Social identity theory and leader–member exchange: individual, dyadic and situational factors affecting the relationship between leader–member exchange and job performance

Purpose – This study aims to integrate social identity and leader–member exchange (LMX) theory to investigate the processes and boundary conditions around LMX–performance relationships. Through the application of two leader–follower subsamples, the authors test three main objectives. What is the effect of multi-dimensional dyad value-congruence on LMX and how does congruence on these dimensions differentially influence leader and follower perceptions of LMX? In a subsample of followers including supervisor-rated performance, the authors develop a model that examines how individual values moderate the effect of dyad contact on supervisor-rated job performance mediated by follower LMX. Design/methodology/approach – The participants for this study include graduate and undergraduate social work students who were taking part in a one-year work placement within a social work organization as well as their immediate supervisors. Across a four-month period, participants filled out measures of their supervisor contact, work values and LMX. Supervisor-rated performance was also included. Findings – Findings from the dyadic subsample show that growth value congruence is a predictor of follower-rated LMX, with value congruence across all values having no effect on leader-rated LMX. Within a subsample of followers, findings suggest that follower-rated LMX mediates the relationship between dyad contact and supervisor-rated job performance, with individual work values moderating this effect. © Gregory Thrasher, Marcus Dickson, Benjamin Biermeier-Hanson and Anwar Najor-Durack. Published in Organization Management Journal. Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial and non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at http://creativecommons.org/licences/by/4.0/legalcode Identity theory and leader– member exchange

There has been a great deal of research supporting the idea that employees' relationships with their supervisors predict job performance (Avolio, Walumbwa, & Weber, 2009;Markham, Yammarino, Murray, & Palanski, 2010). Within the literature on leader-follower interactions, these supervisor-subordinate relationships are often conceptualized as leadermember exchange (LMX) relationships. LMX involves the formation of a job concept that revolves around aspects of both job tasks and relational components of the position (Graen, Novak, & Sommerkamp, 1982). This definition has mainly evaluated the relationship from the subordinate's perspective; however, more recent research has examined the leader's perspective, as well as the dyadic interplay of leader and follower perspectives on these relationships (Gooty, Serban, Thomas, Gavin, & Yammarino, 2012). Further, much of LMX research to date has focused on pre-existing dyads and there is a need for more research on factors that affect both members' perspectives in newly formed dyads (Markham et al., 2010;Park, Sturman, Vanderpool, & Chan, 2015).
Understanding how relationships develop among newly formed dyads requires an understanding of the psychosocial process that underlies workplace interactions. Social identity theory offers an established framework of empirically supported propositions that can be directly applied to questions around LMX. Social identity theory states that individuals naturally categorize themselves and others into multiple categories based on a variety of identities at any given time (Ashforth & Mael, 1989;Tajfel & Turner, 1979). Individuals use social categorization to understand their own identity, the identity of others in their environment and the identification of in-and out-group members (Tajfel & Turner, 1979). Individuals are then motivated to engage in behaviors that solidify membership within valued identity groups through interactions with other ingroup members (Ashforth & Mael, 1989;Tajfel & Turner, 1979). Ashforth and Mael (1989) highlight that within organizations, social identification can manifest through various targets and levels based on an employee's experience and salient identities. For example, the authors highlight that although traditionally discussed at the group level (e.g. Canadian, an employee of university), identification can be associated with a person (e.g. mother, friend and leader) or a reciprocal relationship (e.g. teacher-student and leader-follower).
In line with the broader application of social identity theory to organizations, research on LMX has identified several factors surrounding the dyad, the situation and the individual that align with social identity theory, leading to positive exchange relationships between leaders and followers. Considering the characteristics of the dyad, research on leaderfollower fit has found that job performance is positively affected by shared values between leaders and followers (Markham et al., 2010). Situational characteristics surrounding the dyad have also shown to affect levels of LMX, with factors such as frequency and intensity of dyad contact playing a role (Gajendran & Joshi, 2012;Sin, Nahrgang, & Morgeson, 2009). Research has also identified a host of individual characteristics (e.g. personality and work OMJ 17,3 values) that affect exchange relationships (Markham et al., 2010;Nahrgang, Morgeson, & Ilies, 2009). Although many of these relationships have been studied in isolation, social identity theory would posit that the development of relationships is a complex process that is influenced by the cumulative effect of these dyadic, individual and situational factors.
Beyond the factors the predict LMX, there is a large body of research supporting the robust positive effect that quality exchange relationships have on multiple dimensions of job performance (Gerstner & Day, 1997;Martin, Guillaume, Thomas, Lee, & Epitropaki, 2015). Although the relationship between LMX and job performance has been well established within the literature through multiple empirical studies and meta-analyses (Martin et al., 2015), there is still a need to further understand the mechanisms and boundary conditions involved in this process. Here again, social identity theory offers an explanatory mechanism for how relationship quality not only develops through shared experiences, but how this relationship quality will in turn influence evaluations of members' behaviors through supervisor-rated job performance.
