Professional Documents
Culture Documents
Using cross-cultural laboratory and field studies with samples of leaders, employees, and students from
the United States and the People’s Republic of China, we examined how team-level stimuli, including
empowering leadership and relationship conflict, combine to influence individual members’ motivational
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
states of psychological empowerment and affective commitment. As predicted, we found that these
This document is copyrighted by the American Psychological Association or one of its allied publishers.
motivational states are individually and jointly influenced by teams’ level of empowering leadership and
relationship conflict and that these motivational states mediate the relationships between team stimuli and
team members’ innovative and teamwork behaviors and turnover intentions. In addition, results held
despite controlling for team members’ nationality and collectivism. We discuss contributions of our study
to the team motivation, conflict, and stress literatures.
To accomplish key objectives, organizations continue to rely on Despite progress in the teams and motivation literatures (see
work teams, defined as “a distinguishable set of two or more Chen & Gogus, 2008; Kanfer, Chen, & Pritchard, 2008; Kozlow-
people who interact, dynamically, interdependently, and adap- ski & Bell, 2003; Mathieu et al., 2008), important gaps remain in
tively toward a common and valued goal/objective/mission” (Sa- the understanding of members’ motivation to contribute to their
las, Dickinson, Converse, & Tannenbaum, 1992, p. 4). The liter- team’s effectiveness. Our article’s purpose is to address three
ature on teams has recognized that team effectiveness is a function specific gaps in the understanding of how and when teams exert
of multilevel processes (Kozlowski & Bell, 2003). In particular, contextual influences on members’ motivation to contribute to
team-level stimuli, such as team leadership and other team pro- their teams.
cesses, exert nontrivial influences on individual members’ behav- First, despite classic research suggesting that team dynamics
ior (Hackman, 1992). In turn, individual members can contribute (e.g., leader behaviors, team norms) can impact team members’
to team effectiveness in various ways (see Mathieu, Maynard, behavior (Hackman, 1992; Lewin, 1947), only limited research has
Rapp, & Gilson, 2008). Among these is generating and implement- directly examined the relationships between team and leadership
ing novel ideas and procedures, termed innovative behavior (e.g., stimuli and team members’ motivational reactions and, particu-
Anderson, De Dreu, & Nijstad, 2004), and helping team members larly, team members’ motivational states (i.e., how members con-
coordinate their efforts to ensure team success, termed teamwork ceive of their work environment and their capacity to function
behavior (e.g., Welbourne, Johnson, & Erez, 1998). Furthermore, effectively in their work environment; see Chen & Gogus, 2008;
members’ willingness to remain with their team and organization Zaccaro, Ely, & Nelson, 2008). This is a critical limitation since
(i.e., low turnover intentions) helps to maintain teams’ viability or motivational states likely serve as key mediators between team-
ability to function well over time (Hackman, 1987). level stimuli and individual team members’ behaviors (Chen &
Kanfer, 2006). As such, there remains a gap in the understanding
of why and how team-level stimuli affect team member behaviors.
To address this first gap, we delineated and empirically tested a
This article was published Online First December 20, 2010.
Gilad Chen, Payal Nangia Sharma, Suzanne K. Edinger, and Debra L.
cross-level model (see Figure 1) that specifies how and when team
Shapiro, Management & Organization Department, Robert H. Smith members’ motivational states relate to team-level stimuli of em-
School of Business, University of Maryland; Jiing-Lih Farh, Department of powering leadership (i.e., extent to which leaders enhance auton-
Management, Hong Kong University of Science and Technology, Hong omy, control, self-management, and confidence in their teams;
Kong, China. Chen, Lam, & Zhong, 2007; Kirkman & Rosen, 1999; Konczak,
The work described in this article was supported by a grant from the Stelly, & Trusty, 2000) and relationship conflict (i.e., tension,
Research Grants Council of the Hong Kong Special Administrative Re- annoyance, and animosity among team members; Jehn, 1995,
gion, China (Project 641507), awarded to Jiing-Lih Farh and Gilad Chen. 1997; Jehn & Mannix, 2001). We focused on two individual-level
We would like to thank Lida Zhang and Melody Chao for their assistance
motivational states likely to be affected by empowering leadership
in conducting this research.
Correspondence concerning this article should be addressed to Gilad and relationship conflict, namely, psychological empowerment
Chen, Management & Organization Department, Robert H. Smith School (i.e., extent to which members feel they have the autonomy and
of Business, University of Maryland, College Park, MD 20742-1815. competence to perform meaningful and impactful tasks; Spreitzer,
E-mail: giladchen@rhsmith.umd.edu 1995) and affective commitment (i.e., extent to which employees
541
542 CHEN, SHARMA, EDINGER, SHAPIRO, AND FARH
Innovative
Behavior
H4a (+)
Psychological
Empowering H1a (+)
Empowerment
Leadership
H4b (+)
H3a (-)
H5a (+)
Affective
Conflict Commitment
This document is copyrighted by the American Psychological Association or one of its allied publishers.
feel a sense of emotional attachment to and identify with their team provide a more complete account of how motivating and demoti-
and/or organization1; Meyer & Allen, 1991). By considering these vating team dynamics combine to exert unique and interaction
motivational states, we hoped to better understand how and why effects (which we dub joint influences) on team members’ moti-
team-level stimuli motivate members to contribute meaningfully to vation and behavior.
their teams, as reflected by higher levels of members’ innovative Finally, the vast majority of the research cited above employed
and teamwork behaviors and lower levels of turnover intentions. samples of U.S. American participants, which prevents one from
Second, prior research has tended to focus on team dynamics knowing whether the cross-level effects that were observed gen-
that either motivate (i.e., exert positive influences on members’ eralize to cultures outside of the United States. Accordingly, a
motivational states) or demotivate (i.e., exert negative influences third purpose of our research was to examine these dynamics in
on members’ motivational states) team members. On the one hand, samples consisting of leaders, employees, and students in two
studies that have examined positive (motivating) team-level influ- countries that have been found to significantly differ from each
ences have theorized and found that leader actions that encourage other in their citizens’ typical cultural values—namely, in the
follower self-development, such as empowering leadership and the People’s Republic of China (PRC) as well as the United States (cf.
related construct of transformational leadership (e.g., Bono & Hofstede, 1980; Kirkman, Lowe, & Gibson, 2006; Kirkman &
Judge, 2003) positively influence team members’ individual mo- Shapiro, 2001). The latter comparison is practically important in
tivation (cf. Chen, Kirkman, Kanfer, Allen, & Rosen, 2007; Seib- light of the increasing frequency with which employees operate in
ert, Silver, & Randolph, 2004). On the other hand, studies that culturally diverse environments; moreover, this culture compari-
have examined negative (demotivating) team-level influences have son enabled us to answer calls for more research that seeks to
proposed and found that relationship conflict and other hindering generalize theories of organizational behavior across cultures (Gel-
stressors in the team negatively influence members’ motivation, in fand, Erez, & Aycan, 2007; Kirkman et al., 2006; Tsui, Nifadkar,
that they lead members to withdraw effort and disengage from & Ou, 2007).
their work and their teams (Jehn, 1995, 1997; Jehn & Mannix,
2001; LePine, Podsakoff, & LePine, 2005; Podsakoff, LePine, & Theory and Hypotheses
LePine, 2007).
Missing from the literature is a more complete account of how In the sections that follow, we delineate a theoretical model (see
motivating team-level stimuli that exert positive influences on Figure 1) that takes into account the joint influences of motivating
team members’ motivational states operate simultaneously with (i.e., empowering leadership) and demotivating (i.e., relationship
other team-level stimuli that demotivate or exert negative influ-
ences on team members’ motivational states (Chen & Gogus,
1
2008; Podsakoff et al., 2007). Given that team members are In this research, we did not differentiate between affective commitment
exposed simultaneously to distinct motivating and demotivating to one’s team and one’s organization because it has been argued that the
team provides the immediate context within which organizational influ-
stimuli in the work setting (Hackman, 1992; Podsakoff et al.,
ences on employees occur (Chen & Kanfer, 2006; Hackman, 1992) and
2007), it is important to examine whether these stimuli types exert because there is empirical evidence that team and organizational commit-
unique (possibly opposite) influences on team members’ motiva- ment are highly related and function similarly in teams (Kirkman, &
tion and possibly also interact to affect team members’ motivation. Rosen, 1999). Thus, we expected that affective commitment to one’s team
Accordingly, our second contribution involves an integration and would operate similarly to affective commitment to one’s organization in
extension of prior models of team motivation and work stressors to the context of our model.
