Leader-Member Exchange and Organizational Citizenship Behavior : the Mediating Impact of Self-Esteem

The aim of this study was to examine the influence of leader-member exchange (LMX) from the perspectives of superior as well as subordinates on organizational citizenship behavior (OCB). Self-esteem was also examined as a mediator variable. Data were gathered from a sample of 300 non-supervisory employees and their 118 superiors, that represented banking organizations situated in the Klang Valley of Malaysia. The analysis indicated that superior-LMX had positive impact on OCB, but the impact of subordinate-LMX was not significant. Contrary to what was hypothesized, self-esteem did not influence both superior-LMX and subordinate-LMX relationships with OCB. Key implications of the survey findings both for theory and for practice are discussed, potential limitations are specified, and directions for future research are suggested.


Introduction
As the world moves toward a single market system, enhancing an organization's competitive ability is increasingly critical and behaviors, which may improve individual and organizational efficiency, become more valuable.Customer service organizations such as banks are often seen as strong and profitable business institutions.The banks are now demanding employees who are "good citizens"--individuals willing to extend themselves to help coworkers and employers (Sloat, 1999).In order to be competitive, the banks need to ensure that their employees are sensitive, thoughtful, and effective when carrying out their work.They need to be encouraged to show their fullest potential.Employees who go the extra mile by performing spontaneous behaviors that go beyond their role prescriptions in assisting customers are especially valued by the management.This phenomenon is critical for organizational effectiveness because managers cannot foresee all contingencies or fully anticipate the activities that they may desire or need employees to perform (Katz & Kahn, 1966;Organ, 1988a).Doing jobs beyond what is required without expecting to be rewarded is what we refer to in this study as "Organizational Citizenship Behavior" (hereinafter, OCB; see Organ, 1988a).
Although there have been many studies of OCB in organizations, but there is a dearth of research on the linkage of leader-member exchange (hereinafter, LMX) from the individual perspectives of supervisors (superior-LMX) and subordinates (subordinate-LMX) separately with OCB where the effect of self-esteem, as a mediator, is included.Given this lack of information, attempts are made to answer four questions.Does quality of subordinate-LMX influence his or her OCB?Does quality of superior-LMX influence subordinate's OCB?Does subordinate's self-esteem mediate the relationship between quality of subordinate-LMX and OCB?Does subordinate's self-esteem mediate the relationship between quality of superior-LMX and OCB?

Theoretical framework and development of hypotheses
This study conceptualizes the subordinate-LMX and superior-LMX relationships with five types of OCB--altruism, courtesy, conscientiousness, sportsmanship, and civic virtue.Further, we believe that this relationship is mediated by self-esteem, as depicted in Figure 1 below.

Organizational Citizenship Behavior (OCB)
In recent years, much interest in OCB has been shown.OCB has been said to enhance organizational performance because they lubricate the social machinery of the organization, reduce friction, and increase efficiency (Bateman & Organ, 1983;Smith, Organ, & Near, 1983).OCB represents individual behavior that is discretionary, not directly or explicitly recognized by the formal reward system, and that in the aggregate promotes the effective functioning of the organization (Organ, 1988a).Most OCB actions, taken singly, would not make a dent in the overall performance of the organization (Organ, 1988b).The effect will be seen with the aggregate summation of OCB performed across time and across persons in the group, department, and organization.The most critical element is that these behaviors are defined at face value.OCB are behaviors that are clearly observable by peers, supervisors, or researchers.
Originally, Smith et al. (1983) proposed a two-dimensional model of OCB: altruism and generalized compliance (later called conscientiousness).Altruism refers to behaviors that are voluntary--for example, being cooperative, helpful, and other instances of extra-role behavior--helps a specific individual with a given work-related problem.Generalized compliance refers to those behaviors that are impersonal that does not provide immediate aid to any particular individual but is indirectly helpful to other people in the organization.It places priority on arriving to work on time, not taking too many breaks, or not leaving early.Later, Organ (1988a) expanded the taxonomy of OCB to include altruism (narrower than the altruism of Smith et al., 1983), conscientiousness (narrower than generalized compliance), sportsmanship, courtesy, and civic virtue.Courtesy refers to behaviors that are directed to the prevention of future problems.It is different from altruism because altruism is helping someone who has a problem, while courtesy is helping to prevent problems, performing thoughtful or considerate gestures towards others.Sportsmanship describes those individuals who tolerate the annoyances that are inevitable in the workplace.The behavior demonstrates tolerance of less than ideal conditions at work without complaining.Finally, civic virtue consists of those behaviors that are concerned with the political life of the organization (e.g., attend meetings, engage in policy debates, express one's opinions in implementing a new policy, etc.).

