Assessing relationship quality
Discussions on how to measure the quality of relationships has since been ongoing. The originally developed scale by Dansereau and colleagues (1975) has been refined multiple times. Liden and colleagues (2006) assessed leader-member exchange quality with the LMX-13 scale from subordinates’ perspective. In a survey prepared by Kacmar et al. (2003) a seven-item scale developed by Scandura, Graen and Novak (1986) was applied. The scale showed strong reliability with a Cronbach alpha value of .87. Another study by Mayer et al. (2008) used a measure by Scandura and Graen (1984) to evaluate LMX in three independent samples of 209 employees (a = .80), 904 employees (a = .93), and 455 individuals (a = .92).
Graen and Uhl-Bien (1995) theorize about two aspects that have mainly driven the evolution of the LMX measure. First, it has undergone multiple phases of refinement and second, the question whether it is uni- or multidimensional has not been solved. As the range of reliability for a single measure is expectedly high, the hypothesis of LMX being unidimensional has often caused controversies. The researchers claim LMX to have three dimensions: respect, trust, and obligation (Graen & Uhl-Bien, 1995, p. 237). As all three sub-dimensions are greatly correlated, they may be employed as a unidimensional instrument. The LMX-MDM, containing 12 items, is applied as a multidimensional scale assessing affect, loyalty, contribution, and professional respect (Liden & Maslyn, 1998). Whenever used, studies revealed no evidence that any of the subscales showed stronger predictive validity (Erdogan & Bauer, 2014, p. 409). A frequently applied LMX measure is the LMX-7 (Lee, Scandura & Sharif, 2014; Zhang, Waldman & Wang, 2012) which is recommended by Graen and Uhl-Bien (1995). The seven-item measure generated reliability values between .80 and .90 and is assumed to constitute the most appropriate measure of leader-member exchange to date. Davis and Bryant (2010) adopted the LMX-7 in their investigation and reported a reasonable coefficient alpha of .83.
The meta-analytic review by Gerstner and Day (1997) included assessments of 79 studies containing 85 independent samples. Checking for reliability, the authors discovered a mean item-number of 7.57 with a mean sample-weighed alpha of .85. Predictably, the seven-item version of the LMX (Graen, Novak & Sommerkamp, 1982) showed a higher alpha coefficient for members’ LMX (a = .89) than all other related scales (a = .83) (Gerstner & Day, 1997, p. 831). The researchers detected a slightly better reliability for members than for leaders (a = .77).
To the question of whether LMX is transactional or transformational, the researchers argue that it entails elements of both dimensions:
LMX is both transactional and transformational: It is a dyadic social exchange process that begins with more limited social “transactions” [...], but for those who are able to generate the most effective LMX relationships, the type of leadership that results is transformational. (Graen & Uhl-Bien, 1995, p. 239)
This definition leads to the understanding that LMX is expected to relate to both, transactional and transformational leadership behavior with transformational leadership being associated with greater LMX quality. LMX development was found to increase with the leader communicating a compelling vision, which is related to the demonstration of transformational leadership behaviors (Wang et al., 2005).
Erdogan and Bauer (2014) argue that team members with a high quality LMX with their supervisors experience a much more favorable work atmosphere. Thus, leaders in high quality LMX relationships tend to challenge followers, whereas lower LMX quality results in much more authority (Fairhurst & Chandler, 1989). Team members involved in a high quality relationship with their supervisor are given several advantages over those who have low quality work relationships. Benefits include generous resources, superior projects and emotional support (Liden & Graen, 1980). If less frequent communication limits the amount of exchange and feedback between leader and subordinate even in high quality relationships, uncertainty may appear and limit performance and subsequently performance ratings by supervisors (Andrews & Kacmar, 2001; Kacmar et al., 2003).