MOTIVATING FORCES IN TEAMS 543
conflict) team stimuli on individual-level outcomes (i.e., members’ apply their capabilities toward work-related contributions in their
innovative behavior, teamwork behavior, and turnover intentions). team. More highly empowering leaders convey to their teams they
We focus on these particular individual-level outcomes since prior are confident in their teams’ capabilities to handle challenging
research has theorized and found that team members’ innovative work by encouraging members to be more involved in key deci-
and teamwork behaviors are positively linked, while turnover sion making, to collaboratively and autonomously self-manage
intentions are negatively linked, to team effectiveness (Chen, their work, and to be accountable for outcomes in their team
2005; Chen, Kirkman, et al., 2007; Harrison, Mohammed, (Chen, Kirkman, et al., 2007; Kirkman & Rosen, 1999; Manz &
McGrath, Florey, & Vanderstoep, 2003; Marrone, Tesluk, & Car- Sims, 1987; Seibert et al., 2004). Such empowering leadership
son, 2007). We examine empowering leadership and relationship behaviors are likely to promote team members’ psychological
conflict as the positive and negative team-level predictor variables, empowerment by enhancing members’ personal beliefs that their
respectively, because prior work has identified these two dynamics work is meaningful and relevant and that they have the autonomy,
to be particularly powerful motivating and demotivating forces in opportunity, and capability to influence key outcomes in their team
teams, respectively (e.g., Chen, Kirkman, et al., 2007; Jehn, 1995, and organization. Empowering leadership behaviors also posi-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Building on and integrating prior models of team motivation haviors lead team members to feel more personally accountable
(Chen & Kanfer, 2006), conflict (Jehn, 1995, 1997; Jehn & Man- and emotionally engaged with work processes and outcomes in
nix, 2001), and stressors (LePine et al., 2005; Podsakoff et al., their team and organization.
2007), we also examine whether individual members’ motivational In line with the above-mentioned theorizing, there is empirical
states, including psychological empowerment and affective com- evidence that, as team leaders engage in more empowering behav-
mitment, mediate the cross-level influences of empowering lead- iors, employees respond to these behavioral cues by feeling more
ership and relationship conflict on individual outcomes. According psychologically empowered and affectively committed to their
to Spreitzer (1995), employees feel more psychologically empow- team and organization (Chen, Kirkman, et al., 2007; Den Hartog &
ered when they believe that they (a) can influence outcomes in De Hoogh, 2009; Kirkman & Rosen, 1999; Seibert et al., 2004;
their organization, hence have impact; (b) are capable of accom- Tjosvold & Sun, 2006; Zhang & Bartol, 2010). Thus, the motivat-
plishing task goals, or have a sense of competence; (c) have choice ing effects of empowering leadership are likely to be reflected via
in how they go about doing their work, or have autonomy; and (d) positive influences on the two mediating motivational states of
feel intrinsic enjoyment in the work they do, hence have a sense of psychological empowerment and affective commitment. Hence,
meaningfulness. Affective commitment captures the extent to the following hypothesis:
which employees feel a sense of emotional attachment to, and
identify with, their work environment (Meyer & Allen, 1991). As Hypothesis 1: Team leaders’ empowering leadership posi-
we explain further below, team members with higher levels of tively influences team members’ (a) psychological empower-
psychological empowerment and affective commitment are likely ment and (b) affective commitment.
to be more motivated to engage in behaviors that contribute
positively to their team and organization. The Demotivating Effects of Relationship Conflict
The Motivating Effects of Empowering Leadership In contrast to the positive (motivating) influences of empower-
ing leadership on motivational states, relationship conflict is likely
Prior research has conceptualized and assessed empowering to demotivate team members. High levels of relationship conflict
leadership as a team-level stimulus—that is, as leader behaviors in a team involve strong interpersonal disagreements and tensions
directed to the team as a whole (Chen, Kirkman, et al., 2007; among members and are usually expressed with negative commu-
Kirkman & Rosen, 1999). Highly empowering leaders share power nication and lack of cooperativeness among members revealing
with their employees by delegating authority to employees, hold feelings of anger, distrust, fear, and frustration (Jehn, 1995, 1997;
employees accountable, involve employees in decision making, Jehn & Mannix, 2001). In contrast, teams characterized by low
encourage self-management of work, and convey confidence in levels of relationship conflict experience more harmonious and
employees’ capabilities to handle challenging work (Kirkman & collegial interpersonal relationships among members, expressed
Rosen, 1999; Konczak et al., 2000). In contrast to highly empow- with positive communication that reveals feelings of trust and
ering leaders, leaders characterized as low in empowering leader- mutual respect.
ship engage in more micromanaging or monitoring behaviors Relative to members of teams with lower relationship conflict,
(Spreitzer, De Janasz, & Quinn, 1999) that discourage team self- members of teams with higher relationship conflict are more likely
management, reduce opportunities for autonomous behavior in the to withdraw effort from their tasks on the team (Jehn, 1995). This
team, and convey to their team members little faith in their tendency is consistent with threat rigidity theory, which suggests that
capabilities to handle challenging work. individuals freeze up, withdraw, and narrow their perceptual field of
Chen and Kanfer (2006) theorized that team-level stimuli, such input when they feel threatened by their environment (Staw, Sand-
as empowering leadership, enhance team members’ willingness to elands, & Dutton, 1981). Furthermore, Seo, Barrett, and Bartunek
contribute to their teams indirectly, through members’ motiva- (2004) theorized that unpleasant affective experiences—as is typ-
tional states such as psychological empowerment and affective ical of high levels of relationship conflict (Jehn, 1995)—are neg-
commitment. That is, leaders who exhibit more empowering be- atively linked to motivational states. On the basis of these theories,
haviors, as compared to leaders who exhibit less empowering we propose that the aversive experiences associated with high
behaviors, are more likely to facilitate and motivate members to levels of relationship conflict negatively influence psychological
544 CHEN, SHARMA, EDINGER, SHAPIRO, AND FARH
empowerment because such experiences likely lower team mem- positively related to group members’ motivation to engage in
bers’ sense of work autonomy, confidence in their ability to safety behaviors when members reported having more positive
accomplish work, sense of impact on outcomes in their team, and relationships with their leader.
intrinsic enjoyment of the tasks they perform in the team. Like- Extending the motivational fit rationale, we propose that higher
wise, heightened relationship conflict in the team is also likely to levels of empowering leadership are less likely to exert positive
lead members feeling less emotionally attached to their team and influences on motivational states in the presence of higher, relative
organization (i.e., having lower affective commitment). to lower, levels of relationship conflict because the demotivating
Although no empirical study to date has documented the pro- influences of relationship conflict are likely to interfere with the
posed negative relationship between relationship conflict and psy- motivating influences of empowering leadership. For team mem-
chological empowerment, theorizing by Staw et al. (1981) and Seo bers to effectively handle the heightened levels of autonomy,
et al. (2004; see also Seo & Ilies, 2009) supports this expected control, and self-management encouraged by highly empowering
negative relationship. Earlier empirical research has likewise doc- leaders, they need to work collaboratively with each other, and
umented a negative association between relationship conflict and such a collaborative and functional work environment is more
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
affective commitment (Mills & Schulz, 2009). Furthermore, meta- likely to be evident in teams with lower, rather than higher, levels
This document is copyrighted by the American Psychological Association or one of its allied publishers.
analytic findings by Podsakoff et al. (2007) indicate that hindrance of relationship conflict. In contrast to members of low relationship
stressors—which, similar to relationship conflict, are perceived by conflict teams, members of teams with higher levels of relationship
individuals to thwart their personal growth and goal attainment— conflict are less likely to work in a cohesive and collegial manner
are negatively linked with affective commitment (cf. Podsakoff et when self-managing and coordinating their work or may avoid
al., 2007). Building on these theories and prior empirical evidence each other altogether (Jehn, 1995, 1997). As such, the motivating
we thus predict the following: influences of higher levels of empowering leadership on team
members’ motivational states are less likely to occur in teams
Hypothesis 2: Relationship conflict in the team negatively experiencing higher, versus lower, levels of relationship conflict.
influences team members’ (a) psychological empowerment Hence, in addition to Hypotheses 2a and 2b, we also predict the
and (b) affective commitment. following:
the influences of empowering leadership and relationship conflict might also serve to enhance the impact of team stimuli on team
on team members’ behaviors and turnover intentions. members. Of course, a third possibility is that both collectivism-
However, the past theorizing and research cited above did not related effects serve to countervail each other, leading to an overall
consider, as we do in this article, the joint influences of empow- null effect for collectivism.
ering leadership and relationship conflict on both positive and Thus, although cultural differences—and particularly collectiv-
negative team members’ outcomes, as mediated by psychological ism—likely impact motivation in teams, there was no clear theory
empowerment and affective commitment (see Chen & Gogus, of how such differences might affect the relationships in our
2008; Chen & Kanfer, 2006; Jehn, 1995, 1997; Podsakoff et al., model. As such, in an exploratory manner, we examined whether
2007). Integrating and extending prior models of team motivation, collectivism directly related to team members’ motivational states
conflict, and stress, we expected that the joint (unique and inter- or moderated the influences of empowering leadership and rela-
action) effects of empowering leadership and relationship conflict, tionship conflict on team members’ motivational states. In doing
as mediated by team members’ psychological empowerment and so, we hoped to not only explore the extent to which relationships
affective commitment, would explain variance in both positive and in our model generalize across cultural differences but also iden-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
negative team members’ outcomes, including members’ innova- tify potential cultural influences on relationships and outcomes in
This document is copyrighted by the American Psychological Association or one of its allied publishers.
tive behavior, teamwork behavior, and turnover intentions. Thus, our model. Therefore, in addition to testing our hypothesized
we hypothesized the following: model, we also sought to explore the following research question:
Hypothesis 4: Team members’ psychological empowerment Research Question: Does collectivism influence team members’ mo-
mediates the joint influences of empowering leadership and tivational states, either directly or by moderating the influences of
relationship conflict on team members’ (a) innovative behav- empowering leadership and relationship conflict?
ior, (b) teamwork behavior, and (c) turnover intentions.
