ASSIGNMENT 7-2

PEER7-4.pdf

Home >Business & Finance homework help >Management homework help >ASSIGNMENT 7-2

Social exchange spillover in leader–member relations: A multilevel model

YING CHEN1*, ZHEN XIONG CHEN2, LIFENG ZHONG3, JOOYEON SON1, XIUJUAN ZHANG4 AND ZHIQIANG LIU5 1School of Labor and Employment Relations, University of Illinois at Urbana-Champaign, Champaign, IL, U.S.A. 2Research School of Management, Australian National University, Canberra, Australia 3School of Business, Renmin University of China, Beijing, China 4School of Business, Sun Yat-sen University, Guangzhou, China 5School of Management, Huazhong University of Science and Technology, Wuhan, China

Summary Drawing on role theory and the cultural theory of collectivism, we developed and tested a multilevel model of social exchange spillover in leader–member relations in the Chinese context. In Mplus analyses of a sample of 213 subordinates from 47 groups, we found that, at the individual level, a dimension of leader–member guanxi (LMG), leader–member personal life inclusion (LMG-P), which is defined as the extent to which leaders and members include each other in their personal or family lives, can spill over to affect subordinates’ contextual performance (i.e., interpersonal facilitation and job dedication); furthermore, this effect was moderated by sub- ordinates’ horizontal collectivism orientation, such that LMG-P spilled over to affect contextual performance only for those who were low in horizontal collectivism orientation. At the group level, the variance of LMG-P within a group, which is referred to as LMG-P differentiation, was related negatively to group performance when the supervisors had a low horizontal collectivism orientation. At the cross level, LMG-P differentiation moderated the relationship between LMG-P and job dedication, such that the relationship was positive only when LMG-P differentiation was low. Copyright © 2015 John Wiley & Sons, Ltd.

Keywords: leader–member guanxi (LMG); leader–member exchange (LMX); collectivism; multilevel; performance

Leader–member relationships have been demonstrated to play a very important role in influencing members’ out- comes in Chinese organizations (e.g., Farh, Tsui, Xin, & Cheng, 1998; Hui & Graen, 1997; Law, Wong, Wang, & Wang, 2000). The current research on leader–member relationships conducted in Chinese organizations has been based primarily on the construct of leader–member exchange (LMX), which was developed initially in the West. However, because of the implicit cultural assumptions and limitations of the LMX construct, current LMX theory cannot capture the indigenous domain of leader–member relations in Chinese organizations fully. By adopting a contextualized research strategy in this study (Tsui, 2006, 2012), we examined a unique aspect of leader–member relationships, which we term leader–member personal life inclusion (LMG-P: cf., Chen, Friedman, Yu, Fang, & Lu, 2009), and developed a multilevel model of social exchange spillover from the private to the work domain in leader–member relations in the Chinese context. Leader–member personal life inclusion, which is one of the three dimensions of leader–member guanxi (LMG), is

defined as the extent to which leaders and members include each other in their family or personal lives (Chen et al., 2009). As indicated in its definition, LMG-P is clearly private domain-specific and thus fits our theoretical inquiry. However, the other two dimensions of LMG in Chen et al.’s (2009) study, affective attachment (LMG-A) and deference to supervisor (LMG-D), are not domain-specific. As such, we treated these two dimensions as controls in our analyses. Leader–member exchange refers to the quality of the exchange relationship between supervisors and subordinates

(Graen & Scandura, 1987). It reflects the Western cultural preference that business relationships should not be

*Correspondence to: Ying Chen, School of Labor and Employment Relations, University of Illinois at Urbana-Champaign, 504 East Armory Avenue, Champaign 61820, IL, U.S.A. E-mail: [email protected]

Revised 07 May 2015, Accepted 07 May 2015

Journal of Organizational Behavior, J. Organiz. Behav. 36, 673–697 (2015) Published online 15 June 2015 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/job.2030

SpecialIssue A rticle

intertwined with personal life, and therefore, it focuses implicitly on working relationships. In contrast, LMG is modeled after family relationships, and the quality of LMG refers to the extent to which a leader–member relationship has become a pseudo-family relationship (Chen et al., 2009). We acknowledge that LMX is a global construct, the prevalence of which has been confirmed in a recent

meta-analysis (Rockstuhl, Dulebohn, Ang, & Shore, 2012). At the same time, we contend that this does not mean that LMX theory has captured all the aspects of leader–member relationships from an international perspective. Leader–member relationship research should explore both etic (transferable) and emic (culture-specific) aspects of those relationships (Rockstuhl et al., 2012). Recently, researchers surveyed the three dimensions of LMG (Chen et al., 2009) in Taiwan, Singapore, and six non-Chinese cultures (Brazil, India, Russia, Saudi Arabia, Turkey, and the UK) and found that two subscales—affective attachment and deference to supervisor—demonstrated full metric invariance across all eight samples, whereas LMG-P appeared to be a Chinese culture-specific construct (Smith et al., 2014). The uniqueness of the LMG-P phenomenon is that it allows or even facilitates interpenetration between the

private and work domains. Although previous research on LMG has studied the relationships between dyadic LMG and “self”-oriented outcomes, such as promotion opportunities (e.g., Law et al., 2000), little research has been conducted to explore whether or not socializing with leaders in the private domain is beneficial to other-oriented work outcomes, such as contextual performance. Furthermore, even though Chinese are very sensitive to informa- tion about who has been included in their leaders’ personal or family lives, and scholars have speculated about the possible negative externality effects of guanxi practices on group and organizational outcomes (Chen & Chen, 2009), a recent review of guanxi research revealed that group-level guanxi research is “almost completely absent” (Chen, Chen, & Huang, 2013, p. 195). Hence, we know little about how the composition of high-quality or low-quality LMG-P affects group effectiveness. The boundary conditions between LMG-P and contextual and group performance are also unclear. Drawing on role theory and the cultural theory of relational collectivism, the purpose of this study was to develop

and test a multilevel model of social exchange spillover in leader–member relations. Role theory suggests that role extension leads to the expansion of role behavior requirements (Kahn, Wolfe, Quinn, Snoek, & Rosenthal, 1964; Katz & Kahn, 1978). The theory suggests further that organizations are essentially a set of work roles, and thus, role differentiation and coordination are critical for collective success (Katz & Kahn, 1978). Looking at LMG-P from the perspective of role extension, we argue that, at the individual level, a subordinate who has high LMG-P with his or her supervisor in the private domain will tend to display higher contextual performance in the work domain. Contextual performance is defined as “a set of interpersonal and volitional behaviors that support the social and motivational context in which organizational work is accomplished” (Van Scotter & Motowidlo, 1996, p. 525). Based on the cultural theory of relational collectivism (Brewer & Chen, 2007), we posited further that, at the

individual level, subordinates’ horizontal collectivism orientation can moderate the spillover effect of LMG-P on contextual performance. At the group level, we proposed that a supervisor’s horizontal collectivism orientation moderates the relationship between LMG-P differentiation, defined as the variance of LMG-P at the group level, as well as group performance. Moreover, LMG-P differentiation moderates the spillover effects of LMG-P on contextual performance. Figure 1 illustrates our hypothesized theoretical framework. We attempted to provide three theoretical extensions to the LMG and LMX literatures. First, research has

demonstrated a positive relationship between LMG and work outcomes that benefit LMG beneficiaries, such as promotion opportunities (Law et al., 2000). However, we know little about whether LMG in the private domain is related to work outcomes that can benefit coworkers and the organization. Studying the relationship between LMG-P and contextual performance can help solve this puzzle. LMG-P also enriches current LMX theory by demonstrating that an indigenous construct (i.e., LMG-P) can contribute to the development of a more global leader–member relationship theory. Second, no prior studies have examined the LMG-P spillover effects concurrently at both the individual and group levels of analysis. The multilevel social exchange model provides an integrated picture that enables us to understand more clearly how LMG-P affects work outcomes across levels, thus deepening our understanding of the complexity of the LMG phenomenon in the Chinese context. Third, we

674 Y. CHEN ET AL.

proposed and tested whether or not spillover effects are contingent upon subordinates’ horizontal collectivism orientation at the individual level, and supervisors’ horizontal collectivism orientation at the group level. A more nuanced understanding of the effects of social exchange spillover in leader–member relations will shed light on the long-term debate about whether or not LMG is good or bad for an organization (Chen & Chen, 2012; Fan, 2002; Warren, Dunfee, & Li, 2004).

Theory and Hypotheses

LMG-P social exchange spillover and contextual performance

Using role theory’s explanation of role expansion, we argue that, in the Chinese context, LMG-P can spill over from the private domain to affect contextual performance in the work domain. According to Katz and Kahn (1978), a role is associated with one’s position in a social system and the behavioral requirements associated with that position. Role theory suggests that roles can either contract or expand (Kahn et al., 1964). We argue that when members’ work roles with the leader cross boundaries and expand from the work to the private domain, their roles expand to include roles as pan-family members. The role obligations related to pan-family members will spill over to affect their behavior requirements in the work domain, which should evolve and expand to include contextual performance in the Chinese context. Role theory suggests that the spillover of values and behaviors learned in one role might influence other roles

(Edwards & Rothbard, 2000). In the work–family literature, it is argued that people will transfer “valenced affect, skills, behaviors, and values from the originating domain to the receiving domain” in a positive manner (Hanson, Hammer, & Colton, 2006, p. 251). A number of studies have supported the influence that behaviors from one role have on other roles, for example, work on family roles, or family on work roles (Crouter, 1984; Pearlin & Kohn, 1966; Ruderman, Ohlott, Panzer, & King, 2002). By extension, in this study, we suggest that, in the setting of leader–member (or supervisor–subordinate) relationships, if subordinates have good personal relationships with

Figure 1. Theoretical framework

SOCIAL EXCHANGE SPILLOVER 675

the supervisor in the private domain (i.e., high LMG-P), the effect of LMG-P can spill over, or be transferred positively, into the work domain. In particular, we argue that, when subordinates have high LMG-P, they will help their supervisor actively in the private setting because they feel they have an obligation to fulfill those role requirements. In turn, this effect can carry over to the work domain, that is, the subordinates will also help the supervisor actively in the organizational setting. Specifically, we proposed that LMG-P on the part of leaders and members is related positively to two dimensions of contextual performance: interpersonal facilitation and job dedication. Interpersonal facilitation includes cooperative, considerate, and helpful behaviors that can facilitate coworkers’ performance. Job dedication includes a set of motivational and self-disciplined behaviors, such as working diligently, which can support the objectives of the organization (Van Scotter & Motowidlo, 1996). Individuals have multiple roles and a natural tendency to achieve role/identity consistency across domains, which

is very important for their well-being (Cross, Gore, & Morris, 2003). In the Chinese context, it is a common cultural practice to model working relationships after family relationships, which not only blurs the boundaries between the private and work domains but also integrates work and family roles through the pan-familization process (Yang, 1992). This process refers to one that extends kin relationships to non-kin (Yang, 1992). Through pan-familization, leaders and members can develop and establish an extended “pan-family” network. In this study, we argue that a subordinate who has a good personal relationship with the supervisor (i.e., high