Leader-Member Exchange (LMX) and Organizational Citizenship Behavior (OCB)
A review of the research reveals that relative to perceived organizational support, LMX is a better predictor of OCB (Settoon, Bennett, & Liden, 1996).Deluga (1994) found a positive relationship between employee OCB and the quality of LMX.LMX describes how leaders develop different exchange relationships over time with their various subordinates as they influence each other (Farouk, 2002).LMX research shows that subordinates reporting high-quality LMX not only assume greater job responsibilities but also express contributing to other units (Liden & Graen, 1980).Hence, the quality of the LMX influences levels of delegation, responsibility, and autonomy and in turn, employees perceive greater latitude, decision influence, and feelings of contribution (Gomez & Rosen, 2001).These feelings of contribution may be those that are beyond the job prescribed in their job descriptions and are referred to as OCB.Liden and Maslyn (1998) found a four-dimensional LMX model comprising of contribution, loyalty, affect, and professional respect.We expect the higher the LMX on each of these dimensions, the higher the subordinates will be in performing OCB.Hence, the hypotheses: Hypothesis 1: Subordinate-LMX has a significant positive relationship with OCB.
Hypothesis 2: Superior-LMX has a significant positive relationship with OCB.

Leader-Member Exchange (LMX) and Self-Esteem
People attach greater value to the way they are treated (Tyler & Folger, 1980).There has been relatively little research examining the relationship between LMX and self-esteem.However, considerable conceptual work suggests that higher LMX employees should have higher levels of self-esteem.Korman (1970) pointed out that one of the ways self-esteem can be derived is from social rewards and feedback from others about one's performance.Thus, having superiors treating subordinates this way makes them feel good about themselves and affects their self-esteem.Self-esteem means a global evaluation of the self, and it is typically measured by the degree to which the person endorses various evaluative statements about self (Baumeister & Tice, 1985).We thus hypothesize that: Hypothesis 3: Subordinate-LMX has a significant positive relationship with self-esteem.
Hypothesis 4: Superior-LMX has a significant positive relationship with self-esteem.

Self-Esteem and Organizational Citizenship Behavior (OCB)
Self-esteem is both conscious and unconscious.It can lead to the most fundamental methods for helping people.It begins in childhood and is developed through experiences of what we can do and what we cannot do.Korman (1970) through his "self-consistency" theory suggested that work behavior is based on implementation of a self-concept.This theory predicts that, all other things being equal, the more a person perceives himself or herself to be competent, skilled, or qualified for a particular job, the higher will be his performance on this job (Weiner, 1973).It is expected that only high self-esteem employees will have the emotional stamina to perform OCB under difficult and challenging situations (Tang & Ibrahim, 1998).There is widespread acceptance of the psychological importance of self-esteem.Although there have been equivocal findings regarding the effects of self-esteem on performance, there has been evidence that mood as one type of job satisfaction, is positively correlated with performance (Kramer, Newton, & Pommerenke, 1993).We expect therefore, self-esteem to be related to OCB, as one type of performance outcome.
Hypothesis 5: Self-esteem has a significant positive relationship with OCB.
To date, there has been little effort made to study the potential mediating effect of self-esteem at the individual level, relating to quality of LMX and OCB.Due to the dearth of studies relating self-esteem to organizational behavior literature, we will here investigate the mediating influence of self-esteem on the relationships between quality of LMX and OCB.The following hypotheses are conjectured: Hypothesis 6: Subordinate's self-esteem mediates the relationship between subordinate-LMX and subordinate's OCB in such a way that the impact of subordinate-LMX on OCB will be smaller (partial mediation) or nonsignificant (full mediation) in the presence of self-esteem.
Hypothesis 7: Subordinate's self-esteem mediates the relationship between superior-LMX and subordinate's OCB in such a way that the impact of superior-LMX on OCB will be smaller (partial mediation) or nonsignificant (full mediation) in the presence of self-esteem.
This study employs social exchange theory to investigate the connection between quality of leader-member exchange (LMX), self-esteem, and OCB.Social exchange theory has been the primary framework for understanding OCB (Konovsky & Pugh, 1994;Organ, 1988a;Smith et al., 1983).People always seek to reciprocate those who benefit them (Adams, 1965;Blau, 1964) to reciprocate the support from the organization, the employee may do so via job performance, but such performance may be limited up to a certain extent only since organization has strict contracts.Thus, the exchange that takes place will be more of an ambiguity thus allowing discretionary acts to be carried out by employees.OCB, from a social exchange theory perspective, becomes an outlet for these positive feelings.When a supervisor engages in helping behavior toward an employee, the employee incurs obligations to repay the supervisor so that the exchange is mutually beneficial.Social exchange relationships have an implicit understanding that a history of extra-role efforts will over time be recognized, appreciated, and rewarded.