The current study applies propositions made by social identity theory to integrate and extend research on the relationship between LMX and job performance through two separate subsamples of social workers within an educational placement. Across these two subsamples, we make inferences around the dyadic effects (Sample 1) and the situational and individual processes and boundary conditions around LMX-performance relationships (Sample 2). Through the application of these two subsamples, we contribute to the literature in three primary ways. First, we investigate the effect of multi-dimensional value congruence on perceptions of LMX from both leader and follower perspectives. Second, we investigate a model that examines the distal predictors and boundary conditions of LMXperformance relationships. Third, we test hypotheses using multi-source, time-lagged data, within newly formed leader-follower relationships.

Brief introduction to leader-member exchange
The construct of LMX grew out of the study of dyadic leader-follower workplace roles and how workers use these roles to meet the social and task-based demands of their job (Dansereau, Graen, & Haga, 1975;Graen, 1976). Early research on dyadic leader-follower relationships began by focusing on the differential relationships leaders develop with their multiple followers (Graen, Dansereau, & Minami, 1972). The study of these dyadic relationships eventually grew into a conceptualization where the focus was based less on differential relationships and more on the exchange process that takes place between leaders and their followers (Dansereau et al., 1975). From this conceptualization, it became important to understand how these roles are defined, as well as how they are affected by both situational and individual factors (Dienesch & Liden, 1986). Further, extensive metaanalytic evidence points to a variety of individual and situational predictors, attitudinal mediators and outcomes that are related to LMX (Gerstner & Day, 1997;Martin et al., 2015;Rockstuhl, Dulebohn, Ang, & Shore, 2012).
Regarding the relationship between the quality of LMX relationships and various organizational criteria, there is a large body of research tying the construct to a variety of desired outcomes (Gerstner & Day, 1997;Martin et al., 2015;Rockstuhl et al., 2012). For example, early research on the outcomes of LMX found interventions aimed at training managers on the LMX model (e.g. active listening, exchanging mutual resources and practicing one-on-one session) had greater effects on the objective job performance of government administrative workers than on groups trained in job design or placebo groups (Graen et al., 1982). Similar positive effects have been seen between LMX and supervisorrated performance of newly hired employees (Liden, Wayne, & Stilwell, 1993), creative Identity theory and leadermember exchange performance behaviors (Olsson, Hemlin, & Pousette, 2012) and group level performance (Zhang, Waldman, & Wang, 2012).
Although these findings highlight the benefit of quality LMX relationships for a variety of outcomes, other research has focused on the underlying mechanisms of LMX and the reciprocal nature of exchange relationships. In a meta-analysis examining the relationship between LMX and job performance, Martin et al. (2015) found that it is through trust, empowerment, motivation and job satisfaction that LMX leads to improved performance. Other research on LMX and various outcomes (i.e. career success, performance, cooperation and organizational citizenship behaviors) has examined moderators of the relationship, such as personality, political skill and job characteristics (Bauer, Erdogan, Liden, & Wayne, 2006;Breland, Treadway, Duke, & Adams, 2007;Sekiguchi, Burton, & Sablynski, 2008).
This body of research clearly shows that strong LMX relationships can affect organizations positively from the bottom up. However, the literature would benefit from a deeper examination of how various factors (dyadic, situational and individual) interact with the LMX-job performance process (Gooty et al., 2012;Liden et al., 1993;Markham et al., 2010;Nahrgang et al., 2009).
Dyadic effects: value congruence and leader-member exchange relationships Social identity theory posits that individuals evaluate their own identities and the identities of others through social categorization based on a wide variety of individual and social factors (Tajfel & Turner, 1979). A key component of this categorization is that our social identities are a strong predictor of our behaviors, goals and relationships (Ashforth & Mael, 1989). For these reasons, LMX development can be discussed as a process where trust and exchanges develop based on perceived in-group similarities. Research on the predictors of LMX has generally supported this thinking with several factors, both individual and situational, positively affecting the quality of leader-follower exchange relationships. Examples include extraversion , positive affect (Bauer & Green, 1996), tenure (Bauer et al., 2006), job characteristics (Dunegan, Duchon, & Uhl-Bien, 1992;Sekiguchi et al., 2008) and contact within the dyad (Sin et al., 2009;Gajendran and Joshi, 2012). Just as LMX is a dyadic phenomenon, predictors can be conceptualized as dyadic in the form of congruence. Research on congruence and LMX has primarily examined factors such as big five personality similarity (Bernerth, Armenakis, Feild, Giles, & Walker, 2008) and work value congruence (Ashkanasy & O'Connor, 1997;Erdogan, Kraimer, & Liden, 2004). Although important, this research has largely been conducted on dyads with firmly developed relationships. As LMX has been defined as a process that develops over several trust-based testing episodes between two individuals (Sin et al., 2009), understanding the effect of congruence on both leaders and followers through the lens of social identification offers an important theoretical contribution.