LGM-P) will play different roles in different settings, that is, a “pan-family” member in the private setting, and a subordinate in the work setting. However, the behavioral tendencies (e.g., to help the leader) as a “pan-family” member in the private setting will spill over to the work setting as a subordinate. In particular, when leaders and members have high-quality LMG-P, their roles include those of members of an extended family. In such extended pseudo-family relationships, their role obligations expand and are subject to the principles of the five fundamental Chinese role relationships (wulun), which are the following: emperor–subject; father–son; husband–wife; elder–younger siblings; and friend–friend (Confucius, 1915). Leaders and members in the extended familial circles are likely to be dominated by role obligations that are similar to father–son or elder–younger siblings, depending on their age, seniority, and personal closeness. The common theme of the principle of the five fundamental relationships is that persons at high levels should take

care of those at lower levels whereas persons at lower levels should respect and serve the interests of those at higher levels. Leaders and members in the familial network have much broader and stronger pan-family obligations to each other than their formal work roles suggest. Because of the influence of cultural norms that are embedded in Chinese family relationships, such as wulun, members who are in the leader’s extended familial network—indicated by high-quality LMG-P—are more likely to have strong role obligations to serve the interests of the leader in the work domain as well. With extended family membership, people will exhibit very strong mutual trust, loyalty, dedication, and role-based obligations to help each other (Hwang, 1987). Thus, subordinates with high LMG-P will be willing to do whatever they can to serve the best interests of the supervisor, both in the private and in the work domains. As a result, their role obligations as members of a pseudo-family will spill over to affect their work behaviors, in terms of helping coworkers and working diligently, which will benefit both the organization and its leaders, who are representatives of the organization. Previous studies have shown empirically that having informal connections with supervisors outside the work

domain has positive effects on subordinates’ outcomes in the work domain, such as enhancing their social and psychological capital, promoting their organizational self-esteem (Liu, Hui, Lee, & Chen, 2013), and enabling them to have more open dialogue with the leader (Chen & Tjosvold, 2007). These findings support our argument that the effect of LMG-P developed in the private domain can spill over to the work domain. Therefore, we hypothesized the following:

Hypothesis 1a: LMG-P is related positively to interpersonal facilitation.

Hypothesis 1b: LMG-P is related positively to job dedication.

676 Y. CHEN ET AL.

Subordinates’ horizontal collectivism orientation as a moderator of spillover effects at the individual level

Overall, although we expected a positive spillover effect of LMG-P on contextual performance, we posited that the relationship would be weaker for those who have a high horizontal collectivism orientation. Horizontal collectivism orientation is “a cultural pattern in which the individual sees the self as an aspect of an in-group” (Singelis, Triandis, Bhawuk, & Gelfand, 1995, p. 244). Recent studies on collectivism have clarified that collectivism in the Chinese context is, in essence, relational collectivism that stresses the importance of interpersonal self and relational interdependence (Brewer & Chen, 2007). In our study, we focused on the horizontal collectivism orientation as the boundary condition for social exchange spillover effects for two reasons. First, horizontal collectivism orienta- tion is more relevant to the guanxi phenomenon. Research has shown that, for Chinese, the notion of in-group members includes only those with whom they have close particularistic ties, but not the abstract institution to which they belong (Chen, Chen, & Xin, 2004), indicating that a vertical collectivism orientation is less relevant to the guanxi phenomenon. Brewer and Chen (2007) also pointed out that collectivism in the Chinese cultural context is relational collectivism, which emphasizes interpersonal interdependence rather than group membership. Second, because subordinates have multiple roles and obligations, the spillover effects on contextual performance of subordinates’ role extension with leaders should be affected by employees’ other role relationships in the group context (i.e., their roles with group members). Horizontal collectivism orientation is essential in determining those who are considered in-group and out-group members; therefore, the horizontal collectivism orientation has the potential to dampen or strengthen the social exchange spillover effects in leader–member relations. Katz and Kahn (1978) stated that human organizations can be understood as sets of work roles, each of which

carries certain responsibilities and functions. In a group setting, role relationships among group members are important in helping members make meaningful contributions to the group (Cheng, 1983). Horizontal collectivism orientation defines how an employee views his or her relationship with other group members, and we argue that an employee’s horizontal collectivism orientation moderates the relationship between LMG-P and interpersonal facilitation, as well as job dedication. Individuals high in horizontal collectivism orientation view themselves as part of an in-group. They view all group

members as being similar to each other and sharing the same in-group identity. Because horizontal collectivists have great concerns for other in-group members and value group harmony and solidarity, we contend that individuals with a high horizontal collectivism orientation should help coworkers, regardless of whether or not they have high levels of LMG-P. Even for those who do not have high quality LMG-P with their supervisors, they may still feel strong obligations to help other group members because they share the same identity; thus, by helping others, they help themselves. In such a context, the quality of LMG-P with their supervisors will have no strong influence on their interpersonal facilitation behaviors. In contrast, members with a low horizontal collectivism orientation are less concerned about other group members and do not value group harmony and solidarity. Instead, they care more about their individual distinctiveness and achievements. Therefore, they should be more motivated by LMG-P to engage in behaviors that help coworkers. As a consequence, LMG-P with their supervisors will have stronger positive effects on interpersonal facilitation behaviors for those subordinates who have a low, compared with those with a high, horizontal collectivism orientation. Job dedication emphasizes self-disciplined behaviors that drive employees to work to promote the organization’s

best interests. We argue that a horizontal collectivism orientation also moderates the relationship between LMG-P and job dedication. As mentioned earlier, horizontal collectivists see themselves as part of an in-group. In-group members care for each other and are willing to cooperate with each other, regardless of whether or not they receive any personal benefits. Horizontal collectivists see each group member similarly, such that each is part of an in-group and shares the same group identity. For individuals with a high horizontal collectivism orientation, who identify themselves as a member of the group in which they work, it is likely that they will be dedicated to their jobs as a norm or a form of cooperation imposed by their horizontal collectivism orientation. Research has shown that collec- tivists work hard when their performance has implications for in-group members’ performance (Earley, 1993).

SOCIAL EXCHANGE SPILLOVER 677

Members with a high horizontal collectivism orientation will accomplish difficult tasks as a way to contribute to the well-being of other group members. Individuals low in horizontal collectivism orientation do not view other group members as in-group members and feel that they do not have generalized obligations or a duty to contribute to the well-being of other group members. As a result, high LMG-P with supervisors will have stronger positive effects on job dedication for those subordinates who have a low horizontal collectivism orientation compared with those with a high horizontal collectivism orientation.

Hypothesis 2a: Subordinates’ horizontal collectivism orientation moderates the positive relationship between LMG-P and interpersonal facilitation, such that the relationships will be weaker when their horizontal collectivism orientation is high rather than low.

Hypothesis 2b: Subordinates’ horizontal collectivism orientation moderates the positive relationship between LMG-P and job dedication, such that the relationships will be weaker when their horizontal collectivism orienta- tion is high rather than low.

Cross-level effects of LMG-P differentiation

One of the major objectives of this study was to provide an integrated multilevel model of social exchange spillover in leader–member relations. To do so, we examined further the cross-level interaction between the effects of LMG-P differentiation at the group level and LMG-P at the individual level on employee contextual performance. In the following section, we first argue that, although LMG-P occurs in the private domain, LMG-P differentiation also can be observed in the work domain in the Chinese cultural context. Second, we argue that LMG-P differentiation creates a clear structure of role differentiation within the group according to differences in the quality of LMG-P. This LMG-P-based differentiation would make justice concerns salient within the group, which could weaken the positive relationships between LMG-P and interpersonal facilitation, as well as job dedication. By definition, LMG-P concerns personal relationships between leaders and members in the private domain. We

contend that LMG-P can be observed by group members and that such information is very likely available to them in Chinese workplaces for the following two reasons: First, China has a “diffuse” culture (Trompenaars, 1993). In contrast to “specific cultures,” such as that of the USA (Trompenaars, 1993), in which there is a distinct line in people’s minds between their work and personal lives, and neither supervisors nor their subordinates assume the right to be involved in each other’s personal lives, in a “diffuse” culture like China, there is no clear line between personal and work life, and people hold a “holistic” view with respect to each other, so that individuals care not only about each other’s work lives but their personal lives as well. In the workplace, Chinese employees exchange information about their personal lives constantly. Furthermore, being included in the leader’s personal life provides valuable evidence of a subordinate’s social capital and status in the workplace, so he or she may reveal such infor- mation either intentionally or unintentionally. Such information is highly salient, and coworkers pay close attention to it. For example, Sanchez-Burks et al. (2003) found that individuals who are members of a collectivistic culture pay much more attention to cues about personal information than do Americans. Second, the unique organization-sponsored housing arrangements in China also make information about LMG-P avail-

able to others. Many Chinese organizations, especially state-owned firms, provide housing or subsidize housing costs. Thus, many subordinates and their supervisors live in the same neighborhoods or apartment buildings (Osigweh & Huo, 1993; Walder, 1986). In such environments, subordinates and supervisors have many opportunities to meet each other on a daily basis, not only at work but outside as well. This type of community arrangement intensifies the exchange of information with regard to the intertwining of work and personal relationships among subordinates and their supervisors. We have argued that LMG-P can be understood in terms of role expansion for both leaders and members, such

that work roles expand from the work to the private domain to include pan-family member roles. When examining

678 Y. CHEN ET AL.

LMG-P at the group level, different qualities of LMG-P create a role differentiation structure within the group. In the con- text of high LMG-P differentiation, leaders socialize only with a few group members and are distant from the majority, which introduces clear role relational boundaries into groups. We argue that this LMG-P-based relational boundary makes justice concerns salient within groups. Studies have found consistently that Chinese leaders have a tendency to allocate organizational rewards based on their relationships with subordinates (Law et al., 2000; Zhou & Martocchio, 2001). Although LMG-P differentiation has a certain degree of cultural legitimacy, recent research has shown that Chinese employees consider performance-based reward to be fair and guanxi-based reward unfair (Bozionelos & Wang, 2007). Chinese employees regard merits and performance, rather than LMG-P, as the primary criteria that should determine the allocation of organizational rewards (Chen, 1995; He, Chen, & Zheng, 2004). Thus, Chinese employees are likely to perceive that high LMG-P differentiation and related favorable treatments based on personal closeness lack institutional legitimacy, and thus regard it as a violation of the fairness norm. As a result, in the context of high LMG-P differentiation, subordinates are more likely to perceive that leaders engage in favoritism and unfair treatment of group members. Further, in the context of high LMG-P differentiation, even those who benefit from having high-quality LMG-P may

still believe that their leaders are unfair, which may compromise their trust of the leaders (Chen et al., 2004). Recent studies have shown that individuals do care about how others are treated and knowing that others receive unfair treat- ment lowers subjects’ own perceptions of fairness (Van den Bos & Lind, 2001). Empirical studies have also shown that managers’ practice of guanxi is related negatively to employees’ perceptions of procedural justice (Chen, Friedman, Yu, & Sun, 2011). As a consequence, we argue that LMG-P may not be related positively to contextual performance when LMG-P differentiation is high. In contrast, when LMG-P differentiation is low, it shows that leaders develop almost equally close personal relationships with each of the group members. Equality may also be reflected in both monetary and nonmonetary resource allocation. This reduces uncertainty, and envy among group members greatly and eventually enhances group member coordination, helping with coworkers, and dedication to the job.

