|
Article Excerpt
Leader-member exchange (LMX) theory contends that leaders develop separate relationships with each of their subordinates through a series of work-related exchanges (Graen & Cashman, 1975; Graen & Scandura, 1987). These relationships range from those which are based on exchanges directly specified by the employment contract (low LMX relationships) to those which include the exchange of both material and nonmaterial goods that extend beyond what is required by the employment contract (high LMX relationships; Liden & Maslyn, 1998). High LMX relationships tend to be characterized by mutual respect, liking, and trust (Dansereau, Graen, & Haga, 1975).
LMX traditionally has been considered a unidimensional construct (Graen & Cashman, 1975; Graen, Novak, & Summerkamp, 1982; Graen & Scandura, 1987). However, several theorists (e.g. Dienesch & Liden, 1986; Liden, Sparrowe, & Wayne, 1997) have argued that treating LMX as a multidimensional construct more appropriately characterizes the leader-member relationship and allows for more complete descriptions of its relationships with important individual and organizational outcomes. Consequently, laden and Maslyn (1998) developed a multidimensional scale (the LMX-MDM) designed to measure subordinate LMX perceptions. However, Liden et al. (1997) noted that future LMX research should assess the generalizability of the LMX-MDM and should also develop a supervisor version of the multidimensional LMX scale. Accordingly, the current study addresses these issues and contributes to the existing LMX research as described below.
The current study has the following objectives. The first objective is to assess the dimensionality and validity of Liden and Maslyn's (1998) LMX-MDM scale with the current study's sample. Establishing the generalizability and validity of the LXM-MDM is important if appropriate inferences are to be drawn about how the LXM-MDM relates to important individual and organizational criteria. It is especially important to replicate the factor structure and validity evidence of laden and Maslyn's (1998) LXM-MDM scale because this is the first scale to operationalize LMX as comprising four distinct dimensions. The second objective of the current study is to create a multidimensional supervisor measure of LMX (the SLMX-MDM). Because LMX theory highlights the importance of exchanges between the supervisor and the subordinate, it is important to measure both perspectives in order to more fully and appropriately capture the LMX relationship (Liden & Maslyn, 1998). The current study develops a measure of supervisor LMX-MDM which may be used in future studies interested in capturing the LMX relationship from the supervisor's perspective. The third objective of the current study is explore whether the dimensions of the LMX-MDM and SLMX-MDM differentially predict individual attitudes and behaviours. A better understanding of which aspects of the LMX relationship impact individual attitudes and behaviours will help us to develop and refine our theories about LMX. Additionally, an understanding of how the LMX-MDM and SLMX-MDM relate to important individual and organizational criteria may help practitioners develop interventions aimed at improving employee attitudes and behaviours. The fourth objective of the current study is to examine whether the LMXMDM and SLMX-MDM predict criteria above scales that are either unidimensional or measured from just one perspective. The comparisons of the various LMX measures will provide information regarding the predictive utility of each measure. This comparison information can help researchers and practitioners select the most appropriate measure to best capture LMX relationships. Each of these objectives is discussed below.
Dimensionality of LMX
Traditionally, LMX has been treated as a global construct that represents a measure of the general quality of the exchange relationship between a supervisor and a subordinate (Graen & Cashman, 1975; Graen et al., 1982; Graen & Scandura, 1987). More recently, however, Dienesch and laden (1986) argued that there is no clear theoretical or empirical justification for the traditional conceptualization of LMX as a unidimensional construct. Rather, they maintained that the theoretical underpinnings of LMX theory (e.g. role theory, social exchange theory) are more consistent with a multidimensional perspective.
Role theory (Kahn, Wolfe, Quinn, Snock, & Rosenthal, 1964) served as the foundation for the formulation of LMX (Graen, 1976). Graen and Scandura (1987) proposed that leader-member relationships develop as a result of a variety of role-making episodes. Essentially, supervisors communicate role expectations to subordinates through work assignments. To the extent that subordinates comply with these role expectations, supervisors reciprocate by providing work-related resources, challenging work assignments, and increasing a subordinate's autonomy (Graen & Scandura, 1987). Accordingly, LMX originally was conceptualized as a unidimensional relationship based on work-related exchanges. However, role theory posits that roles are inherently multidimensional (Katz & Kahn, 1978). For example, concepts such as role ambiguity and role conflict (Kahn et al., 1964) are indicative of the differential roles people can assume in the workplace (Liden & Maslyn, 1998). Mintzberg (1973) has argued that the manager alone plays the following roles within an organization: figurehead, leader, liaison, monitor, disseminator, spokesman, entrepreneur, disturbance handler, resource allocator and negotiator. Similarly, while some subordinates may focus on the task-related aspects of their positions, others may concentrate on non-job specific behaviours such as helping co-workers or showing dedication (Borman & Motowidlo, 1993; Organ, 1997). Thus, different types of LMX relationships are likely to emerge depending on the roles assumed by dyadic members (Dienesch & Liden, 1986; Liden & Maslyn, 1998; Liden et al., 1997).