Overview of Studies
Hypothesis 5: Team members’ affective commitment medi-
ates the joint influences of empowering leadership and rela- We designed both an experimental study (Study 1) and a field
tionship conflict on team members’ (a) innovative behavior, study (Study 2) to test our model of relationships (see Figure 1).
(b) teamwork behavior, and (c) turnover intentions. Given that prior research has examined collectivism as either
individual-level or national-level differences, in both studies we
measured collectivism directly at the individual level but also
The Potential Roles of Cultural Differences contrasted individuals from the United States and the PRC, as
As noted at the outset of the article, most research to date on these two societies are known to differ in collectivism (Hofstede,
motivation in teams has been conducted in Western-based societ- 1980; Kirkman & Shapiro, 2001). In the experiment, we manipu-
ies, and it remains unclear whether such findings generalize across lated empowering leadership and relationship conflict and exam-
cultures (cf. Gelfand et al., 2007; Tsui et al., 2007). Given our ined their effects on the motivational states and outcomes using
focus on motivation in teams, one cultural attribute— collectiv- student samples from the United States and the PRC. The second
ism—seemed particularly important to take into account, as it study utilized a multilevel, cross-sectional survey design using a
captures “differences in the extent to which individuals prefer to sample of managers and their employees from organizations in the
act as members of a group and are motivated to maintain positive United States and the PRC. Thus, the two studies sought to
image of their group” (Jackson, Colquitt, Wesson, & Zapata- maximize internal validity as well as test the generalizability of
Phelan, 2006, p. 884). Indeed, a meta-analysis by Taras, Kirkman, findings across student and organizational samples in two different
and Steel (2010) documented positive relationships between col- national cultures.
lectivism and a variety of positive team-oriented outcomes, such as
greater concern for others, ingroup favoritism, conflict avoidance, Study 1
and helping and cooperative behaviors.
Following prior work (e.g., Earley, 1989; Jackson et al., 2006;
Method
Kirkman & Shapiro, 2001; Taras et al., 2010), we could envision
two opposite ways in which team members’ collectivism might Sample, design, and procedure. Undergraduate students en-
affect the relationships in our hypothesized model. On the one rolled in management courses in one U.S. university (N ⫽ 57) and
hand, higher levels of collectivism might lead members to be more one university in the PRC (N ⫽ 79) were randomly assigned to four
motivated in teams regardless of team stimuli, hence reducing the conditions, in a fully crossed, 2 (high vs. low empowering leader-
impact of team stimuli on team members. That is, higher levels of ship) ⫻ 2 (high vs. low relationship conflict) factorial design. After
collectivism might lead members to remain highly motivated in the completing a measure of collectivism, participants read one of four
context of teams regardless of motivating and demotivating team scenarios, which instructed participants to imagine they were mem-
stimuli, such as empowering leadership and relationship conflict, bers of a critical student-run task force in their school and provided
due to the group orientation (i.e., the desire to help a group’s them with a set of e-mails they were told had been sent to them by
welfare) that is part of being collectivistic. On the other hand, their team’s leader and two additional task force members. The
higher levels of collectivism might also lead members to be more e-mails’ content distinguished the four conditions by containing our
in tune with what happens in their teams and care more about manipulation of high versus low levels of empowering leadership and
maintaining positive group atmosphere, hence making them more relationship conflict. The scenarios were realistic and relevant to our
susceptible to team stimuli. That is, higher levels of collectivism study outcomes, in that they reflected a student-run task force partic-
546 CHEN, SHARMA, EDINGER, SHAPIRO, AND FARH
ipants might actually encounter, which would require students to Manipulation checks. We also administered measures of the
engage in innovative and teamwork behaviors and avoid turnover, to task force leader’s empowering leadership and the level of rela-
maximize task force success. Following the scenarios, participants tionship conflict in the task force. Participants indicated how
completed the manipulation checks, measures of psychological em- strongly they agreed with Kirkman and Rosen’s (1999) 14-item
powerment and affective commitment, and, given the hypothetical scale, with the task force’s leader as the referent (e.g., “My Task
nature of scenario-based experiments, measures of participants’ in- Force leader gives my work unit many responsibilities”; ␣ ⫽ .94).
tentions to engage in innovative and teamwork behaviors and to quit To assess the degree of relationship conflict in the task force,
the task force. The scenarios and measures were administered in participants responded to Jehn and Mannix’s (2001) three-item
English to the U.S. participants and in Chinese to the PRC partici- measure (e.g., “How much relationship tension is there in your
pants, after employing standard translation and back-translation pro- Task Force?”; 1 ⫽ none, 5 ⫽ a lot; ␣ ⫽ .94). Analyses of variance
cedures (Brislin, 1980).2 tests of the empowering leadership manipulation check measure as
Manipulations. outcome indicated a significant main effect for the empowering
Empowering leadership. In the high empowering leadership leadership manipulation, Mhigh empowering leadership ⫽ 4.08, Mlow
empowering leadership ⫽ 2.68, F(1, 132) ⫽ 232.42, p ⬍ .05, ⫽ .64,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
condition, the two e-mails from the team leader that participants 2
This document is copyrighted by the American Psychological Association or one of its allied publishers.
read reflected leadership behaviors such as expressing confidence but nonsignificant main effect for the relationship conflict manip-
in the team’s ability to carry out its task successfully, allowing the ulation, Mhigh relationship conflict ⫽ 3.34, Mlow relationship conflict ⫽
team to self-manage its work and make decisions on its own, and 3.39, F(1, 132) ⫽ 0.11, ns, 2 ⫽ .00. In contrast, analyses of
highlighting the relevance of their task to the school’s leadership. variance tests of the relationship conflict manipulation check mea-
In contrast, in the low empowering leadership condition, the two sure as outcome indicated a significant main effect for the rela-
e-mails from the team leader reflected leadership behaviors such as tionship conflict manipulation, Mhigh relationship conflict ⫽ 4.06, Mlow
expressing doubts about the team’s ability to carry out its task relationship conflict ⫽ 2.26, F(1, 132) ⫽ 176.59, p ⬍ .05, ⫽ .57,
2
successfully, prohibiting the team from making decisions without but nonsignificant main effect for the empowering leadership
the leader, closely monitoring the team’s work, and informing the manipulation, Mhigh empowering leadership ⫽ 3.12, Mlow empowering
team that the school’s leadership would likely rely on the leader’s leadership ⫽ 3.20, F(1, 132) ⫽ 0.12, ns, ⫽ .00. These results
2
rather than the team’s recommendations. provide strong evidence for the efficacy and validity of the two
Relationship conflict. The high and low relationship conflict manipulations.
conditions also involved two e-mail messages, reflecting an ex- In addition, participants completed the following three questions
change between two task force members. In the high relationship after reading the scenarios: (a) “It is realistic that I might experience
conflict condition, the e-mails included highly contentious per- a supervisor like J.P.,” (b) “It is realistic that I might experience
sonal disagreements and attacks between the two members. In team-events like those described above in the Task Force,” and (c)
contrast, in the low relationship conflict condition, the e-mails “At some point during my career I will probably encounter a situation
reflected a more cordial and collegial exchange between the two like the one described above.” The mean score across the three items
task force members. was high (means ranged from 3.92 to 4.05 out of 5), and multivariate
Measures. Descriptive statistics, internal consistency reliability
analysis of variance tests indicated no significant differences in ratings
(alpha) coefficients, and correlations involving measures are reported
due to the empowering leadership condition, F(3, 130) ⫽ 2.18, ns; the
in Table 1. Unless otherwise noted, all measures used a 5-point
relationship conflict condition, F(3, 130) ⫽ 1.72, ns; or the interaction
Likert-type scale (1 ⫽ strongly disagree, 5 ⫽ strongly agree).
between the conditions, F(3, 130) ⫽ 0.37, ns. These results indicate
Dependent variable measures. Using two different four-item
that the scenarios were deemed highly realistic by participants, irre-
scales modified from Welbourne et al. (1998), participants rated
spective of condition.