In their meta-analysis on multi-level issues in LMX research, Gooty et al. (2012Gooty et al. ( , p. 1083) state "A leader and follower who share similar mindsets, work values and/or experience momentous relationship-building events together might develop such convergence quicker than leader-follower dyads that do not." Giberson, Resick, Dickson, Mitchelson, Randall, and Clark (2009) mirror this thought, suggesting that it is the values of lower-level leadership along with the personality and values of top-echelon leaders that affect how followers will perceive their place within an organization. Schleicher, Hansen, and Fox (2010) formally define work-values as prioritized, transituational and guiding beliefs that persons hold about desired end states or behaviors at work. Situations in which groups of employees hold congruent values have been shown to lead to positive outcomes at all levels of the organization. For example, Brown and Treviño (2006) OMJ 17,3 looked at value congruence at the team level and found it to moderate the relationship between charismatic leadership and deviant workplace behaviors. At the organizational level, Gutierrez, Candela, and Carver (2012) found person-organization value congruence to predict levels of organizational commitment in nurses. At the individual level, Markham et al. (2010) found that when supervisor and subordinate agreement on LMX relationships was present, work-value congruence between the dyad members predicted job performance among media service provider employees. The relationship between value congruence and work outcomes is an important factor for organizations to consider, but how value congruence affects the development of workplace relationships needs further evaluation.
Followers are not just indirectly affected by the values of their leader. Workplace relationships tend to move from individual interests to common interests based on factors such as trust, respect and obligation to each other (Avolio et al., 2009). Early in the relationship development process, these similarities are more likely to be judged on non-behavioral criteria (i.e. values), as behavioral criteria have not yet been established (Dienesch & Liden, 1986). Within the literature on work values and LMX, several different taxonomies have been applied (Ashkanasy & O'Connor, 1997;Markham et al., 2010). For example, Markham et al. (2010) apply the organizational culture profile (Chatman, 1991) to investigate congruence on 54 value-related items, as where Ashkanasy and O'Connor (1997) apply multi-dimensional scaling analysis to Schwartz and Bilsky's (1987) taxonomy that evaluates instrumental and terminal values (e.g. freedom, achievement, mateship, obedience and coping). As the current study focuses on the development of LMX within newly formed dyads, we opted for a taxonomy that focuses on values likely to be especially salient for new career individuals. As such, we apply a multi-dimensional values taxonomy of growth, status and security values (Leuty & Hansen, 2011;Manhardt, 1972). Specifically, these three values have been defined (Leuty & Hansen, 2011;Meyer, Irving, & Allen, 1998) as: (1) Growth: Work provides feelings of accomplishment, development of knowledge and skills and making a social contribution.
(2) Status: Work provides advancement to higher positions, supervising others and respect from others. (3) Security: Work provides security, ample leisure time and comfortable conditions. The role of job level, or leader-follower status, is an important topic in LMX research, as exchanges and trust-building episodes are often viewed differently depending on the individuals' role in the relationship (Martin et al., 2015;Gooty et al., 2012). Research on LMX suggests that early on in relationships, leaders and followers are further apart on their perceptions of LMX, with convergence occurring over time ). More specifically, early relationships are generally characterized by lower levels of leader-rated LMX and higher levels of follower sensitivity to leader traits Sin et al., 2009).
Considering the potential differences in how value-congruence influences the LMX of leaders and followers, we offer two distinct hypotheses based on the rater: H1a. There is a positive relationship between leader-follower value congruence and follower-rated LMX.
H1b. There is a positive relationship between leader-follower value congruence and leader-rated LMX.

Identity theory and leadermember exchange
Leader-member exchange and performance: the role of situational and individual factors There are a host of studies, empirical and meta-analytic, showing positive relationships between LMX and job performance outcomes such as task and supervisor-rated performance (Gerstner & Day, 1997;Martin et al., 2015), citizenship behaviors (Ilies, Nahrgang, & Morgeson, 2007), creative behaviors (Olsson et al., 2012) and team performance (Zhang et al., 2012). However, research on the processes and boundary conditions of such relationships has been less common. Supporting this assertion, Martin et al. (2015, p. 32) discuss the limits of the current literature, stating "[. . .] there are surprisingly few empirical studies that have directly tested mediational models of LMX despite the frequent calls in the literature." Dyad contact as a distal predictor of performance From an organizational socialization perspective, individuals forge identities at work through "verbal and non-verbal interactions" with individuals (Ashforth & Mael, 1989, p. 27). Research on the relationship between dyad interactions and LMX has predominantly focused on the role of frequency and intensity of contact in predicting positive LMX relationships (Gajendran and Joshi, 2012;Sin et al., 2009). The relationship between dyad contact and LMX is described as existing through a higher frequency of testing episodes that allow for the establishment of trust (Sin et al., 2009). As more time is spent within a dyad, and relevant testing episodes play out, other identities will become salient as a result of increasing verbal and non-verbal interactions. For example, learning that someone has a family may lead to a shared identity of "father or mother." Learning about someone's goals will give insight into potentially shared identities around career values. As such, it is likely that the more time spent within a dyad, the more diverse the available shared social identities become.
