensions was found to be generalizable across gender types, races, and industry types, but might be affected somewhat by group size.In addition, this dissertation explores other factors (e.g., LMX developmental stage, information processing styles, level of concept, and attributional biases, etc.) that can explain the perceptual differences between leaders and their followers. At the end, recommendations and suggestions for future investigations of supervisor-subordinate agreement on LMX reports are provided.


Business Administration, Management

Link to Full Text