;. “‘4‘ u-uw . . ~_, g . fj’fiT)‘: . x 1.13;} 1-13.: - $2325.} 6%.. “ ~. #92:; ~ ~ ' “1- 9 ‘."‘ ,4, " 2 " ”(naufin‘u . r H- :‘K' ‘r"-'u:~.: «9‘4" 3‘3 3‘33“}? :1 fl 2A,; “5'3!“ h u. ' 3 w. ‘3‘, This is to certify that the dissertation entitled ANTECEDENTS OF LEADER-MEMBER EXCHANGE presented by Carl Peter Borchgrevink has been accepted towards fulfillment of the requirements for Ph . D degree in Communication glam/735m Mljor professor Date July 14, 1994 MS U i: an Affirmative Action/Equal Opportunity Institution ‘ 3 0- 12771 iiiiiiiiiiiiiiiii 3 1293 010 LIBRARY Michigan State University me: u RETURN BOXto n'ifiav'; (in; Edam: hum your room. TO AVOID FINES Mom on or baton duo duo. DATE DUE DATE DUE DATE DUE ANTECEDENTS OF LEADER-MEMBER EXCHANGE BY Carl Peter Borchgrevink A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Communication 1994 ABSTRACT ANTECEDENTS OF LEADER-MEMBER EXCHANGE BY Carl Peter Borchgrevink In this monograph the theoretical and empirical foundations of Leader-Member Exchange are reviewed. Leader- Member Exchange is an important interpersonal construct with great relevance for organizational functioning, management, and leadership. We have limited knowledge about potential antecedents to Leader-Member Exchange quality. The intent of the research is to move toward filling this theoretical and empirical vacuum. Following the review of literature surrounding Leader-Member Exchange (LMX), LMX is discussed within the context of leadership. LMX is posited to develop through interaction between leader and member. Social exchange theory is used as explanatory framework for why leaders and members interact beyond their minimum role requirements. Homophily and similarity-attraction is used to suggest why particular interactional choices are made, allowing the LMX to develop. Power is shown to be reflected in the LMX, and power distance reduction theory is used to suggest why leaders and members when interacting will tend to magnify the relative quality of their LMX. Surveys are used for data collection. Regression analyses and path analyses are used and the results show that socioeconomic differences and power differences are related to LMX which in turn influences the communicative relationship. Racial similarity is also found positively related to communication. Copyright by CARL PETER BORCHGREVINK 1994 iv This dissertation is dedicated to my wife Barbara Borchgrevink for her endless love and support and to our children Peter Charles and Laura May, who helped keep my priorities straight. ACKNOWLEDGMENTS In completing my dissertation I realize that there are many who helped and encouraged me along my route towards the Ph.D., and I wish to thank them all. The following deserve particular recognition for their contributions. First, I wish to thank my parents Theodor and Gerd Borchgrevink for instilling in me the belief that all goals are within reach. I also wish to thank Dr. Odd H. Zahl-Begnum who during my undergraduate years encouraged my scholarly interests and strengthened my resolve to attend graduate school when I started to hesitate. Thanks are also extended to Dr. Robert C. Lewis, for inspiring and encouraging me towards an academic career and for providing me with a role model of a successful hospitality researcher. Much gratitude is expressed to the Department of Communication and the School of Hotel, Restaurant and Institutional Management. The former for accepting and educating this non-traditional student and to the latter for making it financially possible. vi I thank the members of my dissertation committee, Dr. Franklin J. Boster, Dr. Ronald F. Cichy, Dr. William A. Donohue, and Dr. J. David Johnson, for their insights, guidance and help in staying focussed. I wish to thank Dr. William A. Donahue for allowing me to gain from working on his research team and for sharing liberally of his experience, insights and wit. In particular I wish to thank Dr. Franklin J. Boster, Chairman of my committee, advisor, and role model for invaluable methodological consultation, theoretical insights, and general guidance. I feel most fortunate, both professionally and personally, to have had the opportunity to work with him. Finally, I wish to thank my wife Barbara Borchgrevink, for her belief in me, support and nurturance, without which I would never had completed this doctorate. vii Chapter 1. TABLE OF CONTENTS Introduction Leader-Member Exchange Leader-Member Exchange and Leadership Leader-Member Exchange: Transaction or Transformation? The Empirical Definition of Leadership Applied Utility and Predictive Ability Social Exchange Theory Subjective Expected Utility Comparison Levels Leader-Member Exchange and Social Exchange Theory Establishing the High Quality LMX Relationship The Similarity-Attraction Paradigm Similarity-Attraction and Organizations Similarity-Attraction and Leader-Member Exchange Organizations as Authority Structures Authoritarianism Authoritarianism Defined Power Power Schemes viii 11 18 19 20 22 23 26 32 35 37 39 4O 41 43 Chapter 2. Chapter 3. Chapter 4. Power Distance Reduction Theory Leader-Member Exchange and Power Distance Reduction Theory Summary Argument and Hypotheses Method Procedures Sample Instrumentation Results The Measurement Model Theoretical Models Causal Models Summary of Results Discussion Summary and Conclusion ix 46 48 50 55 55 56 59 62 62 75 78 81 83 92 Table Table Table Table LIST OF TABLES Demographic Data by Organization 58 Final Factor Solutions 64 Descriptive Statistics for Difference Variables 74 Correlations of LMX, Intradyadic Communication and the Difference Variables 76 Figure Figure Figure Figure LIST OF FIGURES Proposed causal model for LMX A causal model of the relationship between similarity, LMX, and intradyadic communication A causal model of the relationship between differences in power, LMX, and intradyadic communication A causal model of intradyadic similarity, LMX, and intradyadic communication xi 54 78 8O 82 Chapter 1 INTRODUCTION Most people work and live in hierarchically arranged social aggregates. As long as hierarchies remain the dominant organizing structure, knowledge about hierarchical relations will be important to those who work in, or study, organizational systems. In organizations this hierarchical relationship is most often referred to as the superior-subordinate relationship. The superior-subordinate relationship is of primary importance as 1) it is through this relationship that most formalized goals, role expectations and behaviors are explicated and negotiated; 2) there are many such relationships, as every subordinate has a superior; 3) research has found that using the superior-subordinate dyad as the unit of analysis provides added insight into leadership. The Leader—Member Exchange (LMX) is one facet of the superior-subordinate relationship, viz., the superior- subordinate dyadic exchange relationship. The Leader-Member Exchange is the interpersonal exchange relationship between two organizational members at different levels of the hierarchy. Leader-Member Exchange Dissatisfaction with leadership research and leadership theory led to viewing leadership from the perspective of the superior—subordinate dyad. Leadership theories and consequent leadership research had focussed upon a leader’s typical or average behavior toward his or her subordinates, under the assumption that the leader behaves uniformly toward subordinates. According to this perspective, each subordinate will eventually be exposed to the same set of leader behaviors over time, although various subordinates may be exposed to various leader behaviors in various settings (Vecchio, 1982). Graen (1976) called this proposed homogeneity of behavior "Average Leadership Style." Given the average leadership style focus, most leadership research used the group, organization or other aggregate as the unit of analysis (Bass, 1990: Kieser, Reber & Wunderer, 1987). Other scholars, however, argued that leader behavior varies across their subordinates. Therefore, they argue that a more appropriate level of analysis is the dyad (Cashman, Dansereau, Graen & Haga, 1976; Dansereau, Graen & Haga, 1975; Duchon, Green & Taber, 1986; Fairhurst & Chandler, 1989; Graen, Dansereau & Minami, 1972; Krantz, 1989; Liden & Graen, 1980). In discussing the flaws of the average leadership style approach, Dansereau et a1. (1975) specifically pointed to the vertical dyad as, " . . . the appropriate unit of analysis for examining leadership processes because the vertical dyad reflects the 2 3 processes linking member and superior" (p.47). The label Vertical Dyadic Linkage (VDL) was used to describe this approach. Concurrently, Graen and Cashman (1975) proposed the Leader-Member Exchange (LMX) label. Currently both labels are used. This practice is confusing as the labels seem to suggest two different constructs: VDL as a static link between superior and subordinate and LMX as the interaction between a leader and a member, but such is not the case, because both labels are used interchangeably (Vecchio, 1985: Vecchio, Griffeth & Hom, 1986). Leader- Member Exchange is the dominant label presently and will be used in this research. Leader-Member Exchange and Leadership Diagram Given that the seeds for Leader-Member Exchange originated in leadership research, it is appropriate to define leadership as it applies to LMX. In the social science literature the term leadership is used in three ways: an attribute of a position, the characteristics of a person, or a category of behavior (Katz & Kahn, 1978). The conceptual definition of the LMX implies that from this perspective leadership is conceived as a category of behavior. This behavior takes place between superior and subordinate as social exchange. Leadership has received further definition as well. In the following, the constructs of formal, informal 4 transactional and transformational leadership will be discussed and used in setting the scope of this inquiry. Formal Versus Informal Formal and informal leadership are frequently distinguished (Gibb, 1947). This research will focus upon formal leadership, while acknowledging that the LMX approach may have relevance for informal leadership. A focus on formally designated leaders can be seen as a focus on headship (Holloman, 1968). The choice of focus is pragmatic. Identifying informal leaders and their followers is a difficult methodological problem. Furthermore, all LMX research and theory to date has studied formal leaders, and the proposed intent of this research is to hypothesize and test for antecedents to LMX quality within LMX's current definition and domain. '0 r o Tgaggggtigngl_ngggr§hipL Leadership is also categorized into transactional leadership (Hollander 1978, 1993: Jacobs, 1970) and transformational leadership (Burns, 1978; Bass & Avolio, 1993). From the transactional leadership perspective leadership is an interactive process between leader and member based on social exchange principles. The transactional view sees leaders as providing benefits to members with an anticipation of reciprocation through heightened responsiveness to the leader’s concerns and needs (Hollander, 1993). This approach sees both leader and member as having the potential 5 to influence each other in the exchange. The leader and member influence each other by acting in response to, or in anticipation of, accruing benefit from their interplay. Examples of benefits provided by the leader may be direction, affect, and the power to make decisions. The member may bestow benefits such as exceeding the leader's behavioral expectations for laudable employees, adhering more closely to the organizational norm-set, or not blocking leader efforts. In essence, anything perceived to be within the control of either interactant can be provided or considered as a benefit to or by the other. IIansfgzmatigngl_L§ggg;§nip; Downton (1973) and Burns (1978) opposed the notion of leadership being a transactional relationship, and presented transformational leadership as an alternative. Transformational leaders focus on the needs of the member (potential follower), and attempt to meet them. In addition, transforming leaders go beyond the needs that the member is experiencing, arouses higher order needs (Maslow, 1943, 1954), and gratifies them. The primary notion is that the transforming leader engages the member further so that immediate short term needs are transcended as longer term self-development becomes the member's focus, and the collective interest replaces self- interest as the member’s primary goal (Bass, 1990). Proponents of this perspective argue that transactional leadership is commonL and much less worthy of attention and research than transformational leadership, the uncommon form 6 (Bass, 1982). Others argue that transformational leaders are not qualitatively different from transactional leaders, but are more effective in their use of strategies and tactics also available to transactional leaders (Chemers, 1993). Hollander (1993) argues that transformational leadership is an extension of transactional leadership and that it is made possible by vast quantities of idiosyncrasy credits at the leader’s disposal. Le de - emb xchan e: s ' 0 ns at ? WWW Graen and Scandura (1987) have put forth a role theory that consists of three stages: role taking, role making, and role routinization. According to their perspective all members desire and seek to negotiate change in their prescribed role after the role taking stage, and it is through role negotiation that the LMX develops (Graen, 1976: Graen & Scandura, 1987). As such, the LMX is most clearly transactional in nature because it develops as the result of role negotiations between leader and member. I turn now to the three stages of the role theory describing how LMX develops. Bglg_§aking; The role taking stage involves communication of initial role expectations to the member by the member's role-set (Katz & Kahn, 1978). The leader is often a primary sender in the role-set, as the leader tends to produce a great deal of role specific messages. The member receives the information and acts accordingly, 7 providing the leader with feedback regarding role acceptance, role understanding, or both. During this stage the member is primarily a passive recipient and internalizer of role information. It is in the next stage, :9lg_mgkingl that the negotiation process between leader and member takes place. Role making, Following role taking, which may take hours, weeks, or perhaps the entire relationship, the role making stage starts. During this stage the member is no longer a passive recipient of role information. Contrary to the previous stage, both the leader and the member are seen as communicating their understanding, preferences, and expectations regarding their respective roles, their relationship, and the organization at large. As such, they jointly develop how they will interact with each other and establish or discover the degree to which they are interdependent. This role communication and negotiation often takes the form of sequences of offers and counter offers regarding work related behaviors, tasks, or communication. The areas about which they talk and negotiate include 1) access to inside information, 2) provision or use of influence within the organizational system, 3) task choices and behaviors, 4) latitude (power) to make decisions, 5) leader (member) support of member (leader) activities and choices, i.e., intradyadic loyalty, and 6) individual attention. 8 Role routinization. Upon implicit or explicit acceptance and settlement of this negotiation process, the leader and member move into a role routinization process, in which the role behaviors and dyadic interdependencies become increasingly routinized and ossified. This regimen allows for clear expectations so that the dyadic partners know what to expect from each other within the negotiated domain. Although not discussed by Graen and Scandura (1987), there are likely aspects of either’s role that are not salient to leader or member, and therefore not discussed or negotiated. Under such circumstances it is reasonable to expect that the experience of the gally_g;ing sets the role expectations and parameters. The leader-member dyads return to role making and role- taking activities as needed. The routinized expectations may become formally institutionalized and entered into documents, such as job descriptions. 9 Transformatien through lndividuelized Attention Although LMX was discussed from the transactional perspective, it nevertheless has transformational components as well. According to Bass and Avolio (1993) individualized consideration that, "followers are treated differently but equitably on a one-to-one basis (p. 52)," is one of the four factors of transformational leadership. Differential treatment of members and the one-to-one (dyadic) focus is the cornerstone of LMX. This monograph works from the premise that transformational leadership and transactional leadership are both relevant for LMX. The continuum of LMX quality is likely related to transactional and transformational leadership in a predictable fashion. The descriptions of high quality and low quality LMX dyads, transformational leadership and transactional leadership suggest that leader- member dyads that have high quality are more likely experiencing transformational leader behaviors, whereas low quality dyads are more dependent on leader behaviors such as contingent reward and management-by-exception, which are transactional. new From the LMX perspective leadership is conceived behaviorally, it may be transactional (Jacobs, 1970) or transformational (Burns, 1978), and it typically considers only formal leaders (Gibb, 1947). Examining the research investigating LMX, however, it is not always apparent lO leadership is the focus of study. Most often when leadership is defined, the empirical definition used tends to be isomorphic with, or similar to, the LMX definition (e.g., Graen & Cashman, 1975: Rosse & Kraut, 1983; Vecchio & Gobdel, 1984). Most investigations of LMX, or the more broad superior- subordinate relationship, do not deal with leaders and followers directly. Transformational behaviors, transactional behaviors, or any other leader behaviors are not assessed. Only if leadership is defined as an attribute of a formally designated position, can it be argued that LMX research in general examines leadership. Consistent with the LMX research to date this study will focus upon formally designated superiors and subordinates between whom "leadership" as a characteristic of a person, or as a category of behavior (Katz & Hahn, 1978) may or may not take place. That is, this research can be seen as focussing on headship rather than leadership, per se (Holloman, 1968). This focus is not necessarily cause for alarm. According to Kotter (1990) most superior-subordinate relationships are not characterized by a great deal of leadership behaviors, particularly not at the "transformational" end of the continuum, but rather reflect the more mundane experience of manager-managed, or supervisor-supervised. Calls for practitioners to exude more leadership rather than management (Bennis & Nanus, 1985: Kotter, 1990) directs 11 researcher attention toward leadership behaviors. Nevertheless, it is still important to understand the more common "managed or supervisory" relationship, and how relationships become "manager-managed" versus "leader- follower." Understanding how relationships may move from one end of the LMX continuum to the other is also important, and may provide the relevance sought by practitioners (Mintzberg, 1982). If, as posited, the LMX measure reflects leader behavior, it should prove to be a useful tool in understanding organizational behaviors. Furthermore, LMX may be helpful in understanding the process involved in moving the leader-member relationship from a manager-managed relationship to a leader—follower relationship. The pragmatic value of LMX is also important to consider in that it demonstrates the benefits that organizations may accrue by paying attention to LMX development. Next, I turn to these utility considerations. i 1' :II!']'