Method
A cross-sectional research design was used to examine the relationships between quality of LMX, self-esteem, and OCB.Data were collected through printed questionnaires personally administered at various banks situated in Klang Valley.The design partly avoids common method variance bias because measurements of subordinate's OCB and measurements of quality of superior-LMX were obtained from superiors and measurements of quality of subordinate-LMX and self-esteem were taken from the subordinates.

Sample and Procedure
The population was from non-supervisory employees, employed in the domestic commercial banks situated in Klang Valley.The assumption here is that the job descriptions for non-supervisory employees in all domestic commercial banks in Malaysia are similar, have minimal variation in formal responsibility, have no supervisory duties, thus can be considered as a homogenous group.Letters seeking for permission to carry out the survey at the various bank branches were sent to each of the ten domestic commercial banks.All letters were addressed to the central offices.Once the banks indicated their interest to participate in the study, the researcher contacted by phone eighty branches from participating banks and provided them with a description of the research project.Appointments were made to visit the branches to personally deliver the set of printed questionnaires to the Branch Managers or the Assistant Branch Managers.A package with two survey questionnaires one questionnaire (Set A) was to be answered by the subordinate and another one (Set B) to be answered by the superior in charge of the subordinate was distributed to participating banks.
The managers were requested to distribute four questionnaires at random to their subordinates and four questionnaires to the officer/s supervising the subordinates.An envelope was also attached to each of the questionnaire.Instructions were written on the envelope for the respondents to place the questionnaire into the envelope and to seal it before giving it back to the manager in charge.The subordinates were not made aware that their respective superiors were evaluating them.The superiors were requested to participate in this study because they are the best persons to observe the employee's daily performance; they authorize the employee's job performance and give instructions, inspect job progress, coach, and advise; and they are the experts in the job requirements as laid down in the employee's job descriptions.We presume that superiors' evaluations on their subordinates here are reasonably good measures of objective performance.
The subordinates were given questionnaire items measuring their self-esteem and quality of subordinate-LMX.The superiors were given questionnaire items rating the subordinates' OCB and quality of superior-LMX.
A total of 320 non-supervisory employees located in Klang Valley were randomly selected for participation in the study.The managers were requested to assist in the distribution of the questionnaires.The "drop-off" and "pick-up" method were employed and arrangements were made for the questionnaires to be collected from the managers two weeks from the date of the "drop-off".Of the questionnaires distributed to 320 non-supervisory employees, 315 were returned, yielding a response rate of 98%.From this total, only 300 responses were deemed usable for analysis.The total of 300 responses comprised of subordinates and we found that 118 subordinates gave responses on the questionnaires relating to these subordinates.Only surveys for which we received both the superior's and the subordinate's responses were included in the analyses.
The demographic characteristics of the respondents showed that 197 (66%) of the non-supervisory employees were females and 103 (34%) were males.Most of the respondents were married (55%), aged below 30 years (79%), highest qualifications were MCE (Malaysian Certificate of Education) (66.7%), and had been in the current position for a period of between four to seven years (46.3%).In terms of ethnicity, the sample consisted of Malays (76.7%),Chinese (10.3%),Indians (9.3%) and others (3.7%).
A total of 118 superiors returned the questionnaires given to them.The reason why we had only 118 supervisors vis-à-vis 300 subordinates was because some of these subordinates reported to the same superiors.Thus, we encountered situations where one particular superior had to answer more than one questionnaire relating to each of the subordinates being supervised by him or her.Each questionnaire was paired with its respective questionnaire answered by their subordinates.This was done through identification of the initials scribbled by the managers before they distributed the questionnaires.44.7% of the superiors were males and 55.3% were females.Majority of them were married (81.9%), aged between 31 years to 40 years (52.1%), had qualifications of bachelor's degrees (27.7%), had been working in the banks for more than seven years (71.8%) and were Malays (65.4%).