LMX also can be understood from a social exchange theory perspective (Dienesch & Liden, 1986; Liden & Maslyn, 1998; Liden et al., 1997). As the term leader-member exchange implies, LMX relationships are grounded in social exchanges. Blau (1964) noted that social exchanges, as opposed to economic exchanges, result in feelings of increased obligation, gratitude and trust. Consequently, as the number of social exchanges between supervisors and subordinates increases, the quality of the leader-member relationship probably becomes stronger. In the organizational environment, a variety of different material and non-material goods are exchanged in social interactions (Liden & Maslyn, 1998). For example, advice, information, effort, social support and friendship have each been identified as potential social currencies (Krackhardt, 1990; Liden et al., 1997; Sparrowe & Liden, 1997). Dienesch and Liden (1986) argued that leader-member relationships can be differentiated based on the social currencies being exchanged. Thus, just as leader-member relationships can result from several different types of social exchanges, these dyadic relationships are most appropriately assessed when they are differentiated based on those exchanges. Accordingly, dyads that appear to be similar when measured with a unidimensional LMX scale (i.e. high in quality) can in actuality be predicated upon different types of exchanges and may be very dissimilar in nature (Liden & Maslyn, 1998; Liden et al., 1997). For example, one LMX relationship may be based on both members being willing to work extra hours to finish projects, while another may be based on both members enjoying each other's company. Both of these leader-member dyads might report a high level of LMX, but their interactions (as well as the antecedents and consequences of those interactions) might be very different. As such, assessing LMX with a unidimensional measure may be deficient in capturing the entire scope or nature of LMX relationships.
Dienesch and Liden (1986) initially suggested that LMX relationships are based on three different exchange 'currencies': perceived contribution to the exchange (contribution), expressions of public support (loyalty), and mutual affection (affect). Subsequent research indicated that a fourth dimension (i.e. professional respect) was required to more fully capture LMX relationships (Liden & Maslyn, 1998). Consequently, Liden and Maslyn (1998) developed and validated the LMX-MDM to measure these four LMX dimensions, which assesses the LMX relationship from the perspective of the subordinate. One of Dienesch and Liden's (1986) general criticisms of LMX research is its reliance on small, narrow samples. Likewise, Liden and Maslyn (1998) noted the need for the LMX-MDM to be validated with different samples and organizations. Accordingly, the first objective of the current study is to assess the dimensionality and validity of the LMX-MDM using the current study's sample.
Measurement of LMX
In addition to the debate concerning the dimensionality of LMX relationships, there are two additional issues related to how LMX relationships should be measured: whose perspective should be measured to capture the LMX relationship? Also, what aspect of the LMX relationship should be measured? The prevailing practice in LMX research has been to measure LMX solely from the subordinate's perspective (Scandura & Schriesheim, 1994). Perhaps the reason for the emphasis on measuring LMX from the subordinate's perspective is the belief that LMX is considered to be a shared understanding, and therefore, assessing supervisor LMX would provide little additional information. However, Gerstner and Day (1997), in a meta-analytic review of LMX correlates, reported a .29 mean sample-weighted correlation between leader and member perceptions of LMX. The magnitude of this correlation is quite low for a relationship characterized by purportedly shared perspectives, indicating that supervisors' and subordinates' views of their LMX relationships are quite discrepant. As such, assessing LMX from only one perspective may provide an incomplete or inaccurate depiction of the LMX relationship. Consequently, several researchers have argued that LMX should always be assessed from the perspective of both the subordinate and the supervisor (Gerstner & Day, 1997; Scandura & Schriesheim, 1994; Schriesheim, Neider, & Scandura, 1998).