their intent to engage in innovative behaviors (e.g., “I would
Control variables. We controlled for participants’ nationality
probably work to implement new ideas”) and in teamwork behav-
and individual-level collectivism. Participant nationality (1 ⫽ PRC
iors (e.g., “I would probably work to make sure the Task Force
Chinese; 2 ⫽ U.S. American) was selected to evaluate the extent
succeeds”). To assess the extent to which participants intended to
to which results held when taking into account culturally distinct
leave the task force, participants were asked to respond to four
country-level differences. Students’ self-reported collectivism was
modified items taken from Kelloway and colleagues’ (1999) mea-
assessed by asking participants to respond to six statements taken
sure of turnover intentions (e.g., “I would probably think about
from Dorfman and Howell’s (1988) measure of individual-level
leaving this Task Force”).
collectivism (e.g., “Group success is more important than individ-
Psychological empowerment and affective commitment. To
ual success”).
assess the extent to which participants felt individually empow-
ered, based on the manipulated scenarios, participants responded Confirmatory factor analyses. We examined the validity of
to 12 statements taken from Spreitzer’s (1995) measure of indi- the measures using confirmatory factor analyses in LISREL
vidual empowerment with the following instructions: “How (Jöreskog & Sörbom, 1993). To form the measurement models, we
strongly do you agree with the statements below regarding how created three parcels each for the psychological empowerment,
you would probably feel about your work on the Task Force?”
(e.g., “I have significant autonomy in determining how I do my 2
Given the small sample size in this study, we did not conduct tests of
job”). Similarly, participants indicated their likely level of affec- measurement invariance to examine the equivalence of measures across the
tive commitment to the task force using eight modified items from two languages. However, we do report evidence for measurement invari-
Allen and Meyer’s (1990) measure (e.g., “I would probably feel a ance across languages in Study 2, which had a larger sample, using the
strong sense of belonging to my Task Force”). same measures (applied to the field setting).
MOTIVATING FORCES IN TEAMS 547
Table 1
Descriptive Statistics and Correlations (Study 1)
Variable M SD 1 2 3 4 5 6 7 8 9
a
1. Nationality 1.42 0.50 —
2. Collectivism 3.57 0.58 ⫺.01 (.71)
3. Empowering leadership 0.49 0.50 ⫺.00 ⫺.18ⴱ —
4. Relationship conflict 0.50 0.50 ⫺.02 ⫺.04 ⫺.02 —
5. Psychological empowerment 3.27 0.73 .28ⴱ .10 .45ⴱ ⫺.18ⴱ (.91)
6. Affective commitment 2.96 0.68 .06 ⫺.08 .20ⴱ ⫺.52ⴱ .52ⴱ (.76)
7. Turnover intentions 3.19 1.07 ⫺.09 ⫺.10 ⫺.29ⴱ .42ⴱ ⫺.59ⴱ ⫺.71ⴱ (.92)
8. Innovative behavior 3.75 0.88 .29ⴱ ⫺.04 .28ⴱ ⫺.14 .52ⴱ .43ⴱ ⫺.37ⴱ (.93)
9. Teamwork behavior 3.68 0.74 .13 .04 .21ⴱ ⫺.32ⴱ .58ⴱ .58ⴱ ⫺.51ⴱ .65ⴱ (.86)
Note. N ⫽ 136. Reliability estimates (coefficient alpha) are on the diagonal. Empowering leadership and relationship conflict conditions were coded as
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
a
1 ⫽ Chinese; 2 ⫽ U.S. American.
ⴱ
p ⬍ .05.
affective commitment, and collectivism measures. The scales for empowerment (Hypothesis 3a) and affective commitment (Hy-
innovative and teamwork behaviors and turnover intentions con- pothesis 3b). Consistent with both hypotheses, relationship conflict
sisted of only three or four items, so these measures were not significantly interacted with empowering leadership to affect both
parceled. The six-factor measurement model, which allowed all psychological empowerment ( ⫽ ⫺.44, p ⬍ .05; Model 2) and
factors to correlate freely with each other, fit the data well, 2(176, affective commitment ( ⫽ ⫺.69, p ⬍ .05; Model 4). To probe the
N ⫽ 136) ⫽ 321.75, root-mean-square error of approximation nature of these effects, we plotted the interactions following Aiken
(RMSEA) ⫽ .08, comparative fit index (CFI) ⫽ .96. An alterna- and West’s (1991) procedures. As shown in Figure 2, in support of
tive one-factor model in which the correlations between all six Hypothesis 3a, empowering leadership affected psychological em-
factors were set to 1.0 fit the data significantly worse than the powerment more positively when relationship conflict was low
six-factor model, ⌬2(15) ⫽ 1,115.17, p ⬍ .05, RMSEA ⫽ .24, (dashed line) versus high (solid line). In addition, supporting
CFI ⫽ .77. The six-factor model also fit the data significantly Hypothesis 3b, Figure 3 shows that empowering leadership also
better relative to an alternative four-factor model in which the affected affective commitment more positively when relationship
correlations between psychological empowerment and affective conflict was low (dashed line) relative to high (solid line).
commitment and between innovative and teamwork behaviors Next, we turn to our mediation hypotheses (Hypotheses 4 and
were set to 1.0, ⌬2(2) ⫽ 174.81, p ⬍ .05, RMSEA ⫽ .12, CFI ⫽ 5). Table 2 summarizes ordinary least squares regression analyses
.93. These results support the validity of the measures. of outcomes. In line with Hypotheses 4a– 4c, after taking into
Analysis strategy. We tested Hypotheses 1, 2, and 3 using account controls and the independent variables, psychological
hierarchical regressions, after grand-mean centering the indepen- empowerment significantly predicted innovative behavior ( ⫽
dent variables (empowering leadership, relationship conflict, and .35, p ⬍ .05; Model 6), teamwork behavior ( ⫽ .42, p ⬍ .05;
their interaction). We tested Hypotheses 4 and 5 using the Model 8), and turnover intentions ( ⫽ ⫺.35, p ⬍ .05; Model 10).
bootstrapping-based test developed by MacKinnon, Lockwood, Similarly, as suggested by Hypotheses 5a–5c, affective commit-
and Williams (2004), which provides a 95% confidence interval ment significantly predicted innovative behavior ( ⫽ .43, p ⬍
for the indirect effect (i.e., the indirect effect is significant when .05; Model 6), teamwork behavior ( ⫽ .43, p ⬍ .05; Model 8),
the 95% confidence interval excludes zero).
and turnover intentions ( ⫽ ⫺.84, p ⬍ .05; Model 10) above and
beyond the independent variables. These results provide initial
Results support for Hypotheses 4 and 5.
Hypotheses tests. Regression analyses of hypotheses are The indirect (i.e., mediated) effects, reflecting the product of the
reported in Table 2. Our first two hypotheses predicted that em- estimated independent variable 3 mediator effect and the estimated
powering leadership would positively influence psychological em- mediator 3 dependent variable effect, as well as their respective
powerment (Hypothesis 1a) and affective commitment (Hypothe- bootstrapped 95% confidence intervals from MacKinnon et al.’s
sis 1b), while relationship conflict would negatively influence (2004) test, are reported in Table 3. As shown in Table 3, all 18
psychological empowerment (Hypothesis 2a) and affective com- indirect effects were statistically significant. In particular, in sup-
mitment (Hypothesis 2b). Supporting Hypotheses 1a and 1b, em- port of Hypotheses 4a, 4b, and 4c, psychological empowerment
powering leadership significantly and positively predicted psycho- significantly mediated the effects of empowering leadership, rela-
logical empowerment ( ⫽ .73, p ⬍ .05; Model 1) and affective tionship conflict, and the Empowering Leadership ⫻ Relationship
commitment ( ⫽ .26, p ⬍ .05; Model 3). In support of Hypoth- Conflict interaction on innovative behavior, teamwork behavior,
eses 2a and 2b, relationship conflict significantly and negatively and turnover intentions, respectively. Furthermore, supporting Hy-
predicted psychological empowerment ( ⫽ ⫺.25, p ⬍ .05; Model potheses 5a, 5b, and 5c, affective commitment also significantly
1) and affective commitment ( ⫽ ⫺.73, p ⬍ .05; Model 3). mediated the effects of empowering leadership, relationship con-
Our next two hypotheses predicted that empowering leadership flict, and the Empowering Leadership ⫻ Relationship Conflict
and relationship conflict would interact to affect psychological interaction on innovative behavior, teamwork behavior, and turn-
548 CHEN, SHARMA, EDINGER, SHAPIRO, AND FARH
⫺.24 (.11)ⴱ
⫺.35 (.12)ⴱ
⫺.84 (.13)ⴱ
3.00
Model 10
.03 (.13)
⫺.21 (.14)
.20 (.15)
⫺.03 (.25)
Turnover intentions
.60ⴱ
Psychological Empowerment
2.50
⫺.68 (.16)ⴱ
.89 (.15)ⴱ
.70 (.31)ⴱ
⫺.14 (.16)
⫺.25 (.14)
Model 9
2.00
.33ⴱ
Low Relationship Conflict
1.50
High Relationship Conflict
.45 (.20)ⴱ
.42 (.09)ⴱ
.43 (.11)ⴱ
.00 (.10)
.02 (.09)
⫺.09 (.11)
⫺.05 (.12)
Model 8
Teamwork behavior
.47ⴱ
1.00
Low Empowering High Empowering
Leadership Leadership
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
.32 (.12)ⴱ
⫺.47 (.10)ⴱ
⫺.03 (.24)
Model 7
.17ⴱ
.35 (.12)ⴱ
.43 (.14)ⴱ
.19 (.27)
Model 6
Innovative behavior
.35ⴱ
.50 (.14)ⴱ
.51 (.14)
.01 (.12)
⫺.23 (.14)
⫺.26 (.28)
Model 5
⫺.69 (.18)ⴱ
.03 (.09)
⫺.08 (.08)
Model 4
Affective commitment
.34ⴱ
.23 (.09)ⴱ
.72 (.10)ⴱ
⫺.25 (.10)ⴱ
⫺.44 (.20)ⴱ
.39 (.10)
Model 2
larly across the two societies and that our model of relationships
empowerment
Psychological
.24 (.09)ⴱ
.73 (.10)ⴱ
⫺.25 (.10)ⴱ
.40 (.10)
Model 1
4.00
.36ⴱ
Affective Commitment
3.50
Empowering Leadership ⫻ Relationship
Empowering leadership
Affective commitment
Conflict interaction
Relationship conflict
2.50
Low Empowering High Empowering
Collectivism
a
Nationality
Leadership Leadership
p ⬍ .05.
