Abstract
The Leader-Member Exchange model (LMX) has not been adequately explored via multi-level analysis, even though its propositions warrant such examination. The theory has been proposed to have both within-group effects as well as between-group effects. Also, the relationship between LMX and performance has been equivocal, and researchers have called for the exploration of moderating variables to understand these conflicting findings. Therefore, the purpose of this dissertation is (a) to examine the inconsistencies between LMX and performance by including situational factors and (b) to explore these relationships from a multiple levels perspective.Four contextual variables were included in this study on the basis of a review of situational leadership theories and the LMX literature: work unit size, cohesiveness, climate (operationalized with measures of conflict vs. cooperation, autonomy, skill variety, supportiveness, and social relations following from the James and James, 1992, conceptualization of psychological climate), and leader power (operationalized with measures of expert, referent, legitimate, reward, and coercive power following from the French & Raven, 1959, conceptualization of power).Direct effects were found for the relationships of cohesiveness and all climate variables with LMX. Average work unit cohesiveness was also related to variance in LMX quality. Cohesiveness moderated the relationship between LMX and performance. Conflict vs. cooperation moderated the relationship between LMX and satisfaction with supervisor-human relations. Autonomy moderated the relationship between LMX and job satisfaction and LMX and satisfaction with supervisor-human relations. Skill variety moderated the relationship between LMX quality and satisfaction with supervisor-technical ability. The last climate variable to interact with LMX was supportiveness, which moderated the relationship between LMX and job satisfaction and the LMX/satisfaction with supervisor-technical ability relationship. Coercive power moderated the relationship between LMX and both satisfaction with supervisor dependent variables.Noteworthy in this research is that several of the relationships found would have been masked if raw score analyses alone had been conducted, underscoring the need for both theorizing and testing of relationships at multiple levels. Second, the majority of the relationships analyzed revealed that effects occurred at both within- and between-group levels, highlighting the fact that both the traditional Average Leadership Style approach as well as the LMX conceptualization of leadership occurred concurrently in work units.Thus, the current research expanded the LMX framework to include situational variables previously omitted, and demonstrated that contextual factors do have an effect on the LMX relationship that develops as well as on the outcomes of this relationship.