! :E i't' “1.! _ WW TheLMXscale has had strong predictive success. For instance, Graen, Liden, and Hoel (1982) investigated the employee withdrawal process and found LMX quality to predict turnover. Specifically, low quality LMX employees tend to terminate employment more frequently than high quality LMX employees. Vecchio and Gobdel (1984) found an inverse relationship between the intention to quit and LMX quality. Ferris 12 (1985) replicated this relationship, finding that LMX quality predicted tenure more accurately than average leadership style. Organizational commitment and voluntary turnover are related (Baysinger & Mobley, 1983, Reichers, 1985). Thus, not surprisingly, organizational commitment correlates highly with LMX quality. Graen, Wakabayashi, Graen, and Graen (1990) report a correlation of z=.4l between subordinates’ organizational commitment and LMX. Borchgrevink and Boster (1994) found strong relationships between two LMX factors and occupational commitment (n=.54 and ;=.61, respectively). Failure to replicate the IMmZand turnover relationship has also occurred (Borchgrevink & Boster, 1994: Vecchio, 1985; Vecchio, Griffeth & Mom, 1986). This result is not surprising because the superior-subordinate relationship is only one of a multitude of potential predictors of turnover (Baysinger & Mobley, 1983). The non-null findings are important, however, as employee retention is a major problem for many industries. For instance, the hospitality industry is particularly concerned as turnover as high as 300% has been reported (Woods & Macauley, 1989). Promotion and career progress within an organization may be related to turnover, tenure, and commitment because neither is possible if the employee leaves the employing organization. Graen et al. (1990) report strong positive correlations between LMX quality averaged across the first 13 three years of employment and several measures of subsequent speed of promotion and leader perception of member promotability. The assessments were made at various stages across 13 years. The correlation coefficients for speed of promotion range from ;=.34 to ;=.41, and the correlation with the promotability measures range from 1:.28 to ;=.40. Graen et a1. (1990) also report strong positive correlations between the initial averaged LMX quality level and various measures of the employment situation thirteen years after initial hire, such as overall job satisfaction (;=.39), satisfaction with supervisor (;-.27), and appraised performance (;=.26). Perfennance. Graen and Cashman (1975) posited a positive relationship between LMX and performance. The results of this research are mixed. Several studies are consistent with this hypothesis (Dansereau et al., 1975: Graen & Ginsburgh, 1977: Graen et al., 1990: Liden & Graen, 1980), yet other studies find weak or non-significant relationships (Rosse & Kraut, 1983: Vecchio 1982, Vecchio & Gobdel, 1984). The tenuousness of this relationship suggests that other variables likely mediate or moderate the relationship. Task characteristics (Dunegan, Duchon & Uhl-Bien, 1992) have been found to be such a moderator. If the task is unpredictable and has high variability, LMX is positively related to performance, and if the task is mundane and routine, LMX again is positively related to performance. 14 Between these two relative extremes LMX quality apparently has no impact on performance. These findings suggest that when a task is variable and unpredictable, direction and assistance is needed. Moreover, if a task is mundane and boring, socioemotional support may be needed. But, it appears that for subordinates involved with tasks that are neither boring and mundane, nor highly variable and unpredictable, the supervisor is less necessary. Hence, the relationship with the supervisor, the LMX, is of less importance and impact. Qennnnieeting at Week, LMX quality has also been found to predict communicative patterns toward the superior. Specifically, the quality of LMX is related positively to administrative and planning communication (Schiemann, 1977). The research of Kim and Organ (1982) may be seen as indicating that leaders will initiate more noncontractual social exchanges with high quality LMX subordinates than with low quality LMX subordinates. Borchgrevink and Boster (1992) have replicated these relationships. They found strong relationships between two LMX factors and measures of job communication (n=.75 and ;=.86) and non-job communication (r=.47 and ;=.60), respectively. Borchgrevink and Boster (1994) report two other findings of relevance to communicating at work. They found a positive relationship between a subordinate’s communicative responsiveness and LMX quality (n=.23 and n=.49), where communicative responsiveness is a perceptual 15 measure of an individual’s empathy and ability to sex the EiQDL_&DiDQ under most circumstances (Stiff, Dillard, Somera, Kim & Sleight, 1988). In addition, the amount of social support provided subordinates by their superior correlated highly with LMX quality (;=.85 and ;=.78). Job Satisfection. Using the Job Descriptive Index (Smith, Kendall & Hulin, 1969) job satisfaction subscale to measure subordinates’ satisfaction with work, Borchgrevink and Boster (1994) report a strong positive relationship with LMX quality (;=.57 and ;-.61). Scandura and Graen (1984), using the Hoppock (1935) job satisfaction instrument, report strong positive correlations as well (;=.79 and ;=.81). Vecchio, Griffeth and Ham (1986) also used Hoppock (1935) and report a relationship of ;=.41, while Rosse & Kraut (1983) devised their own global job-satisfaction measure which correlated .46 with LMX. Lagace (1990), using Indsales (Churchill, Ford & Walker, 1974), a satisfaction scaled designed for sales managers, found Leader-member Exchange positively related to global job-satisfaction and satisfaction with manager. McClane (1991a) also provides support in reporting positive relationships between LMX and leader, task, and co-worker satisfaction. These findings are important for organizations to the degree that organizations value job satisfaction among their members. For the individual organizational member it may be reasonable to assume that job satisfaction is of prevailing interest. 16 Q;genize§ienel_§lineeee Working from the assumption that organizational climate is an individual level phenomenon impacted by intradyadic leader behavior, Kozlowski and Doherty (1989) investigated the relationship between several climate subscales and LMX quality. They report correlations ranging from ;=.24 to ;=.60. These findings are important in that many argue climate perceptions mediate the relationship between the organizational context and individual responses such as affect, motivation, and behavior at work (Litwin & Stringer, 1968: Schneider, 1983a, 1983b). EEEDQQ£& Occupational tedium or burnout (Freudenberger, 1974) is a pervasive concern in many industries (Miller, Stiff and Ellis, 1988: Krone, Tabacchi & Farber, 1989a, 1989b: Tabacchi Krone 8 Farber, 1990). Borchgrevink (1993) investigated the relationship between Maslach's (1982) three dimensions of burnout and LMX quality. The three dimensions of burnout, i.e., emotional exhaustion, depersonalization, and personal accomplishments, all had strong relationships with two LMX factors (correlations of zs-.60/;=-.66, ;=-.78/;=-.89 and ;=.20/n=.44, respectively). These findings are important in that burnout has negative effects on both employee and organization (Miller et al., 1988: Miller, Zook & Ellis, 1989). Employee consequences include physiological and psychosomatic effects, general decreased job satisfaction and negative l7 behavioral adaptations such as heavy drinking. For the organization the consequences may be lost profits, dissatisfied employees and clients, increased turnover and negative work attitudes (Miller et al., 1988, 1989). peeision Influence. Scandura, Graen and Novak (1986) found that member’s perceptions of having decision influence, as well as leader’s perceptions of members’ decision influence, was positively related to LMX quality (;=.45, and ;=.25). LMX quality was found to interact with the leaders’ rating of member performance so as to indicate that members of high quality LMX dyads perceived high decision influence regardless of performance levels, whereas low quality LMX members perceived themselves as having high decisional influence only when they were rated as high performers. The above review has discussed the relationships between LMX and many variables that may be considered relevant for organizational functioning. Adding to the importance of the reviewed relationships is research that has shown that LMX quality can be improved through training in role making behaviors (Graen, Novak 8 Sommerkamp, 1982: Scandura 8 Graen, 1984). If the relationships reviewed are not totally spurious, the finding that LMX can be changed through training is important because a change in LMX may be related to a change in the variables discussed. Having reviewed the theoretical and empirical literature surrounding LMX, I shall now turn to Social 18 Exchange Theory which provides an explanatory framework for the occurrence and development of LMX. Furthermore, I will review literature that will prove important for developing hypotheses regarding antecedents to LMX. In doing so, I will discuss the Similarity Attraction Paradigm, organizations as authority structures, authoritarianism and, power. Social Exchange Theory The above discussion of LMX and role making indicated that LMX and Social Exchange Theory (Blau, 1964: Homans, 1958: Roloff, 1981: Thibaut 8 Kelley, 1959) are related. As such, it is appropriate to discuss Social Exchange Theory further, making evident how LMX can be subsumed under the rubric of Social Exchange Theory. Social Exchange Theory is a theory that considers the transfer of resources between and among actors under the assumption that the guiding force is the actors’ self interest (Roloff, 1981). Hence, Social Exchange Theory is a utility perspective because the motivation for actors to interact lies in the perceived rewards or costs associated with interacting, or not interacting, with (a) specified individual(s) at a specific point in time. People exchange resources through interaction. A resource is essentially any commodity, physical or ephemeral, which may be transmitted through interpersonal behaviors. Foa and Foa (1974) identify six categories: love, status, services, goods, information, and money. 19 Resource exchanges are typically perceived as rewarding, however, it is important to keep in mind that an exchange also has the potential of being costly. Williamson (1981) argues that there are transaction costs involved when we initiate exchange relationships. Transaction costs are the social systems version of friction in mechanical systems. Will the social exchange partners operate harmoniously, or can misunderstandings, conflicts, and delays be anticipated? Furthermore, when a particular exchange of resources takes place, other resource exchange opportunities may have to be forfeited. That is, we incur opportunity cost. Considering the scarcity of certain resources, when an individual engages in exchange with person A, the individual may no longer be able to maintain an exchange with person B, the normative expectations surrounding romantic relationships being an example. However, opportunity cost is also relevant for the, presumably non-romantic, leader- member relationship. 5 1° l'v E t 1 H!°Ji! An exchange will likely take place if potential interactants perceive that the product of reward value and reward probability exceed the product of cost value and cost probability. For example, if an organizational member anticipates that rejection and acceptance are equally as likely to result from proposing a change in work (role behaviors), and the reward of acceptance is greater than the 20 cost of rejection, the individual is likely to make the proposal. The result of this comparison is referred to as subjective expected utility. When utility assessments are not probability estimates, they are based on sampling outcomes of tentative initial interaction. That is, if an organizational member (leader) has no basis for establishing a priori the relevant success ratio in interacting with a particular supervisor (subordinate), the member (leader) will make assessment of prospective rewards or costs by engaging the supervisor (subordinate) in a small sample of interaction. This process can most clearly be anticipated to take place in newly formed leader-member dyads. W Furthermore, individuals use two potential comparison standards to determine whether to engage in, or maintain, a [social] exchange relationship (Thibaut and Kelley, 1959). These standards are the comparison level (CL), and the comparison level of alternatives (CLA). CL draws from the individual’s, or referent other’s, experiential base, i.e., the expected rewards or costs based on past experience. If the perceived reward of a potential social exchange is at minimum equal to the CL, then the exchange is attractive. CLA refers to the perceived rewards or costs of potential exchange alternatives that will induce a change in exchange (relational) partner. 21 Mills and Clarke (1982) warn against characterizing all relationships as exchange relationships. They point out that relationships in romance, friendship and family tend not to be exchange based, but rather based on a concern for the welfare of the relational partner. They label this relationship as eennnnel. Within communal relationships provision of services or products and general interaction takes place with an eye upon the needs of the relational partner, not on potential and anticipated reciprocation. They have demonstrated that treating a communal relationship as if it were an exchange relationship compromises the relationship and leads to feelings of exploitation and relational deterioration. The distinction between exchange and communal relationships may limit the utility of considering all superior-subordinate relationships as exchange relationships. However, it is important to note that Mills and Clark (1982) suggest that the superior- subordinate relationship (boss-secretary) is a primary example of a long-term exchange relationship, and that it is possible to have both a communal and an exchange relationships with the same individual, keeping the communal and exchange aspects of the relationship distinct. In describing the latter they use the context of a business organization as example. Superior-subordinate relationships are indeed exchange based, but neye in addition, be communal. If the superior-subordinate relationship is both an exchange and a communal relationship, then the communal 22 expectations are kept distinct from the exchange based expectations. The argument and approach of Mills and Clark (1982) does not invalidate considering the superior- subordinate relationship, e.g., the LMX, from the social exchange theory perspective. At this point I will examine the relationship between Social Exchange Theory and LMX. Leader-Member Exchange and Social Exchange Theory Graen and Scandura (1987) posit that all organizational members desire to seek change in their prescribed roles, and attempt to negotiate these changes. It is a reasonable pre- supposition that most organizational members will not be motivated to seek changes that entail perceived negative summary consequences in their role behavior. Working from this pre-supposition, it is reasonable to assume that the subjective expected utility of change drives the desire for change. LMX is posited to develop through interaction, and social exchange theory provides a useful framework for explaining why superiors and subordinates are motivated to interact beyond the minimum requirements of their roles.x Leader-member communication beyond that required by a position is central in explaining how subordinates holding identical positions develop different levels of LMX. Additionally, from the LMX perspective the negotiable domains are agnate to the resources that are central to Social Exchange Theory. Love is the only category of the resources proposed by Foa and Foa (1974) that does not have a clear fit with the negotiable domains (Graen 8 Scandura, 23 1987). It can be, for the most part, argued that love is a resource extraneous to interpersonal relationships within organizations. However, if the resource of love can be more broadly defined as positive affect, it can be argued, as do Dienesch and Liden (1986), that it is relevant for leader- member relations. Several subsequent reports also suggest a relationship between affect and LMX (Salzmann and Grasha, 1991: Schriesheim, Neider, Scandura and Tepper, 1992: Turban, Jones 8 Rozelle, 1990). WWW Most discussions of LMX acknowledge that a leader cannot develop high quality relationships with all members, because of the time constraints inherent in establishing and maintaining a high quality LMX relationship. Ostensibly, leaders cannot interact with all members at levels above the role requirements. The demand on the leader’s time is an opportunity cost. The time constraints force leaders to choose specific members from all their members with whom to establish high quality LMX relationships. This choice may not be a conscious selection by the leaders, and may perhaps entail self-selection by members. Regardless, most often it is impossible for the leaders to establish high quality relationships with all members, and the need for a differentiation among them becomes important. Social Exchange theory supplies a reasonable explanation of how this differentiation occurs. 