Table 2
Table 3
Indirect Effects of Empowering Leadership and Relationship Conflict on Outcomes (Study 1)
Empowering leadership 3 affective commitment 3 innovative behavior (Hypothesis 5a) .11ⴱ [.02, .24]
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Relationship conflict 3 affective commitment 3 innovative behavior (Hypothesis 5a) ⫺.31ⴱ [⫺.54, ⫺.10]
Leadership ⫻ Conflict 3 affective commitment 3 innovative behavior (Hypothesis 5a) ⫺.30ⴱ [⫺.58, ⫺.09]
Empowering leadership 3 affective commitment 3 teamwork behavior (Hypothesis 5b) .11ⴱ [.02, .23]
Relationship conflict 3 affective commitment 3 teamwork behavior (Hypothesis 5b) ⫺.31ⴱ [⫺.51, ⫺.15]
Leadership ⫻ Conflict 3 affective commitment 3 teamwork behavior (Hypothesis 5b) ⫺.30ⴱ [⫺.54, ⫺.11]
Empowering leadership 3 affective commitment 3 turnover intentions (Hypothesis 5c) ⫺.22ⴱ [⫺.41, ⫺.05]
Relationship conflict 3 affective commitment 3 turnover intentions (Hypothesis 5c) .61ⴱ [.38, .88]
Leadership ⫻ Conflict 3 affective commitment 3 turnover intentions (Hypothesis 5c) .58ⴱ [.26, .96]
ⴱ
p ⬍ .05.
cultural differences—at least, differences likely to be observed pants’ intent to behave, as opposed to actual behavior. Second, our
between U.S. Americans and Chinese. laboratory study was conducted using an undergraduate student
population in a classroom context. Thus, this study has limited
Discussion construct and external validity. To overcome these limitations, we
designed a second study with a sample of managers and employees
In summary, the results from the experimental study provide
in a field setting.
strong support for our hypothesized model. In particular, the
average (absolute) effects of empowering leadership and relation-
ship conflict on the two motivational states were nearly identical Study 2
(average b ⫽ .49, .50, respectively) yet in the opposite direction,
suggesting that the motivating effects of empowering leadership
on motivational states were very similar to the demotivating ef- Method
fects of relationship conflict. Additionally, we found that the Sample and procedures. Participants were 144 leaders en-
effects of empowering leadership on both psychological empow- rolled in executive master’s of business administration leadership
erment and affective commitment were more positive when rela- development courses taught by a large public U.S. university in the
tionship conflict was low, rather than high. Furthermore, both United States and in the PRC, as well as their respective followers.
psychological empowerment and affective commitment helped to
Leaders identified up to five subordinates (i.e., direct reports) from
link the unique and joint influences of empowering leadership and
their work teams, generating a sample of 566 potential followers
relationship conflict on our dependent variables of innovative
from 144 teams. The 566 followers represented 60% of the 945
behavior, teamwork behavior, and turnover intentions.
members belonging to the 144 teams (average team size ⫽ 6.56).
Importantly, our results held even after controlling for differ-
Teams consisted of management, production, project, and service
ences in participants’ nationality (United States vs. PRC) and
collectivism and generalized across the U.S. and PRC samples. teams, and prestudy interviews with leaders indicated that these
Furthermore, we found that cultural differences at the individual teams required members to work interdependently toward com-
level and societal level exerted different effects on motivation, in mon goals. The final sample was restricted to teams where one
that individual-level collectivism related positively to psycholog- leader and at least two followers provided complete data, yielding
ical empowerment, whereas psychological empowerment was sig- a usable sample of 105 leaders (out of 144, or 73% survey response
nificantly higher among U.S. than PRC participants (i.e., in a less rate) and 386 followers (out of 566, or 68% survey response rate),
collectivistic society). with an average of 3.68 (range ⫽ 2–5) members per team/leader.
Although this study involved random assignment into experi- On average, the percentage of members sampled within team (i.e.,
mentally controlled conditions and thus high internal validity, team response rate, calculated as the number of completed team
there are two notable limitations. First, because the experimental members’ surveys received divided by team size) was 65%. Of
design manipulated empowering leadership and relationship con- these, 41 leaders were Chinese, 38 teams included all-Chinese
flict in a controlled and hypothetical environment, we were only members (leader and followers), 54 teams were all-U.S. members,
able to test our model’s outcome variables by measuring partici- and 13 included a mixture of U.S. and Chinese members (mixed
550 CHEN, SHARMA, EDINGER, SHAPIRO, AND FARH
teams largely consisted of expatriate leaders and/or followers Average intermember agreement (rwg(j), with a uniform expected
working in the PRC).3 variance distribution) across units was .98, and in addition,
Leaders were told that their followers would be asked to anon- ICC(1) ⫽ .14, F(104, 281) ⫽ 1.62, p ⬍ .05, and ICC(2) ⫽ .40.
ymously provide feedback about the leaders’ leadership style. These results indicate strong within-team agreement (LeBreton &
Given that leaders received developmental feedback in return for Senter, 2008) and sufficient intermember reliability (Bliese, 2000),
participation, they were encouraged to identify representative team supporting the aggregation of empowering leader behaviors scores
members (not only those who might hold favorable views of to the work unit level.
them). Followers were contacted directly by the researchers with Relationship conflict in the team. To assess the degree of
an e-mail note explaining that as part of a leadership development relationship conflict in their team, followers completed Jehn and
exercise, they were being asked to provide anonymous feedback Mannix’s (2001) three-item scale (1 ⫽ none, 5 ⫽ a lot). Scores
about their leader. This e-mail stated that (a) participation was were aggregated using team means to reflect team-level relation-
voluntary, (b) the survey website was managed by independent ship conflict. In support of aggregation, average intermember
researchers (and not their organization’s management), and (c) agreement (rwg(j), with a uniform expected variance distribution)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
leaders would not have access to each follower’s individual re- across units was .88, ICC(1) ⫽ .19, F(104, 281) ⫽ 2.05, p ⬍ .05,
This document is copyrighted by the American Psychological Association or one of its allied publishers.
sponse but, rather, would receive feedback in aggregate form only. and ICC(2) ⫽ .51.
Leaders and followers completed their online surveys in their Followers’ psychological empowerment and affective commit-
native language (English or Chinese). Prior to the study, English- ment. Followers indicated the extent to which they personally
version surveys were translated and back-translated into Chinese felt psychologically empowered, using the 12-item scale devel-
in accordance with established cross-cultural translation proce- oped by Spreitzer (1995). Followers also reported the extent to
dures (Brislin, 1980). which they felt affectively committed to their organization, using
Followers completed measures of psychological empowerment,
the eight-item scale developed by Allen and Meyer (1990). In line
affective commitment, and turnover intentions and rated their
with our theory, followers’ psychological empowerment and af-
leader’s empowering leadership and their team’s level of relation-
fective commitment were examined at the individual level of
ship conflict. Given conceptual arguments (Chen & Kanfer, 2006;
analysis.
