Leader-Member Exchange (LMX) measurement: Evidence for consensus, construct breadth, and discriminant validity Article

Joseph, DL, Newman, DA, Sin, HP. (2011). Leader-Member Exchange (LMX) measurement: Evidence for consensus, construct breadth, and discriminant validity . 6 89-135. 10.1108/S1479-8387(2011)0000006012

cited authors

  • Joseph, DL; Newman, DA; Sin, HP

authors

abstract

  • Purpose - This chapter (a) summarizes leader-member exchange (LMX) measurement practices since the influential reviews by Schriesheim, Castro, and Cogliser (1999) and Gerstner and Day (1997), (b) clarifies the status of LMX as a broad construct from a hierarchical factor model, (c) conducts multitrait-multimethod (MTMM) analyses on leader and follower reports of multidimensional LMX, and (d) investigates discriminant validity between Member LMX and satisfaction with supervisor.Methodology/Approach - We used (a) a literature search of LMX measurement practices, (b) a combination of meta-analysis and factor analysis to specify the broad LMX construct underlying Liden and Maslyn's (1998) (LMX-MDM) multidimensional instrument, (c) MTMM analyses of leader and member ratings of the LMX-MDM, and (d) a combination of meta-analysis and multiple regression to assess incremental validity of Member LMX beyond satisfaction with supervisor. Findings - Since 1999, 85% of LMX studies now use one of two dominant LMX scales (LMX-7, Scandura, & Graen, 1984; LMX-MDM, Liden & Maslyn, 1998). These two measures are correlated (rcorrected = 9), suggesting the LMX-7 and the LMX-MDM are alternate forms of the same instrument. 94% of studies that used these two measures treat LMX as a single, broad construct rather than as a multidimensional set of constructs. MTMM analyses suggest Leader LMX and Member LMX are two, separate-but-related constructs (i.e., confirming two source factors and no lower-order trait factors). Last, Member LMX meta-analytically correlates with satisfaction with supervisor at rcorrected = 8. There is some incremental validity of LMX, but the pattern is inconsistent across samples. Social Implications - We point out that LMX researchers have now moved toward standard measurement of LMX - as a broad, higher-order factor that varies between leader and follower. By doing so, we reveal that the stage is set for cumulative and replicable research on leadership as a dyadic, follower-specific phenomenon.Originality/Value of Paper - Our chapter is the first to reveal consensus in LMX measurement across studies; to summarize the standard treatment of LMX as a single, broad factor; and to apply MTMM analyses to demonstrate separate Leader LMX and Member LMX source factors.

publication date

  • December 1, 2011

Digital Object Identifier (DOI)

start page

  • 89

end page

  • 135

volume

  • 6