24 Making tne cnoice Superiors and subordinates consider their dyadic partner(s), make subjective expected utility assessments, and determine whether they engage in the interaction and role negotiation requisite for high quality LMX relationships to develop. The subjective expected utility assessment is tied into their previous experiential base and the currently available dyadic interaction partners, i.e., the leader and the various members of all potential leader- member dyads. Most members typically have significantly less discretion than leaders in choosing dyad partners. However, both leader and member can influence the degree and quality of intradyadic interactions. The previous experiential base consists of all previous superior-subordinate relationships and the rewards or costs associated with said relationships. This experiential base provides the comparison level (CL) for superiors (leaders) and subordinates (members), respectively. The comparison level of alternatives (CLA) is a utility assessment of potential social exchange with the various other potential dyadic partners. Leagen choice. From the leader’s perspective it may be reasonable to envision that they perform an initial assessment of all their members, using their CL as sorting criteria. Choices to engage members in exchanges beyond the minimum requirements of their role are likely based on the relative degree to which they meet or exceed the current CL. 25 The CLA is relevant when many members are at or above the leader’s CL, so that further differentiation is required. The members will be compared to each other by the leader to make choices yielding the highest potential sum utility. The CLA will also be considered when leaders receive "new" organizational members and a resorting is required. Mennez_eneieee From the subordinates’ perspectives, the CLA may be moot. Only if they have a choice of superiors, or multiple superiors will the CLA be relevant in determining social exchange behaviors, i.e., LMX behaviors. Subordinates’ CL will determine the degree to which the subordinates will be motivated to choose to engage their superior beyond the minimum role requirements. At this point it is reasonable to ponder how the utility assessment is determined, beyond being based on the past and current experiential base and trial-and-error as discussed above. Superiors and subordinates are faced with a myriad of cues and behaviors, many of which may be totally irrelevant in terms of role behaviors. The need for a sense making mechanism, a tool, by which to establish the subjective expected utility effectively and efficiently from these cues may arise. The similarity-attraction paradigm provides such mechanisms. The Similarity-Attraction Paradigm Following a rigorous and extensive program of research, Byrne (1971) developed what he called ene_ene;eee19n penegigne The primary finding is a positive relationship 26 between the similarity of two individuals and the attraction among them. The initial relationship investigated was attitudinal similarity, however, the research program also considered similarity in terms of opinions, personality, intelligence, emotional state, behaviors, socioeconomic status, and physical characteristics. In all instances, they replicated the primary finding of a positive relationship between similarity and attraction. However, subsequent research has primarily focussed on the relationship between attitudinal similarity and attraction. Three primary challenges to the similarity-attraction paradigm have been put forth, respectively proposing that the theoretical interpretation of the similarity-attraction relationship is erroneous (Condon 8 Crano, 1988): that the relationship is in fact dissimilarity-repulsion, not similarity-attraction (Rosenbaum, 1986a): and that the similarity-attraction relationship is a myth (Sunnafrank, 1992). The dominant theoretical explanation of the similarity- attraction relationship is that consensual validation acts as a higher order conditioned stimulus. The notion is that interpersonal agreement (attitude similarity) provides consensual validation of the beliefs held by the interactants. The consensual validation is seen as providing (partial) satisfaction of individuals’ drive for logic, order and predictability in environmental and general functioning. The individual who provides the drive 27 satisfaction is thereby the recipient of positive affect. Drawing upon research (Aronson 8 Worchel, 1966: McWhirter 8 Jecker, 1967) indicating that individuals infer a positive evaluation of self by agreeing others, and that individuals tend to like those who like them, Condon and Crane (1988) suggest that this inferred positive evaluation, rather than consensual validation and drive satisfaction, explains the similarity-attraction relationship. Their findings include a strong similarity-attraction relationship (n“=.64) that is significantly reduced when inferred evaluation was held constant (:3,_,=.18) . The inferred evaluation-attraction relationship is also strong (n“=.81), and maintains significant strength when attitude similarity is partialled (rnmg=.66). This result indicates that of the two explanations, inferred evaluation received the most support. Rosenbaum (1986a) challenges the nnlgeneel_eeneeneien (p. 1156) that attitudinal similarity leads to attraction and suggests as alternate that dissimilarity leads to repulsion. Although his explanation has been criticized (Byrne, Clore 8 Smeaton, 1986), he provides much support, and proposes that dissimilarity-repulsion and similarity- attraction may work in a two-stage sequence. In response Rosenbaum (1986b) argues further for his alternate hypothesis, but finds it reasonable that both dissimilarity- repulsion and similarity-attraction may be operative. This proposal appears to receive support by some (McDermott, 1991), while others continue to deny a dissimilarity- 28 repulsion association (Singh 8 Tan, 1992: Smeaton, Byrne 8 Murnen, 1989). Sunnafrank (1991, 1992: Sunnafrank 8 Miller, 1981) is the primary voice arguing to discard the similarity- attraction paradigm altogether. He acknowledges that there is a similarity-attraction relationship, but that it exists only in pre-acquaintance relationships (Sunnafrank 1991), and that a causal relationship between attitude similarity and attraction is most unlikely (Sunnafrank, 1992). Sunnafrank’s research (1991, 1992) shows that if the research subjects are provided opportunity to communicate then any differential attraction among similar and dissimilar interactants is greatly reduced and often eliminated. His review of the literature (Sunnafrank 1991, 1992) provides much additional support for this finding. Bochner (1991) argues that attraction is a likely causal antecedent to similarity. He proposes that attraction leads to a desire for a relationship which leads to interaction. The functional purpose of initial and subsequent communication is seen as ". . . fostering perceptions of attitudinal and personality similarity . . . (p. 487)." Furthermore, Bochner (1991) points out that virtually all attitude similarity-attraction work focusses on non- interpersonal relationships and non—interpersonal attraction (Miller 8 Steinberg, 1975): that it uses a global "attraction as liking" measure far removed from interpersonal bonding: and that the interactional 29 circumstances are atypical in that the subjects are unilaterally informed of the other’s attitudes through written communication prior to measuring attraction to the other under the assumption that they will jointly be working on some atypical task. Bochner (1991) clearly demonstrates that the conditions of attitude similarity-attraction research are far removed from naturally occurring situations, impugning the generalizability of this research. Sunnafrank’s (1991, 1992: Sunnafrank 8 Miller, 1981) approach alleviates the concerns mentioned by Bochner (1991) somewhat by instructing the subjects to talk about their attitudes prior to measuring attraction. However, the interaction time was limited to five minutes and the subjects were at times given direction as to interactional content, somewhat atypical for normal (get-acquainted) interactions. Cappella and Palmer (1990) provide a much more natural situation in that subjects were paired and allowed to interact for thirty minutes without any direction as to communication content. Half of the subjects knew their dyad partner for at least a semester and interacted regularly, while the others were strangers. Those who knew each other were paired based on self-perception of their relative similarity, while the strangers were paired based on their responses to an attitude questionnaire. Only the stranger dyads were pre-appraised about the relative attitudinal similarity of the interactant prior to the study. Furthermore, they measured a broad range of 30 attitudes in establishing relative similarity. Their research findings contradict Sunnafrank’s in that the attitude-similarity relationship was not eliminated after the interaction. Moreover, they found that non-verbal involvement behaviors explained variance in attraction above and beyond attitudinal similarity, and that similarity in non-verbal behaviors explains variance in attraction but eliminates the effect of attitudinal similarity. Cappella and Palmer (1992) make a compelling argument in explaining the contradictory results by referring to the initial and short term nature of the interactions and tendencies to equivocate and avoid face threats (Bavelas, Black, Chovil 8 Mullett, 1990: Brown 8 Levinson, 1987) when interacting, unless impending intrapersonal goals or the social circumstances require conflictual statements. The tendency toward pleasantries may account for the elimination of the attitude similarity attraction effect, whereas the longer interaction time of Cappella and Palmer’s (1990) research made it more difficult for the interactants not to reveal differences, attitudinal or other, in their exchange of verbal and non-verbal communication. Both Sunnafrank (1992) and Bochner (1991) argue that the more closely the research setting mirrors natural conditions, the more other factors come into play in determining interpersonal attraction. Sunnafrank (1992) and Cappella and Palmer (1992) have attempted to rebut each other’s findings based on 31 methodological concerns. Neither of the three challenges to the similarity-attraction paradigm discussed have been firmly settled. However, of the three challenges only the Sunnafrank-Cappella dispute bears directly on this proposal. Condon and Crano’s (1988) alternate theoretical explanation is interesting, but will not change the hypothesized empirical relationships. Neither dissimilarity-repulsion (Rosenbaum, 1986a) as the primary relationship, or the dissimilarity-repulsion and similarity-attraction sequence (Byrne, Clore 8 Smeaton, 1986: Rosenbaum,1986b) are inconsistent with this proposal’s argument from the social exchange theory perspective. Should Sunnafrank’s argument (1991, 1992: Sunnafrank 8 Miller, 1981) prevail, there would seem to be little reason to expect attitudinal similarity to be used as comparison level for establishing the subjective expected utility of exchange. However, it could be argued that after a few minutes of interaction a perception of relative similarity develops, but that as interactions take place over a longer period of time real and perceived differences become apparent and salient. This argument is consistent with the research of Cappella and Palmer (1990). It is also important to note that the challenges were based on ennienginel similarity only, and that other forms of similarity were not discussed. However, there is apparent agreement that other forms of similarity are relevant in terms of interpersonal attraction and evaluation (Byrne, 1971: Byrne et al., 1986: Cappella 8 Palmer, 1992: 32 Condon 8 Crano, 1988: Sunnafrank, 1992: Tsui 8 O’Reilly, 1989). Sinilezity-Attzaction eng ngenizatiene Quite recently, the similarity-attraction paradigm has received some attention in the organizational literature in terms of demographic similarity (Tsui 8 O’Reilly, 1989: Zenger 8 Lawrence, 1989). Demographic variables have long been important in researching intraorganizational behaviors and organizational outcomes (Lawrence, 1988: Morgan, 1986: Pfeffer, 1982: O’Leary 8 Hansen, 1983: Tsui 8 O’Reilly, 1989: Zedeck 8 Cascio, 1984: Zenger 8 Lawrence, 1989: Duchon, Green 8 Taber, 1986). Tsui and O’Reilly (1989) point out that most research considered simple demographic effects and not composite effects or relative dissimilarity among interactants. They argued that this modal approach of studying independent demographic effects does not adequately capture the potential richness and predictive power of demographic variables. To improve the relevance and impact of demographic research in organizations they propose a construct labelled neleeienel_genegxenny, when referring to the similarity-attraction paradigm (Byrne, 1971: Byrne, Clare 8 Smeaton, 1986). The construct, relational demography refers to the degree of demographic similarity between two individuals or groups of individuals. A relational demography score is the squared difference in values on a demographic variable for 33 two individuals or groups‘. 'The larger the score the larger the difference between the two units of analysis. Tsui and O’Reilly (1989) investigated relational demography of superior-subordinate dyads (n=272) in reference to subordinate role ambiguity, superior’s perceptions of subordinate effectiveness, and the superior’s attraction to (liking of) the subordinate. The postulated impact of relational demography on work related perceptions and attitudes was assumed to work through increased interpersonal attraction and increased communicative interaction due to intradyadic demographic similarity. They provide much empirical evidence for the proposed similarity effect. They investigated the impact of relational demography considering age, gender, race, education, job tenure and company tenure. At the global level, they found that increased dissimilarity within the dyad was related to role ambiguity for the subordinate (BE=.32), the effectiveness rating of subordinates by superiors (R_2=.27) , interpersonal 1 The process of squaring the difference between superior and subordinate scores provides an absolute difference score with an exponential curve. However, this also has the potential of distorting the relationship among variables in that a relationship that is originally linear may become curvilinear, non-linear or appear to have no relationship upon squaring the scores. Although it breaks with precedence, this research uses a non-manipulated difference score in order to maintain the inherent character of the relationship. The initial intent was to maintain the tradition of squaring the scores. However, squared difference scores in this instance proved indeed to mask the relationship. 34 attraction on the part of superiors for the subordinates (BE=.16), and subordinate role conflict (EE=.14). They also report multiple specific findings that bolster the relevance of superior-subordinate similarity for this proposal. Subordinates in mixed gender dyads were rated as performing poorer (§=-.38), were liked less by superiors (§=-.28), and experienced greater role ambiguity (n.37) and role conflict (§=.32) than subordinates in same gender dyads. Dissimilarity in race did not lead to a performance rating effect. However, superiors with dissimilar race subordinates had greater affect (fi=.17) for these subordinates than superiors in same race dyads had for their subordinates. In terms of role ambiguity and role conflict they found an apparent counter-intuitive negative relationship (Q=-.22: fi=-.18), which upon closer investigation proved to depend upon the ordinal placement of race within the superior-subordinate hierarchy. Age differences within the superior-subordinate dyad were positively related to role ambiguity only (fi=.10). Educational differences were negatively related to supervisory affect (§=-.16) and, counterintuitively, to subordinate role ambiguity as well (fi:.-.11). Ratings of subordinate performance effectiveness and supervisor liking of subordinates was negatively related to the measure of job tenure difference (es-.19: =-.24) while subordinate role ambiguity had a positive association with job tenure difference (§=.33). 35 Although Zenger and Lawrence (1989) do not use the label, relational demography, they considered demographic variables in much the same fashion as Tsui and O’Reilly (1989). However, they reversed the sign so that their demographic similarity score increases with similarity, rather than decreases. They investigated the relationship of technical communication with various age and tenure measures within and across various organizational groups. Their findings include relationships between: age similarity and intragroup technical communication (;=.43): group tenure and intragroup technical communication (;:.30): organizational tenure and intragroup technical communication (;=-.25): career level and intergroup technical communication (;=.51): organizational tenure and intergroup technical communication (;=.33). Therefore, it appears that superior-subordinate similarity has an impact on a variety of organizationally relevant phenomena. Considering that the LMX is posited to result from role negotiations (Graen 8 Scandura, 1987) these findings relating role ambiguity and role conflict to technical communication have the most direct relevance for this dissertation. Similarity-Attraction and Leader-Member Exchange There is a positive relationship between perceived similarity with another individual and interpersonal attraction to that person (Byrne, 1971: Hollander, 1976: Rogers 8 Bhowmik, 1971: Tsui 8 O’Reilly, 1989: Zahl-Begnum, 36 Koldingnes 8 Moog-Begnum, 1988), and a proposed relationship between interpersonal attraction and LMX development (Dienesch 8 Liden, 1986: Dockery 8 Steiner, 1990: Schriesheim, et al., 1992). The LMX is posited to develop through interaction, and there is a tendency for people to communication mostly with similar others (Johnson, 1993: Rogers 8 Bhowmik, 1971). Further, LMX is a facet of the superior-subordinate interpersonal relationship, and development of the superior-subordinate relationship is impeded and sometimes facilitated by demographic and privilege differences (Johnson, 1993: Tsui 8 O’Reilly, 1989). Finally, the Duchon et a1. (1986) finding of gender and various status measures as predictors of out-group versus in-group status, and Lagace (1990) finding that leaders’ gender is related to LMX quality, reinforce the importance of similarity on LMX. In sum, measures of intradyadic similarity, such as relational demography, appear most likely to be related to LMX quality in a predictable fashion. In fact, Steiner (1988) and McClane (1991b) provide some initial support for this prediction. Steiner (1988) reports that high quality LMX dyads are more similar in intrinsic and extrinsic work values than their low quality counter-parts. McClane (1991b) found that leader-member similarity in need for power predicted LMX Quality, whereas similarity in terms of gender, locus of control, least 37 preferred co-worker score and need for achievement, provided no predictive utility. Subsequently, I shall argue that the psychological constructs of authoritarianism and power are relevant to consider when investigating LMX quality because they provide requisite situational relevance (Rogers 8 Bhowmik, 1971). To do so adequately I need first to establish that organizations are in essence authority structures in which individuals’ bend toward authoritarianism, i.e., the manner in which they accept and execute authority, and their perceived power, may impact their perception and display of behavior within the organization. 