Through the creation of shared identities, we argue that dyad contact has a necessary but insufficient effect on supervisor ratings of job performance. There are situational factors that may cause leaders to spend time with both high-and low-performing employees. Especially early on in employment, a leader may spend time correcting a poor employee's behavior, while spending just as much time fostering an effective employee's good behaviors. We argue that dyad contact has an indirect effect on performance as mediated by follower perceptions of LMX based on the development of shared identities created by contact. Although one could argue the reciprocal effects of LMX on dyadic contactwe suggest that within newly formed dyads, contact is likely a precursor to LMX during the role routinization phase. As such, we hypothesize dyad contact leading to LMX and in turn performance outcomes: H2. The relationship between dyad contact and job performance is mediated by LMX.

Moderating effect of individual values
Research on the boundary conditions of the LMX-job performance relationship has identified several factors, both individual and situational, that act as moderators (Bauer et al., 2006;Breland et al., 2007;Dunegan et al., 1992;Haynie, Cullen, Lester, Winter, & Svyantek, 2014;Sekiguchi et al., 2008). In line with the conceptualization of LMX as a relationship based on shared identity and the exchange of resources, studies examining potential moderators consistently show that quality LMX relationships have the potential to OMJ 17,3 protect against factors that may otherwise lead to lower performance outcomes resulting from an increased dependence on their immediate leader.
For example, individuals high in extraversion have been shown to experience increased levels of mentoring and organizational knowledge (Turban, Moake, Wu, & Cheung, 2017), suggesting these individuals are less dependent on their immediate leaders. Research on the moderating effect of extraversion on the LMX-performance relationship shows that when LMX quality is high, extraversion does not affect task performance; however, when LMX quality is low, individuals low in extraversion experience much worse performance outcomes (Bauer et al., 2006). This effect is further mirrored for individuals low in political skill (Breland et al., 2007) and empowerment (Harris, Wheeler, & Kacmar, 2009). For example, Breland et al. (2007) found that individuals who rated themselves as having low political skill and LMX displayed the lowest levels of subjective career success. However, the authors found that when LMX was high the effect of political skill on subjective career success was drastically decreased. As political skill is defined as an ability to influence others toward one's political goals (Ahearn, Ferris, Hochwarter, Douglas, & Ammeter, 2004), low political skill individuals may be more reliant on their immediate leaders as they are unable to establish external organizational networks. Similar protective effects are also seen for individuals who are in situations that increase dependence on one's immediate leader, with high job-embeddedness (Sekiguchi et al., 2008) and low leader relational identity (Chang & Johnson, 2010) leading to much worse performance outcomes in the presence of poor quality LMX relationships.
Individuals who value different components of work may also experience different performance outcomes based on their LMX relationships because of dependence on an immediate leader. As previously noted, workplace values are stable beliefs individuals hold about desired workplace outcomes and the behaviors necessary to achieve them (Schleicher et al., 2010). Within the current study, we have adopted a taxonomy of growth, status and security values (Leuty & Hansen, 2011;Manhardt, 1972), and based on their definitions, it would seem that some values, and resulting goals, are more dependent on an immediate leader (e.g. growth and status). While security values are rooted in organizational factors (e.g. compensation systems), both growth and status are products of an individual's leader. For example, status within an organization is largely dependent on receiving promotions or other extrinsic rewards. Followers who value status, likely look to their leaders to offer guidance and goals around achieving status. As such, for individuals who value status, LMX relationships likely offer an avenue for accumulating the resources and opportunities necessary to achieve status increasing the effect LMX has on supervisor-rated performance. Similarly, growth within newly established roles is largely a function of opportunities offered by one's immediate leader. LMX is largely rooted in trust, and followers who are trusted may be more likely to take advantage of growth opportunities leading to increased supervisor-rated performance. Based on the important role supervisor plays in satisfying growth and status values, we hypothesize the following: H3a. Growth values moderate the LMX-performance relationship, such that individuals high in growth values experience worse performance outcomes when LMX is low.
H3b. Status values moderate the LMXÀperformance relationship, such that individuals high in status values experience worse performance outcomes when LMX is low.