Organizational Citizenship Behavior and In-role Behavior
Previous studies have examined the construct of OCB based on either the leader's reports or peer's reports.This study examined OCB from the standpoint of the subordinate's superior.Superiors have also been the source of choice in OCB literature (Bateman & Organ, 1983;Organ & Konovsky, 1989;Podsakoff, MacKenzie, Moorman, & Fetter, 1990;Smith et al., 1983).Thus, if superiors see early attendance as an extra-role behavior, a subordinate who comes to work earlier than usual is said to engage in OCB regardless of how the subordinate sees his/her behavior (Morrison, 1994).The 24-item OCB scale developed by Podsakoff et al. (1990) was utilized to assess five dimensions of OCB proposed by Organ (1988a).These dimensions were altruism, conscientiousness, courtesy, sportsmanship, and civic virtue.Support for this scale has been cited by several previous studies of OCB (Moorman, 1991;Nasurdin, 2000;Niehoff & Moorman, 1993;Rioux & Penner, 2001).The item ratings were obtained from a 7-point Likert-type ranging from 1= "Strongly Disagree" to 7= "Strongly Agree".The ratings indicated the extent that each of the behaviors was characteristic of the employee's behavior.
Previous research suggested that OCB measures may assess in-role performance (Schnake 1991).Thus, in-role behavior scale was included in the questionnaire as a control variable so as to isolate variance in OCB measures that was not associated with performance of in-role behaviors (Moorman et al., 1993).This approach was used instead of including in-role behavior in the criterion set because the theory supporting OCB-predictor relationships was not applicable to in-role behaviors; thus in-role behaviors were not expected to be related to the predictors (Williams & Anderson, 1991).The in-role items were adapted from Williams and Anderson (1991).The 7-item in-role scale and the 24-item OCB scale were together submitted to a principal components analysis with varimax rotation (N = 300).Factors with eigenvalues greater than or equal to 1.00 were selected.Within a particular factor, items were chosen if their factor loadings were greater than or equal to 0.35.Referring to Hair, Anderson, Tatham, and Black (1998), the cut-off point of 0.35 is acceptable for a sample size of 250 and above.Most part of the items loaded cleanly into the five factors explaining a total of 69.94% of the variance.All extracted items had factor loadings greater than .35.Ten items such as--"He/she attends meetings that are not compulsory, but are considered important", "He/she adequately completes assigned duties", and "He/she is aware of how his/her behavior affects other people's jobs"--were dropped from the measure due to cross factor loadings.".Factor 1 was made up of five of the original altruism items.The name "Altruism" was thus retained.Factor 2 also contained all the five original items of sportsmanship.The name for this factor was thus maintained as "Sportsmanship.Factor 3 comprised four items reflecting in-role behaviors.This factor is named "In-role".Factor 4 had two items reflecting courtesy, one item reflecting civic virtue and one item reflecting conscientiousness.The common thread across the items was to always consider others and not to create problems with them.Hence, we named this factor as "Courtesy".Factor 5 composed of three items from the original civic virtue items and one item on conscientiousness.This factor was named "Civic Conscious".The "Conscientiousness" items from the original OCB scale loaded in both Factor 4 and Factor 5. Factor 4 (Courtesy) had one conscientiousness item, "He/she does not take extra time for breaks" and Factor 5 (Civic Conscious) had also one conscientiousness item, "He/she is confident that if he/she does his/her job honestly, he/she will be rewarded accordingly."The other conscientiousness items were dropped from further analysis due to cross loadings.The coefficient alphas for the OCB dimensions were: altruism .92,sportsmanship .89,courtesy .67,civic conscious .74,and in-role .90.