0 ' ' s ut t t "The development of organizations is the principal mechanism by which, in a highly differentiated society, it is possible to get tnings gone, to achieve goals beyond the reach of the individual" (Parsons, 1960, p. 40, italics in original). These goals are achieved through a variety of basic social control processes such as socialization, the forming and taking of norms, the attainment and exercise of power, goal setting and goal attainment. All social aggregates display social control processes, however, it is within organizations that the social control mechanisms have taken their most highly developed form, viz., the hierarchical authority structure (Scott, 1992). Despite the authority structure, organizations need not be authoritarian in nature, but need to have established a 38 system to make and implement decisions. This system can range from radically democratic to highly autocratic, as long as the participants in the system accept the system as proper and the decisions as legitimate. Legitimacy of directives is the essence of the organizational authority structure (Katz 8 Kahn, 1978) in that, " . . . the corresponding right to exercise authority . . . is fixed by {enignelly_ee§epliened norms, by enactments, decrees and regulations, in such a manner that the legitimacy of the authority becomes the legality of the general rule . . . " (Weber, 1958/1915, p. 299, emphasis in original). Katz and Kahn (1978) argue that acceptance of the legal rules of a social system are requisite for participation in the social system. This acceptance of authority, i.e., acceptance of power as legitimate, comes in part from recognizing the objective social reality, and in part from the socialization process. Compliance with the authority structure is motivated by fear of the negative consequences of not complying as well as from the gratification stemming from being a leg;eniging member of the organization. There seems to be much room for individual difference in the degree to which persons accept the legal rule and execution of the authority structure. In addition, there are individual differences in the degree to which the anticipated or experienced consequences of relative adherence to the authority structures are seen as rewarding, and thus reinforce or dissuade adherence. Given the 39 similarity-attraction paradigm discussed earlier. The LMX relationship may be impacted by the relative degree of intradyadic congruence in the perceptual valence and acceptance of authority structures. Authoritarianism is the appropriate label for this psychological individual difference variable. Authonitenienisn Altemeyer (1988) offers a reconceptualization of authoritarianism that is psychometrically sounder and conceptually much clearer than earlier work. Therefore, his definition and scale will be used in this dissertation. The relative degree of intradyadic difference in authoritarianism is important for this research. Following Lewin, Lippitt, and White’s (1939: White 8 Lippitt, 1968) seminal research, much leadership theory and research considered the relationship between authoritarian, democratic and laissez-faire leadership behavions and many organizationally important variables (Argyle, Gardner 8 Ciofi, 1958: Cammalleri, Hendrick, Pittmen, Blout 8 Prather, 1973: Day 8 Hamblin, 1964: Hespe 8 Wall, 1976: Ley, 1966: Muringham 8 Leung, 1976: Rudin, 1964: Shaw, 1955: Vroom 8 Mann, 1960). However, authoritarianism as an attitude and the relative disparity of the authoritarian attitude within the leader-member (superior-subordinate) dyad has not received consideration in organizational and leadership research. 40 Antnozitenienism Defined. Altemeyer’s (1988) definition of authoritarianism consists of three factors: Authoritarian Submission, a high degree of submission to the authorities who are perceived to be established and legitimate in the society in which one lives: Authoritarian Aggression, a general aggressiveness, directed against various persons that is perceived to be sanctioned by established authorities: Conventionalism, a high degree of adherence to the social conventions perceived to be endorsed by society and its established authorities. This definition describes authoritarianism as the attitudes individuals have toward authority structures and enacted authority in eeciety en lerge. However, there is no reason to believe that the attitude would not carry over into smaller social structures. An organization is just such a smaller social structure. Reconsidering the definition, exchanging society in which one lives with organization in which one works, the only factor that does not appear immediately relevant to organizational functioning is authoritarian aggression. However, organizations are one of the few places, beyond athletics, law-enforcement agencies, and the military, in which aggression may find an outlet and relative endorsement. For example, there is a range of actions a manager may take in response to (perceived) poor performance, or intraorganizational (interdepartmental) competition for scarce resources, or interorganizational 41 competition in the marketplace. Each of these contexts provide sanctioned outlets for aggressive behaviors. Perhaps this opportunity for sanctioned aggression can partially explain the noted attraction to positions of authority within organizations by individuals with authoritarian personalities (Katz 8 Kahn, 1978). Of interest in this monograph is the notion that relative attitudinal similarity within the leader-member dyad is related to LMX quality. The attitudes of interest refers to a) the appropriateness of submitting to those with legitimate authority within the organization, b) the appropriateness of adhering to the norms established and endorsed by the organization, and c) approving of beneylng eggzeeeiyely toward non-conformers, poor performers, and other threats to the organization. I anticipate that interdyadic differences in authoritarianism are negatively related to LMX quality. The authoritarian attitude is distinct from the degree of relative authority or influence that organizational actors have. This authority or influence is most often referred to as (social) power. The relative degree of power across organizational actors is also likely to impact LMX. Therefore I now turn to a discussion of power and power’s relationship to LMX. Power In considering the organization literature it becomes apparent that power, and power research is not addressed as 42 frequently as one would anticipate for such a pervasive and important topic. Dahl (1957) argued that power as concept is " . . . as ancient and ubiquitous as any that social theory can boast" (p. 201). Pfeffer (1981) argues that power is neglected in the organizational literature for three reasons. First, the concept of power is considered problematic within the social science literature. Second, although power is an important construct within the organizational domain, it is not omnipresent, and other constructs that adhere more to the dominant paradigm of the rational perspective often offer competing explanations for organizational decision making. Finally, power has negative social connotations for many (see also McClelland, 1970). None of the three reasons provided warrant continued neglect. Combined with the noted pervasiveness of the power construct, the apparent neglect adds to the relevance of this research. In this monograph, power is defined as Emerson (1962) did, acknowledging that it is not an attribute of an individual, but a property of relationships: "Thus, it would appear that the power to control or influence the other resides in control over the things he values, which may range from oil resources to ego-support, depending upon the relation in question. In short, ow side innlicitly in ene othen’e genengency" [emphasis in original] (p. 32). 43 Pow e s Soc' De e t Pow The most predominantly used conceptual scheme for understanding power is the concept of the five socially dependent dimensions of power offered by French and Raven (1959), who argue power is ". . . potential influence-or, conversely, influence is kineeie power" (Raven, 1965, p. 371, emphasis in original). The five dimensions are coercive power, reward power, expert power, referent power, and legitimate power. They also identified a sixth dimension, informational influence, which is seen as socially independent in that the informational content, not the communicator, is seen as inducing the relevant behavior. I will be focussing on the five socially dependent dimensions of power, which take the form of public dependent or private dependent influence. Englie Dependent, In some instances it is necessary for the power holder, or influence agent, to be able to observe the social actor perform the desired behaviors in order for the power dimension to engage the social actor in question. That is, the social actor’s motivation to perform the desired behavior is assumed positively related to the power holder’s opportunity to observe the behavior take place, or opportunity to observe and identify actor-specific outcomes. Coercion and reward are the two sources of this public dependent influence. Coercive power comes from a social actor’s perception that an interactant has the 44 ability to administer some form of punishment to the social actor if the social actor does not behave as desired. Reward power lies within the social actor’s perception that the interactant has the ability to provide something of value in exchange for performing the desired behaviors. EnivaEe erendent. When the social actors in question are influenced to perform by an interactant, but not concerned about the interactant’s ability to observe their behaviors or behavioral outcomes, the influence is considered to be private-dependent. This form of influence is seen as being manifested in three ways: expertness, reference, and legitimacy. Expert power exists when social actors perceive an interactant as having knowledge, skill or ability that is superior to theirs in relation to some task or issue at hand. As such, they will be motivated to perform in a manner consistent with what they perceive the interactant as desiring, under the assumption it will lead to the desired outcome. Individuals may be seen as having referent power when others identify with them, and adopt their point of view or behaviors in order to emulate them. The converse type of referent power is invoked when persons disassociate themselves from particular others, taking care not to adopt their behaviors or characteristics. Legitimate power refers to the perceived appropriateness of influence or, the rign; of one to 45 prescribe another’s behaviors. This form of power is based on one’s acceptance of the relationship with another, and acknowledgment that it is a normative requirement to behave as desired (see also Dornbusch and Scott, 1975). Normative Bowen Dornbusch and Scott (1975: Scott, 1992) use a norm based approach to identifying power, and label power as authorized and endorsed, respectively. " . . . A’s power over B is authorized to the extent that beliefs held by groups superior to A legitimate A’s power over B: . . . A’s power over B is endorsed to the extent that beliefs held by B’s colleagues who are also subject to A’s power legitimate A’s control over B" (p. 41). Authorized power is seen as leading to greater vertical reach, i.e., the leader who has authorized power is likely to have greater access to, and communicate more frequently with, superiors. This access is what Pelz (1952) described, and has since been labelled the Eele_§fjeee. Jablin (1987) reports the Pelz Effect is related to LMX development through role individualization success. Successful role ’>< individualization is conceptually identical with successful role making in Graen and Scandura’s (1987) terminology. Endorsed power by definition is power given to an individual by subordinates. It may be that the power is given to the superior in exchange for increased negotiating latitude for the subordinate regarding issues at work. From Coleman’s (1975) perspective subordinates have an Abelian 46 interest relaEion in outcomes the superior can provide (e.g., negotiating latitude) for which they enter a control exchange, i.e., the subordinate endorses the superior’s power to obtain the outcome of interest. This concept is identical with the notion of entering into a social exchange based on subjective expected utility discussed previously. In dyads with high quality LMX the superior provides the subordinate with increased negotiating latitude. However, it may be reasonable to expect that the relationship is reciprocally determinant (Sims 8 Manz, 1984). That is, although high quality LMX may be a result of endorsing the superior’s power, the degree of endorsed power the superior has may also be a result of superior- subordinate interactions that could be characterized as high quality from the LMX perspective. This argument is a plausible explanation for how power is a route to more power. Mulder’s (1976) Power Distance Reduction Theory provides additional support for a positive relationship between power and LMX quality, and is relevant for the discussion of power as antecedent to LMX development. Power Distance Reduction Theory It was partly in response to noticing that many participative management systems failed that Mulder (1976, 1977) developed Power Distance reduction Theory. He found that introducing participative systems in organizations paradoxically led to increases in power distances between the powerful and the least powerful, rather than reduction 47 in power differential as anticipated (Mulder, 1971, 1977). The Power Distance Reduction Theory is based upon fifteen hypotheses for which Mulder (1977) provides substantial empirical support. The entire theory and set of hypotheses will not be reviewed here. The discussion will be limited primarily to the components of Power Distance Reduction Theory that appear to have relevance for LMX development. E0 231' !' The primary hypothesis and premise of Power Distance Reduction Theory is that the mere exercise of power provides gratification. However, in contrast to notions of the economically rational man, and the predictions of need deprivation and gratification theories (Maslow, 1943, 1954: Herzberg, Mausner 8 Snyderman, 1959), the exercise of power does not satisfy. On the contrary, power is seen as addictive, and the wielding of power is seen as leading to a desire for more power. The theory acknowledges that individuals try to increase their power, but argues that the desire for increased power is positively related to the amount of power an individual has at any point in time. As such, the pOwerless are seen as having a minimal drive toward obtaining power, and should exhibit a minimal amount of behavior aimed at reducing the power distance. fl . ! . 'n E E' ! On the other hand, those who wield substantial power will strive to keep, or increase, the power distance to those less powerful. In fact, contrary to the predictions 48 of Schelderup-Ebbe’s (1922) pecking order, Power Distance Reduction Theory argues that "The greater the distance from the less powerful person, the stronger the striving of the powerful to increase it (Mulder, 1977, p. 4)." Furthermore when the power distance becomes small, the more powerful " . . . give in to reality and accept the intruders in their power realm (Mulder, 1976, p. 81)." In other words, the greater the distance, the more power maintaining or power increasing behavior should be displayed by the powerful toward the relative powerless. In terms of those individuals close to the more powerful it would be reasonable to expect that they would display aggressive behavior toward the more powerful in their addictive strive for power. However, research has shown that they will fiercely compete with referent others, and show extreme consideration and pandering to the more powerful rather than fight them. (Mulder 8 Stemerding, 1963: Mulder, 1977). Leader-Member Exchange and Power Distance Reduction Theory Considering the descriptors used for high quality versus low quality LMX dyads, it appears that high quality dyads reflect more power-balanced relationships and low quality LMX dyads reflect less balance. Moreover, there are data consistent with this assertion. In an investigation of LMX between one leader and three of his members Fairhurst and Chandler (1989) found that use of power in language 49 varied across three LMX levels. Furthermore, Fairhurst, Rogers, and Sarr (1987) found that manager dominance was negatively related to LMX (fi=-.43). s ' v o m The issue is how this power difference is related to LMX development. From the perspective of Power Distance Reduction Theory the following argument is plausible: When a new superior-subordinate dyad is created, each dyad member has perceptions of their respective relative power. The dyad members display their power, and power distance reduction tendencies according to this perception. Therefore, subordinates that perceive themselves as relatively powerless will exhibit virtually no power displays or power distance reduction tendencies in their interactions with the superior. The more powerful superior will, according to Power Distance Reduction Theory, struggle to maintain and possibly increase the power distance, and should exhibit a great deal of powerful behavior toward the low-power subordinates. On the other hand, if superior-subordinate dyads contain subordinates who perceive themselves as relatively powerful, they should employ more powerful displays when interacting with the superior. The subordinates perception of power may be imaginary or real. Nevertheless, according to Power Distance Reduction Theory the superior should not contest this power display as readily, but perhaps accept the subordinates into the superior’s power realm. 