The sample Public sector universities consisted of 54% males and 46% females. The private sector sample universities consisted of 59% Males and 41% females. Frequencies of the gender similar and dissimilar dyads are elaborated in Table 3.

The frequency of dyadic tenure with the current supervisor is more in the sample public sector universities as compared to private sector universities as depicted in Table 4

The dependent variable of the study is OCB which was measured through a standardized 12-item OCB scale by MacKenzie et al. (1993). This explicit self-report scale gathers information from subordinates (N=743) pertaining to the four dimensions of OCB namely: civic virtue, altruism, sportsmanship, and conscientiousness on a 7-point Likert scale. Out of the 12 items, items 1, 2, 3 measure civic virtue. Items 4, 5, 6 measure Sportsmanship. Items 7, 8, 9 measure Altruism. Items 10, 11, 12 measure Conscientiousness. Sample items include: "I attend functions that are not required but that help the university's image", "I help orient new employees even though it is not required".

The independent variable of the study is LMX which was measured through a 12-Item LMX-MDM (Leader-Member Exchange - Multi Dimensional Measure) developed by Liden and Maslyn (1998). The questionnaire elicited subordinates' (N=743) self-reports for assessing the quality of their relationship with their supervisor across four facets of LMX-MDM: affect, loyalty, contribution and professional respect using a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Sample items include: "I like my supervisor very much as a person"; "My supervisor would come to my defense if I were criticized by others in the university".

The study is based on self- reports from the subordinates because they are in a better perspective to evaluate the quality of their association with their supervisor. Further the nature of this study warrants an elicitation of the subjective experiences of subordinates' engagement in OCB.

Note. M= Mean; SD = Standard Deviation; LMX =Leader Member exchange

Table 5 shows descriptive statistics, alpha reliability coefficients, and skewness and kurtosis of OCB, CMS, and LMX along with its subscales. The table also shows the Mean and Standard deviation of the scales. All the scales are adequately reliable indicating that the scale are internally consistent. Table 1 also shows skewness and kurtosis values that shows data are normally distributed.

Hypothesis 1a posited that the subordinates in gender similar LMX dyads would report a high quality of LMX in comparison to those in gender dissimilar LMX dyads. Hypothesis 1b posited that female subordinates in the gender similar LMX dyads will report a high quality of LMX in comparison to male subordinates in gender similar LMX dyads. The result of one way ANOVA for the public and private sector universities is depicted in Table 6 and Table 7 respectively.

Note: M=Mean; SD=Standard Deviation; LMX=Leader Member Exchange

Table 6 shows the group mean differences on OCB and its subscales among similar and cross-gender dyads for public sector universities. Results show that there are significant differences among four types of dyads on LMX. The results further demonstrate that among the gender similar dyads, female gender similar LMX dyads exhibit significant differences (M=66.62, SD=11.59) with the male gender similar dyads (M=61.38, SD=15.0), as well as with both male gender dissimilar dyads (M=62.90, SD=14.46) and female gender dissimilar dyads (M=62.87, SD=14.14) on LMX quality (F (3, 483) =2.91, p = .03).

Note: M=Mean; SD=Standard Deviation; LMX=Leader Member Exchange

Table 7 shows the group mean differences on OCB and its subscales among similar and cross-gender dyads for private sector universities. Results show that there are significant differences among four types of dyads on OCB and its subscales. The results further demonstrate that among the gender similar dyads, Male gender similar LMX dyads (M=65.56, SD=12.36) exhibited significant differences not only with the female gender similar dyads (M=57.60, SD=10.04) but also with both male gender dissimilar dyads (M=65.44, SD=10.97) and female gender dissimilar dyads (M=61.69, SD=12.94) on LMX quality (F (3, 252) =3.77, p=.01). The ANOVA results for the public and private sector universities provided partial support for hypotheses 1a and 1b.

Note. OCB= organizational Citizenship Behavior, civic =Civic Virtue, sports=Sportsmanship, Altr=Altruism, con=Conscientiousness

Table 8 shows the group mean differences on OCB and its subscales among similar and cross-gender dyads for public sector universities. Results show that there are non-significant differences among four types of dyads on OCB and its subscales across both public and private sector universities thereby rejecting hypothesis 2.

Note. OCB= organizational Citizenship Behavior, civic =Civic Virtue, sports=Sportsmanship, Altr=Altruism, con=Conscientiousness

Table 9 shows the group mean differences on OCB and its subscales among similar and cross-gender dyads for private sector universities. Results show that there are non-significant differences among four types of dyads on OCB and its subscales

Multiple regression analyses were conducted for sample private sector universities to assess each component of the proposed moderation model. The results are presented in Table 11 and figure 2. The results exhibits the association between LMX and OCB (a-path) and the interaction effect of LMX and dichotomous gender dyads on OCB.

The overall model was insignificant F (3,252) = 4.00 p=.008, R² =.21. Under the condition of the significance of gender similar dyads (p=.00) his model significantly explained 21% of the variance in OCB. The resultant slope of the moderation analysis is exhibited in Fig 2

Table 11 illustrates that LMX exhibited an insignificant effect on OCB (β = .084, t (252) = 1.36, p = .174). Gender similarity exhibited an insignificant effect on OCB (β = -5.51, t (252) = -1.04, p = .298). The interaction effect of LMX and gender similarity was insignificant (β = .087, t (252) = 1.06, p= .288).