Hackman, 1992) that the team provides the immediate context in
Control variables. We controlled for several variables in this
which most organizational experiences occur and empirical evi-
study. First, leaders indicated their nationality (Chinese or U.S. Amer-
dence (Kirkman & Rosen, 1999) suggesting that commitment to
ican) to control for possible cultural differences in leadership behav-
one’s team and to one’s firm operate similarly, measures of affec-
iors and leader–follower relationships. Second, we also controlled for
tive commitment and turnover intentions used the organization,
leaders’ organizational tenure (number of years since being hired by
rather than the team, as referent. Leaders rated their followers’
innovative and teamwork behavior.4 We also included several the organization) since this could explain differences attributed to
control measures, as indicated below. organization-related experience. Leader nationality and tenure were
Measures. Descriptive statistics, internal consistency reliabil- measured at the team level, as there was only one leader per team.
ity (alpha) coefficients, and correlations involving measures from Third, as in Study 1, we also controlled for follower nationality
Study 2 are reported in Table 4. Note that we employed the same (Chinese or U.S. American; provided by followers) to control for
measures of all substantive constructs as in Study 1, except that we possible cultural differences in followers’ reactions to leaders. Fourth,
used the original versions of these measures in this study rather we also controlled for followers’ collectivism, by asking followers to
than the versions modified to fit the experimental scenarios devel- complete the same six-item measure of collectivism employed in
oped in Study 1. Unless otherwise noted, measures used a 5-point Study 1 (Dorfman & Howell, 1988). Finally, we measured followers’
Likert-type scale (1 ⫽ strongly disagree, 5 ⫽ strongly agree). organizational tenure (number of years since being hired by the
Followers’ innovative and teamwork behaviors. Using Wel-
bourne et al.’s (1998) four-item measure of innovative behavior 3
Given that team response rates differed across teams, we reanalyzed
and four-item measure of teamwork behavior, leaders rated the
the data with team response rate as control. Results (available upon
level of each of their followers’ innovative behavior and teamwork request) indicated that differences in team response rates did not signifi-
behavior. The individual follower was the focal referent for these cantly relate to any of the outcomes and that controlling for team response
ratings of innovative and teamwork behaviors, and hence, these rate did not affect any of the findings we report.
measures were treated at the individual level of analysis, consistent 4
To ensure the individual-level outcomes were relevant to team effec-
with our theoretical focus on individual innovative and teamwork tiveness in this sample, we examined the correlations between average
behaviors in teams. levels of each outcome in the team and an index of team effectiveness.
Followers’ turnover intentions. To assess the extent to which Specifically, team members completed a five-item measure of team per-
followers intended to leave their firms, followers responded to formance developed by Kirkman and Rosen (1999; e.g., “My work team
Kelloway, Gottlieb, and Barham’s (1999) four-item measure. Con- produces high quality work”; 1 ⫽ strongly disagree, 5 ⫽ strong agree).
sistent with our theory, followers’ turnover intentions were treated Team members’ ratings yielded high interrater agreement (average rwg(j) ⫽
.89) and a sufficient level of interrater reliability, ICC(1) ⫽ .20, ICC(2) ⫽
at the individual level of analysis.
.48, F(104, 281) ⫽ 1.93, p ⬍ .05, supporting aggregation to the team level.
Empowering leadership behaviors. Followers rated their team Team-level analyses revealed significant correlations between team per-
leader’s empowering leadership using Kirkman and Rosen’s (1999) formance and average member innovative behavior in the team (r ⫽ .28,
14-item scale. The team’s leader was the focal referent for the em- p ⬍ .05), average member teamwork behavior in the team (r ⫽ .31, p ⬍
powering behaviors ratings, and hence, consistent with our theoretical .05), and average member turnover intentions in the team (r ⫽ ⫺.20, p ⬍
focus, this measure was treated at the team level of analysis. .05), supporting the relevance of each outcome in this sample.
MOTIVATING FORCES IN TEAMS 551
Table 4
Descriptive Statistics and Correlations (Study 2)
Variable M SD 1 2 3 4 5 6 7 8 9 10 11 12
a
1. Leader nationality 1.60 0.49 —
2. Leader tenure 6.67 4.60 ⫺.09 —
3. Follower nationalitya 1.50 0.50 .81ⴱ ⫺.01 —
4. Follower tenure 4.76 5.61 .06 .30ⴱ .11ⴱ —
5. Follower collectivism 3.44 0.59 ⫺.35ⴱ .02 ⫺.37ⴱ ⫺.04 (.71)
6. Empowering leadership 3.98 0.30 ⫺.10ⴱ ⫺.08 ⫺.16ⴱ .11ⴱ .10ⴱ (.89)
7. Relationship conflict 2.16 0.48 .13ⴱ .12ⴱ .10ⴱ .06 ⫺.06 ⫺.35ⴱ (.85)
8. Psychological empowerment 4.09 0.54 .13ⴱ ⫺.02 .17ⴱ .17ⴱ .08 .27ⴱ ⫺.08 (.90)
9. Affective commitment 3.56 0.69 ⫺.05 .00 ⫺.02 .21ⴱ .17ⴱ .30ⴱ ⫺.17ⴱ .45ⴱ (.81)
10. Turnover intentions 2.10 1.03 .07 ⫺.01 .02 ⫺.09 ⫺.15ⴱ ⫺.25ⴱ .24ⴱ ⫺.33ⴱ ⫺.54ⴱ (.96)
11. Innovative behavior 3.42 0.98 .22ⴱ .03 .25ⴱ .02 ⫺.06 .16ⴱ ⫺.05 .25ⴱ .08 ⫺.09 (.97)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
12. Teamwork Behavior 3.94 0.80 .19ⴱ .00 .21ⴱ .05 ⫺.06 .15ⴱ ⫺.03 .18ⴱ .15ⴱ ⫺.15ⴱ .52ⴱ (.95)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Note. N ⫽ 386 followers. Reliability estimates (coefficient alpha) are on the diagonal.
a
1 ⫽ Chinese; 2 ⫽ U.S. American.
ⴱ
p ⬍ .05.
organization) to control for possible performance differences due to confirmatory factor analysis conducted using LISREL on the Chi-
differences in organizational experience. nese and English subsamples indicated that the six-factor model in
Confirmatory factor analyses. Although we followed Bris- which the factor correlations and factor loadings were set to be
lin’s (1980) translation– back-translation procedures, we examined equal across the two subsamples fit the data sufficiently well,
further the validity of the measures using several confirmatory factor 2(327, N ⫽ 170,216) ⫽ 688.53, RMSEA ⫽ .08, CFI ⫽ .94. These
analyses in LISREL. We focused only on measures collected from results support the psychometric equivalence of the measures
followers since there was a sufficient sample size (N ⫽ 386) for across the two languages (Vandenberg & Lance, 2000).
testing these measures and common-method biases were more likely Analysis strategy. Due to the multilevel nature of our theo-
among these measures. For all analyses, we created three parcels of retical hypotheses and data, our individual-level data were likely
items per scale by randomly combining items into three parcels per
nonindependent. Indeed, ICC(1) values for innovative behavior
scale (except for relationship conflict and turnover intentions, which
(.41), teamwork behavior (.34), and turnover intentions (.24) in-
were measured by only three and four items per scale, respectively).