50 Given the above scenario, over time, a dyad with an initial large power distance should self-perpetuate or increase this distance. A dyad with a small(er) power distance should decrease the distance and eventually have equal levels of power. It is expected that this negotiation of the power relationship takes place through communication, and it is well documented that power is evidenced in the language choices or speech patterns of social actors (Berger 8 Bradac, 1982: Fairhurst 8 Chandler, 1989: Sherer 8 Giles, 1979: Fairhurst et al., 1987: Donohue, Rogan, Ramesh 8 Borchgrevink, 1990). Thus, I expect that the interaction that takes place within dyads with high quality and low quality LMX should reflect the proposed power difference, and that the power relationship should become more evident over time. This hypothesis is consistent with the addiction hypotheses of Power Distance Reduction Theory. Summary Argument and Hypotheses LMX is one aspect of the interpersonal relationship of superior-subordinate dyads. From the LMX perspective leadership is seen as a dyadic-level phenomenon in that leaders behave differently toward their various subordinates. The behavioral differentiation across dyads may be quantitative or qualitative. The LMX is expected to develop through interpersonal interaction which takes the form of role negotiation between leader and member.° In most instances leaders do not have the time, opportunity, or perhaps desire to interact at such a level with all 51 subordinates as to allow high quality LMX to develop with all of them. Social exchange theory provides the theoretical foundation for the LMX. According to social exchange theory individuals choose to interact and engage in exchanges based on the subjective expected utility (SEU) of those interactions/exchanges. Individuals use their comparison levels (CL) and comparison levels for alternatives (CLA) to establish the SEU needed to differentiate the potential interaction or exchange partners, and choose those with whom to interact. The variety and profusion of cues available to potential interactants upon which to establish Cls and CLAs makes it impossible to attend to every cue. The potential interactants need to reduce the cues to a comprehensible set in order to be able to make sense of them. In order to reduce the data, interactants focus on the potential exchange partners’ behavioral, psychological, socioeconomic, and physical cues. The similarity attraction paradigm suggests that similarity between two potential exchange partners will predict attraction among them. Similarity and resulting attraction provide cues that may be used to estimate CL and CLA. Consistent with this reasoning, communication scholars suggest that interpersonal similarity is related to interpersonal communication. The relevance of demographic similarity to interpersonal relationships within organizations has already been established tentatively by 52 others. In addition, socioeconomic, cultural, and ethnic cues may be used in establishing perceptions of similarity. Moreover, psychological similarity may also be relevant considering that the majority of the empirical support for the similarity-attraction paradigm was attitudinal in nature. Given that organizations are authority structures, and that relative authority is often displayed and experienced within the superior-subordinate relationships, authoritarianism is a relevant psychological construct to consider. Authoritarianism taps the degree to which individuals believe it is appropriate to submit and conform to the rule of legitimate authorities and the prevailing norms of the social structure (i.e., the organization), as well as the appropriateness of aggressing towards non- conformers. Similarity in authoritarianism between supervisors (leaders) and subordinates (members) should be related to attraction as predicted by the similarity- attraction paradigm. Furthermore, differences in authoritarianism may lead to differential expectations and divergence in CL and CLA. Demographic, socioeconomic, and attitudinal similarity are assumed related to the LMX. More precisely, similarity is seen to lead to interpersonal attraction which in turn is used as a cue in establishing the relevant CL and CLA to determine the subjective expected utility of interacting with the available LMX dyad partners. 53 Power is also relevant. The aspect of power considered important is the relative power distance within leader- member dyads. The earlier discussion pointed out how the description of LMX in theoretical and empirical terms indicated that LMX is negatively related to intradyadic power differences. Power Distance Reduction Theory provides an explanation for why and how this result may occur. Power Distance Reduction Theory suggests that leaders and members who perceive themselves as having substantial differential power will maintain or increase the power differential, whereas leaders and members that perceive themselves as having very similar amounts of power will minimize or eliminate their power differential. Perceived power differences within leader-member dyads are anticipated negatively related to LMX quality. If the two dyad members perceive each other as similar, they are likely to be attracted to each other, interact, and develop high quality LMX through the role negotiation process. If both dyad members perceive the other as dissimilar, they are not likely to be attracted to each other, and are not likely to interact beyond their minimum role requirements. Thus, they will not develop high quality LMX. If there is disparity in the perception of relative intradyadic similarity, one dyad member is likely to be attracted to and attempt to interact beyond the minimum role requirements with the dyad partner, who may not be attracted and thus not reciprocating that interest. To the extent 54 that either dyad member makes the relative similarity- dissimilarity salient to the dyad partner they are likely to move to perceptual congruence. Even if interaction is limited to the minimum role requirement, i.e., contractual social exchange, the dyad members are likely to move towards perceptual congruence. This argument is consistent with Bochner’s (1991) position that one of the main functions of communication in early stages of a relationship is to establish intradyadic similarity. Figure 1 describes the proposed causal structure expected among these variables. A study designed to test this model is presented in the next chapter. Leader- Member ; Interaction Similarity Leader- Member Exchange Perceived ”’//,,,,,,,,,///4> Leader-Member Power Difference Figure 1 Proposed causal model for LMX. Chapter 2 METHOD oc e Potential companies were introduced to the research request by a letter addressed to a company executive identified as an alumnus of the Michigan State University School of Hotel, Restaurant and Institutional Management. The initial letter explained the general nature and importance of the research and contained a copy of the questionnaire. The letter also explained the nature of the data the companies would be provided with if they chose to participate. Within two weeks a follow-up call was made to inquire further about access, to answer any questions, and to explain in greater detail the degree of involvement required by the company were they to take part in the study. Participating companies were provided surveys and response envelopes. Participating companies were asked to prepare a numbered list of supervisors which respondents would use to identify their supervisor and themselves if they worked in a supervisory capacity. A label on the response envelope stated that the list of supervisors was not to be included with the completed questionnaire. The response envelopes were labelled so that respondents could mail the 55 56 questionnaire directly to me if they chose to. However, all the companies collected sealed questionnaires and mailed them to me in bulk. The cover letter each respondent received with the questionnaire stated that they were guaranteed confidentiality, that the questionnaire would ask questions about them and their supervisor, but that they would not be identified in the results, and that the results would only be available in summary form. $31321; Access to the employees of 19 hospitality companies was obtained. There are six hotels, two restaurants, seven foodservice companies, three food service suppliers, and one hospitality education/training company in the sample. The companies are in California, Colorado, Illinois, Michigan, Ohio, and Texas. A request to canvas each organization was made in order to tap as many hierarchical levels as possible. However, actual level of access varied across organizations. Ten of the organizations agreed to canvassing and nine provided access to executive level only. The organizations in this study will be provided aggregate level information regarding their superior-subordinate relationships in terms of LMX. They will also be provided aggregate level information regarding the employees’ perceived relative power and superior-subordinate power differences. The respondents in the participating organizations have not been provided an incentive to respond. Upon access, follow-up calls have been placed/made 57 with each organization approximately two weeks and four weeks after issuing the questionnaires to maximize participation. A total of 965 surveys were sent to the 20 participating companies, from which 353 usable responses were obtained. This provides a response rate of 36.6%. Of the 353 responses 15 did not identify their racial/ethnic category. Thirteen are American Indian or Alaskan Native, four are Asian or Pacific Islanders, 16 are African Americans, 293 are Caucasian and 15 are Hispanic. In terms of gender 176 are female and 168 are male. Nine did not answer the gender question. The mean number of years of schooling is 14.8 with standard deviation 2.72 and minimum schooling identified as 2 years and maximum schooling identified as 23 years (n=330). Tenure with the present employer obtained a 5.8 year mean with 5.7 year standard deviation. The minimum number of years is one and the maximum is 32 years (n=338). Tenure in the respondent’s present position has a 4.1 year mean with a 4.1 standard deviation. The reported range is 1 year to 27 years. Supervisor capacity is claimed by 148 respondents while 195 indicate that they do not serve in a supervisor capacity. Ten respondents declined to answer this question. Of the 148 respondents claiming supervisory status, 99 provided the 58 6 «w 66 6 6H («NH H6 Hume” 6 6H 6H 6 6H 6 6H HmHHdozm o 6 6 H 6 6 6 moH>uomooom 6 6 6 H H 6 6 moH>ummooom 6 m 6 H H 6 6 umHHddsm 6 6 HH 6 6 6 HH Hope: 6 6 HH 6 N 6 HH HmcoHumoscm 6H RH 66 6 6H H6 66 moH>uomooom o 6 6 H m e 6 Hope: H6 6 66 H 66 6H 66 moH>uomooo6 6 6 6H 6 HH 6 6H HmHHdoam H H6 H6 H 6H NH 66 moH>ummooom 6 H 6 o 6 H 6 H666: 6n 6 66 6 66 66 66 accusmummm 6H 6 6H 6 HH 6H «6 H666: NH 6 6 6 6 6 6H moH>ummooom 6 6 a H 6 6 6 moH>ummooom 6H m 6H m 6 6H 6H 6666566666 6H 6H 66 H 6H 6H 56 Hope: 5 2 HO m cowucnuccouo an sumo oHsncHooaoa H oana 59 requested identification, while 202 subordinates provided their supervisors’ identification number. The mean number of subordinates per supervisor is 3.6, with a range of 1 to 11. Table 1 contains sample statistics per organization. There is some demographic bias in terms of supervisory capacity, as two Asian or Pacific Islanders and 134 Caucasians indicated they were supervisors. None of the other racial/ethnic categories indicated that they serve in a supervisory capacity. Furthermore, 33 are female while 103 are male. Supervisors have 16.3 years mean schooling with a minimum of 10 years and a maximum of 22 years. Average tenure with the current employer is 7.5 years with a range from one to 21 years and a standard deviation of 6.2 years. Finally, the position specific tenure mean is 4.9 years, with standard deviation 4.4 years and a 1 to 20 year range. The noted bias is consistent with population parameters. We]: This study uses demographic measures consistent with those Tsui and O’Reilly (1989) report having used. However, race and ethnicity are collected using categories consistent with those specified in the standards for federal statistics (U.S. Department of Commerce, 1978). Job tenure and company tenure is measured in years. Consistent with Byrne (1971), socioeconomic status is estimated based on reports of discretionary spending. The subjects are asked to indicate their monthly entertainment 60 and clothing expenditures, as well as the amount of money they spend each month for food, housing, and housing related expenses. The "1986 Right-Wing Authoritarianism” scale as reported by Altemeyer (1988) is used to assess authoritarianism. This scale has thirty items that consistently provides alpha reliabilities above .80. The most widely used power typology is that of French and Raven (1959). This typology is also used in Power Distance Reduction Theory (Mulder, 1977). The first conceptually consistent and psychometrically sound measure of French and Raven’s (1959) social power typology was a 20- item scale (each item is measured on a five point Likert type scale) developed by Hinken and Schriesheim (1989). This scale is written for subordinate perception of supervisor’s power. These items are employed as are a set of 20 similar items designed to tap a subordinate’s self- perception of power. Currently two LMX measurement models exist for which validation has been performed. Borchgrevink and Boster (1994) offer a two factor second order unidimensional model based on Graen’s LMX-14 scale (personal communication, September 20, 1990). One factor is a relational_e§timate tapping negotiating latitude, mutual trust, mutual support, mutual understanding and work relationship effectiveness. The other factor is an estimate of in;ggennne§§. Schriesheim et a1. (1992), working from the Dienesch and 61 Liden (1986) perspective offer a LMX-6 scale with 3 two-item factors. The three factors offered by Schriesheim et a1. (1992) are penceived CQDEIIQBEIOD to Ehe excnange, loyelty and nnnnel_efifieene Both scales are used in this research as there appears to be some conceptual distinction between the two measures of LMX. . The four item Sneeke_nien_§npezyiee: scale used by Borchgrevink and Boster (1994) will be extended and used in this study as a measure of intradyadic communication. They found the scale to be internally and externally consistent and to provide alpha reliability .72 (n=62). leen Respense Scelee, The leader-member exchange, power, authoritarianism and communication scales are all 5- item Likert nype scales ranging from strongly agree to strongly disagree (Likert, 1932). The socioeconomic scale is also a five item scale ranging from very much more than average to very much less than average. Race/ethnic category and gender are both nominal scales, while schooling and tenure are assessed using a single item unbound ratio level request. Chapter 3 RESULTS Ih§_M§§§BI§m§n£_MQQ§l Initially, confirmatory factor analyses (CFA) were performed to assess the content validity of the measuring instrument. The final solution consisted of fifteen factors. Internal consistency was achieved with zero Spearman product rule deviations and 11 deviations from flatness with 22 deviations expected by chance alone (p<.01). The test for external consistency produced 37 product rule deviations and 35 deviations from flatness with 64 deviations expected by chance alone (p<.05). The factors, factor items, and factor item statistics can be found in Table 2. The LMX items produced three clusters which were second order unidimensional and combined as one cluster for further analysis. The LMX cluster consisted of six items from Borchgrevink and Boster (1994) and four items from Schriesheim, Neider, Scandura, and Tepper (1992). The reliability of this scale was estimated by coefficient 3 to be .91. The index formed from summing these items was distributed approximately normally with a mean of 38.39 and a standard deviation of 7.06. 62 63 The intradyadic communication items produced two clusters. The factors are second order unidimensional, and were combined in the remaining data analysis. The distribution of the index formed by summing the items closely approximated the normal distribution with a mean of 22.76 and a standard deviation of 4.23. Coefficient g was employed to assess the reliability of the measure,and was found to be .81. The power items clustered into five factors consistent with Hinken and Schriesheim’s (1989) analysis and proposed factor structure. The factors are supervisor reward power, supervisor coercive power, supervisor legitimate power, supervisor expert power, and supervisor referent power. The distribution of the index formed by summing the supervisor reward power items closely approximated the normal distribution with a mean of 11.56 and a standard deviation of 2.83. Coefficient g was employed to assess reliability and found to be .85. The index formed by summing the supervisor coercive power items obtained an approximated normal distribution with a mean of 10.49 and a standard deviation of 3.48. The reliability was assessed by employing g; which was found to be .91. The distribution of‘ the index formed by summing the supervisor legitimate power items closely approximated the normal distribution with a Table 2 Leader-Member Exchange 64 Final Factor Solutions ILQE I have enough confidence in my supervisor that I would stand up for my supervisor’s decisions if my supervisor was not present to do so. My working relationship with my supervisor is better than average. My working relationship with my supervisor is very effective. I feel close to my supervisor. I am very satisfied with the way my supervisor and I understand each other. If my supervisor had to divide the workers into two groups, with one being the most preferred, I would be a member of the most preferred group. My supervisor recognizes my potential. The way my supervisor sees me he/she would probably say that my ability to do my job is exceptional. My supervisor would probably say that my work goals and his/hers are the same. I feel that my work goals and those of my supervisor are the same. SD .87 .99 .91 .88 .90 .94 Table 2 (cont’d) Intradyadic Communication 65 Item, I speak often with my supervisor about job related issues. I speak often with my supervisor about issues not related to work. My supervisor only speaks with me when he/she tells me what to do (reverse coded). I often share my good ideas with my supervisor. My supervisor and I speak with each other about job operations. My supervisor and I speak with each other about management issues. Supervisor Reward Power “QED 3.90 SD .94 .85 .79 ILQE My supervisor can increase my pay level. My supervisor can influence my getting a pay raise. My supervisor can influence my getting a promotion. SD 1.17 .98 Table 2 (cont’d) Supervisor Coercive Power 66 Iten My supervisor can make my work difficult for me. My supervisor can make things unpleasant here. My supervisor can make being at work distasteful. Supervisor Legitimate Power Item My supervisor can make me feel I should satisfy my job requirements. My supervisor can give me the feeling that I have responsibilities to fulfill. My supervisor can make me recognize that I have tasks to accomplish