Hypothesis 3a: LMG-P differentiation moderates the relationship between LMG-P and interpersonal facilitation negatively, such that the relationship is weaker when LMG-P differentiation is high.

Hypothesis 3b: LMG-P differentiation moderates the relationship between LMG-P and job dedication negatively, such that the relationship is weaker when LMG-P differentiation is high.

LMG-P differentiation and group performance

In addition to its function as a contextual factor that moderates the relationship between LMG-P and performance, we ar- gue that LMG-P differentiation can also influence group effectiveness contingent upon supervisors’ collectivism orienta- tion. As we discussed earlier, even though LMG-P occurs in the private domain, it is observable in the work domain. Employees notice differential leader–member relationships and tend to compare themselves with other group members (Lam & Huang, 2012); therefore, it is important to study how leader–member relationship differentiation affects group per- formance. Although it has long been speculated that LMG-P differentiation may affect group performance (e.g., Tsui & Farh, 1997), to the best of our knowledge, no study has yet been conducted to explore the relationship between LMG-P differentiation and group performance. Therefore, we referred to the LMX differentiation literature for insights. Recently, several studies have explored the relationship between LMX differentiation and group outcomes, with

inconclusive results. Some studies have found that LMX differentiation was related positively to group performance (Naidoo, Scherbaum, Goldstein, & Graen, 2011), while others reported that LMX differentiation was related negatively to team coordination and, in turn, team performance (Li & Liao, 2014). Some studies have suggested that the direct effects of LMX differentiation on performance are context-based. For example, Liden, Erdogan, Wayne, and Sparrowe found that LMX differentiation was related positively to group performance for groups with a low median LMX or high job interdependence (2006).

SOCIAL EXCHANGE SPILLOVER 679

Chinese society is extremely complex and is influenced by multiple cultural traditions and practices (Chen & Chen, 2009). On the one hand, China is a country with a cultural heritage of relational collectivism; thus, supervisors who develop a different quality of LMG-P with their group members have cultural legitimacy. On the other hand, Mao’s socialist tradition and modern Western-oriented human resource management practices have fought hard against guanxi in the workplace. These factors are entangled, and it is difficult to argue that LMG-P differentiation has a negative main effect on group performance. Even though we did not find direct empirical evidence, a recent study of LMG differentiation and individual outcomes did not find that LMG differentiation had a main effect on helping; rather, the relationship between LMG and helping behavior was contingent on the degree of LMG differentiation (Chen, Yu, & Son, 2014). We contend, therefore, that the effects of LMG-P differentiation on group performance are contingent upon contextual factors, one of which is supervisors’ horizontal collectivism orientation. Supervisors’ cultural orientation of horizontal collectivism can serve as an important group context. As we

discussed earlier, the horizontal collectivism orientation is a cultural pattern that emphasizes harmony, equality of each group member, and caring for each group member’s interests. Supervisors who are high in horizontal collec- tivism orientation consider all their group members as in-group members and stress their interdependence and the common goals of the group. We contend that LMG-P differentiation will be related negatively to group performance only when supervisors have a low horizontal collectivism orientation. We have argued that LMG-P differentiation can make justice concerns salient within the group, but supervisors’

horizontal collectivism orientation can weaken those concerns. If a supervisor is high in horizontal collectivism orientation, he or she will care about group members’ interests and well-being. He or she will view each group member as similar to the others and also will promote the norm of harmonious relationships and a team-oriented group culture. In this context, even if LMG-P differentiation is high, such that a supervisor socializes with only a few subordinates and is distant from other group members, group members are not likely to attribute LMG-P differentiation to favoritism or the leader’s unfairness. Thus, they should still be able to cooperate with each other and coordinate with group members easily. Recent studies have shown that one important mechanism by which LMX differentiation affects group performance is that it leads to poor team coordination (Li & Liao, 2014). When supervisors have a high horizontal collectivism orientation, however, group members are considered to be a single cohesive entity, which is unlikely to promote animosity within the group; thus, cooperation and coordination should not be affected negatively. In contrast, when a leader has a low horizontal collectivism orientation, he or she will not perceive that each group

member is similar and will care little about the well-being of group members. When there is high LMG-P differenti- ation, group members may believe that the differentiation serves the supervisor’s own interests and hurts the group’s interests and goals. Group members are more likely to perceive that leaders engage in favoritism and share the percep- tion that a climate of injustice exists within the group. Animosity between in-group and out-group members is likely to occur in those circumstances, cooperation and coordination decrease greatly, and group performance suffers as a result.

Hypothesis 4: Supervisors’ horizontal collectivism orientation moderates the relationship between LMG-P differentiation and group performance, such that the relationship will only be negative when the supervisor has a low horizontal collectivism orientation.

Methods

Sample and procedures

The data for the study were collected in China. Participants included managers/supervisors who were part-time MBA students from a leading university in Northern China and their subordinates. All respondents were full-time

680 Y. CHEN ET AL.

employees and were asked to choose five of their immediate subordinates from their work groups randomly to participate in the survey. Each survey respondent was given a stamped envelope preaddressed to one of the researchers. Separate questionnaires were sent to supervisors and subordinates. In a cover letter attached to the questionnaires, we informed the respondents about the voluntary nature of participation, the objectives of the survey, and the confidentiality of their responses. In the subordinates’ survey, respondents were asked to assess their perceptions of the three dimensions of LMG

(LMG-P, LMG-A, and LMG-D), as well as LMX and collectivism. Supervisors were asked to evaluate each of the five subordinates’ contextual performance and their own collectivism values. We created a coding scheme to match the supervisor–subordinate data. Completed usable data were obtained from 226 subordinates, with a response rate of 94 percent. Forty-seven supervisors of 48 groups returned completed usable questionnaires, with a response rate of 97.9percent. Among the 48 groups, 37 had a 100percent response rate, 8 had an 80 percent response rate, and 3 groups had a 60 percent response rate, for a mean within-group response rate of 94 percent.

Measures

With the exception of the scales of LMG-P, LMG-A, and LMG-D, the measures were constructed originally in English. We followed standard translation and back-translation procedures (Brislin, 1980) to translate the measures into Chinese. Unless otherwise indicated, the Likert scale response options ranged from “1” = “strongly disagree” to “7” = “strongly agree.”

LMG-P (LMG-personal life inclusion) Leader–member personal life inclusion was one dimension on a 12-item scale of LMG with three dimensions (i.e., affective attachment, personal life inclusion, and deference to superiors) developed by Chen et al. (2009). We conducted exploratory factor analysis (EFA) for LMG, which confirmed that there were three dimensions. The four items of LMG-P were as follows: “My supervisor asks me to help him/her deal with some personal/family errands,” “After the office hour, I have social activities together with my supervisor such as having dinner together or having entertainment together which go beyond work duties,” “During holidays, my supervisor and I would call each other or visit each other,” and “I am well acquainted with the family members of my supervisor and I have personal contact with these members.” The alpha coefficient was 0.81. The sample items for the other two dimensions are included in the control variable section.

Horizontal collectivism orientation We used a four-item scale developed by Triandis and Gelfand (1998) to measure this variable. Sample items included “I care about the well-being of my co-workers” and “If a co-worker gets a prize, I would feel proud.” The alpha coefficient for subordinates’ horizontal collectivism orientation was 0.93, while for supervisors, it was 0.74.

Contextual performance1

We used a 15-item scale to measure the two dimensions of contextual performance (Motowidlo & Van Scotter, 1994)—interpersonal facilitation and job dedication. Each group’s direct supervisor rated his or her selected subordinates on these two dimensions. Interpersonal facilitation was measured by seven items, while job dedication was measured by eight items. Sample items for interpersonal facilitation included “Praise co-workers when they are successful” and “Help someone without being asked,” while sample items for job dedication included “Puts in extra

1We tested whether two dimensions of contextual performance should be considered as individual level variables. With regard to interpersonal facilitation, ICC(1) = 0.21; ICC(2) = 0.58, and the mean of rwg = 0.61. With regard to job dedication, ICC(1) = 0.15; ICC(2) = 0.47, and the mean of rwg = 0.44. The low values of ICC(2) and rwg demonstrated that two dimensions of contextual performance should be considered as individual- level, rather than group-level variables.

SOCIAL EXCHANGE SPILLOVER 681

hours to get work done on time” and “Persists in overcoming obstacles to complete a task.” The alpha coefficient for interpersonal facilitation was 0.91, and that for job dedication was 0.94.

LMG-P differentiation We calculated within-group variance (standard deviation [SD]) in individual-level LMG-P scores to operationalize LMG-P differentiation.

Group performance Group performance was measured using each group’s direct supervisor ratings on five performance criteria that have been used in prior research (Van der Vegt & Bunderson, 2005). These five criteria were efficiency, quality, overall achievement, productivity, and mission fulfillment. The response set for these items ranged from “1” = “far below average” to “7” =“far above average.” The alpha coefficient was 0.84.

Control variables The control variables had two levels: individual and group. At the individual level, LMX, LMG-A, LMG-D, and vertical collectivism orientation as rated by subordinates were used as controls. Recent meta-analysis has demonstrated that LMX is related to contextual performance (Ilies, Nahrgang, & Morgeson, 2007). Vertical collectivism orientation was also reported as being related to organizational citizenship behavior (OCBs) (Hui, Lee, & Wang, 2014), so we controlled that in the analyses as well. LMG-A and LMG-D were related significantly to two dependent variables in the zero-order correlations (Table 2).

LMX We measured LMX quality using the 12-item LMX-MDM scale (Liden & Maslyn, 1998). As there are no theoretical reasons to expect that the four dimensions of LMX function differently, and EFA analysis showed that it had only one dimension, we averaged the scores of the subdimensions into one score. Sample items included “I like my supervisor as a person” and “I do work for my supervisor that goes beyond what is specified in my job description.” The alpha coefficient was 0.96. Leader–member guanxi affective attachment and LMG-D are the other two dimensions of the three-dimensional

LMG measure. Affective attachment refers to the extent to which leaders and members have emotional connections and are willing to care for each other. Deference to supervisor refers to the degree of obedience and devotion a member has toward his or her leader (Chen et al., 2009). We controlled for LMG-A (alpha coefficient= 0.90) and LMG-D (alpha coefficient = 0.90) at the individual level in all analyses that included LMG-P.

LMG-A (LMG-affective attachment) The sample items included “I would feel sorry and upset if my supervisor decided to work for another company” and “My supervisor and I always share thoughts, opinions, and feelings toward work and life.” The alpha coefficient was 0.90.