Identity theory and leadermember exchange
Method Participants Full institutional review board approval was received from the institution where the primary research was conducted. Participants for this study include graduate and undergraduate social work students who were taking part in a one-year work placement within a social work organization. This sample comes from a high-fidelity situation where aspiring social workers are engaged in an employment role under the supervision of a certified social worker. Placement students are granted the authority to conduct professional staff functions, including independent management of cases, along with other responsibilities that a non-student social worker would normally engage in. As such, even though the followers in this sample are enrolled as students, their roles during the sampling of this study reflect that of an entry-level social worker under the supervision of an experienced manager. Although supervisors were responsible for multiple placement students, each supervisor in the matched sample was randomly assigned one of their placement students. The supervisor was instructed to reference this individual when responding to measures about their follower. Although this method decreased our sample size, it allowed us to ensure that all dyads were independent of one another. Along with their supervisors, the placement employees comprise two subsamples used in the current study: (1) a matched sample of independent leaderÀfollower dyads; and (2) a larger unmatched sample which included follower self-report data that was linked with the end of semester supervisor-rated job performance.
The final total sample included 155 students and 113 supervisors. For our hypotheses surrounding multi-dimensional value congruence (H1a and H1b), we were limited to a sample of 34 matched dyads. We acknowledge that our sample of matched dyads is relatively smallpreventing us from including value congruence in the larger moderated mediation model. As discussed in the results and discussion sections, this small number of dyads displayed strong effect sizes for the relationships tested in H1a and H1bsuggesting adequate power. H2, H3a and H3b were tested using the unmatched sample of placement students (N = 155). The demographic characteristics of the matched and unmatched samples of placement students and supervisors are shown in Table 1.

Procedures
After initial information sessions with all students during their class orientations, students and supervisors were contacted via email and invited to take part in a survey aimed at examining the role of work values in the development of work relationships. Students and supervisors were emailed separately and matched later using unique codes given to dyad pairs. This method allowed for dyad matching, as well as the inclusion of unmatched individuals in the non-value congruence analyses. Within the survey, both leaders and followers were asked to complete a work value inventory consisting of 21 work values, a measure of LMX, questions about their weekly contact within the dyad and demographics items. All survey data for both leaders and followers (work values, LMX and dyad contact) were collected approximately three months into the one-year placement. Supervisor-rated performance data were obtained for followers based on the students' first-semester review, which occurs approximately four months into the placement. This time frame is in alignment with previous time lags applied in research on LMX development (8-12 weeks) (Bauer & Green, 1996;Nahrgang et al., 2009). Although we acknowledge that 16 weeks is longer than this previous period, social work students were evaluated by their placement supervisors at the end of the semesterlimiting our control of the exact timing of this evaluation.

Measures
Leader-member exchange. LMX was evaluated using Liden and Maslyn's (1998) LMX measure. The measure consists of 12 items measured on a 1-7 Likert scale ranging from strongly disagree to strongly agree. The measure evaluates personal relationships with supervisors with statements such as "I like my supervisor very much as a person" and "I admire my supervisor's professional skill." For the unmatched sample, this measure displayed an internal consistency of 0.94 for followers and 0.90 for leaders. For the matched sample, the LMX scale showed an internal consistency of 0.92 for followers and 0.93 for leaders.
Work values. Work values were measured using Manhardt's (1972) Work Values Inventory. In an evaluation of the validity of values measures, Leuty and Hansen (2011) found this measure to be valid across multiple dimensions of work values, such as working environment, being challenged, opportunities for income and support. The measure was also found to be comparable to other popular values measures (Leuty & Hansen, 2011), suggesting it is an acceptable measure for this study. The measure asks each participant to rate 21 work-related values on how important they are to the participant's ideal job, using a scale of 1-5 that ranges from unimportant to very important. This measure has also been shown to have a three-factor structure, measuring values of security, growth and status. Sample items include, "provides comfortable working conditions" (security), "encourages continued development of knowledge and skills" (growth) and "is respected by other people" (status). We conducted two CFAs to evaluate the three-factor value structure against a onefactor structure. Results of this chi-square difference test suggest that the three-factor model (x 2 = 416.56, df = 186, p < 0.001, RMSEA = 0.09, CFI = 0.75, TLI = 0.72) of values fit the data better than a one-factor model (x 2 = 505.00, df = 189, p < 0.001, RMSEA = 0.10, CFI = 0.66, TLI = 0.62, Dx 2 = 88.44, Ddf = 3). This measure displayed an internal consistency of Identity theory and leadermember exchange 0.86 for followers and 0.78 for leaders within the unmatched sample. Within the matched sample, the scale showed an internal consistency of 0.75 for followers and 0.80 for leaders.
Job performance. Supervisor-rated job performance data were gathered using the Field Instructor Assessment of Student Competency (FIASC). The FIASC is an institutionally developed and nationally accredited assessment that evaluates the extent to which placement students achieve competencies in their field placement. This form is completed each semester by social work field instructors who supervise social work students in field placement. Field instructors rated students' level of competency on indicators that align with essential practice behaviors. Field instructors rated the student on each behavior as (4) exceeds competency, (3) meets competency, (2) marginally meets competency and (1) does not meet competency. The current study applied the supervisor-rated performance evaluations from placement students' first-semester review, which occurs approximately four months into the one-year placement. The FIASC displayed an internal consistency of 0.93.