Leader-member Exchange
The quality of LMX was measured using the 12-item LMX questionnaire that was adapted from Liden and Maslyn (1998).Four dimensions made up LMX: contribution, loyalty, affect, and professional respect.Quality of LMX was looked at from the views of the subordinates and their superiors; thus we ran factors identified for quality of subordinate-LMX and quality of superior-LMX separately.
Factor analysis was carried out on the superior-LMX items and one factor emerged with eigenvalues greater than 1.00, explaining a total variance of 57.34%.All of the items were retained.We named this factor "LMX".This factor encompassed all the four dimensions of LMX found in past LMX literature.The coefficient alpha of the scale was .93.This is comparable to those reported by Liden and Maslyn (1988) that showed reliability coefficients ranging from .80 to .92.
Factor analysis was also carried out on the subordinate-LMX items and two factors emerged with eigenvalues greater than 1.00, explaining a total variance of 69.20%.Factor 1 had eight items, which was named "Devoted Appreciation".These items were reflected in the original "Affect" and "Professional Respect" in the original LMX items.Factor 2 consisted of two items, which were similar to the original items of "Contribution".The name was thus maintained.Two items "My supervisor defends my work, actions to his/her superior even without complete knowledge of the issue" and "I do not mind working my hardest for my supervisor" were dropped due to cross-loadings.The two factors: devoted appreciation and contribution, documented coefficient alphas of .94 and .51respectively.Factor 2 (Contribution) was eliminated from our analysis from hereon due to the low coefficient alpha.

Self-esteem
In this study, self-esteem was hypothesized to mediate the link between LMX and OCB.A mediating variable is an independent variable that explains all the variance in a dependent variable previously explained by another independent variable in a model.Once a mediator is introduced, there should be no unique variance explained by the first independent variable (Baron & Kenny, 1986).If employees felt the presence of self-esteem, the effect of LMX on OCB became non-significant.
Self-esteem was assessed by responses to the Rosenberg Self-Esteem Scale (1965).This measure of self-esteem is a reliable global measure (Rosenberg, 1979) and has received more psychometric analysis and empirical validation than any other self-esteem measure.It is considered to measure enduring or trait self-esteem.One of its greatest strengths is the amount of research that has been conducted with a wide range of groups on this scale over the years.This Rosenberg Self-Esteem Scale, in the strictest, most parsimonious sense, measures willingness of the individual to endorse favorable statements of him or herself.Ten items indicate this construct.Respondents respond by indicating degree of agreement or disagreement on a 7-point scale.
A principal varimax rotation was carried out and two factors emerged with eigenvalues greater than 1.0 explaining a total of 52.79% of the variance.Although the two factors had eigenvalues above 1.0,only Factor 1 was taken to represent the self-esteem measure.This factor seemed to contribute a significant percentage of variance compared to the other.Furthermore self-esteem measure by Rosenberg is an established measure that has always been unidimensional and the difference in eigenvalues between Factor 1 and Factor 2 was quite significant.The coefficient alpha of the scale was .64.

Demographic Variables
Age, tenure, and gender were also added as control variables to prevent potential confounding effects.By including them as controls, we ensured that we obtained the effects of the focal independent variables even when considering the effects of these specified control variables.

Results
Table 1 reports the means, standard deviations, and zero-order correlations for all variables.We can see that the coefficients of Pearson's correlations were wide ranging (-.05 to 0.73) across all variables in the sample.Subordinate's devoted appreciation was significantly correlated with all variables, except courtesy.Superior's LMX was significantly correlated with all variables with the only exception of self-esteem.Self-esteem was found to correlate significantly with only one variable involved in the study--subordinate's devoted appreciation.In-role behavior correlated significantly with all of OCB dimensions, devoted appreciation, and LMX but not with self-esteem.In summary, we saw that self-esteem correlated with most of the other variables at very low correlation levels.Not surprisingly, most research evidence from field studies had similarly reported a large number of low, bivariate correlations between self-esteem and measures of job characteristics, behaviors, and attitudes (Tharenou & Harker, 1982).
Our findings from Table 2 indicated the results of two hypotheses.Contrary to expectation, subordinate-LMX (devoted appreciation) was not found to have significant relationship with OCB.Thus, Hypothesis 1 was not substantiated.It was however found that LMX, which was taken from the perspective of the superior, was positively related to all four of the OCB dimensions.These dimensions were altruism, sportsmanship, courtesy, and civic conscious.Thus, Hypothesis 2 was fully supported.In essence, the results showed that quality of superior-LMX was more influential to the subordinates to perform OCB compared to the quality of subordinate-LMX which they, the subordinates themselves, have over their superiors.
Table 3 and Table 4 show the results of the tests required for mediated regression analyses.Three conditions for mediation were examined (Baron & Kenny, 1986).The first condition is that the predictors must be significantly related to the mediating element.The second condition is that the predictors must relate to the criterion in the absence of the mediator.The final condition is that, when both the predictors and mediator element are included, the direct relationship between predictors and criterion should become significantly smaller (partial mediation) or non-significant (full mediation).We regressed the mediator (self-esteem) on the predictors (devoted appreciation, contribution, and LMX).We found that only the beta weight ( = .20)for LMX was significant for self-esteem but not for devoted appreciation.Thus no support was provided for Hypothesis 3 but support was found for Hypothesis 4.
Next, we regressed the criterion on the self-esteem (Table 4).There had to be a significant relationship between the two in order to proceed to the next step of mediation testing.It was however evidenced that self-esteem did not have any significant relationship with OCB dimensions (Table 4).Hypothesis 5 was thus not substantiated.
These regression equations had provided the tests of the linkages of the mediational model.The conditions for mediation had not been met.It was not required for us to further proceed with testing the effect of self-esteem on the relationships between the predictors and the criterion.We thus concluded that the predictors (except contribution) did not cause self-esteem and that self-esteem did not cause the criterion.Both Hypothesis 6 and Hypothesis 7 were not supported.