LMG-D (LMG-deference to supervisor) The sample items included “I am willing to obey my supervisor unconditionally” and “While I disagree with my supervisor, I would still support his/her decisions.” The alpha coefficient was 0.90.

Vertical collectivism orientation A four-item scale developed by Triandis and Gelfand (1998) was used to measure vertical collectivism orientation. Sample items included “Parents and children must stay together as much as possible” and “It is my duty to take care of my family, even when I have to sacrifice what I want.” The alpha coefficient for subordinates’ vertical collectivism orientation was 0.85, while that for supervisors was 0.75.

682 Y. CHEN ET AL.

Further, employee age, sex, education, and supervisor tenure are included typically in OCB research (LePine, Erez, & Johnson, 2002); therefore, we also controlled for these demographic variables. Age and supervisor tenure were measured in years. Education was measured in five ordered categories: high school; some professional training; some college education; a bachelor’s degree; and a graduate degree. Sex was coded 1 for male and 0 for female. In all of our analyses of LMG-P and LMG-P differentiation at the group level, we controlled not only for LMX

differentiation and the LMX mean but also for the LMG-P mean, and the other two dimensions of LMG at the group level, including the LMG-A mean, LMG-A differentiation, the LMG-D mean, and LMG-D differentiation. Except for LMG-D differentiation, those variables were related significantly to two dependent variables in the zero-order correlations. In this way, when analyzing the effects of LMG-P and LMG-P differentiation, we took into account LMX differentiation, and the LMX and LMG-P mean as well, together with the other two dimensions of LMG. We also controlled for supervisor’s vertical collectivism orientation, and original group size (by design, each group in this study had five members; original group size was operationalized as the total number of employees in the work group from which the five members were chosen). This was reported by each group’s direct supervisor.

LMX mean To operationalize the LMX mean, we calculated the within-group mean of individual-level LMX scores.

LMX differentiation Following Chan’s (1998) dispersion model and prior examples of measures of LMX differentiation (e.g., Liden et al., 2006), we calculated within-group variance (SD) in individual-level LMX scores to operationalize LMX differentiation.

LMG-A mean We calculated the within-group mean of individual-level LMG-A scores to operationalize the LMG-A mean.

LMG-A differentiation To operationalize LMG-A differentiation, we calculated within-group variance (SD) in individual-level LMG-A scores.

LMG-D mean We calculated the within-group mean of individual-level LMG-D scores to operationalize the LMG-D mean.

LMG-D differentiation To operationalize LMG-D differentiation, we calculated within-group variance (SD) in individual-level LMG-D scores.

LMG-P mean We calculated the within-group mean of individual-level LMG-P scores to operationalize the LMG-P mean.

Confirmatory factor analyses2

Because the data had a nested and multilevel structure, it was more appropriate to conduct a two-level confirmatory factor analysis (CFA) to examine the discriminant validity of the constructs. We used the “TWOLEVEL” and MLR (maximum likelihood with robust standard errors) methods in Mplus (Muthén & Muthén, 1998–2011) to conduct the two-level CFA, as it can analyze both the individual-level and group-level variables simultaneously, as well

2We conducted content validity tests of LMG and paternalism using a sample of 20 PhD students from a leading Chinese university. We also conducted EFA and CFA using two separate samples to examine discriminant validity between LMG and paternalism. The results showed that LMG and paternalism are two distinct constructs. Detailed results are available from the first author upon request.

SOCIAL EXCHANGE SPILLOVER 683

as consider the nonindependence of cases in the same group using the corrected standard errors for the highest level of clustering (Muthén, 1991; Muthén & Muthén, 1998–2011). In order to demonstrate that LMX and the three dimensions of LMG are distinct constructs, we conducted a two-level

CFA analysis with 11 variables (eight variables reported by subordinates and three reported by supervisors) that included both substantive and control variables. The individual-level variables included LMG-P, LMG-A, LMG-D, LMX, vertical collectivism orientation, horizontal collectivism orientation, interpersonal facilitation, and job dedication; these were set as within-level variables. At the group level, variables included supervisors’ horizontal collectivism orientation, vertical collectivism orientation, and group performance; these were set as between-level variables. The model had 60 item-level indicators in total. According to Jöreskog and Sorbom (1986), when a model has too

many indicators and the sample is not large enough, it is difficult to achieve model fit even when the model has strong theoretical support. In addition, item parceling is often suggested when researchers opt for an indicator structure, because first, the more the number of indicators, the higher chance for cross-loadings among the indicators (Hall, Snell, & Foust, 1999). Thus, we created three parcels for both interpersonal facilitation and job dedication, four parcels for LMX along its four dimensions, and two parcels for subordinates’ horizontal and vertical collectivism orientation. We used full items for three dimensions of LMG and all three group-level variables. We conducted a two-level CFA for the hypothesized 11-factor model using Mplus 6.11 (Muthén & Muthén, 1998–

2011). The results indicated that all indicators (items) loaded correctly to the corresponding constructs with significant factor loadings. We then compared the hypothesized model with six alternative models. The results showed that our hypothesized 11-factor model produced the best fit (χ2 =621.80, df=333, p<0.001, comparative fit index=0.93, Tucker–Lewis index=0.91, root-mean-square effort of approximation=0.06, within-group standardized root-mean-square residual = 0.03, and between-group standardized root-mean-square residual = 0.04). The results for the two-level CFA with 11 factors showed that the model captured distinct constructs. Detailed results can be found in Table 1.

Analytic strategy

The data used in this paper contained a hierarchical structure in which responses for individual-level variables were nested within groups. The dependent variables included two individual-level variables (i.e., interpersonal facilitation

Table 1. Results of the two-level confirmatory factor analyses.

Model χ2 df CFI TLI RMSEA SRMRw/b

Model 1—hypothesized 11-factor model 621.80 333 0.93 0.91 0.06 0.03/0.04 Model 2—10-factor model (combined LMX and LMG-P) 855.83 340 0.87 0.85 0.08 0.05/0.04 Model 3—10-factor model (combined interpersonal facilitation and job dedication)

817.10 340 0.88 0.86 0.08 0.04/0.04

Model 4—10-factor model (combined subordinate’s horizontal and vertical collectivism orientation)

671.64 340 0.92 0.90 0.06 0.04/0.04

Model 5—10-factor model (combined supervisor’s horizontal and vertical collectivism orientation)

654.14 335 0.92 0.90 0.06 0.03/0.5

Model 6—9-factor model (combined LMG-P, LMG-A, and LMG-D)

1023.38 346 0.83 0.80 0.09 0.05/0.04

Model 7—8-factor model (combined LMX and LMG) 1211.88 351 0.78 0.75 0.10 0.06/0.04

Note: Two-level confirmatory factor analysis was conducted with supervisor’s horizontal and vertical collectivism and group performance as level 2 variables and LMX, LMG-A, LMG-D, LMG-P, subordinates’ vertical and horizontal collectivism orientation, interpersonal facilitation, and job dedication as level 1 variables. LMX = leader–member exchange. LMG-A = leader–member guanxi affective attachment = LMG-D = LMG-deference to leaders; LMG-P = LMG-personal life inclusion; CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square effort of approxima- tion; SRMR = standardized root-mean-square residual; SRMRW = refers to within group SRMR; SRMRb = refers to between group SRMR.

684 Y. CHEN ET AL.

and job dedication) and one group-level outcome (i.e., group performance). Given the multilevel nature of the data and the two levels of dependent variables, we tested our hypothesized multilevel relationships simultaneously using the estimator of MLR (maximum likelihood with robust standard errors) in Mplus 6.11 (Muthén & Muthén, 1998–2011) with a random slope.

Results

Means, SDs, and intercorrelations among variables are presented in Table 2.

Model estimation

To estimate the hypothesized multilevel model (Figure 1), at level 1 (i.e., individual level), we estimated the relationships of LMG-P on interpersonal facilitation, as well as job dedication, which were moderated by subordinates’ horizontal collectivism orientation. We specified the slopes of LMG-P on interpersonal facilitation and job dedication to be random. In addition, LMX and the other two dimensions of LMG (i.e., LMG-A and LMG-D), as well as subordinates’ vertical collectivism orientation, were included as control variables with fixed effects on interpersonal facilitation and job dedication. We also controlled for subordinates’ demographics, including age, sex, education, and years with supervisor at the individual level with fixed effects. At level 2 (i.e., group level), we specified the relationship between LMG-P differentiation and group performance, which was moderated by super- visor’s horizontal collectivism orientation. We controlled for the original group size, supervisor’s vertical collectivism orientation, and the mean and differentiations of all individual-level variables (i.e., LMX, LMG-P, LMG-D, and LMG-A) at the group level.3 At the cross level, we tested the moderating effect of group level LMG-P differenti- ation on the individual level relationships between LMG-P and interpersonal facilitation, as well as job dedication. To facilitate the interpretation of the findings, all individual-level variables were group-mean centered (Enders & Tofighi, 2007), and all level 2 variables were grand-mean centered (Hofmann & Gavin, 1998). There are no overall model indices available to test two-level random slope models, because there is not a single

population covariance matrix to test the model fit. Instead, we calculated pseudo-R2 using Snijders and Bosker’s (1999) formulas for the model that reflected the proportional reduction of level 1 and level 2 errors due to the inclusion predictors in the model, compared with the model with only control variables. The results showed that the predictors accounted for 6percent of the total variance in interpersonal facilitation, 3.4 percent of the total variance in job dedication, and 11.4 percent of the total variance in group performance, above and beyond the control variables. The results demonstrated that LMG-P differentiation and LMG-P indeed played significant roles in predicting subordinates’ contextual performance.

Hypothesis tests

Hypotheses 1a and 1b predicted that LMG-P would be related positively to interpersonal facilitation (1a) and job dedication (1b). Table 3 shows that LMG-P was related positively to interpersonal facilitation (γ= 0.22, p <0.05). Table 3 also shows that the relationship between LMG-P and job dedication was positive and significant (γ= 0.20, p <0.05). Hence, Hypotheses 1a and 1b were supported.

3Without controls, the results are similar to the results with all controls in terms of significant patterns and coefficient magnitude. Detailed results are available from the first author upon request.

SOCIAL EXCHANGE SPILLOVER 685

T ab le

2 . M ea n s an d st an d ar d d ev ia ti o n (S D ) an d in te rc o rr el at io n s am

o n g v ar ia b le s.