Additional variables. Along with the variables of interest, both leaders and followers were asked questions regarding their age, race, gender (male = 1 and female = 2) and hours of dyad contact per week. Followers-rated average hours spent together per week were used as the dyad contact variable in subsequent analyses. Followers were asked questions about their year in the program and if they have engaged in a placement before. A dummy-coded leader-follower status variable (follower = 1 and leader = 2) was also created to be used in a supplemental interaction analysis.
Data analysis. To analyze the hypotheses concerning value congruence, each matched dyad member's value scores across the three dimensions were correlated with the other dyad members, to achieve a congruence score. We acknowledge that polynomial regression and surface modeling is accepted best practice for evaluating congruence (Edwards, 1994) and are aware of the limitations of our application of a difference score method. Because of our small sample of matched dyads, we were limited in power to conduct this analysis. However, we believe that our current analysis does offer insights into the differential effects of value dimensions across leader-and follower-rated LMX.
As discussed in Ashkanasy and O'Connor (1997), this method of jointly evaluating independent values ratings offers a more objective assessment of value congruence than asking dyads to rate their similarity with members. The value congruence index was then used in all subsequent regression analyses applying the matched sample to examine all questions involving value congruence. Subsample one analyses around value congruence and LMX (H1a and H1b) were conducted using regression in the base package of R. Subsample two analyses (H2, H3a and H3b) applied follower-rated LMX and supervisorrated performance and were tested within a bootstrapped (N = 1,000) path analyses in Mplus 8 (Muthén & Muthén,1998/2017. Tables 2 and 3. All effects reported as unstandardized coefficients.

Descriptive statistics and correlations for both the matched and unmatched samples can be seen in
Value congruence and leader-member exchange (Sample 1) H1a and H1b state that there is a positive relationship between value congruence and follower-and leader-rated LMX. Results suggest that, there is a significant positive relationship between follower-rated LMX and growth value congruence (b = 0.91, SE = 0.41, p = 0.03), and non-significant relationships between follower-rated LMX and status value congruence (b = 0.27, SE = 0.43, p = 0.54) and security value congruence (b = 0.49, OMJ 17,3 SE = 0.31, p = 0.12)offering mixed support for H1a. Results show a non-significant relationship between leader-rated LMX and growth value congruence (b = À0.78, SE = 0.42, p = 0.07), status value congruence (b = À0.37, SE = 0.43, p = 0.39) and security value congruence (b = 0.11, SE = 0.32, p = 0.75)failing to support H1b. To further examine the above differences, we conducted supplemental analysis to test the interaction between leader-follower status and value congruence on LMX. We found that the relation between value congruence and LMX is moderated by leader-follower status. This moderation was statistically significant for growth value congruence (b = À1.64, SE = 0.59, p < 0.01), but not for status (b = À0.64, SE = 0.61, p = 0.30) or security value congruence (b = À0.39, SE = 0.44, p = 0.39). Taken together, these results suggest that shared growth values are uniquely important for followers' perceptions of LMX (Figure 2).
Leader-member exchange-performance model (Sample 2) H2 and H3 pertain to our model proposing that while dyad contact has an indirect effect on supervisor-rated job performance mediated by follower-rated LMX, this relationship is moderated by follower work values. These hypotheses were tested within an extended and separate sample from that used to test the value congruence hypotheses. H2 stating that dyad contact has an indirect effect on job performance mediated by LMX was supported (b = 0.007, SE = 0.04, CI = 0.003, 0.011 p < 0.01). H3a and H3b state that follower individual growth and follower status values moderate the LMX-performance relationship. H3a states that valuing growth moderates the LMX-performance relationship, with high growth Identity theory and leadermember exchange values resulting in worse performance in the presence of low LMX. This hypothesis was not supported (b = À0.04, SE = 0.08, p = 0.59). H3b states that valuing status moderates the LMX-performance relationship, with high-status values resulting in worse performance in the presence of low LMX. This hypothesis was supported (b = 0.13, SE = 0.06, p = 0.02), with individuals high in status values exhibiting weaker LMX-performance relationships than those low in status values (Figure 3).