Supervisor Expert Power SD .87 .8O .73 ILEE My supervisor can give me good technical suggestions. My supervisor can share with me his/her considerable experience and/or training. My supervisor can provide me with needed technical knowledge. SD .89 .87 .93 Table 2 (cont’d) Supervisor Referent Power 67 Item My supervisor can make me feel valued. My supervisor can make me feel like he/she approves of me. My supervisor can make me feel personally accepted. Subordinate Reward Power lMean 4.01 SD .86 .87 .90 ILEE I can increase pay levels at work. I can influence people at work getting a pay raise. I can influence people at work getting a promotion. Subordinate Coercive Power Mean 3.02 SD 1.22 Item I can make work difficult for people at work. I can make things unpleasant for people at work. I can make being at work distasteful for people at work. Mean 3.31 SD 1.20 68 Table 2 (cont’d) Subordinate Legitimate Power lten Mean I can make people at work feel like they should 3.68 satisfy their job requirements. I give people at work the feeling that they 3.82 have responsibilities to fulfill. I can make people at work recognize that they have 3.90 tasks to accomplish. Subordinate Expert Power SD .95 .82 .81 Item, Mean I can give people at work 3.98 good technical suggestions. I can share my considerable experience and/or training 4.13 with people at work. I can provide people at work with needed technical 3.99 knowledge. Subordinate Referent Power SD .73 .77 .80 Item, Mean I can make people at 4.22 work feel valued. I can make people at work feel like I approve 4.24 of them. I can make people at work feel personally 4.13 accepted. SD .71 .72 .74 Table 2 (cont’d) Authoritarianism 69 IEQE It is important to fully protect the rights of radicals and deviants (reverse coded). "Free Speech" means that people should be allowed to make speeches and write books urging the overthrow of the government (reverse coded). There is absolutely nothing wrong with nudist camps (reverse coded). It is best to treat dissenters with leniency and an open mind, since new ideas are the lifeblood of progressive change (reverse coded). High school and university students must be encouraged to challenge their parents’ way, confront established authorities and in general criticize the customs and traditions of our society (reverse coded). The way things are going in this country, it’s going to take a lot of "strong medicine" to straighten out the troublemakers, criminals and perverts. Some of the worst people in our country nowadays are those who do not respect our flag, our leaders, and the normal way things are supposed to be done. Man 2.91 70 Table 2 (cont’d) One reason we have so many troublemakers in our society nowadays is that parents and other authorities have 2.51 1.16 forgotten that good old- fashioned physical punishment is still one of the best ways to make people behave properly. Our country will be great if we honor the ways of our forefathers, do what the authorities tell us to do 2.60 1.05 and get rid of the "rotten apples" who are ruining everything. Socioeconomic Factors Socioeconomic-Necessities Item Mean so The average American spends about $150 per month on 3.07 1.14 foods and groceries for home consumption. I spend: The average American spends about $315 per month on 3.83 1.20 housing and housing related expenses. I spend: Socioeconomic-Entertainment Item Mean SD The average American spends about $50 per month on 3.23 1.18 entertainment. I spend: The average American spends about $80 per month on 2.55 1.09 clothing and shoes. I spend: The average American spends about $70 per month on 2.85 1.24 food in restaurants. I spend: 71 mean of 12.10 and a standard deviation of 2.11. Coefficient e was employed to assess the reliability of the measure, and was found to be .85. The supervisor expert power items were also summed to form an index. The distribution of the index closely approximated the normal curve with mean of 11.54 and a standard deviation of 2.42. To assess reliability of the measure, coefficient e was employed and found to be .87. The distribution of the index formed by summing the supervisor referent power items closely approximated the normal curve with a mean of 12.01 and a standard deviation of 2.42. Coefficient g was employed to assess the reliability of the measure, and was found to be .92. Identical to the supervisor power items, the subordinate power items clustered into the five Hinken and Schriesheim’s (1989) factors. The distribution of the index formed by summing the subordinate reward power items closely approximated the normal distribution with a mean of 9.06 and a standard deviation of 3.01. Coefficient g was employed to assess the reliability of the measure, and was found to be .80. The subordinate coercive power items were also summed, and the distribution of this index closely approximated the normal distribution with a mean score of 9.52 and a standard deviation of 3.43. Coefficient g was employed to assess the reliability of the measure, and was found to be .93. The distribution of the index formed by summing the subordinate legitimate power items closely approximated the normal distribution and obtained a mean of 11.40 and a standard 72 deviation of 2.31. Reliability of the measure was assessed by employing coefficient g, which was found to be .88. The subordinate expert power items were also summed to form an index. The distribution of this index closely approximated the normal distribution with a mean of 12.09 and standard deviation of 2.05. Coefficient e was employed to assess the reliability of the measure, and was found to be .87. The final power index was formed by summing the subordinate referent power items. The distribution of the index closely approximated the normal distribution with a mean of 12.60 and standard deviation of 1.96. Reliability of the measure was assessed by employing coefficient g, which was found to be .89. Nine of Altemeyer’s (1988) authoritarianism items clustered together in a final solution. The distribution of the index formed by summing the authoritarianism items closely approximated the normal distribution with a mean of 25.87 and a standard deviation of 5.64. Coefficient g was employed to assess the reliability of the measure, and was found to be .80. The five items selected from Byrne (1971) for use as socioeconomic indicators clustered into two factors. The factors are labelled §2212e22n2m122Ne2e§s1tx and S22igecenemiszzntertainment to reflect the nature of the clustered items. The distribution of the index formed by summing the socioeconomic-entertainment items closely approximates the normal distribution with a mean of 8.62 and 73 a standard deviation of 2.92. Coefficient g was employed to assess the reliability of the scale, and was found to be .54. The index obtained by summing the socioeconomic- necessity items is also a close approximation of the normal distribution with a mean score of 6.90 and a standard deviation of 1.96. Coefficient g was employed to assess the reliability of the measure, and was found to be .57. Difference scores were created for all variables. The power difference score was created by subtracting the relevant subordinate scale scores from supervisor scale scores. Calculating difference scores for the other variables depended upon being able to identify the relevant superior-subordinate relationship, which often resulted in substantial amounts of missing data. In each instance the subordinate’s score was subtracted from the supervisor’s score. The racial/ethnic and gender measures were coded as 0 if the superior-subordinate dyad was not the same and 1 if they were the same. The difference scores for authoritarianism, gender, company tenure, and position tenure had no relationship with LMX or intradyadic communication and were discarded from further analysis. The distribution of difference scores closely approximated the normal distribution. Descriptive statistics for the difference variables can be found in Table 3, and a 74 unamoOE amuw onch « >DHHMAflflHmIAMMmmm 66H 8 66. 66. 66H « 6H.m 66.Hu moocououuwo H6206umoocm 66H 66. ~H.~ 66.: moocououuwa >uHmmoOGZIOHEocooooHoom 66H 66. 66.6 66.: moocououuwo ucoschuHousuuoHaocooooHoom H66 66. 66.6 oo.H moocououuwo Hoaom o>HouooO H66 66. H6.~ 66. moosououuwa Hozom ovoEHuHuoH 666 66. 66.6 66.: mmocwumuuHo bosom 666666 666 66. 66.6 06.: mmocoumuuao uo3om ucwuwuom «on H6. oH.n 66. moocououuHa H6306 pucsom z o ucwwmduuwoo om 66w: mummmma moHnowuc> mocououuHa you uOHumHumum o>fiuawuomma n manna 75 correlation matrix of the measures of LMX, intradyadic communication, and the difference variables can be found in Table 4. Theoretical Models The similarity hypothesis is that increased intradyadic similarity will lead to increased intradyadic attraction. Subsequently, increased attraction will lead to increased communication within the dyad, allowing for LMX development. Thus, similarity affects attraction which affects communication which affects LMX. To test this hypothesis, both the measurement of LMX and intradyadic communication were individually regressed on the measures of similarity. Intradyadic communication was also regressed on LMX. With LMX as the dependent variable, differences in socioeconomic-necessities and intradyadic communication were strong predictors with B=.14 (t=2.20: df=127: p=.03) and e=.68 (e=1o.93; gf=127; p<.0001), respectively, producing a multiple correlation of .71 (£=64.15: df=2,127: p<.0001). With intradyadic communication as the dependent measure, the significant predictors were LMX (§=.69: t=11.06: §f=126: p<.0001) and Race (§=.14: t=2.30: p=.02) which generated R=.71 (2:64.34: g:=2, 126: p<.0001). Subsequently, both sets of regression analysis were run using the correlation matrix corrected for measurement error (N=119)- With LMX as the dependent measure the multiple 76 .LOLLfi Uikafllg OH 3 Suuguu. LOW RUDOLLOU G. 0‘2.th LED —oo.vnccc «pa-vac. .mo.vauo -. cccpn. No. we. -- ccc0n. no. he. -- mp. mm. mm. NN. NN. pp.- No.- 0~.- m~.- n~.- 00.. 06.. OO. o0. 0N. MN.. 06.. oo- 0~.- 0o.- 0M. mp.- No. 0m. .—p .o— .0 up. ccc0N. coc0N. no. MO.- MP. NP. 0’.- op. Np. .0 asp. cop. op.- so. pp.- NF.- «- up.- on.- ON.- .N ~o-- con~.. comm.- no.- or.- o.- n0.- 0'.- NN. mo. .0 o-.- oo- no.- chN. 00.. amp. N—. no. c0p.- n-. 0—. ccaNN. .. amp. ¢~.. .- 00. pm. 00. 0—. «a.- 0M. .m .0 app": 6:.- ..6—3 ~—- 60830663 .5306 26:06 .2 no.- no.- mo. 635333 .5306 32.3.63 .3 co. ccc0N. Cocoa. «OOCOLOVVVO Luzon «605060: .0 ca—.- no. pp. «ounceoee_o Luzon ueoaxw .6 3.- 2.3..- .6... 6856623 6266 629.08 .5 6—.- co~. no. >u.t-._E.6 .6.oo¢ .o no. mo. .3.- 60958023 1:63.636 .6. cac¢~.- op. ccn~. noocotoee_o no.».mnoooz-omsocoooo_oow .6 - .8.- uof 635.333 HEHoutoucm-Escooofioow .n 66.- - 28:. 8366.55.66 36266.55 .~ 3.- n6. - our-:65 tacos-tutu».— ._. .n .~ .— moHanH6> mucououuwo on» can .cofiucoflcoeeoo oficcaocnucH .xzu mo mCOHuoaouuoO 6 «Hana 77 correlation was .87, and the beta weights for intradyadic communication and differences in socioeconomic-necessities improved to .79 and .25 respectively. With intradyadic communication as dependent, the coefficients were B=.85, §=.82 for LMX and §=.18 for Race. The power hypothesis is that differences in power within the superior-subordinate dyad are negatively related to LMX. Specifically, as the power distance decreases, LMX should increase. To examine this hypothesis both LMX and intradyadic communication were regressed separately on the power difference variables. For LMX the best regression equation had intradyadic communication (g=.68: 3:18.84: 918330: p-.0001) and differences in referent power (§=.24: e=6.54: e;=330; n=.0001) as predictors (B=.79: £=271.57: df=2, 330, p=.0001). Regressing intradyadic communication on the difference variables and LMX produced significant coefficients for LMX (§=.74: 3:20.33: g:=327; p<.0001) and differences in coercive power (§=-.09: ;--2.35: g1=327; p<.02). The multiple correlation was .76 (£=229.42: Qf=2, 327, p<.0001). The regression analysis was repeated using correlations corrected for attenuation. The regression of LMX onto intradyadic communication and differences in referent power produced an 3=.91 and beta weights of .82 for intradyadic communication and .18 for referent power differences. Considering intradyadic communication as the dependent 78 variable produced R=.89, beta weight of .87 for LMX, and beta weight of -.08 for coercive power. Cau odel These results suggest a causal model. The similarity data suggests the model found in Figure 2. It is based on 129 identified superior-subordinate dyads. This model implies that increased socioeconomic difference leads to increased LMX, which in turn leads to increased intradyadic communication. It also specifies that similar racial identity leads to increased intradyadic communication. The model fits the data well. Neither the predicted correlation between socioeconomic differences (n=1.07: p=.283) nor the predicted correlation between race and LMX (z=-.05; pe.963) differed significantly from the obtained value. The obtained multiple correlation coefficient for intradyadic Socio- "3—1’9 LMX —.) .82 Intradyadic economic Communication ifferences .16 Racial similarity N=129 Figure 2 A causal model of the relationship between similarity, LMX, and intradyadic communication. 79 communication was .84, whereas LMX obtained .35. Moreover, the chi-square test of the fit of the model showed that the model fit the data closely (13(2)=1.16: p=>.05). The power data suggest the path model presented in Figure 3. It is based on a larger number of superior- subordinate dyads (N=328) than the similarity model, as the questionnaire contained scales for supervisor and subordinate power. This model suggests that the difference in referent power is positively related to LMX, and the difference in coercive power is negatively related to LMX. LMX in turn is positively related to intradyadic communication, so that the effects of referent and coercive power differences on intradyadic communication are mediated by LMX. Neither of the correlations predicted by the causal model differed significantly from their obtained values (2?- .068: p=.495. z=-.82: p=.410), and the obtained multiple correlation coefficients were .89 for intradyadic communication and .53 for LMX. Moreover, the global test of the model fit indicated that the data were consistent with the model (13(2)=1.14: p>.05). Although the argument for the similarity and the power models came from different theories, both deal with intradyadic similarity, and thus, are not contradictory. Moreover, the models share common variables, viz., LMX and intradyadic communication. Given these facts, a merger of the two approaches was attempted, producing the model presented in Figure 4. This model posits that socioeconomic 80 difference and differences in referent power both have positive paths to LMX, and difference in coercive power has Referent Power Difference .49 - o 17 5\ LMX . 89 Intradyadic ///////;7 ‘fi‘TT—Tabommunication Coercive Power -.14 Differences N=328 Figure 3 A causal model of the relationship between differences in power, LMX, and communication. a negative path. LMX in turn is positively related to intradyadic communication which also has racial similarity as a predictor variable. In order to test this model the path analysis was first run using correlations based on list-wise deletion (3:127), followed by a path analysis using the correlations based on pair-wise deletion (n:l34). Both approaches indicate a reasonable fit. The listwise approach produced no significant errors, and was not statistically significant (13(4):3.12: p>.05). This approach produced multiple correlation coefficients of .84 for intradyadic communication and .63 for LMX. Pair-wise analysis produced no statistically significant residuals and chi-square was not statistically significant (13(4)=3.63: p>.05). The multiple correlation coefficients were .90 for intradyadic communication and .59 for LMX. 81 Summary of Results All scales used in this study were submitted to confirmatory factor analysis to ascertain construct validity for the measurement model. The final solution produced fifteen scales. Difference scores were created by subtracting subordinate scores from supervisor scores, with the exception of race and gender where differences were coded as 0 and similarity as 1. Relationships were apparent among LMX, intradyadic communication, power differences, socioeconomic differences, and race. Difference scores that did not indicate a relationship were discarded from further analyses. LMX and intradyadic communication were regressed on all retained difference scores, as well as on each other. Socioeconomic differences, differences in referent power, and intradyadic communication proved to be strong predictors of LMX. Racial differences, differences in coercive power and LMX emerged as powerful predictors of intradyadic communication. Causal models were developed and tested for fit. Three models resulted, one for the power hypothesis, one for the similarity hypothesis and a third model merging the two. All models provided reasonable fit. 82 .cowumoHcoEEoo Uchmouuucw can .xzn .muwunawsau oHochonuucw no Hocoa Humane < v ousmwm Aanv 6~Huz moocmuomufia/ 030m 0>H0Hmoo vH.I moocmu0uuH uozom ucououmm COfiUflOHCfi—HEOU P \ oHocxocuucH/ N6. ilmzq ’vH.I owfiocooooHoom ha. “UV muHHnHHaHm HcHoo Chapter 4 DISCUSSION Generally the results are consistent with the hypotheses. Several power difference and similarity variables are related to LMX and intradyadic communication. Contradicting expectations, however, the power difference variables and the socioeconomic difference variable do not have a direct effect on intradyadic communication, but rather their effect is mediated by LMX. Racial differences do lead to less intradyadic communication, and there is a relationship between intradyadic communication and LMX. Contrary to expectations, however, communication between supervisors and subordinates does not lead to an improvement in LMX. Rather, the nature of the LMX relationship affects communication within the dyad! The socioeconomic differences scale that is related to LMX measures difference in spending on housing and housing related expenses, and food and groceries for home consumption. Food and shelter are primary need items, and one might expect that this type of expenditure would be neither visible and salient nor have an impact on the superior-subordinate relationship. They are low profile expenses, and it is likely seldom that supervisors and subordinates discuss their shopping habits. Nevertheless, where they shop, and perhaps some preferred brands, may become points of discussion, and thereby act as cues for 83 84 social class allegiance. The symbolic value of housing is more evident. Relative financial success is often expressed in housing choice. The physical structures provide a cue of relative socioeconomic status. Simply knowing a person’s address, or home town, without having seen the actual house is itself a cue as