V ar ia b le

M ea n

S D

1 2

3 4

5 6

7 8

9 1 0

1 1

L ev el

1 (N

= 2 1 6 to

2 2 6 )

1 A g e

3 0 .3 1

6 .7 2

2 S ex

0 .5 0

0 .0 6

3 E d u ca ti o n

4 .6 8

0 .9 9

�0 .0 9

�0 .0 4

4 Y ea rs

w it h su p er v is o r

3 .2 0

3 .9 2

0 .4 0 * *

0 .0 9

�0 .2 7 * *

5 L M X

5 .3 2

1 .2 3

�0 .1 0

0 .1 2

�0 .0 4

0 .0 3

6 L M G -A

5 .0 3

1 .4 0

0 .0 0

0 .1 6 *

�0 .0 5

0 .0 7

0 .7 4 * *

7 L M G -D

4 .3 1

1 .4 2

�0 .0 1

0 .2 0 * *

�0 .1 4 *

0 .0 5

0 .5 2 * *

0 .5 9 * *

8 L M G -P

3 .7 9

1 .5 0

0 .1 0

0 .3 3 * *

0 .0 0

0 .2 4 * *

0 .4 0 * *

0 .5 4 * *

0 .4 8 * *

9 S u b o rd in at e’ s h o ri zo n ta l

co ll ec ti v is m

o ri en ta ti o n

5 .7 5

1 .1 2

�0 .1 3 *

0 .0 8

�0 .1 6 *

0 .0 4

0 .4 0 * *

0 .4 7 * *

0 .3 7 * *

0 .1 7 * *

1 0

S u b o rd in at e’ s v er ti ca l

co ll ec ti v is m

o ri en ta ti o n

6 .1 1

0 .9 6

�0 .0 9

0 .0 9

�0 .2 2 *

0 .0 9

0 .3 0 * *

0 .2 9 * *

0 .1 8 * *

0 .0 6

0 .7 0 * *

1 1

In te rp er so n al

fa ci li ta ti o n

5 .0 0

0 .9 9

0 .0 0

�0 .0 3

�0 .0 6

0 .0 8

0 .3 1 * *

0 .2 4 * *

0 .1 6 *

1 2

Jo b d ed ic at io n

5 .0 6

1 .2 3

�0 .0 5

0 .1 0

�0 .1 5 *

0 .1 4 *

0 .2 5 * *

0 .2 2 * *

0 .1 9 * *

0 .1 6 *

0 .6 3 * *

L ev el

2 (N

= 4 6 to

4 7 )

1 O ri g in al

g ro u p si ze

1 7 .5 8

1 9 .6 8

2 L M X

m ea n

5 .3 0

0 .8 4

0 .1 6

3 L M X

d if fe re n ti at io n

0 .9 2

0 .4 2

�0 .1 0

�0 .3 2 * *

4 L M G -A

m ea n

5 .0 0

0 .9 6

0 .0 4

0 .8 9 * *

�0 .3 * *

5 L M G -A

d if fe re n ti at io n

1 .0 8

0 .4 5

�0 .0 3

�0 .1 6 *

0 .6 5 * *

�0 .1 9 * *

6 L M G -D

m ea n

4 .2 8

0 .8 3

0 .1 3

0 .6 0 * *

�0 .2 7 * *

0 .7 2 * *

�0 .2 7 * *

7 L M G -D

d if fe re n ti at io n

1 .1 8

0 .5 2

0 .0 5

0 .1 4 *

0 .3 * *

0 .1 6 *

0 .3 6 * *

0 .1 7 * *

8 L M G -P

m ea n

3 .7 7

0 .9 5

�0 .0 1

0 .5 0 * *

�0 .2 8 * *

0 .6 3 * *

�0 .3 4 * *

0 .6 4 * *

�0 .0 1

9 L M G -P

d if fe re n ti at io n

1 .2 1

0 .4 7

0 .2 6

0 .1 9 * *

0 .2 6 * *

0 .1 3 *

0 .4 * *

0 .1 8 * *

0 .4 4 * *

�0 .0 4

1 0

S u p er v is o r’ s h o ri zo n ta l

co ll ec ti v is m

o ri en ta ti o n

6 .4 4

0 .5 5

�0 .1 4

0 .3 2 *

� 0 .2 4

0 .3 3 *

�0 .0 8

0 .1 6

0 .0 4

�0 .0 3

0 .0 1

1 1

S u p er v is o r’ s v er ti ca l

co ll ec ti v is m

o ri en ta ti o n

6 .4 5

0 .6 0

�0 .1 4

0 .0 9

�0 .1 0

0 .2 2

0 .0 0

0 .1 1

�0 .2 9

0 .1 1

�0 .0 7

0 .1 1

1 2

G ro u p p er fo rm

an ce

5 .4 7

0 .8 8

�0 .8 5 * *

�0 .0 4

�0 .2 2

0 .0 3

�0 .2 3

0 .0 6

�0 .3 8 * *

�0 .0 1

�0 .2 3

�0 .3 0

�0 .3 7

N o te :

S ex

w as

m ea su re d as

0 if fe m al e an d 1 if m al e.

L M X = le ad er – m em

b er

ex ch an ge ; L M G -A

= le ad er – m em

b er

g u a n xi

af fe ct iv e at ta ch m en t; L M G -D

= L M G -d ef er en ce

to le ad er s; L M G -P

= L M G -p er so na l li fe

in cl us io n .

* p < .0 5 ;

* * p < .0 1 .

686 Y. CHEN ET AL.

T ab le

3 . R es u lt s o f h y p o th es es

te st b y M p lu s.

In te rp er so n al

fa ci li ta ti o n

Jo b d ed ic at io n

G ro up

p er fo rm

an ce

M o d el

1 M o de l 2

M o de l 1

M o d el

2 M o de l 1

M o de l 2

C o ef fi ci en t

s. e.

C o ef fi ci en t

s. e.

C o ef fi ci en t

s. e.

C o ef fi ci en t

s. e.

C o ef fi ci en t

s. e.

C o ef fi ci en t

s. e.

L ev el

1 A g e

0 .0 0

0 .0 1

�0 .0 1

0 .0 1

0 .0 0

0 .0 1

S ex

�0 .0 8

0 .1 8

�0 .1 2

0 .1 8

0 .0 2

0 .2 2

�0 .0 2

0 .2 2

E d u ca ti o n

�0 .0 2

0 .0 5

�0 .0 3

0 .0 5

0 .0 4

0 .0 5

0 .0 2

0 .0 5

Y ea rs

w it h su p er v is or

0 .0 0

0 .0 1

�0 .0 1

0 .0 1

0 .0 2

�0 .0 1

0 .0 2

L M X

0 .1 5 *

0 .0 9

0 .1 5 *

0 .0 9

0 .1 3

0 .1 2

L M G -A

0 .0 2

0 .0 7

0 .0 1

0 .0 7

0 .0 1

0 .1 1

0 .0 0

0 .1 2

L M G -D

�0 .0 8

0 .0 7

�0 .0 8

0 .0 6

0 .0 1

0 .0 8

0 .0 1

0 .0 8

L M G -P

0 .1 1 **

0 .0 5

0 .2 2 ** *

0 .0 6

0 .0 5

0 .0 7

0 .2 0 * *

0 .0 9

S u b o rd in at e’ s v er ti ca l

co ll ec ti v is m

o ri en ta ti o n

0 .0 8

0 .1 2

0 .0 8

0 .1 2

0 .1 8

0 .1 4

0 .1 7

0 .1 3

S u b o rd in at e’ s h o ri zo n ta l

co ll ec ti v is m

o ri en ta ti o n

0 .1 2

0 .1 0

0 .0 9

0 .1 1

�0 .0 5

0 .1 6

�0 .0 9

0 .1 5

L M G -P

o n * em

pl o y ee

h o ri zo n ta l co ll ec ti v is m

o ri en ta ti o n

�0 .1 5 **

0 .0 7

�0 .1 8 * *

0 .0 9

L ev el

2 In te rc ep t

5 .0 2 ** *

0 .0 7

5 .0 3 ** *

0 .0 7

5 .0 6 ** *

0 .0 9

5 .0 6 * **

0 .0 9

0 .0 0

0 .1 1

0 .0 1

0 .1 1

G ro up

si ze

�0 .0 1 ** *

0 .0 0

�0 .0 1 ** *

0 .0 0

�0 .0 0

0 .0 0

0 .0 1

�0 .0 0

0 .0 0

L M X

m ea n

0 .4 3 ** *

0 .1 3

0 .4 2 ** *

0 .1 3

0 .1 4

0 .2 8

0 .1 4

0 .2 8

�0 .4 0

0 .3 0

�0 .3 5

0 .2 8

L M X

d if fe re n ti at io n

�0 .0 8

0 .3 1

�0 .0 8

0 .3 1

0 .1 8

0 .2 6

0 .1 8

0 .2 6

�0 .2 1

0 .5 2

�0 .2 3

0 .5 1

L M G -A

m ea n

�0 .2 5

0 .1 6

�0 .2 5

0 .1 6

�0 .0 1

0 .3 1

�0 .0 1

0 .3 1

0 .4 7

0 .1 7

0 .3 3

0 .3 8

L M G -A

d if fe rr en ti at io n

�0 .2 0

.0 .2 9

�0 .2 0

0 .2 9

�0 .4 2

0 .3 4

�0 .4 2

0 .3 4

�0 .1 2

0 .4 9

�0 .0 5

0 .4 6

L M G -D

m ea n

0 .3 5 **

0 .1 8

0 .3 5 **

0 .1 8

0 .4 2 **

0 .1 8

0 .4 2 * *

0 .1 8

0 .1 9

0 .1 7

0 .2 4

0 .1 6

L M G -D

d if fe rr en ti at io n

�0 .1 3

0 .1 3

�0 .1 2

0 .1 3

�0 .1 1

0 .1 9

�0 .1 1

0 .1 9

�0 .6 5 *

0 .3 8

�0 .5 6

0 .3 4

L M G -P

m ea n

�0 .1 7

0 .1 2

�0 .1 7

0 .1 2

�0 .0 7

0 .1 9

�0 .0 7

0 .1 9

�0 .3 2

0 .2 1

�0 .3 0

0 .2 0

L M G -P

d if fe rr en ti at io n

�0 .0 6

0 .2 0

�0 .0 7

0 .2 0

�0 .1 9

0 .2 4

�0 .2 0

0 .2 5

�0 .1 3

0 .3 4

�0 .2 1

0 .3 1

L M G -P

d if fe re nt ia ti o n *

L M G -P

�0 .2 0

0 .1 5

�0 .3 0 *

0 .1 8

S u p er v is o r’ s h o ri zo n ta l

co ll ec ti v is m

o ri en ta ti o n

�0 .1 5

0 .1 9

�0 .0 2

0 .2 1

S u p er v is o r’ s v er ti ca l

co ll ec ti v is m

o ri en ta ti o n

�0 .0 0

0 .1 7

0 .0 3

0 .1 7

L M G -P

d if fe re nt ia ti o n *

su p er vi so r’ s h o ri zo n ta l

co ll ec ti v is m

o ri en ta ti o n

0 .2 1 *

0 .1 1

N o te : n (l ev el

1 ) = 2 1 3 ; n (l ev el

2 ) = 4 7 .

S ex

w as

m ea su re d as

0 if fe m al e an d 1 if m al e.

L M G -A

= le ad er – m em

b er

af fe ct iv e at ta ch m en t; L M G -D

= d ef er en ce

to le ad er s; L M G -P

= le ad er – m em

b er

p er so n al li fe

in cl us io n .