The second issue regarding the measurement of the LMX relationship centres on which aspect of the relationship to measure (i.e. the target of the measurement). LMX measures were initially developed to assess subordinates' perceptions of their relationships with their supervisors. These existing scales largely ask subordinates to report, for example, how much they like or respect their supervisors (e.g. Liden & Maslyn, 1998). In the few cases where the supervisor's perspective has been measured, researchers generally have created measures that were 'mirrors' of the subordinate scale. For example, an item from the LMX-7 asks the subordinate the question: 'How well do you feel that your immediate supervisor recognizes your potential?' This item is then mirrored and presented to the supervisor as: 'How well do you feel that you recognize this subordinate's potential?' This mirroring of items is 'common practice' (Schriesheim et aL, 1998, p. 304) when developing a measure of supervisor version of LMX (i.e. SLMX). Although mirroring the item in this manner achieves the objective of measuring the LMX relationship from both the subordinate's and supervisor's perspective, this mirroring approach captures only one aspect of the LMX relationship (e.g. whether the supervisor recognizes the subordinate's potential). That is, the mirroring approach neglects to capture the exchanges implicit in LMX theory (Liden & Maslyn, 1998; Sparrowe & Liden, 1997). The majority of past approaches that have measured SLMX primarily draw on what the supervisor does for the subordinate, rather than what the subordinate does for the supervisor (Maslyn & Uhl-Bien, 2001).
LMX relationships are described as partnerships and as reflecting 'individuals' assessments of each other in terms of professional capabilities and behaviors' (Graen & Uhl-Bien, 1995, pp. 237-238). As noted above, the mirroring approach fails to capture these partnerships and the interdependencies inherent in LMX relationships. As such, to more fully capture the dynamics of LMX relationships, we believe it is more appropriate to measure the supervisor's and subordinate's assessments of each other in terms of the exchanges inherent in their dyadic relationships. Using the above item as an illustration, instead of asking the supervisor 'How well do you feel that you recognize this subordinate's potential?', we believe it is more appropriate to ask a parallel (but not mirrored) item (e.g. 'How well do you feel that the subordinate recognizes your potential?'). In doing so, rather than both perspectives using the subordinate as the target (and thereby providing conceptually redundant information but from different perspectives), each perspective reports on a unique aspect of the LMX exchange relationship (conceptually independent information).
Few studies have measured SLMX other than by simply mirroring the LMX items (for an exception, see Maslyn & Uhl-Bien, 2001). Further, there exists no published multidimensional scale that measures supervisors' perceptions of LMX. Although the multidimensionality of subordinate LMX perceptions has been demonstrated (Liden & Maslyn, 1998; Schriesheim, Neider, Scandura, & Tepper, 1992a; Schriesheim, Scandura, Neider, & Eisenbach, 1992b), no research has been conducted to evaluate whether supervisors also conceptualize LMX in a multidimensional manner. The second objective of the current study is to assess the multidimensionality of supervisor LMX perceptions by developing a multidimensional scale (i.e. the SLMX-MDM) that captures the exchanges implicit in LMX theory.
Relationships between LMX, SLMX and employee job attitudes and behaviours
The third objective of the current study is to assess the convergent and criterion-related validity of the LMX-MDM and SLMX-MDM. We also investigate whether the dimensions of the LMX-MDM and SLMX-MDM differentially predict individual job attitudes and behaviours. Liden and Maslyn (1998) presented validity evidence for the LMX-MDM. The present investigation adds to this evidence by assessing the LXM-MDM's validity with the current sample of employees and extends this research by assessing the validity of the SLMX-MDM. The criterion variables (described below) were chosen based on theoretical linkages to LMX relationships and to allow comparisons between the current study's results and those of Liden and Maslyn (1998).
Convergent validity
As observed in Liden and Maslyn (1998), we expect that each dimension of the LMXMDM will significantly correlate with the LMX-7 (a unidimensional measure), given that both purport to measure the same construct. Likewise, because the SLMX-MDM and SLMX-7 purport to measure the same construct, we expect them to be significantly correlated with one another.
Criterion-related validity
We examined subordinate satisfaction with supervisor, job involvement, affective organizational commitment, in-role performance, and organizational citizenship behaviours (OCBs) as important attitudinal and behavioural correlates of LMX relationships. Theoretically, we expect these criteria to be important because LMX reflects the quality of the supervisor-subordinate relationship. The quality of this relationship, in turn, probably influences the amount of resources and emotional support exchanged between members of the...
|