dicated that team differences accounted for between 24% and 41%
A first set of analyses examined the extent to which the six
of the total variance in individual-level outcomes. Therefore, we
measures obtained from followers (relationship conflict, empow-
ering leadership, psychological empowerment, affective commit- tested the hypotheses with random coefficient modeling (RCM)
ment, turnover intentions, and collectivism) were sufficiently dis- using the hierarchical linear modeling statistical package (Rauden-
tinct. The hypothesized six-factor model fit the data very well, bush, Bryk, & Congdon, 2004; see also Hofmann, Griffin, &
2(137, N ⫽ 386) ⫽ 216.74, RMSEA ⫽ .04, CFI ⫽ .99. An Gavin, 2000). RCM first partitions the total variance in individual-
alternative one-factor model in which the correlation among all level outcomes into within-unit (Level 1, or individual level) and
factors was set to 1.0 fit the data significantly worse than the between-unit (Level 2, or team level) components. Next, RCM
hypothesized six-factor model, ⌬2(15) ⫽ 3,026.20, p ⬍ .05, regresses the Level 1 variance in the outcome on Level 1 predic-
RMSEA ⫽ .24, CFI ⫽ .63. The six-factor model also fit the data tors, and the Level 2 variance in the outcome on Level 2 predic-
significantly better than an alternative five-factor model, in which tors. In our analyses, Level 2 predictors included leader national-
the correlation between psychological empowerment and affective ity, leader organizational tenure, empowering leadership,
commitment was set at 1.0, ⌬2(1) ⫽ 241.97, p ⬍ .05, RMSEA ⫽ relationship conflict, and the Empowering Leadership ⫻ Relation-
.08, CFI ⫽ .96. These results support the discriminant validity of ship Conflict interaction, whereas Level 1 predictors included
measures collected from followers. follower nationality, organizational tenure, collectivism, psycho-
A second set of analyses was conducted to support the psycho- logical empowerment, affective commitment, innovative behavior,
metric equivalence of measures collected in Chinese (from 170 teamwork behavior, and turnover intentions.5
followers) versus English (from 216 followers). First, we found
that the hypothesized six-factor model fit very well in the Chinese
subsample, 2(137, N ⫽ 170) ⫽ 195.04, RMSEA ⫽ .05, CFI ⫽ 5
.98, and the English subsample, 2(137, N ⫽ 216) ⫽ 236.54, Given that there were only two nationalities and that team members
were not nested neatly under the two nationalities (i.e., leaders from either
RMSEA ⫽ .06, CFI ⫽ .97. In addition, the hypothesized six-factor
nationality led teams whose members were either all U.S. Americans, all
model fit the data significantly better than the alternative one- Chinese, or from both nations), we could not model nationality as a Level
factor model in which all the constructs correlated at 1.0 in both 3 variable. However, we captured differences in nationality appropriately
the Chinese subsample, ⌬2(15) ⫽ 1,601.30, p ⬍ .05, RMSEA ⫽ by including leader nationality as a Level 2 (team-level) dummy variable
.25, CFI ⫽ .54, and the English subsample, ⌬2(15) ⫽ 1,637.35, and follower nationality as a Level 1 (individual-level) variable in all
p ⬍ .05, RMSEA ⫽ .23, CFI ⫽ .58. Finally, a multigroup analyses (cf. Bliese, 2002; Hofmann et al., 2000).
552 CHEN, SHARMA, EDINGER, SHAPIRO, AND FARH
⫺.22 (.08)ⴱ
⫺.64 (.07)ⴱ
.36 (.10)ⴱ
.77 (.21)ⴱ
Model 10
⫺.09 (.19)
.01 (.01)
⫺.09 (.09)
.09 (.17)
.00 (.01)
⫺.20 (.19)
Hypotheses 1–3. To test for the mediation hypotheses (Hypotheses
Turnover intentions
.33ⴱ
4 and 5), we employed MacKinnon et al.’s (2004) test, given that
this test is also appropriate for multilevel mediation models (see
Bauer, Preacher, & Gil, 2006). In all analyses, we grand-mean-
⫺.26 (.10)ⴱ
⫺.71 (.20)ⴱ
.48 (.12)ⴱ
1.12 (.24)ⴱ
⫺.26 (.21)
⫺.01 (.01)
.14 (.21)
.00 (.01)
centered the variables, as in Study 1. Although estimates of effect
Model 9
.15ⴱ
sizes are only tenuous in cross-level models, we nonetheless report
Snijders and Bosker’s (1999) pseudo-R2 (⬃R2), which is based on
proportional reduction of total (both Level 1 and Level 2) errors
due to predictors in the model.
.14 (.06)ⴱ
.13 (.17)
.00 (.01)
.01 (.09)
.05 (.08)
.25 (.18)
.01 (.01)
.34 (.22)
⫺.01 (.13)
⫺.05 (.32)
Model 8
Teamwork behavior
.07ⴱ
Results and Discussion
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Note. N ⫽ 105 leaders/work units and 386 followers. Table entries represent unstandardized parameter estimates with standard errors in parentheses.
.46 (.21)ⴱ
This document is copyrighted by the American Psychological Association or one of its allied publishers.
.18 (.16)
.00 (.01)
.05 (.09)
.23 (.17)
.01 (.01)
⫺.04 (.13)
⫺.13 (.32)
Model 7
.06ⴱ
respectively, empowering leadership was found to have a signifi-
cant cross-level relationship with psychological empowerment
(␥ ⫽ .50, p ⬍ .05; Model 1) and affective commitment (␥ ⫽ .56,
p ⬍ .05; Model 3). In contrast to Hypothesis 2a, relationship
⫺.02 (.01)ⴱ
.31 (.10)ⴱ
.53 (.42)
.01 (.08)
⫺.02 (.07)
.02 (.41)
.02 (.01)
.48 (.27)
⫺.08 (.15)
⫺.10 (.45)
Model 6
conflict did not exhibit a significant cross-level relationship with
Innovative behavior
.10ⴱ
psychological empowerment (␥ ⫽ .00, ns; Model 1), yet in support
of Hypothesis 2b, relationship conflict did significantly relate to
affective commitment (␥ ⫽ ⫺.12, p ⬍ .05; Model 3). Furthermore,
the cross-level influence of the Empowering Leadership ⫻ Rela-
.62 (.26)ⴱ
.61 (.41)
⫺.01 (.01)
.07 (.08)
.02 (.40)
.01 (.01)
⫺.09 (.15)
⫺.14 (.44)
Model 5
tionship Conflict interaction on psychological empowerment was
.08ⴱ
nonsignificant (␥ ⫽ ⫺.14, ns; Model 2), failing to support Hy-
pothesis 3a. However, in support of Hypothesis 3b, the Empow-
ering Leadership ⫻ Relationship Conflict interaction exhibited a
.02 (.01)ⴱ
.21 (.07)ⴱ
.60 (.12)ⴱ
⫺.17 (.07)ⴱ
⫺.46 (.18)ⴱ
.14 (.12)
⫺.05 (.13)
.00 (.01)
Model 4
.17ⴱ
Hierarchical Linear Modeling Analyses of Mediating and Dependent Variables (Study 2)
.56 (.12)ⴱ
⫺.12 (.06)ⴱ
.16 (.12)
⫺.08 (.13)
⫺.01 (.01)
Model 3
.16ⴱ
els 5–10), which tested whether the mediators (psychological
empowerment and affective commitment) uniquely related to the
outcomes—a precondition for mediation. In line with Hypotheses
.27 (.08)ⴱ
.01 (.00)ⴱ
.14 (.05)ⴱ
.51 (.09)ⴱ
.00 (.08)
.00 (.01)
⫺.01 (.06)
⫺.14 (.15)
4a and 4c, psychological empowerment significantly predicted
Model 2
.15ⴱ
innovative behavior ( ⫽ .31, p ⬍ .05; Model 6) and turnover
empowerment
Psychological
.50 (.09)ⴱ
⫺.01 (.08)
.00 (.01)
.00 (.06)
Model 1
Empowering leadership
Conflict interaction
Relationship conflict
p ⬍ .05.
Tenure
Table 6
Indirect Effects of Empowering Leadership and Relationship Conflict on Outcomes (Study 2)
enced greater psychological empowerment (see Table 5, Model 1). moderated the positive influence of empowering leadership on
Follower collectivism was also related positively to affective com- psychological empowerment in our Study 1 and on affective
mitment (see Table 5, Model 3) and negatively to turnover inten- commitment in both Studies 1 and 2. We thus conclude from our
tions (see Table 5, Model 9). No other effects were detected with findings that an optimal mix between motivating and demotivating
follower nationality or collectivism. Interestingly, leader national- team stimuli can exist, such that high levels of empowering lead-
ity was not related to any of the outcomes in the model. Finally, as ership are more likely to promote individual team members’ mo-
in Study 1, additional analyses indicated that follower nationality tivational states when relationship conflict is low, rather than high.
and follower collectivism did not significantly moderate the These moderation effects support prior theorizing by Kanfer and
unique and interaction effects of empowering leadership and rela- her colleagues (Chen & Kanfer, 2006; Kanfer & Heggestad, 1997),
tionship conflict on psychological empowerment or affective com- which suggest that a motivational fit between different team stim-
mitment (p ⬎ .10). uli is required to produce the most motivating context for team
In summary, as in Study 1, individual-level and societal-level members.
cultural differences related differently to psychological empower- Second, further extending Chen and Kanfer’s (2006) theory and
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
ment—that is, collectivism related positively to psychological related empirical research (e.g., Chen, Kirkman, et al., 2007;
This document is copyrighted by the American Psychological Association or one of its allied publishers.
empowerment, whereas psychological empowerment was higher Seibert et al., 2004), our study also linked team-level stimuli to
among the less collectivistic U.S. team members relative to the both positive (innovative and teamwork behaviors) and negative
more collectivistic PRC team members. Also as in Study 1, results (turnover intentions) individual-level outcomes by specifying two
in Study 2 indicated that cultural differences do not seem to affect motivational states (psychological empowerment and affective
the validity of our theoretical model of relationships. commitment) as mediators. Interestingly, although Study 1 sup-
ported all hypothesized relationships in our model, the data in
General Discussion Study 2 provided more mixed support for the model. In particular,
although empowering leadership significantly related to psycho-
In this article, we have sought to answer three specific research logical empowerment and affective commitment in both studies,
questions pertaining to individuals’ motivation. First, we sought to relationship conflict related significantly to affective commitment
uncover individual-level mechanisms through which team stimuli in both studies but related also to psychological empowerment
affect team members’ motivation to contribute to their team; only in Study 1. Also, although psychological empowerment and
second, we considered what dynamics in work teams motivate affective commitment mediated influences of team stimuli on
versus demotivate members to contribute to their teams; and third, turnover intentions in both studies, in Study 2 psychological em-
we asked whether such team dynamics generalize across different powerment linked team stimuli to innovative behavior but not to
aspects of cultures (nationality and individual collectivism). The teamwork behavior, whereas affective commitment linked team
findings of our cross-cultural laboratory and field studies provide stimuli to teamwork behavior but not to innovative behavior.