Discussion
We apply a social identity theory framework to advance the literature on LMX and job performance by first examining the role that dyadic value congruence plays in fostering quality LMX relationships. In an extended sample, we then assessed the extent to which LMX relationships act as a mediator between dyad contact and job performance ratings. Finally, we tested whether this process is moderated by followers' work values. Overall, we offer a test of LMX theory that applies established psychological theory to test multi-dimensional/multitarget value congruence effects in our first sample and then lagged situational effects and LMX to supervisor-rated performance effects in our second sample. Through this approach, the current study offers several contributions to the literature on LMX and job performance. First, in our dyadic leader-follower sample, we extend propositions made by social identity theory around value congruence and LMX (Ashkanasy & O'Connor, 1997;Markham et al., 2010) by offering support for a multi-dimensional and multi-target approach to questions of values and LMX. Second, within our larger non-dyadic sample, we offer insights into previous conflicting findings around dyad contact and LMX, by offering support for the indirect effect of dyad contact on supervisor-rated performance via LMX. Third, within this second sample, we also Theoretical implications Factors affecting leader-member exchange. We found that growth value congruence has a positive effect on follower-rated LMX, which highlights the role social identity plays in fostering quality exchange relationships. Social identity theory posits that individuals use social categorization to understand their own identity, the identity of others in their environment and the identification of in-and out-group members. Individuals are then motivated to engage in behaviors that solidify membership with other ingroup members (Ashforth & Mael, 1989;Tajfel & Turner, 1979). Our finding that it is specifically growth value congruence that has a positive effect on follower-rated LMX suggests that the context within which a relationship exists may play a role in how identification influences relationship development. Our sample consisted of social work students within a work placement under a licensed social work supervisorcreating an inherently developmental relationship. Within this context, it may be that development or growth values are especially salient for followers. In this situation, leaders with values that align with their followers' growth values, likely foster a stronger identification because of the underlying development nature of that relationship. Identity theory and leadermember exchange Previous research approaching LMX from a social identity perspective has identified factors such as increased organizational identification as resulting from perceived similarities between dyads (Loi, Chan, & Lam, 2014). Our results suggest that growth value congruence may also be in alignment with university values (participants were university students and supervisors were university affiliated) influencing overall organizational identificationcreating consistency through their educational institutional values, placement supervisor values and individual values. Further, Jackson and Johnson (2012) found that the fit between leaders' and followers' various identities (relational, individual and collective) lead to higher-quality LMX relationships. Our finding that growth value congruence displays a strong relationship with LMX suggests that it is through these shared underlying values that dyads can foster positive relationships through the formation of a shared identity.
Although we found that value congruence had a positive effect on follower perceptions of LMX, this was not the case for leader perceptions. The null effects of value congruence on leader perceptions of LMX seem to contradict prior findings that over time both leader and follower perceptions of LMX should increase . One explanation for this seemingly incongruent finding may arise from the short-relationship tenure (approximately three months) of the dyads within our sample at the time LMX data were gathered. It may be that early in dyad relationships, before experiencing many "testing episodes," leaders and followers tend to show less agreement around the quality of their relationship, with leaders being less sensitive to some characteristics than followers. It is also likely that the relationship of the student-placement supervisor forms a high-stakes relationship for the follower, but not the leader. This further supports the idea that the differential nature of leader-follower relationships is an important consideration in LMX formation, and thus in LMX research.
Leader-member exchange-job performance. The model presented in the current study extends the literature on LMX and job performance through the inclusion of both situational and individual factors. Although there have been several potential mediators identified to explain the LMX-job performance relationship, there has been less research examining the role LMX plays as a mediator between various individual and situational characteristics and job performance. Jackson and Johnson (2012) found that the relationship between shared identities and job performance was mediated by LMX. Research on the relationship between contact frequency and LMX has displayed somewhat conflicting findings. In a metaanalysis on LMX agreement, Sin et al. (2009) found that, while contact intensity and tenure predicted quality LMX, contact frequency did not. However, Gajendran and Joshi (2012) found that in geographically dispersed teams, contact frequency interacted with LMX quality in the prediction of performance (specifically, innovation). Our finding that contact frequency is an important factor in predicting LMX supports the idea that the nature of relationship structures is important in LMX relationships.The current study took place during a terminal student placement program. Although dyads were not necessarily geographically dispersed, the nature of this program may have caused followers to exist within an environment where contact may not be as consistent as in traditional job settings. This suggests that when there is no regular contact between supervisors and subordinateswhich can occur in many work settings, including in geographically dispersed teamscontact frequency becomes an important factor. Our finding that the relationship between dyad contact and job performance is mediated by LMX suggests that, although factors such as shared identities are important for fostering a collective identity leading to LMX and performance, dyad contact may be a necessary but insufficient factor. This makes intuitive sensealthough there is no theoretical reason for contact alone to predict job performance, OMJ 17,3 contact frequency creates the opportunity for trust-building, which in turn fosters LMX, and ultimately job performance.
Our findings pertaining to the moderating effect of individual values on the relationship between LMX and job performance extend research on the potential boundary conditions of LMX-performance relationships. Research on the moderators of LMX-performance relationships consistently finds that in the presence of quality LMX relationships, individual and situational differences have little effect on job performance. However, individual and situational differences that create a greater reliance on one's leader for desired resources lead to much lower levels of job performance in the presence of low-quality LMX. Based on this consistent finding, we hypothesized that growth and status values would be tied to a subordinate's relationship with their leader, which would lead to negative performance outcomes.