* p < 0 .1 0 ;

* * p < 0 .0 5 ;

* * *p

< 0 .0 1 .

SOCIAL EXCHANGE SPILLOVER 687

Hypotheses 2a and 2b predicted that subordinates’ horizontal collectivism orientation would moderate the positive relationship between LMG-P and interpersonal facilitation (2a) and job dedication (2b), such that the rela- tionship would become weaker when the horizontal collectivism orientation was high. The multilevel modeling re- sults demonstrated a negative effect of the interaction between LMG-P and subordinates’ horizontal collectivism orientation on interpersonal facilitation (γ =�0.15, p < 0.05). Following Cohen, Cohen, West, and Aiken’s (2003) approach, we plotted this interaction at conditional values of 1 SD above and below the mean (Figure 2). Simple slope analyses showed that LMG-P was related positively to interpersonal facilitation when subordinates’ collectiv- ism orientation was low (γ= 0.37, t= 3.53, p< 0.01), whereas LMP-G was not related to interpersonal facilitation when collectivism orientation was high (γ =0.08, t =0.84, p = 0.40). Therefore, Hypothesis 2a was supported. The multilevel modeling results also demonstrated a negative effect of the interaction between LMG-P and

subordinates’ horizontal collectivism orientation on job dedication (γ =0.18, p< 0.05). We plotted this interaction as well at conditional values of 1 SD above and below the mean (Figure 3). Simple slope analyses showed that

Low LMG-P High LMG-P

In te

rp er

so n

al F

ac il

it at

io n

Low Subordinate's Horizontal Collectivism Orientation High Subordinate's Horizontal Collectivism Orientation

Figure 2. Subordinate’s horizontal collectivism orientation as a moderator of the relationship between leader–member personal life inclusion (LMG-P) and interpersonal facilitation

Low LMG-P High LMG-P

Jo b

D ed

ic at

io n

Low Subordinate's Horizontal Collectivism Orientation High Subordinate's Horizontal Collectivism Orientation

Figure 3. Subordinate’s horizontal collectivism orientation as a moderator of the relationship between leader–member personal life inclusion (LMG-P) and job dedication

688 Y. CHEN ET AL.

LMG-P was related positively to job dedication when the individual’s horizontal collectivism orientation was low (γ= 0.38, t =2.62, p <0.01), whereas LMP-G was not related to job dedication when the horizontal collectivism orientation was high (γ =0.02, t =0.21, p = 0.83). Hence, Hypothesis 2b was supported. Hypotheses 3a and 3b predicted that LMG-P differentiation moderates the relationship between LMG-P and in-

terpersonal facilitation, as well as job dedication, such that the relationships would become weaker when LMG-P differentiation was high. The cross-level modeling results showed a negative effect of LMG-P differentiation on the random slope between LMG-P and interpersonal facilitation, but the effect was not significant (γ= �0.20, p> 0.10) Hence, Hypothesis 3a was not supported, although it was in the direction we predicted. The cross-level modeling results showed that the cross-level interaction effect on job dedication was negative and

marginally significant (γ= �0.30, p < 0.10). We plotted this interaction at conditional values of 1SD above and below the mean (Figure 4). Simple slope analyses showed that LMG-P was related positively to job dedication when LMG-P differentiation was low (γ= 0.50, t= 2.15, p < 0.05) but was not when LMG-P differentiation was high (γ= �0.10, t= �0.67, p = 0.50). Hence, Hypothesis 3b was partially supported. Hypothesis 4 predicted that supervisors’ horizontal collectivism orientation would moderate the relationship

between LMG-P differentiation and group performance, such that the relationship would be negative only when the supervisor was low in horizontal collectivism. The multilevel modeling results showed a positive effect of the interaction between LMG-P differentiation and supervisors’ horizontal collectivism orientation on group performance (γ= 0.21, p< 0.10). We plotted this interaction at conditional values of 1 SD above and below the mean (Figure 5). Simple slope analyses demonstrated that, when the supervisors’ horizontal collectivism orientation was high, LMG-P differentiation was not related to group performance (γ= 0.00, t =0, p = 1.0), whereas when horizontal collectivism orientation was low, LMG-P differentiation was related negatively to group performance (γ= �0.42, t= �2. 05, p< 0.05). Hence, Hypothesis 4 was also partially supported.

Discussion

In this paper, we developed and tested a multilevel social exchange spillover model in leader–member relations in the Chinese context. Specifically, we proposed that, at the individual level, the effects of LMG-P in the private

Low LMG-P High LMG-P

Jo b

D ed

ic at

io n

Low LMG-P differentiation

High LMG-P differentiation

Figure 4. Leader–member personal life inclusion (LMG-P) differentiation as a moderator of the relationship between LMG-P and job dedication

SOCIAL EXCHANGE SPILLOVER 689

domain could spill over to the work domain to influence contextual performance. We argued that such spillover effects would be contingent upon individuals’ cultural values of horizontal collectivism orientation. At the cross level, LMG-P differentiation moderated the relationship between LMG-P and contextual performance. At the group level, we proposed that LMG-P differentiation would be related negatively to group performance when supervisors had a low horizontal collectivism orientation. Our results generally supported the hypothesized spillover effects of LMG-P on contextual performance. In terms of cultural values as contingencies of LMG-P, the positive relationships between LMG-P and interpersonal facilitation and job dedication were significant only for those who had a low horizontal collectivism orientation. At the group level, LMG-P differentiation was related negatively to group performance when supervisors had a low horizontal collectivism orientation. At the cross level, LMG-P differentiation weakened the positive relationship between LMG-P and job dedication.

Theoretical implications

This study has three major theoretical implications. First, it enriches current LMX theory by demonstrating that an indigenous construct (i.e., LMG-P) can contribute to the development of a more global leader–member relationship theory. Our findings showed that LMG-P, as a unique aspect of leader–member relationships in China, explains additional variance in interpersonal facilitation and job dedication, beyond the effects of LMX. Thus, it is critical to study leader–member relationships in Chinese organizations using both the universal (i.e., LMX) and indigenous aspects (i.e., LMG-P) to examine the effects of leader–member relationships on work outcomes. Moreover, although both LMX and LMG build their theoretical arguments on social exchange theory, neither has adopted a boundary spillover perspective through the lens of role expansion and role differentiation. Thus, the theory lacks some of the complexities and dynamics of the interpenetration of leader–member relations across the private and work domains. Our results showed that LMG-P was related positively to two dimensions of contextual performance. These findings demonstrate that when supervisors and subordinates develop high-quality LMG-P, role expansion of subordinates from the work to the private domain can indeed lead to the expansion of role behavior requirements in the work domain. By socializing with a supervisor, subordinates’ work roles are redefined to include pan-family roles, and their role behavior requirements also expand to include contextual performance, because of the strong ob- ligations associated with pan-family roles.

Low LMG-P Differentiation

High LMG-P Differentiation

G ro

u p

P er

fo rm

an ce

Low Supervisor's Horizontal Collectivism Orientation High Supervisor's Horizontal Collectivism Orientation

Figure 5. Supervisor’s horizontal collectivism orientation as a moderator of the relationship between leader–member personal life inclusion (LMG-P) differentiation and group performance

690 Y. CHEN ET AL.

Second, this study highlighted the boundary conditions of social exchange spillover from the private to the work domain, thus deepening our understanding of the relationship between LMG-P and contextual performance. The majority of previous studies has stressed either the positive or negative relationship between LMG and work outcomes, ignoring its possible dampening or enhancing effects on such relationships as employees’ cultural value orientations. The findings of this study showed that employees’ horizontal collectivism orientation serves as a boundary condition for the positive spillover from LMG-P to contextual performance. Our moderated findings add to the LMG literature and suggest that for those who have a high horizontal collectivism orientation, socializing with their supervisors may not be an effective relational motivator for them to increase contextual performance in the workplace. Another important boundary condition is LMG-P differentiation. The results of our multilevel model with random slope analysis showed that LMG-P was related positively to job dedication when LMG-P differentia- tion was low. Under those conditions, subordinates were less concerned about favoritism and uncertainty about their own insider status. Thus, when they were included in the supervisor’s personal life, the subordinates were motivated to work harder to reciprocate with the supervisor by enhancing their contextual performance. This study sheds light on the long-term debate about whether or not LMG is good or bad for an organization (Chen & Chen, 2012; Warren, Dunfee, & Li, 2004; Fan, 2002). Our findings showed that the effects of LMG-P on work outcomes are contingent upon contextual factors, such as the degree of LMG-P differentiation and supervisor’s horizontal collectivism orientation. Third, our findings demonstrated the importance of exploring the effects of LMG on work outcomes from a multi-

level perspective. To date, group-level LMG has not been well studied. A recent paper on guanxi reviewed more than 200 journal articles and found that there was only one study on group-level LMG (Chen et al., 2013). Although there was a call long ago to study the effects on group performance of the existence of multiple informal guanxi ties among multiple group members (e.g., Tsui & Farh, 1997), to the best of our knowledge, our study is the first to explore the relationship between group-level LMG, that is, LMG-P differentiation and group performance. When only the dyadic level of LMG is considered, after controlling for LMX, it has been reported to relate positively to certain work attitudes and behaviors, such as job satisfaction (Wong, Tinsley, Law, & Mobley, 2003), organizational affective commitment (Chen et al., 2009), and organizational citizenship behaviors (Shih & Lin, 2014). However, when both the dyadic LMG-P and LMG-P differentiation at the group level were considered, as we did in this study, our more nuanced find- ings demonstrated the complexities of the LMG phenomenon. In this study, we found that LMG-P was associated pos- itively with contextual performance at the dyadic level. However, at the group level, LMG-P differentiation was related negatively to group performance when supervisors had a low horizontal collectivism orientation.

Managerial implications

The findings of our study have important managerial implications for organizations that operate in China. First, su- pervisors may want to pay attention to the boundary constraints of the effects of LMG-P on work outcomes. As part of both folk wisdom and cultural traditions, Chinese leaders and members recognize the importance of social ex- change in the nonwork domain. They are well motivated to build guanxi by including each other in their social and private lives to achieve personal gains or address job or organizational concerns (Zhang, Deng, & Wang, 2013). This study showed that the effects of LMG-P on work outcomes have constraints. On the one hand, LMG-P is related positively to contextual performance that contributes to organizational accomplishment. On the other hand, LMG-P may not necessarily relate to contextual performance among those subordinates who have a high horizontal collectivism orientation. In order to promote the contextual performance of subordinates with high hori- zontal collectivism orientation, supervisors may want to stress the importance of group goals and interests rather than spending too much time socializing with them in the private domain. Second, Chinese organizations may need to be aware of both the negative and positive outcomes of LMG-P.

Because of the spillover effects of LMG-P in the work domain, organizations may make use of the benefits that high-quality LMG can bring to the effectiveness of the organization while establishing human resource policies to

SOCIAL EXCHANGE SPILLOVER 691

prevent the potential detrimental effects that LMG may cause. At the group level, the relational structure of LMG-P, which is LMG-P differentiation, may affect group outcomes adversely when supervisors have fewer concerns about members’ interests and well-being.