to the person’s socioeconomic status. Thus, supervisors and subordinates may well be aware of each other’s socioeconomic level, and this difference variable taps its relative disparity. The empirical relationship is positive which is contrary to the hypothesis. Similarity-attraction is not operating here, rather it is the supervisor’s success relative to the subordinate. Subordinates may be attracted by apparent success and wish to emulate the supervisor. Therefore, the subordinate desires and attempts to develop a good relationship with the supervisor, and perhaps pander the supervisor. The supervisor enjoys this positive attention and reciprocates this behavior by working on maintaining a good relationship. The data indicate that as the socioeconomic distance approaches zero, or becomes negative, the LMX relationships deteriorate. This result could indicate that when a supervisor is assigned a subordinate of higher socioeconomic 2 class than the supervisor’s , the supervisor may find it 2 A relevant example may be when business owner’s place their offspring or nein annaren; in the company to learn it from the bottom-up. 85 difficult to interact, and the subordinate may not be as receptive to the supervisor as others would be. Hence, the relationship may strain and suffer. Another form of desire for emulation may be evoked by the differences in referent power. Referent power here is empirically defined as being able to make someone feel they have their approval, and are valued and accepted. To the extent that supervisors have more referent power than subordinates, subordinates should desire to have a relationship with their supervisors, wish to interact with them, and attempt to build or maintain an approving, accepting relationship. This argument may explain why differences in referent power are related positively to LMX quality. Furthermore, when the distance in referent power approaches zero, or becomes negative, two things may be occurring. Subordinates may find interactions with their supervisor less rewarding, and be less inclined to maintain or develop a good LMX. As a result, supervisors may be chagrined to see subordinates model their peers, and may feel threatened by subordinates who are as good as, or better than, them at bolstering subordinates’ self-worth and esteem. If these circumstances occurred, the supervisors would be less inclined to desire a good dyadic relationship. The negative association between coercive power difference and LMX is consistent with the power hypothesis. The relationship may be as outlined earlier, i.e., given perceptions of equal power and similar power displays a 86 supervisor may include the subordinate in his/her power realm, and maintain distancing behaviors towards those perceived as less powerful. The equal partners in power would develop a positive LMX relationship. Coercive power is the abili;y to make being at work and working unpleasant, difficult and distasteful. If supervisors’ and subordinates’ jobs are highly interdependent they both have a high degree of coercive power over each in that they can each make the others work life miserable, should they so choose. Thus they should be motivated to develop and maintain a good working relationship, i.e., high quality LMX. The results indicate that LMX is strongest when the coercive power difference is negative. When the subordinate has the most coercive power, the supervisor may be highly motivated to build a good working relationship. On the other hand, if supervisors have a great deal of coercive power relative to the subordinates, the subordinates may find relating and interacting risky and therefore choose not to engage in relationship building. Race was not related to LMX, but it was associated with intradyadic communication. This result implies that LMX quality is not influenced by racial differences. It is inconsistent with the similarity hypothesis, yet it has intuitive appeal. Given the prominence of equal employment opportunity it is important and gratifying to note that good working relationships are not racially determined. Not finding a race effect may seem to be inconsistent with 87 earlier research that found superiors’ affect for subordinates positively related to differences in race (Tsui 8 O’Reilly, 1989). Nonetheless, intradyadic affect may not be related to LMX, as scholars working from the Dienesch and Liden perspective (1986) suggest. Indeed, the LMX-Affect dimension suggested by Schriesheim, Scandura, Eisenbach and Neider (1992) did not cluster with the remaining LMX scales in this research. Similarity in race is related to intradyadic communication. This relationship is an extension of earlier research that found demographic similarity related to technical communication within organizations (Zenger 8 Lawrence, 1989). Race is a potentially salient symbol of identity and reference group. It may indicate common ground and common experience, both of which provide impetus for interaction. This is consistent with the tendency for people to communication mostly with similar others (Johnson, 1993: Rogers 8 Bhowmik, 1971). This research proposed that LMX quality is positively related to intradyadic interaction, and this prediction proved to be the case. The direction of the causal hypothesis was, however, in error. LMX quality does not appear to be dependent on role negotiation and other communication. Rather, the research suggests that a high quality LMX relationship provides impetus for interaction. Supervisors and subordinates who develop a good relationship tend to talk more with each other, while supervisors and 88 subordinates who have a poor quality LMX relationship talk less. Although this research indicated that intradyadic communication was not antecedent to LMX, it is possible for intradyadic communication and LMX to be interdependent. LMX may impact intradyadic communication, but intradyadic communication may subsequently impact LMX. The nature of this research does not allow an investigation of that possibility, as longitudinal data would be needed. The null relationships are inherently less exciting, but equally important to discuss. Differences in legitimate, expert, and reward power were not related to LMX quality. Although the supervisors and subordinates both may have some input on reward decisions they may not make the reward decisions because the reward power may be centralized, thus controlled by others in the organizational hierarchy. Furthermore, it is likely that reward decisions are not based on issues relevant to the supervisor- subordinate relationship, but are more clearly tied to other more objective organizational outcome measures. If this reasoning is correct, the relative difference in reward power between supervisors and subordinates should have little impact on the LMX. Legitimate power is based upon the normative expectations of a particular position. That is, an individual in a particular position has a particular level of legitimate power due to the position, not the individual 89 in the position. Thus, legitimate power is much less relationally determined than it is a function of the position. Differences in legitimate power between supervisors and subordinates may therefore be acknowledged but static, and thus not influence the LMX relationship. It is surprising that differences in expert power were related neither to talking nor to LMX. It is reasonable to expect that differences in job expertise would be related to intradyadic communication, because intradyadic communication is a most likely medium for the transfer of needed technical knowledge and skill. It is possible, however, that differences in expertise are salient, but that the expertise is not shared. If expertise was hoarded, one would expect the relationship to suffer, i.e., there should be a positive association between differences in expertise and LMX. This relations was not obtained, however. The subordinates may have no expectations of attaining expert knowledge from their supervisors, and supervisors may not desire, be expected, or directed to minimize differences in expertise. Caution should be used in over-explaining this null- relationship. It may reflect an industry wide attitude that training is not necessary, or that it is somebody other than the immediate supervisors’ duty. If relative disparity in expertise is not considered relevant to the superior- subordinate, it should not impact the LMX. The relational-demography measures of schooling, company tenure, and position specific tenure were not 90 related to LMX or intradyadic communication. This is inconsistent with the similarity hypothesis and earlier research (Tsui 8 O’Reilly, 1989). Tsui and O’Reilly’s associations were between affect and role ambiguity. As discussed previously, affect may be irrelevant to LMX. Furthermore, these similarity measures differ from those that were related to LMX, in that they have no evident socioemotional component. Whereas rewards are presumed gratifying and coercive behaviors not, differences in tenure or schooling carry no connotation of socioemotional reward or cost. It may be argued that company or job tenure or degree of schooling provide cues as to success, and may provide an impetus for interaction and emulation. Job tenure may be such a cue, but rather than a success cue, it may imply lack of success or moderate success. That is, relative immobility in a position may indicate that one is not promotable or laterally transferable, but rather competent or comfortable, in the current position. Company tenure could be a relevant success indicator, however, in this instance it correlates very highly with position tenure (;:.65). Considering that these are both single item measures, the relationship is strong, and may reflect the same phenomenon. Supervisors and subordinates may not be aware of the number of years each attended school. Thus, even if schooling is a relevant indicator, it may not be able to be used. 91 Differences in the authoritarian attitudes were not related to LMX as hypothesized. Inconsistent with the similarity hypothesis, no effects were found for this variable. One reasonable explanation is that organizational actors do not display their authoritarian attitudes within the organization because not doing so improves their chances of staying employed. Furthermore, it is reasonable that a selection effect may have occurred, limiting the range of the authoritarian attitudes. Specifically, applicants who display behaviors consistent with attitudes at the fringes of the authoritarian scale would not likely get hired. The measure of socioeconomic-entertainment was not related to LMX or intradyadic communication. The degree of entertainment related spending is likely not discussed within the leader-member dyad, and therefore, not a salient cue for differentiation. One of the items of this index taps spending on clothing and shoes, which would appear to be an accessible cue. The sample for this research was taken, however, from the hospitality industry. Within the hospitality industry uniforms are the norm. Non-uniformed employees are executive level, for whom there usually are specific expectations as to dress code as well. Thus, clothing is not a relevant cue. Finally, gender differences were not related to LMX or intradyadic communication. This result implies that the quality of LMX relationships, and the degree of intradyadic communication are not based upon the gender of the dyad 92 partners. This datum is inconsistent with similarity- attraction and homophily expectations. It is consistent with the earlier finding regarding race differences and LMX, yet inconsistent in that there is also no relationship with intradyadic communication. A possible explanation is that the hospitality industry has always provided job opportunities for both males and females, thereby making gender of low salience to leader-member dyads within hospitality organizations. This null-relationship needs to be investigated further to ascertain whether it is idiosyncratic to this sample, industry, or due to an artifact of this study. Summary and Conclusion This research contributes to the literature in that it identifies antecedents of LMX and clarifies the LMX- communication relationship. Specifically, socioeconomic differences, differences in coercive power and differences in referent power are antecedents of LMX. It is important to note that a variable apparently extraneous to work impacts the superior-subordinate work relationship. Socioeconomic differences are positively related to LMX. In other words, as the intradyadic differences increase, the more likely the dyadic partners are to develop high quality LMX relationships. The relative success of managers or supervisors is attractive to subordinates and increases the value of the supervisor as a role model. In terms of building the LMX 93 relationship supervisors should not conceal potential success cues. It would be beneficial if future research would also seek to uncover other variables extraneous to organizations that may be related to LMX quality. The intradyadic difference in referent power is a strong predictor of LMX quality. Referent power has been defined empirically as being able to make others feel approved, valued and accepted. This definition emphasizes the potential for work place impact that supervisors have when they attend to the socioemotional well being and self- image of their subordinates. Moreover, differences in coercive power are negatively related to LMX. The scale items do not tap the degree of coercive behaviors, but rather the potential for this. This definition points to the potential impact of displaying potential coercive behaviors through unveiled threats and potential sanctions. One would not claim that supervisors and managers should make consequences unclear, or eschew coercive power when warranted, but rather that they should be cautious in invoking it. It may have significant negative effects on the LMX. Finally, it is an important finding that the LMX drives the communicative relationship, and that the nature of the communication does not cause LMX. This important result raises a question about the validity of the proposed linkage between Graen and Scandura’s (1987) role building theory and LMX. The relevance of the role theory and the proposed 94 linkage needs further investigation. It is possible that LMX and the communicative relationship are interdependent. To test this possibility longitudinal research is needed. Furthermore, it would be important to expand our knowledge about the LMX-communication relationship. For example, does LMX impact all aspect of communication, or can certain dimensions of communication be differentiated by LMX? It is roughly twenty years since the LMX construct was introduced. The last ten years LMX has received a great deal of attention, and most recently several scholars have focussed on the validity of the LMX construct. This monograph has expanded that work by initiating the inquiry into the antecedents of LMX. LIST OF REFERENCES LIST OF REFERENCES Altemeyer, B. (1988). Enenies of Ezeeden: e st n ' Ri 'n ut o 't ' 's . San Francisco: Jossey-Bass Inc. Aronson, E. 8 Worchel, S. (1966). Similarity versus liking as determinant of interpersonal attractiveness. Rsychononic Science, 5‘ 157-158. Argyle, M., Gardner, G. 8 Ciofi, F. (1958). Supervisory methods related to productivity, absenteeism, and labor turnover. uman e a 'ons 11,_e;;gei Bass, B. M. (1982). Intensity of relation, dyadic- group considerations, cognitive categorization, and transformational leadership. In J. G. Hunt, U. Sekaran 8 C. A. Schriesheim (eds.), Leggensnin: Reyong Estebliennent Views, Carbondale, IL: Southern Illinois University Press. Base. 8- M- (1990)- WWW Lendersnin: Theony, Reseancn, eng Renegeziel Applisatiensl 3rd_edll New York: The Free Press- Macmillan, Inc. Bass, B. M. 8 Avolio, B. J. (1993). Transformational leadership: A response to critiques. In M. M. Chemers 8 R. Ayman (eds.), Leedenship Tneeny eng Reeeanch. New York: Academic Press. Bavelas, J. B., Black, A., Chovil, N. 8 Mullett, J. (1990). Egnivecal Qommunication, Newbury Park, CA: Sage Publications, Inc. Baysinger, B. D. 8 Mobley, W. H. (1983). Employee turnover: individual and organizational analysis. In K. M. Rowland 8 G. R. Ferris (eds.), Research in e so uma so es e ° 5 ch AnnualI ll 269-319. Greenwich, CT: JAI Press Inc. Bennis, W. G. 8 Nanus, B. (1985). Leaders: The Strategies for Taking Qhenge. New York: Harper 8 Row. 95 96 Berger, C. R. 8 Bradac, J. J. (1982). Lengnage and §ocial Rnewledge: gncerteinEy in Intenpersonal Relatiene, London, England: Edward Arnolds (Publishers) Ltd. Blau, P. M. (1964). e a d Powe ' oc'al ' e. New York: John Wiley. Bochner, A. P. (1991). On the paradigm that would not die. Censunieatign_xearheekl 111 484-491- Borchgrevink- C- P- (1993)- Burngut_Ameng_N2n: Menageriel Rospiteliey Employees. Paper presented August 5, 1993 at the 1993 CHRIE (Council on Hotel, Restaurant 8 Institutional Education) annual conference in Chicago, IL,. Borchgrevink, C. P. 8 Boster, F. J. (1994). Leader- Member Exchange: A Test of the Measurement Model. Rosnitality Researen Journel, l7(3), 75-100. Brown, P. 8 Levinson, S. (1987). P en ss' om nnivensels in Lengnege Usege, Cambridge: Cambridge University Press. Burns, J. M. (1978). Leegenenipe New York: Harper 8 Row. Byrne, D. (1971). Ine AEEIacgien Benedigm, New York: Academic Press, Inc. Byrne, D., Clore, G. L., Jr. 8 Smeaton, G. (1986). The attraction hypothesis: Do similar attitudes affect anything? Jonrnel ef Rersoneliey end §QQi§l 2.22h212921 511 1167-1170- Cammalleri, J. A., Hendrick, H. W., Pittmen, W. C., Jr., Blout, H. D. 8 Prather, D. C. (1973). Effects of different leadership styles on group accuracy. Regenel Q: Applleg Rsyenolegy, 57, 32-57. Cappella, J. N. 8 Palmer, M. T. (1990). Attitude similarity, relational history, and attraction: The mediating effects of kinesic and vocal behaviors. Qammunieatien_uenesraph§l 511311 161-183- Cappella, J. N. 8 Palmer, M. T. (1992). The effects of partners’ conversation on the association between attitude similarity and attraction. gennnnieeeien Monogrepns , .21211 180- 189. 97 Cashman, J., Dansereau, F. Jr., Graen, G. 8 Haga, W. J. (1976). Organizational understructure and leadership: A longitudinal investigation of the managerial role- making process. Qrganizatienal_Behaxier_and_Human Perfermance- 15. 