The current study adopted a gender focus in analyzing the relationship between LMX and OCB primarily owing to the fact that gender is a pertinent factor that has the propensity to render influence on social processes within the organizational contexts (Wayne et al., 1994). The current study contextualized in the higher education sector of Pakistan is anchored in the similarity-attraction paradigm which provided a reference for theorizing the relationship of gender similarity in relation to LMX and OCB. Therefore while the study sought to contribute to the knowledge base pertaining to the similarity attraction paradigm, it also aimed at elucidating the mechanism through which LMX predicts OCB.

Hypothesis 1a posited that the subordinates in gender similar LMX dyads would report a high quality of LMX in comparison to those in gender dissimilar LMX dyads. Hypothesis 1b posited that female subordinates in the gender similar LMX dyads will report a high quality LMX in comparison to male subordinates in gender similar LMX dyads. The results demonstrated that in the sample public sector universities, female gender similar LMX dyads reported a high quality of LMX not only in comparison with their male counterpart but also with both male and female gender dissimilar dyads. Interestingly for the sample private sector universities, contrary to our hypothesis male gender similar LMX dyads reported a high quality of LMX not only in comparison with their counterpart female gender similar dyads but also with both male and female gender dissimilar dyads.

The equivocal findings yielded by the study with reference to hypothesis 1a, 1b are also supported by the literature (Thomas, 2013; Lo et al., 2009). The findings, equivocal as they may be, yet brings forth a pattern of gender oriented differential treatment by the supervisors towards their subordinates across the selected public and private sector universities of Rawalpindi and Islamabad. While the finding of high quality LMX reported by male subordinates in gender similar dyads in male dominated private universities lend support to hypothesis 1a; it also alludes to a 'boys club' with a tendency towards gender differential treatment. In the similar vein while the high quality LMX reported by female subordinates in gender similar dyads in male dominated sample public universities lend support to hypothesis 1a and 1b, it also alludes to a female support network whereby female supervisors developed an trusting relationship with female subordinates. More studies are warranted to elucidate the nature of relationships among both the male and female gender similar dyads across the private and public sector sample universities.

In the sample public sector universities within the female gender similar dyads, 68.8% female faculty reported a dyadic tenure that ranged from 1-5 years with the current female supervisor while 31.1% reported a dyadic tenure that ranged from 6-10 years with the current female supervisor. In the sample private sector universities within the female gender similar dyads, 82.6% female faculty reported a dyadic tenure that ranged from 1-5 years with the current female supervisor while 17.3% reported a dyadic tenure that ranged from 6-10 years with the current female supervisor. It may well be possible that female subordinates in gender similar dyads in the sample private universities had a relatively short dyadic tenure with their supervisors in comparison to those in the public sector sample. As a consequence the affect based trust between the female supervisor and female subordinates in LMX dyads at the private sector universities may not have been adequately developed to a point where it manifested as OCB. Trust develops over a period of time (Harrison et al., 1998) before its effects transpires at the organizational level. The finding that the female subordinates in gender similar dyads in private sector did not report high quality of LMX in sample private sector universities may be explained by the social identity theory (Gaertner & Dovidio, 2008). As per the theory, the female supervisors owing to their numerical minority may have identified themselves with the male majority for maintaining their positive identity. By doing so they may have enacted a leadership role non-stereotypically in an agentic manner like their male colleagues. The female subordinates consequently could not identify themselves with such supervisors and hence the trusting LMX relationship could not have forged which could subsequently motivate them to engage in OCB.

The current study failed to provide support for hypothesis 2 that female subordinates in gender similar LMX dyads will report high OCB in comparison to their counterparts across the sample public and private universities. This is evident from the fact that no mean differences were found in the OCBs of both the male and female gender similar dyads. Therefore the study rejects the hypothesis 2. While the finding is in conformity with a study by Thomas (2013) in the US context it however contrasted with a s study by Tourigny et al. (2017) in the context of Chinese manufacturing industry where gender similarity in female LMX dyads predicted OCB. Findings of the present study also contrasted with a study by Farrel and Finkelstein (2007).

The study reported mixed findings for hypothesis 3 which posited a moderation effect of gender similarity in a relationship between LMX and OCB. While the study reported a significant moderation effect of gender similarity in in the sample public sector universities, it however alluded to an insignificant moderation effect of gender similarity for the sample private sector universities thereby discounting the similarity effect of gender of both the supervisor and the subordinates in intervening the effect of LMX on OCB. This finding is consistent with studies conducted elsewhere (e.g. Soldner, 2009; Estiri at al., 2017; Brownell, 2010).

As the study contributed findings to the relevant knowledge base, in due process it also made a practical contribution by validating the psychometric properties of the scales used in the study. While this was warranted to ascertain the relevance of scales in Pakistani context, yet it was also carried out in response to the evoked urge (Liden & Maslyn, 1998) for ascertaining the generalizability of the construct across various contexts. A pertinent finding by the current study in this regard was the validation of unidimensionality of LMX-MDM rather than the multidimensionality as initially conceptualized by Liden and Maslyn (1998). This finding corroborate with that of other studies (e.g. Joseph, Newman & Sin, 2011; Scherbaum et al. 2006) whereby the factor analysis resulted in the loading of all four facets of LMX-MDM onto a single factor, thereby suggesting strong item correlation.

The findings of the study must be interpreted by taking into consideration the limitations of the research. The current study employs a cross-sectional design and therefore imbibes the limitations that are inherent in cross-sectional designs. It is suggested that for conclusive findings longitudinal study designs must be employed for examining the interactive effect of gender in influencing the quality LMX and its subsequent effect in engagement of OCB. While the subordinate self-reports form the basis of present research; it would be worthwhile to incorporate the perceptions of the immediate supervisors as well for more conclusive findings.