Limitations and future research

There are several limitations to this research. First, all of the group-level variables come from the group supervisor, while the individual-level independent variables were derived from the subordinates, which may raise concerns about common method variance. Although this is a legitimate concern, all of our individual-level dependent variables were rated by the group supervisors. Further, several hypotheses focused on the moderating effects of horizontal collectivism orientation and moderating effects are less likely to be affected by common method variance (Chang, Van Witteloostuijn, & Eden, 2010). Still, in future research that explores the relationship between LMG-P differentiation and group performance, it would be worthwhile to ask the group supervisor’s leader to evaluate group performance. Second, we regarded LMG-P as an indigenous construct, but similar phenomena of LMG-P may also exist in

other cultures (Smith, Huang, Harb, & Torres, 2011; Zorn, 1995). Thus, it would be beneficial to examine the spillover effects found in this study in other cultural contexts to see whether they hold. We share the belief that indigenous studies are meaningful, not only to local people and in local contexts but also to build and develop universal theories (Yang, 2000). We speculate that social exchange spillover effects in leader–member relations are universal, but the extent of such spillover may depend on cultural norms, organizational policies, and individual characteristics. Third, we explored the horizontal collectivism orientation of subordinates and supervisors only as value-related

contingency factors. However, individuals may also differ in their preferences for the integration between their work and private lives, as well as their power distance orientation. Future research should explore the possible moderating effects of these individual differences when examining the relationship between LMG-P and employee outcomes.4

Fourth, we assumed that supervisor-rated contextual performance would reflect how well subordinates engage in extra-role behaviors that can benefit the effectiveness of the organization. However, it is possible that supervisor- rated contextual performance is inflated because of supervisor bias, as has been shown for task performance (Ma & Qu, 2010). Thus, when studying the relationship between LMG-P and contextual performance, future research should try to obtain objective data for measuring contextual performance. Another potential limitation of this study is that supervisors rated the performance of their own groups, and the

fact that this was used as a dependent variable when examining the moderating effects of supervisory collectivism could influence the results. Regardless of the boundary condition, supervisors who possess a high horizontal collec- tivism orientation are more likely to evaluate their group’s performance higher compared with those with a lower horizontal collectivism orientation. Thus, we suggest that, in future research that explores the relationship between LMG-P differentiation and group performance, it would be worthwhile to use objective or third-person measures of group performance. Moreover, in this study, we focused not only on individual-level performance but on group- level performance as well. The rationale for shifting our focus from individual to group performance was based on the assumption that group-level performance is dependent on members’ individual collective efforts to create a better outcome; thus, group-level performance can be attained through cooperation and coordination among team members (Li & Liao, 2014). Given the importance of cooperation and coordination in group outcomes (e.g., Bass, 1985; Hackman & Morris, 1975; Weingart, 1997), future research may want to explore the boundary condition of the relationship between individuals’ coordination and cooperation, and group performance, such as the types of tasks (e.g., interdependent or independent) used as the basis for evaluations of group performance.5

4We thank an anonymous reviewer for this insight. 5We thank an anonymous reviewer for this insight.

692 Y. CHEN ET AL.

In addition, our data are cross-sectional, and thus, we cannot make statements about causality. It is possible that employees’ contextual performance may gain leaders’ attention and further invite the high performers into leaders’ personal lives as a reward or to demonstrate affective trust. However, ethnographic studies have shown that employees are motivated to contribute to the interests of the group and organization when they initiate, develop, and maintain high-quality LMG with the leader (Jiang, Chen, & Shi, 2013). Future studies should employ a multistate, longitudinal design that includes newcomers to the organization in order to explore fully the spillover effects of leader–member relations from the private to the work domain or the converse. Finally, our study examined the spillover effects only from the subordinates’ perspectives. Future research should

also explore this effect from the supervisors’ perspectives. It is possible that high-quality LMG-P may enhance the relationship a leader has with a member and such a positive effect may spill over to influence managerial decisions. More research is also needed to explore further the mechanism between LMG-P differentiation and group outcomes.

Conclusions

In this study, we developed and tested a social exchange spillover model in leader–member relations based on role theory and the cultural theory of relational collectivism. Our findings demonstrated that the effects of LMG-P in the private domain can spill over to affect members’ contextual performance in the work domain. Moreover, LMG-P differentiation is related negatively to group performance when supervisors have a low horizontal collectivism orientation. We believe this line of research is important to deepen our understanding of the interpenetration of leader–member relations across the private and the work domains.

Acknowledgements

The authors thank Bor-Shiuan Cheng, Ding-Yu Jiang, Amit Kramer, Jiade Luo, Dan Newman, Joey Tsai, Andrew Wang, and Tsung-Yu Wu for their help and valuable comments on this study. This research is partially supported by two grants from the National Natural Science Foundation of China (71372143 and 71272123).

Author biographies

Ying Chen is an assistant professor at the School of Labor and Employment Relations at the University of Illinois at Urbana-Champaign. She received her PhD in organization studies at Vanderbilt University. Her research interests include leader-member exchange, Chinese guanxi, and labor relations. Zhen Xiong Chen is Professor of Management at the Research School of Management, The Australian National University. He received his PhD from Hong Kong University of Science and Technology. His research centers on organizational behavior in cross-cultural contexts. His work has been published in Academy of Management Journal, Journal of Applied Psychology, Journal of Organizational Behavior, and others. Lifeng Zhong is an associate professor in the School of Business at Renmin University of China. He received his PhD at the Chinese Academy of Sciences. His research focuses on strategic human resource management, employee–organization relationships, and leader-follower relationships.

SOCIAL EXCHANGE SPILLOVER 693

Jooyeon Son is a PhD candidate at the School of Labor and Employment Relations at the University of Illinois, Urbana-Champaign. She conducts research in the areas of gender, team diversity, leader-member exchange, and talent management. Xiujuan Zhang is an associate professor at the School of Business, Sun Yat-sen University of China. She received her PhD in business administration from Sun Yat-sen University. Her research interests focus on leadership, organizational justice, employee emotion, and voice. Zhiqiang Liu is an associate professor in the School of Management, Huazhong University of Science and Tech- nology, where he also received his PhD. His current interests include leadership, status competition, and creativity. His papers have appeared in the Journal of Organizational Behavior, Journal of Vocational Behavior, Asian Pacific Journal of Management, Journal of Business Ethics, among others.

References Bass, B. M. (1985). Leadership and performance beyond expectations. New York: Free Press. Bozionelos, N., & Wang, L. (2007). An investigation on the attitudes of Chinese workers towards individually based performance-related reward systems. The International Journal of Human Resource Management, 18(2), 284–302.

Brewer, M. B., & Chen, Y. R. (2007). Where (who) are collectives in collectivism? Toward conceptual clarification of individualism and collectivism. Psychological Review, 114(1), 133–151.

Brislin, R. W. (1980). Translation and content analysis of oral and written material. Handbook of cross-cultural psychology, 2(2), 349–444.

Chan, D. (1998). Functional relations among constructs in the same content domain at different levels of analysis: A typology of composition models. Journal of Applied Psychology, 83(2), 234–246.

Chang, S. J., Van Witteloostuijn, A., & Eden, L. (2010). From the editors: Common method variance in international business research. Journal of International Business Studies, 41(2), 178–184.

Chen, C. C. (1995). New trends in rewards allocation preferences: A Sino-US comparison. Academy of Management Journal, 38(2), 408–428.

Chen, C. C., & Chen, X. P. (2009). Negative externalities of close guanxi within organizations. Asia Pacific Journal of Management, 26(1), 37–53.

Chen, C. C., Chen, X. P., & Huang, S. (2013). Chinese guanxi: An integrative review and new directions for future research. Management and Organization Review, 9(1), 167–207.

Chen, X. P., & Chen, C. C. 2012. Chinese guanxi: The good, the bad, and the controversial. In X. Huang, & M. H. Bond (Eds.), The handbook of Chinese organizational behaviour: Integrating theory, research and practice (pp. 415–435). Cheltenham: Edward Elgar.

Chen, C. C., Chen, Y. R., & Xin, K. (2004). Guanxi practices and trust in management: A procedural justice perspective. Organization Science, 15(2), 200–209.

Chen, Y., Friedman, R., Yu, E., Fang, W., & Lu, X. (2009). Developing a three-dimensional model and scale for supervisor- subordinate guanxi. Management and Organization Review, 5(3), 375–399.

Chen, Y., Friedman, R., Yu, E., & Sun, F. (2011). Examining the positive and negative effects of guanxi practices: A multi-level analysis of guanxi practices and procedural justice perceptions. Asia Pacific Journal of Management, 28(4), 715–735.

Chen, N. Y. F., & Tjosvold, D. (2007). Guanxi and leader member relationships between American managers and Chinese employees: Open-minded dialogue as mediator. Asia Pacific Journal of Management, 24(2), 171–189.

Chen, Y., Yu, E., & Son, J. (2014). Beyond leader–member exchange (LMX) differentiation: An indigenous approach to leader–member relationship differentiation. The Leadership Quarterly, 25(3), 611–627.

Cheng, J. L. C. (1983). Interdependence and coordination in organizations: A role-system analysis. Academy of Management Journal, 26, 156–162.

Cohen, J., Cohen, P., West, S., & Aiken, L. (2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Hillsdale, NJ: Erlbaum.

Confucius (1915). The ethics of Confucius. New York: Putnam. Cross, S. E., Gore, J. S., & Morris, M. L. (2003). The relational-interdependent self-construal, self-concept consistency, and well-being. Journal of Personality and Social Psychology, 85(5), 933–944.

Crouter, A. C. (1984). Spillover from family to work: The neglected side of the work-family interface. Human Relations, 37(6), 425–441.

694 Y. CHEN ET AL.

Earley, P. C. (1993). East meets West meets Mideast: Further explorations of collectivistic and individualistic work groups. Academy of Management Journal, 36(2), 319–348.

Edwards, J. R., & Rothbard, N. P. (2000). Mechanisms linking work and family: Clarifying the relationship between work and family constructs. Academy of Management Review, 25, 178–199.

Enders, C. K., & Tofighi, D. (2007). Centering predictor variables in cross-sectional multilevel models: A new look at an old issue. Psychological Methods, 12, 121–138.

Fan, Y. (2002). Ganxi’s consequences: Personal gains at social cost. Journal of Business Ethics, 38(4), 371–380. Farh, J. L., Tsui, A. S., Xin, K., & Cheng, B. S. (1998). The influence of relational demography and guanxi: The Chinese case. Organization Science, 9(4), 471–488.

Graen, G. B., & Scandura, T. A. (1987). Toward a psychology of dyadic organizing. Research in Organizational Behavior, 9, 175–208.

Hackman, J. R., & Morris, C. G. (1975). Group tasks, group interaction process, and group performance effectiveness: A review and proposed integration. In L. Berkowitz (Ed.), Advances in experimental social psychology (vol. 8, pp. 1–99). New York: Academic Press.