Although valuing growth did not moderate the LMX-performance relationship, highstatus values displayed an interaction pattern similar to that found with other moderators examined in the literature (e.g. political skill). Our results show that the more the subordinates value status, the worse their performance outcomes are in the presence of lowquality LMX. However, quality LMX relationships create a buffering effect for this relationshipperformance does not differ based on status values when LMX was high. Further, our conditional effects model shows that the indirect effect of dyad contact on supervisor-rated performance via LMX was only significant for individuals high in status values. This effect highlights that the values an individual holds will influence their need for contact with their leader. From a social-identity standpoint, individuals high in status likely look to contact with their leader as a way to foster their individual goals through an identity around status. Taken together, these findings suggest that individuals' reliance on their leader for desired outcomes can be affected by factors more distal than personality or job characteristics. Furthermore, these effects were seen in this sample of newly formed dyads with performance ratings taking place after a four-month period. This suggests that the negative effects of poor LMX on the performance of certain individuals happen fairly early in relationship development.

Practical implications
The overarching implication of these findings for practitioners is that quality LMX relationships buffer the negative effects of various individual and situational factors. The buffering effect of quality LMX is likely because of mediators (i.e. trust, empowerment, motivation and job satisfaction) in the LMX-performance relationship (Martin et al., 2015). These results suggest that regardless of individual or situational factors, those who perceive high levels of LMX have the resources necessary to perform effectively. These findings offer an important implication for human resource practitioners across all job types, in that promoting quality LMX relationships can offer protective effects from a host of factors that may otherwise decrease job performance.
To benefit from the positive effects of quality LMX relationships, leaders need to first foster these relationships. A first step to fostering LMX is simply through leaders engaging in sufficient contact with their followers. Although contact alone may not lead to trust, leaders who make an effort to engage in quality contact with their followers will be better able to develop trust through these testing episodes, leading to increases in LMX and eventually job performance.
Another implication of these findings surrounds the short tenure of the dyads studied. Many of the moderators discussed in the LMX literature (e.g. empowerment, extraversion and leader relational identity) may be heightened in the early days and weeks of a new job or placement. It Identity theory and leadermember exchange may be especially important for leaders to engage with new employees to encourage identification with both leadership and the organization as a whole. Individuals hold a wide array of values, and leaders who are able to understand those values early on will likely benefit from increased follower performance through the development of trust-based LMX relationships.

Limitations and future directions
Although the current study offers several contributions within both social identity and LMX theory, it is not without limitations. First, the cross-sectional nature of the current study created endogeneity concerns. For instance, there may be a reciprocal effect between LMX and dyad contact, where leaders and followers who share quality relationships are more likely to spend time together. However, research on the role of trust-building through testing episodes suggests that dyad contact will initially precede LMX development (Sin et al., 2009). Value congruence may also be influenced by LMX, with values converging throughout a relationship. Further, in our full mediation model, we did apply a lagged model where supervisor-rated performance was evaluated from a separate source and several months after the follower survey. Future research should examine the effects of dyad contact and value congruence on LMX through the application of longitudinal methods. Second, although we were able to test a moderated-mediation model of the role dyad contact and individual values play in the LMX-job performance relationship, the small sample of matched leaders and followers prevented us from including value congruence in our moderated-mediation model. Although the small matched sample prevented us from including shared values in our larger model, the large effects found for value congruence on LMX suggest that this effect is a strong one. Previous research has demonstrated the effect of shared characteristics on job performance through LMX as a mediator (e.g. identity fit; Jackson & Johnson, 2012); future research should extend this through examinations of the indirect effect value congruence may have on job performance through LMX.
Third, although we incorporated dyadic-level LMX, Gooty et al. (2012) point to the need for more research on the multi-level analysis of LMX relationships. Although we only examined factors affecting independent leader-follower relationships, LMX is rooted in the differential treatment of followers by their leaders. Future research should examine the effects of team-level LMX in relation to team value congruence and team performance.
Lastly, our small sample of matched dyads limited our power to test our hypotheses using current best practices. Specifically, concerning our congruence hypotheses, we were unable to test them using polynomial regression, which is the preferred analysis as opposed to difference scores (Edwards, 1994). Although our small sample size puts our results at risk of type I error, we believe that the significant results found reflect robust effects around value congruence, LMX and performance. We acknowledge this limitation and encourage future researchers to use this method to examine the effect of multi-dimensional value congruence on LMX and performance outcomes.

Conclusion
The current research contributes to the literature on the relationship between LMX and job performance through an examination of the individual, dyadic and situational factors affecting the process of quality exchange relationships leading to job performance. Our findings suggest that, although contact within a dyad is necessary for LMX formation, the nature of this contact is what drives LMX relationships. Beyond this, we provide evidence for the buffering effects of quality LMX relationships in the presence of otherwise negative traits. Taken together, our findings suggest that, even though increased LMX consistently leads to increases in performance, several factors both individual and situational affect this process.