Discussion and Conclusion
There are several implications for supervisors and organizations.From a theoretical perspective, the study looked at the impact of quality of LMX coming from two different perspectives: the superior-LMX and the subordinate-LMX.By looking at both perspectives, we could understand better the many complexities relating to the study of LMX in relation to OCB.Self-esteem was also included to see how it influenced the relationships between LMX and OCB.However results from this study did not support the hypotheses that subordinate-LMX and superior-LMX were associated with OCB through the mediation of self-esteem.
Practitioners can also benefit from the study through the understanding of how relationships portrayed by the superiors can play a major part in influencing subordinates to perform work that goes beyond their job scope without expecting to be rewarded.
In a nutshell, the results from this study help us to understand the importance of good leader-member relationship shown by the superiors to the subordinates.We found evidence in the present study that superior-LMX has direct effect to making employees perform beyond their job scope.This is especially true when the subordinates see their superiors giving them support and encouragement to them at work.In an environment in which relationships are important, superiors' emotional support and guidance appeared to assist subordinates in attaining higher levels of performance.Interestingly, when subordinates demonstrate good leader-member relationship towards their superiors, this somehow has no bearing towards making the subordinates perform OCB.Subordinate-LMX did not appear to be related to OCB.Future research is thus needed to more fully understand the role of subordinate-LMX and OCB.
The effect of self-esteem as a mediator in the relationships between subordinate-LMX with OCB and superior-LMX was not evidenced.It is plausible to say that self-esteem seems to be more closely related to motivational concerns to protect oneself rather than being affected by social elements (Schroth & Shah, 2000).However, in defense of the social exchange theory, it can be argued that subordinate-LMX and superior-LMX may affect other types of self-esteem.
The main weakness of the study is that the results pertaining to LMX and OCB may be susceptible to common method variance.Another weakness of our study was the cross-sectional design, which does not allow for an assessment of causality.As a consequence, our results are mute where issues of causality are concerned.
These limitations aside, the results of this study highlight the important and complex role of understanding LMX and OCB in the Malaysian scenario.We also envision several other avenues for future research.The most obvious avenue is to explore the relationship between subordinate-LMX and OCB and also the relationships between self-esteem and other organizational variables.It is our firm belief that the pursuit of this line of research will greatly benefit our understanding of the reciprocity that functions in relationships.
In conclusion, the results of the study suggest that within the framework of social exchange theory, superior-LMX directly influence subordinate's OCB and that subordinate-LMX does not.Superiors through their supervisory relationships seem to have a strong influence on subordinates to perform OCB.It may be because the quality of relationships from the superiors causes the subordinates to identify with and be more involved in their work.It has been said that it is primarily the superior who explains the organization to the employee and explains the employee to the organization (Farouk, 2002).The results also reveal that the subordinate's own LMX towards the superiors do not influence them in performing OCB.The influence of self-esteem is also examined as a mediator in these relationships.Self-esteem signifies an attitude that evaluates the self.Much research has found self-esteem to cause and predict behavior (1989).Interestingly, we do not find this true in this study.Self-esteem does not influence (mediate) the relationships between superior-LMX, subordinate-LMX and OCB.In our view, this result should in no way diminish the value or importance of studies using self-esteem as a variable of interest.Individual differences in responses to self-esteem may be far more social than has been assumed.

Table 2 .
Summary of Linear Regression Analysis for OCB with Predictors