Hall, R. J., Snell, A. F., & Foust, M. S. (1999). Item parceling strategies in SEM: Investigating the subtle effects of unmodeled secondary constructs. Organizational Research Methods, 2(3), 233–256.

Hanson, G. C., Hammer, L. B., & Colton, C. L. (2006). Development and validation of a multidimensional scale of perceived work–family positive spillover. Journal of Occupational Health Psychology, 11(3), 249–265.

He, W., Chen, C., & Zheng, L. (2004). Rewards allocation preferences of employees in Chinese state-owned enterprises: Effects of ownership reform and collectivism. Organization Science, 15(2), 221–231.

Hofmann, D. A., & Gavin, M. B. (1998). Centering decisions in hierarchical linear models: Implications for research in organizations. Journal of Management, 24(5), 623–641.

Hui, C., & Graen, G. (1997). Guanxi and professional leadership in contemporary Sino-American joint ventures in mainland China. The Leadership Quarterly, 8(4), 451–465.

Hui, C., Lee, C., & Wang, H. (2014). Organizational inducements and employee citizenship behavior: The mediating role of perceived insider status and the moderating role of collectivism. Human Resource Management, 54(3), 439–456.

Hwang, K. K. (1987). Face and favor: The Chinese power game. American Journal of Sociology, 92, 944–974. Ilies, R., Nahrgang, J. D., & Morgeson, F. P. (2007). Leader-member exchange and citizenship behaviors: A meta-analysis. Journal of Applied Psychology, 92(1), 269.

Jiang, X., Chen, C. C., & Shi, K. (2013). Favor in exchange for trust? The role of subordinates’ attribution of supervisory favors. Asia Pacific Journal of Management, 30(2), 513–536.

Jöreskog, K. G., & Sorbom, D. (1986). LISREL IV: Analysis of linear structural relationships by maximum likelihood, instrumental variables and least squares methods (4th ed.). Mooresville, IN: Scientific Software.

Kahn, R. L., Wolfe, D. M., Quinn, P. R., Snoek, J. D., & Rosenthal, R. A. (1964). Organizational stress: Studies in role conflict and ambiguity. New York: Wiley.

Katz, D., & Kahn, R. L. (1978). The social psychology of organizations (2nd ed.). New York: Wiley. Lam, C., & Huang, X. (2012). Managing social comparison processes among Chinese employees. In Huang, X., & Bond, M. (Eds.), Handbook of Chinese organizational behavior: Integrating theory, research and practice (ed., pp. 118–139). Cheltenham: Edward-Elgar.

Law, K. S., Wong, C. S., Wang, D., & Wang, L. (2000). Effect of supervisor–subordinate guanxi on supervisory decisions in China: An empirical investigation. International Journal of Human Resource Management, 11(4), 751–765.

LePine, J. A., Erez, A., & Johnson, D. E. (2002). The nature and dimensionality of organizational citizenship behavior: A critical review and meta-analysis. Journal of Applied Psychology, 87(1), 52.

Li, A. N., & Liao, H. (2014). How do leader–member exchange quality and differentiation affect performance in teams? An integrated multilevel dual process model. Journal of Applied Psychology, 99(5), 847–866.

Liden, R. C., Erdogan, B., Wayne, S. J., & Sparrowe, R. T. (2006). Leader-member exchange, differentiation, and task interdependence: Implications for individual and group performance. Journal of Organizational Behavior, 27(6), 723–746.

Liden, R., & Maslyn, J. (1998). Multidimensionality of leader-member exchange: An empirical assessment through scale development. Journal of Management, 24(1), 43–72.

Liu, J., Hui, C., Lee, C., & Chen, Z. X. (2013). Why do I feel valued and why do I contribute? a relational approach to employee’s organization-based self-esteem and job performance. Journal of Management Studies, 50(6), 1018–1040.

Ma, L., & Qu, Q. (2010). Differentiation in leader–member exchange: A hierarchical linear modeling approach. The Leadership Quarterly, 21(5), 733–744.

Motowidlo, S. J., & Van Scotter, J. R. (1994). Evidence that task performance should be distinguished from contextual performance. Journal of Applied Psychology, 79(4), 475.

Muthén, B. O. (1991). Multilevel factor analysis of class and student achievement components. Journal of Educational Measurement, 28, 338–354.

SOCIAL EXCHANGE SPILLOVER 695

Muthén, L. K., & Muthén, B. O. (1998–2011). Mplus user’s guide (5th ed.). Los Angeles, CA: Muthén & Muthén. Naidoo, L. J., Scherbaum, C. A., Goldstein, H. W., & Graen, G. (2011). A longitudinal examination of LMX, ability, differentiation and team performance. Journal of Business and Psychology, 26(3), 347–357.

Osigweh, C. A. B., & Huo, P. Y. (1993). Conceptions of employee responsibilities and rights in the United States and the People’s Republic of China. International Journal of Human Resource Management, 4(1), 85–112.

Pearlin, L., & Kohn, M. (1966). Social class, occupational, and parental values: A cross national study. American Sociological Review, 31, 466–479.

Rockstuhl, T., Dulebohn, J. H., Ang, S., & Shore, L. M. (2012). Leader–member exchange (LMX) and culture: A meta-analysis of correlates of LMX across 23 countries. Journal of Applied Psychology, 97(6), 1097–1130.

Ruderman, M. N., Ohlott, P. J., Panzer, K., & King, S. (2002). Benefits of multiple roles for managerial women. Academy of Management Journal, 45, 369–386.

Sanchez-Burks, J., Lee, F., Choi, I., Nisbett, R., Zhao, S., & Koo, J. (2003). Conversing across cultures: East–west communica- tion styles in work and nonwork contexts. Journal of Personality and Social Psychology, 85(2), 363–372.

Shih, C. T., & Lin, C. C. T. (2014). From good friends to good soldiers: A psychological contract perspective. Asia Pacific Journal of Management, 31(1), 309–326.

Singelis, T. M., Triandis, H. C., Bhawuk, D. P., & Gelfand, M. J. (1995). Horizontal and vertical dimensions of individualism and collectivism: A theoretical and measurement refinement. Cross-cultural research, 29(3), 240–275.

Smith, P., Huang, H. J., Harb, C., & Torres, C. (2011). How distinctive are indigenous ways of achieving influence? A comparative study of guanxi, wasta, jeitinho, and “pulling strings”. Journal of Cross-Cultural Psychology, 43(1), 135–150.

Smith, P. B., Wasti, S. A., Grigoryan, L., Achoui, M., Bedford, O., Budhwar, P., et al. (2014). Are guanxi-type supervisor– subordinate relationships culture-general? An eight-nation test of measurement invariance. Journal of Cross-Cultural Psychology, 45(6), 921–938.

Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. London, England: Sage.

Triandis, H. C., & Gelfand, M. J. (1998). Converging measurement of horizontal and vertical individualism and collectivism. Journal of Personality and Social Psychology, 74(1), 118–128.

Trompenaars, F. (1993). Riding the waves of culture: Understanding cultural diversity in business. London: The Economist Books.

Tsui, A. S. (2006). Contextualization in Chinese management research. Management and Organization Review, 2(1), 1–13. Tsui, A. S. (2012). Contextualizing research in a modernizing China. In X. Huang & M. H. Bond (Eds.), Handbook of Chinese organizational behavior: Integrating theory, research and practice (pp. 13–28). Cheltenham: Edward Elgar Publishing.

Tsui, A. S., & Farh, J. L. L. (1997). Where guanxi matters relational demography and guanxi in the Chinese context. Work and Occupations, 24(1), 56–79.

Van den Bos, K., & Lind, E. A. (2001). The psychology of own versus others’ treatment: Self-oriented and other-oriented effects on perceptions of procedural justice. Personality and Social Psychology Bulletin, 27(10), 1324–1333.

Van Der Vegt, G. S., & Bunderson, J. S. (2005). Learning and performance in multidisciplinary teams: The importance of collective team identification. Academy of Management Journal, 48(3), 532–547.

Van Scotter, J. R., & Motowidlo, S. J. (1996). Interpersonal facilitation and job dedication as separate facets of contextual performance. Journal of Applied Psychology, 81(5), 525–531.

Walder, A. G. (1986). Communist neo-traditionalism: Work and authority in Chinese industry. Berkeley: University of California Press.

Warren, D. E., Dunfee, T. W., & Li, N. (2004). Social exchange in China: The double-edged sword of guanxi. Journal of Business Ethics, 55(4), 353–370.

Weingart, L. (1997). How did they do that? The ways and means of studying group process. In L. L. Cummings & B. Staw (Eds.), Research in organizational behavior (vol. 19, pp. 189–239). Greenwich, CT: JAI Press.

Wong, C. S., Tinsley, C., Law, K. S., & Mobley, W. H. (2003). Development and validation of a multidimensional measure of guanxi. Journal of Psychology in Chinese Societies, 4(1), 43–69.

Yang, K. S. 1992. Chinese social orientation: From the social interaction perspective. In K. S. Yang & A. B. Yu (Eds.), Chinese psychology and behaviour (pp. 87–142). Laurel, Taipei: Kew Gong Book.

Yang, K. S. (2000). Monocultural and cross-cultural indigenous approaches: The royal road to the development of a balanced global psychology. Asian Journal of Social Psychology, 3(3), 241–263.

Zhang, L., Deng, Y., & Wang, Q. (2013). An exploratory study of Chinese motives for building supervisor–subordinate guanxi. Journal of Business Ethics, 124(4), 1–17.

Zorn, T. E. (1995). Bosses and buddies: Constructing and performing simultaneously hierarchical and close friendship relationships. Understanding relationship processes, 6, 122–147.

Zhou, J., & Martocchio, J. J. (2001). Chinese and American manager’s compensation award decisions: A comparative policy capturing study. Personnel Psychology, 54(1), 115–145.

696 Y. CHEN ET AL.

Appendix

Three-dimensional leader–member guanxi measure (Chen et al., 2009)

Affective attachment 1. My supervisor and I always share thoughts, opinions, and feelings toward work and life. 2. I feel easy and comfortable when I communicate with my supervisor 3. I would feel sorry and upset if my supervisor decided to work for another company. 4. If my supervisor has problems with his/her personal life I will do my best to help him/her out.

Personal life inclusion 1. My supervisor would ask me to help him/her deal with some personal/family errands. 2. During holidays my supervisor and I would call each other or visit each other. 3. After the office hour, I have social activities together with my supervisor such as having dinner together or having

entertainment together which go beyond work duties. 4. I am well acquainted with the family members of my supervisor and have personal contact with these members.

Deference to supervisor 1. I am willing to obey my supervisor unconditionally. 2. While I disagree with my supervisor, I would still support his/her decisions. 3. When my goal is in conflict with my supervisor’s goal, I am willing to give up my goal in order to fulfill my supervisor’s goal. 4. I am willing to sacrifice my interests in order to fulfill my supervisor’s interests.

SOCIAL EXCHANGE SPILLOVER 697

Copyright of Journal of Organizational Behavior is the property of John Wiley & Sons, Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.