An Empirical Testing of A Multi-route Model on LMX - Job Satisfaction Relationship

With more than enough empirical evidence supporting a positive relationship between Leader-Member Exchange (LMX) and job satisfaction (Gerstner et al., 1997), LMX literature suggests that a subordinate is more likely to be satisfied with a job if s/he is in good relationship with a supervisor than those who are not. However, this suggestion is not quite intuitive, because LMX captures an interpersonal relationship with a supervisor while job satisfaction reflects a job incumbent’s attitude toward his or her own job. Unfortunately, the current literature provides no theoretical explanation why a good interpersonal relationship captured by LMX would lead an individual to experience a positive attitude toward his/her work. To fill this theoretical gap, a premise that employees’ job satisfaction level would be influenced by some job characteristics and role stress factors, which are in turn influenced by some supervisory activities, was specified in a multi-route model that incorporated six variables as mediators. This premise can be thought of as an application of situational constraints perspective (Peters, O’Connor, & Rudolf, 1980) to job satisfaction. This perspective suggests that situational constraints may prevent people from successfully accomplishing a task. Although the major focus of situational constraints research has been on performance (Peters & O’Connor, 1980), a series of empirical studies showed that employees who perceived high levels of constraints tended to be dissatisfied with their jobs (Jex & Gudanowski, 1992; Keenan & Newton, 1984; O’Connor, 1984; Spector, Dwyer, & Jex, 1988). While situational constraints may come from various aspects of a job, one of most likely aspects would be a supervisor (Spector, 1997: 38). Supervisors exercise a significant amount of influence on subordinates’ perception of situational factors (Tetlock, Skitka, & Boettger, 1989) as well as behaviors (Tetlock, 1985). And, subordinates’ satisfaction with the supervisory activities are shown to be more highly correlated with perceived situational constraints (O’Connor, Peters, Rudolf, & Pooyan, 1982) and with perceived role conflict and role ambiguity (Jackson & Schuler, 1985) than any other job satisfaction facet. Taken altogether, this literature suggests that supervisory activities entailed by a certain quality of LMX are likely to determine various job aspects, which in turn constrain or facilitate job satisfaction level. To further elaborate this premise, the current model incorporated six variables (i.e., initiated task interdependence, supervisor feedback, autonomy, task significance, role conflict, and role ambiguity) as the mediators between LMX and job satisfaction. Initiated task interdependence and supervisory feedback were incorporated as the proxy variables of the results or outcomes of two representative supervisory activities, that is, work assignment and supervisory communication. And, autonomy, task significance, role conflict, and role ambiguity were set as the direct antecedents of job satisfaction, drawing on the literature of job characteristics (Fried & Ferris, 1987; Hackman & Oldham, 1975; 1976; Kiggundu, 1981; 1983) and role theory (House & Rizzo, 1972; Jackson & Schuler, 1985; Katz & Kahn, 1978). With these variables, the model states that a subordinate with high quality of LMX would experience a high level of job satisfaction, possibly 1) because s/he is given more autonomy by being assigned to a job that initiates a work for other people (high initiated task interdependence), 2) because s/he is assigned to a job perceived as important and meaningful (high task significance), or 3) because s/he experiences low levels of role conflict and role ambiguity by getting more information from the supervisor. The model was empirically tested and compared with four alternative models on a data set collected from 409 supervisor-subordinate dyads by using LISREL. In general, the result supported the premise of the current study. That is, job satisfaction level was determined by work-related factors, which were in turn affected by the quality of LMX either directly or indirectly through supervisory activities. This result extends our knowledge by showing how and why an individual’s attitudes toward his or her own work are influenced by his or her interpersonal relationship. And, such an extended knowledge provides us with more specific guidelines for how to design and administer organizational intervention programs to improve work attitudes. For example, the finding that LMX does not have a direct effect on job satisfaction suggests that being a nice boss would be necessary, but not sufficient to improve an employee’s work attitudes. Not only does a supervisor need to understand what kinds of working conditions an employee would value, s/he also needs to deliver such conditions through his or her supervisory activities, such as work assignment and communication. And, these activities are carried out in directions that subordinates perceive their job autonomous and/or important and experience less role conflict and ambiguity. Also, the finding that the effect of LMX on autonomy is fully mediated by initiated task interdependence hints at that a supervisor must take account of work structure among subordinates into his or her supervisory activities. When work is structured in a way that a job incumbent needs to depend on other people’s work outcome, coworkers of the employee would communicate role-expectation that is consistent with such work structure (Bravo, Peiró, Rodriguez, & Whitely, 2003). Such role expectations may not be compatible with the role expectations communicated by a supervisor, if s/he personally granted job autonomy to a subordinate based on the quality of LMX without any corresponding change of work structure. And, the inconsistent role expectations from different sources may lead the focal person to experience role conflict