278-296. Chemers, M. M. (1993). An integrative theory of leadership. In M. M. Chemers 8 R. Ayman (eds.), a er ' heo se r h° t've a Qinections, New York: Academic Press, Inc. Churchill, G. A., Jr., Ford, N. M., 8 Walker, 0. C. (1974). Measuring the job satisfaction of industrial salesmen. I_urn_l_ef_u_rketing_8esear_h1 Augustl 254- 250. Coleman, J. S. (1975). Social structure and a theory of action. In P. M. Blau (ed.), 0 c s o gengy e: §oeial Signetnne, New York: The Free Press. Condon, J. W. 8 Crano, W. D. (1988). Inferred evaluation and the relation between attitude similarity and interpersonal attraction. o o s and_§221al_£_xsh_12sxl 5.1511 789- 797. Dahl, R. A. (1957). The concept of power. Renegienel Scienee, 21 201-215. As quoted by Pfeffer (1981). Dansereau, F. Jr., Graen, G., 8 Haga, W. J. (1975). A vertical dyad linkage approach to leadership within formal organizations. A longitudinal investigation of the rolemaking process. r on av Hunan Rerfiormence, l_, 46- 78 Day, R. C. 8 Hamblin, R. L. (1964). Some effects of close and punitive styles of supervision. Anenieen l_urnal_2f_§2§ieles¥1 69 499- 510- Dienesch, R. M. 8 Liden, R. C. (1986). Leader-member exchange model of leadership: A critique and further development- Agedemx_of_nanasement_8exieul 111 618-634- Dockery, T. M. 8 Steiner, D. D. (1990). The role of the initial interaction in leader-member exchange. 6:999 g Organiga gign Stu gieg, 15(5), 393- -413. Donohue, W. A., Rogan, R., Ramesh, C. 8 Borchgrevink, C. P. (1990). t o ev m Hostage Regotiation. Paper presented to the panel on Special Topics in Conflict Management at the Central States Communication Association, Detroit. Dornbusch, S. M. 8 Scott, W. R. (1975). tio d 98 the Exercise of Aunnezity. San Francisco: Jossey-Bass Publishers. Downton- J V (1973) Bebel_Lead_rshin1_Qemmitment and_Qharisma_1n__he_Be_Qluti_narx_2roee_§1 New York: Free Press. Duchon, D., Green, S. G. 8 Taber, T. D. (1986). Vertical dyad linkage: A longitudinal assessment of antecedents, measures and consequences. Journel of Applieg Rsyenology, 71, 56-60. Dunegan, K. J., Duchon, D. 8 Uhl-Bien, M. (1992). Examining the link between leader-member exchange and subordinate performance: The role of task analyzability and variety as moderators- Journal_ef_nanagementl l8(1), 59-76. Emerson, R. M. (1962). Power-dependence relations. 'c c'olo ' ev' w 27(1), 31-41. Fairhurst, G. T. 8 Chandler, T. A. (1989). Social structure in leader-member interaction. Qennnnieeeien uQanrfighgl fig: 215-2390 Fairhurst, G. T., Rogers, L. E. 8 Sarr, R. A. (1987). Manager-subordinate control patterns and judgments about the relationship. In McLaughlin, M. L. (ed.) gemmunieatien_xearbeekl 101 395-415- Newbury Park. CA: Sage Publications, Inc. Ferris, G. R. (1985). Role of leadership in the employee withdrawal process: A constructive replication. Jounnel of Applieg Rsyenglogy, 79(5), 777-781. Foa- U. G. & Foa- E- B- (1974)- fiesietal_§trugture§_ef tne Rind. Springfield, IL: Charles C. Thomas. French, J. R. P., Jr. 8 Raven, B. H. (1959). The bases of social power. In D. Cartwright (ed.), finngiee_in Sociel Powen. Ann Arbor, MI: University of Michigan Press, 150-167. Freudenberger, H. J. (1974). Staff burn-out. Jennnel of Secial Issues, 221 159-165. Gibb, C. A. (1947). The principles and traits of leadership- J2urnal_2f_Abnermal_§egial_£§xehelegx- __ 267-284. Graen, G. (1976). Role-making processes within complex 99 organizations. In M. Dunnette (ed.), fiendbooR of Q;ganize§ional Psycnelegy (pp. 1201-1245). Chicago, IL: Rand McNally Graen, G. 8 Cashman, J. F. (1975). A role-making model of leadership in formal organizations: A developmental approach. In J. G. Hunt 8 L. L. Larson (eds.), LQQQ§I§h12_££QDLi§I§1 OH: Kent State University Press. Graen, G. 8 Ginsburgh, S. (1977). Job resignation as a function of role orientation and leader acceptance: A longitudinal investigation of organizational assimilation. Or ' v um £§£I2IE§DQ§1 121 1'17- Graen, G. B., 8 Scandura, T. A. (1987). Toward a psychology of dyadic organizing. In L. L. Cummings 8 B. M. Staw (eds ) _Q§2Q_Qh_iD_Q£Q§D132L19n91_flénéxi_£1 51 175-208. Greenwich, CT: JAI Press Inc. Graen, G., Dansereau, F. Jr, 8 Minami, T. (1972). empirical test of the man in the middle hypothesis among executives in a hierarchical organization employing a unit set analysiS- Organizational_flehaxier and H m n Pe fo a , R, 262-285. Graen, G. B., Liden, R. C. 8 Hoel, W. (1982). Role of leadership in the employee withdrawal process. gennnel ef Applieg Rsycnolegy, §7(6), 868-872. Graen, G., Novak, M. A. 8 Sommerkamp, P. (1982). The effects of leader-member exchange and job design on productivity and satisfaction: testing a dual attachment model. Qrganizational_fiehaxier_and_fluman Eerformance- 32. 109-131- Graen, G. B., Wakabayashi, M, Graen, M. R, 8 Graen, M. G. (1990). International generalizability of American hypotheses about Japanese management progress: A strong inference investigation. L§3Q§I§h12_QBQIL£IlYI l(1), 1- 23. Herzberg, F. I., Mausner, B. 8 Snyderman, B. B. (1959). Ine netiv vetio on :0 Week, New York: John Wiley 8 Sons. Hespe, G. 8 Wall, T. (1976). The demand for participation among employees. finnen_Relgeien§L 29, 411-428. Hinken, T. R. 8 Schriesheim, C. A. (1989). Development 100 and application of new scales to measure the French and Raven (1959) bases of social power. a 'ed £§Y2h219911.111111 551-557- Hollander- E- P- (1976) Er1nc1ples_and_uethods__f So oeiel Reychology (3rd ed.) Oxford, G. B.: Oxford University Press, Inc. Hollander. E- P- (1978) Leadersh12_nxnam12§__a 222021931 guide te Effie eeeiv e BEIQEiQDfihiES . New York: Macmillan, Free Press Hollander, E. P. (1993). Legitimacy, power, and influence: A perspective on relational features of leadership. In M. M. Chemers 8 R. Ayman (eds. ), Lea_er_h12_Theerx_and_Besearch1 New York: Academic Press, Inc. Holloman, C. R. (1968). Leadership and headship: There is a difference. Eer_onnel_Adm1n1§trati_nl 311411 38- 44. Homans, G. C. (1958). Social behavior as exchange. Amer1can_lournal_gf_§291_los¥1 £31 597- 606- H0ppock- R- (1935)- Jeb_§atisfaction1 New York: Harper. Jablin, F. M. (1987). Organizational entry, assimilation and exit. In F. M. Jablin, L. L. Putnam, K. H. Roberts 8 L. W. Porter (eds.), HandbooR e: Onganieeeienel ; 1 !° . E I ! 1° . J' E !° 0 . Newbury Park, CA: Sage Publications, 679-740. Jacobs. T. 0- (1970) Leader§h12_and_fixchange_1n Formal_Qrganizat12n_1 Alexandria, VA: Human Resources Research Organization. Johnson- J-D- (1993)- Qrganizat19nal_92mmunication ELEEELBIQI Norwood, NJ: Ablex Publishing Corporation. KatZ- D. & Kahn. R- L- (1978)- The_§22111_2s12h21292_2f engenineeiene, (2nd ed.). New York: John Wiley 8 Sons. Kieser, A., Reber, G. 8 Wunderer, R. (1987) (eds.). ndw uch de b de 5 Stuttgart, Germany: C. E. Poeschel Verlag. Kim, K. I. 8 Organ, D. W. (1982). Determinants of leader-subordinate exchange relationships. gnoup e Onganizational Stugies, 1(1), 77-89. Kozlowski, S. W. J. 8 Doherty, M. L. (1989). 101 Integration of climate and leadership: Examination of a neglected issue. denpnal on Applied Rsycnology, 74(4)I 546-553. Kotter- J P- (1990) A_Eorce_for_ghanse1_now Leadersh12_Diff_rs_from_uanagemen11 New York: The Free Press, Macmillan, Inc. Krantz,J. (1989). The managerial couple: Superior -subordinate relationships as a unit of analysis. Reman Reseupces Renegenene, 28(2), 161-175. Krone, C., Tabacchi, M. 8 Farber, B. (1989a). Manager burnout The_Q2rnell_fletel_an__gestaurant Adm1ni_tr__1_n_Quarterlxl 121111 58- 63- Krone, C., Tabacchi, M. 8 Farber, B. (1989b). A conceptual and empirical investigation of workplace burnout in foodservice management. fleepineliey EducaEion and Reseepen d onnnel, 13(3), 83-91. Lagace, R. R. (1990). Leader-member exchange: Antecedents and consequences of the cadre and hired hand. dournal of Bensonel Selling e Selee Ranegenent, 121111 11'19- Lawrence, B. S. (1988). New wrinkles in the theory of age: Demography, norms and performance ratings. Academx_of_nanasement_leurnall 311211 309-337- Lewin, K., Lippitt, R. 8 White, R. K. (1939). Patterns of aggressive behavior in experimentally created "social climates" J2urnal__f_§ecial_2_xch212921 1.1 271- 299. Ley, R. (1966). Labor turnover as a function of worker differences, work environment, and authoritarianism of foremen. Journal e: Applied Rs yeno lggy, 50 , 497- 500. Liden R. C. 8 Graen, G. (1980). Generalizability of the vertical dyad linkage model of leadership. Aeedemy ef Renegement Joupnel, 23(3), 451-465. Likert, R. (1932). A technique for the measurement of attitudes. Anchivee ef Psyenolegy, 159. 44-53. Litwin, G. H. 8 Stringer, R. A. (1968). Meeixeeien_end '0 Boston: Harvard University, Graduate School of Business Administration. 102 Maslach, C. (1982). Bu : e ost r' . Englewood Cliffs, NJ: Prentice-Hall. Maslow, A. H. (1943). A theory of human motivation. Es yghglggic Cpl Be vigg, 59, 370- 396. MasloW. A. H. (1954). u9t1291192_and_zer§2n111121 New York: Harper Bros. McClane, W. E. (1991a). Implications of member role differentiation. fizpup 8 Organizapigp Studies, 16(i)l 102-113. McClane, W. E. (1991b). The interaction of leader and member characteristics in the leader-member exchange (LMX) model of leadership- 91111.9:992_22§911911 22(3). 283-300. McClelland, D. C. (1970). The two faces of power. J_urn_1_2f.12121na§19n11.8£fair§1 211111 29- 47 McDermott, S. T. (1991). The generalizability of the communication to attraction relationship to intercultural communication: Repulsion or attraction? In J. a. Anderson (ed ). Qemmun1sa§19n.¥earbe_k1 111 492- 497. McWhirter, R. M. 8 Jecker, J. D. (1967). Attitude similarity and inferred attraction. Egygnpppmig fisiense1 11 225-226- Miller, G. R. 8 Steinberg, M. (1975). Between people: A new analysis of interpersonal communication. Palo Alto, CA: Science Research Associates. Cited by Bochner, A. P. (1991). On the paradigm that would not die. In J A. Anderson (ed ). 92mmun121112n_191122231 19. 484- 491. Miller, K. I., Stiff, J. B. 8 Ellis, B. H. (1988). Communication and empathy as precursors to burnout among human service workers. Qgmmun122112n_nenegranh§1 £5; 250- 265. Miller, K. I., Zook, E. G. 8 Ellis, B. H. (1989). Occupational differences in the influence of communication on stress and burnout in the workplace, nan19ement_ngmun1satign_guarterlx 11211 166- -190. Mills, J. 8 Clark, M. S. (1982). Exchange and communal relationships. In L. Wheeler (ed.), Review pf Personality and Sogiai Psyghology, 11 121-144. Mintzberg, H. (1982). If you're not serving Bill and 103 Barbara, then you're not serving leadership. In J. G. Hunt, U. Sekaran 8 C. A. Schriesheim (eds.), Leadersnip: Beyond Estnblisnnen; Views, Carbondale, IL: Southern Illinois University Press. Morgan. G. (1986)- Image§_o1_grganizations1 Newbury Park, CA: Sage Publications. Mulder, M. (1971). Power equalization through participation? Adm1n1s1rat1xe.§219n29.Quarterlx1 191 31-38. Mulder, M. (1976). Reduction of power differences in practice: The power distance reduction theory and its implications. In G. Hofsteede 8 M. S. Kassem (eds.), o e 'b to Assen/Amsterdam, The Netherlands: Koninklijke Van Gorcum 8 Co. B. V. Mulder. M- (1977)- Tne_Dailx_Eower_§ame1 Leiden. The Netherlands: Martinus Nijhoff Social Sciences Division. Mulder, M. 8 Stemerding, A. (1963). Threat, attraction to group and need for strong leadership. nnmdn Reiatipns, 16(4)I 317-334. Muringham, J. K. 8 Leung, T. K. (1976). The effects of leadership involvement and the importance of the task on subordinates' performance. Beha212r_and_numan_£_rferman2e1 111.299'310- O’Leary, V. E. 8 Hansen, R. D. (1983). Performance evaluation: A social psychological perspective. In F. Landy, S. Zedeck 8 J. Cleveland (eds.), £gp£pzmdngg Measurement_and_1heorx1 Hillsdale. NJ: Lawrence Erlbaum Associates Publishing, 197-218. Parsons, T. (1960). S tu and s ' o Spgieties. Glencoe, IL: The Free Press. Pelz, D. (1952). Influence: A key to effective leadership in the first line supervisor. e o e 221 209-217. Pfeffer, J. (1981). w ' s. Marshfield, MA: Pitman Publishing. Pfeffer. J- (1982)- Qr9an111112n§_and_grganization Tneopy. Marshfield, MA: Pitman Publishing. Raven, B. H. (1965). Social influence and power. In 104 I. D. Steiner 8 M. Fishbein (eds.), Current Studies in Social Psychologyl NY: Holt, Rinehart 8 Winston, 371- 382. Reichers, A. E. (1985). A review and reconceptualization of organizational commitment. AW 1.9.1.1)... 465-476- Rogers, E. M. 8 Bhowmik (1971). Homophily-heterophily: Relational concepts for communication research. prlie Opinion Quarterly. 34(4). 523-538. Roloff, M. E. (1981). t sona C u 'c ion: Social Encnange Approacn, Beverly Hills, CA: Sage Publications, Inc. Rosenbaum, M. E. (1986a). The repulsion hypothesis: On the non-development of relationships. W W 5119.1... 1156- -1166 Rosenbaum, M. E. (1986b). Comment on a proposed two stage theory: First repulsion; Then attraction. Joupnal of Personality and Social Psychology, 51(6), 1171-1172. Rosse, J. G. 8 Kraut, A. I. (1983). Reconsidering the vertical dyad linkage model of leadership. Jpnpnel_e; Oecupetional Psycnelogyll §§(l). 63-71. Rudin, S. A. (1964). Leadership as psychophysiological activation of group members: A case experimental study. Psycnelogieel Repopts, 1E1 577-578. Salzmann, J. 8 Grasha, A. F. (1991). Psychological size and psychological distance in manager-subordinate relationships. 'a s o o l§l(§). 629- -646. Scandura, T. A. 8 Graen, G. B. (1984). Moderating effects of initial leader-member exchange status on the effects of a leadership intervention. denrnal e: AW. 9.29).. 428-436- Scandura, T. A., Graen, G. B, 8 Novak, M. A. (1986). When managers decide not to decide autocratically: An investigation of leader-member exchange and decision influence. gournal pf Applied Esyenelegy, Zl(§), 579- 584. 105 Schelderup-Ebbe, T. (1922). Beitrage zur sozialpsychologie des haushuhns [Contribution to the Social Psychology of the Henhouse]. Zeitchr. Psychol., 88, 225-252. Cited in A. Sjolund, fixnppep_ykelegi [Group Psychology], Oslo, Norway. Fabritius Forlagshus. Schiemann. W- (1977)- 81ructural_and_lnternersonal fects o tter s f Mana e ' l C u i t' ' on it ' Inve t'o . Unpublished doctoral dissertation. IL: University of Illinois. Schneider, B. (1983a). Interactional psychology and organizational behavior. In L. L. Cummings 8 B. M. staw (eds.). Besear2h_1n_Organ1zation11_aeha!1911 51 1- 31. Greenwich, CT: JAI Press. Schneider, B. (1983b). Work climates: An interactionists perspective. In N. W. Feimer 8 E. S. Geller (eds.), v' o e : ' ' s Eepspeepives, 106-128. New York: Praeger. Schriesheim, C. A., Neider, L. L., Scandura, T. A. 8 Tepper, B. J. (1992). Development and preliminary evaluation of a new scale (LMX-6) to measure leader- member exchange in organizations. Ednenpienel_end 2sxch_1291211_Nea_urement1 521 135- 147 Schriesheim, C. A., Scandura, T. A., Eisenbach, R. J. 8 Neider, L. L. (1992). Validation of a new leader- member exchange scale (LMX-6) using hierarchically- nested maximum likelihood confirmatory factor analysis. Educepionel and Psyenologieel Meesnpement. 52‘ 983-992. Scott. W- R- (1992). QIgan111112n§1_31112n111_flatural end Open Syspens, 3rd ed. Englewood Cliffs, NJ: Prentice Hall. Shaw, M. E. (1955). A comparison of two types of leadership in various communication nets. Jen;nel_p§ Abn_rmal_and__22111_£_xchologx1 291 127- -134- Sherer, K. R. 8 Giles, H. (1979) (eds. ) Enpepeen tu- es , oc . °s ,. -- ' . ’ _--- ,, Cambridge, England: Cambridge University Press. Sims, H. P. 8 Manz, C. C. (1984). Observing leader verbal behavior: toward reciprocal determinism in leadership theory. gelely 222-232. 106 Singh, R. 8 Tan, L. S. C. (1992). Attitudes and attraction: A Test of the similarity-attraction and dissimilarity-repulsion hypotheses. 't's a f §QC Cidl Eeyenology, 11‘ 227-238. Smeaton. G., Byrne, D. 8 Murnen, S. K. (1989). The repulsion hypothesis revisited: Similarity Irrelevance or Dissimilarity Bias? J2urnal_of_2ersonalitx_and 822111.29191219921 591111.54-59- Smith, P. C., Kendall, M. 8 Hulin, C. L. (1969). The asure e t of Sa c on W d at eme t. Chicago, IL: Rand McNally 8 Co. Steiner, D. D. (1988). Value perceptions in leader member exchange- The_J2urnal_of_fiogial_£sxshologxl 12§(5). 611-618. Stiff, J. B., Dillard, J. P., Somera, L., Kim, H. 8 Sleight, C. (1988). Empathy, communication and prosocial behavior. Qommunication_nonograeh§1 151211 198-213. Sunnafrank, M. (1991). Interpersonal attraction and attitude similarity: A communication-based assessment. In J a. Anderson (ed ). Sommunication_xearbo_kl 111 451- 433. Sunnafrank, M. (1992) On debunking the attitude similarity myth- Qommunication_NQDogranh_1 221211 164- 179. Sunnafrank, M. J. 8 Miller, G. R. (1981). The role of initial conversations in determining attraction to similar and dissimilar strangers. finnen_gemnnnieepipn Reseaggn, 8(1), l6-25. Tabacchi, M., Krone, C. 8 Farber, B. (1990). Workplace and social support in ameliorating managerial burnout in the food and beverage industry. Reseazcn dournal, 15(2), 553- -560. Thibaut, J. 8 Kelley, H. H. (1959). Ine_§eeiel 2122121292_91_§rgunsl New York: John Wiley- Tsui, A. s. & O’Reilly, c. A., III. (1989). Beyond simple demographic effects: The importance of relational demography in superior-subordinate dyads. Academy e: Menagemenp qurnel, 32(2), 402-423. 107 Turban, D. B., Jones, A. P. 8 Rozelle, R. M. (1990). Influences of supervisor liking of a subordinate and the reward context on the treatment and evaluation of that subordinate. u211211192_and_fimot12n1 111111.215- 233. U.S. Department of Commerce: Office of Federal Statistical Policy and Standards (1978). Spepietieel _I_>g>_l__iey_fiendppek_L Washington, DC: U.S. Government Printing Office. Vecchio, R. P. (1982). A further test of leadership effects due to between-group variation and within-group variation. Jeurnel e: Applied Eeyenelegy, 91, 200-208. Vecchio, R. P. (1985). Predicting employee turnover from leader-member exchange: A failure to replicate. Aeademy of Menegenenp Jonznel, 28, 478-485. Vecchio, R. P. 8 Gobdel, B. C. (1984). The vertical dyad linkage model of leadership: Problems and prospects. Qrg1n11at19n11_fienaxior_and_numan Performanee. 31y 5-20. Vecchio, R. P., Griffeth, R. W. 8 Hom, P. W. (1986). The predictive utility of the vertical dyad linkage approach. Ine geu upnel of So ciel Psycnpl ogy, l2§(5 ), 617- 625. Vroom, V. H. 8 Mann, F. C. (1960). Leader authorization and employee attitudes. Bergennel 2&22h212921 121 125-140- Weber, M. (1958/1915). The social psychology of the world religions. In H. H. Gerth 8 C. W. Mills (eds. and trans.). Er2m_Nax_Neber__E§§121_1n_§osiologx (pp- 267-301). New York: Galaxy. (Original work published 1915). White, R. 8 Lippitt, R. (1968). Leader behavior and member reactions in three "social climates." In D. Cartwright 8 A. Zander (eds.), grenp_nynemieey 318-335. New York: Harper Row. Williamson, O. E. (1981). The economics of organization: The transaction cost approach. Anepieen Journal_of_§eciologxl 811 548-577. Cited in Scott (1992). Woods, R. H. 8 Macaulay, J. F. (1989). R for turnover: Retention programs that work. Ine_gepnell_nppel_end ' 's 'o ua 39(1). 78-90. 108 Zahl-Begnum, O. H., Koldingnes, W. 8 Moog-Begnum, S. (1988). Appeids pg Qrgenisasjons Eeykelogi [Work and Organizational Psychology]. Oslo, Norway: NKS Forlaget. Zedeck, S. 8 Cascio, W. F. (1984). Psychological issues in personnel decisions. Annnel_3eyiey_ef Esyenology. 35. 461-518. Zenger, T. R. 8 Lawrence, B. S. (1989). Organizational demography: The differential effects of age and tenure distributions on technical communication. Aeedeny of Ma a e e Journa 32(2). 353-376.