evidence for the motivating effects of empowering leadership and Thus, our findings suggest that it is possible that empowering
the demotivating effects of relationship conflict, as well as for the leadership and relationship conflict relate to individual outcomes
roles of team members’ motivational states (psychological em- through somewhat different motivational mechanisms, at least in
powerment and affective commitment) in linking empowering some settings. That is, while empowering leadership may relate to
leadership and relationship conflict to important individual-level individual outcomes through both psychological empowerment
outcomes in teams. Additionally, by collecting data in both the and affective commitment, it is possible that relationship conflict
United States And the PRC, we have been able to make some is more likely to relate to outcomes through affective commitment
preliminary statements regarding the role of cultural differences— than through psychological empowerment. These latter effects are
especially those associated with collectivism—in members’ moti- consistent with scholars’ findings that relationship conflict is likely
vational reactions in teams. to exert stronger, more direct effects on more affective-based
outcomes, such as liking of other group members and affective
Theoretical Implications commitment (cf. Jehn, 1995, 1997; Jehn & Mannix, 2001). Still,
our findings broaden the understanding of how, why, and when
The present findings contribute to existing theory in four ways. motivating and demotivating team stimuli lead members to con-
First, our research extends prior theoretical models of employee tribute to their teams.
motivation in teams by explicating the processes through which Third, by considering both motivating and demotivating team
ambient, team-level stimuli exert both motivating and demotivat- stimuli, as well as positive and negative individual-level contribu-
ing influences on team members’ motivation. Although some prior tions to their teams, our study also helps to integrate theories of
research has considered motivational processes in teams (Chen & team motivation, conflict, and stress. As we noted at the outset of
Gogus, 2008; Chen & Kanfer, 2006), most of this work has the article, to date, the team motivation literature has emphasized
focused on either motivating or demotivating team influences on motivating influences and positive behaviors and outcomes (Chen
team members. Extending prior work, a key contribution of our & Gogus, 2008; Chen & Kanfer, 2006), whereas the conflict and
research involves finding (a) that empowering leadership (a moti- stress literatures have emphasized demotivating influences and
vating team stimulus) and relationship conflict (a demotivating negative behaviors and outcomes (e.g., Jehn, 1995, 1997; Jehn &
team stimulus) uniquely and oppositely influence team members’ Mannix, 2001; LePine et al., 2005; Podsakoff et al., 2007). Our
motivation and (b) that the positive influence of empowering theorizing and empirical findings clearly show that integrating
leadership on team members’ motivation is greater when relation- these perspectives allows for greater understanding of motivational
ship conflict is low. In particular, relationship conflict negatively reactions and behaviors in teams. Indeed, in line with our assertion
MOTIVATING FORCES IN TEAMS 555
that team members can potentially be exposed to both motivating Limitations and Future Research Directions
and demotivating stimuli, we in fact found that considering both
As with any research study, ours is not without limitations. First,
types of team stimuli can explain more variance in team members’
the findings in our field study (Study 2) were not as strong as those
motivation and behavior than considering only one type of stimuli.
of our experimental study (Study 1). One explanation for these
Finally, answering repeated calls for more cross-cultural re-
differences in findings may be that the field study, relative to the
search on motivation (Gelfand et al., 2007; Tsui et al., 2007), our lab study, may have lacked sufficient statistical power to detect our
exploratory analyses did not detect significant moderating effect hypothesized effects. Another possible reason is that we measured
for nationality or collectivism on any relationships in the model. behavioral intentions in the lab study but ratings of actual behav-
Although this could be conceived as testing a null hypothesis (i.e., iors in the field sample. Hence, it could be that the direct and
that results do not differ across cultural differences) and hence mediated influences of team stimuli were stronger when consid-
should be interpreted with caution, these results offer little reason ering behavioral intentions, which are more proximal to team
to expect that models of team motivation would not generalize to stimuli and motivational states than actual behaviors (cf. Chen &
non-Western societies. In addition, across both studies, higher Kanfer, 2006). Yet other possibilities for the inconsistency in
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
levels of collectivism were related to higher levels of psycholog- findings might be the diversity in team types represented in the
This document is copyrighted by the American Psychological Association or one of its allied publishers.
ical empowerment and, to a lesser extent (i.e., in Study 2 but not field sample (i.e., our hypothesized effects might be stronger in
in Study 1), affective commitment. These findings suggest that some types of teams than others), as well as the low ICC(2) values
more collectivistic members tend to be more motivated regardless we obtained for measures of team stimuli in Study 2, which could
of the team-level stimuli they may be experiencing. Additionally have attenuated our findings (cf. Bliese, 2000). Thus, although we
and perhaps paradoxically, we also found across the two studies attempted to maximize internal validity in Study 1 and external
that individuals from a less collectivistic society (the United validity in Study 2, clearly, replications that take these measure-
States) were also more likely to feel psychologically empowered ment and sampling differences into account are needed.
than individuals from a more collectivistic society (the PRC; cf. Additionally, we collected data in only two countries: the United
Hofstede, 1980). These findings are consistent with Kirkman, States and the PRC. While these two countries are dissimilar on
Chen, Farh, Chen, and Lowe (2009), who also found that cultural many cultural variables, they do not represent the entire range of
influences at the individual level and society level may not be cultures available to study. Research conducted in a greater array
identical. Clearly, more research is needed to extend the under- of cultural contexts would permit more conclusive statements to be
standing of how cultural differences might affect motivational made regarding the role of culture in the interplay among team
processes in teams and the level of analysis at which cultural stimuli, work motivation, and employee behaviors.
differences might affect such processes. Finally, while our choice of predictor, mediating, and outcome
variables was theoretically grounded, additional research could
expand the group of tested variables. For example, use of different
Practical Implications sets of challenge and hindrance stressors (such as abusive super-
vision or role overload; cf. LePine et al., 2005), as well as con-
Our research also offers two important implications for manag- sideration of different motivating team stimuli (e.g., transforma-
ers and organizations. First, our findings indicate that team leaders tional leadership, team cohesion, team empowerment), would
who engage in empowering leadership behaviors are likely to exert allow for greater integration between models of team motivation
motivating influences on their team’s members but that such and employee stress. Additionally, motivational states other than
positive influences are more likely to occur when relationship psychological empowerment and affective commitment (such as
conflict in teams is kept to lower levels. Managers who encounter perceptions of justice or intrinsic motivation), as well as inclusion
moderate to high levels of relationship conflict in their work teams of affective states (e.g., negative and positive emotional states),
should thus seek to mitigate its effects on employees by develop- should be considered in subsequent models.
ing a cohesive and supportive team environment among team
members (Jehn & Mannix, 2001). On the basis of these findings, Conclusion
managers need to have a solid understanding of the interpersonal In conclusion, the present research takes a significant step
dynamics in their teams to ensure that their efforts to motivate their toward greater understanding of the dynamics in work teams that
team members are not nullified by dysfunctional team dynamics, motivate versus demotivate individual members to contribute to
such as high levels of relationship conflict among members. their teams. In doing so, we have offered contributions to the team
Second, although tentative, our findings involving cultural in- motivation, conflict, and stress literatures by considering the si-
fluences on motivation suggest further that interventions that mo- multaneous positive and negative influences of teams on their
tivate individuals in teams—at least those directed at enhancing members’ motivation and behavior. We have also offered a pre-
employee empowerment and commitment— can in fact generalize liminary look at how cultural differences may influence these
across culturally distinct societies. However, given prior findings motivational states and outcomes. Hopefully, the common findings
(e.g., Kirkman et al., 2009; Kirkman & Shapiro, 2001), as well as across our studies, as well as their differences, will prompt more
our own findings pertaining to collectivism suggesting that scholars to examine the combined effects of motivating and de-
individual-level differences in cultural values can also impact team motivating effects of team stimuli, as well as mediating mecha-
members’ motivation, managers also need to realize that some nisms that can explain how team members respond to positive and
employees may embrace (or, rather, resist) their motivational in- negative team-related stimuli—a mixture of events that is inevita-
terventions and motivating behaviors more so than others. ble yet understudied.
556 CHEN, SHARMA, EDINGER, SHAPIRO, AND FARH
Konczak, L. J., Stelly, D. J., & Trusty, M. L. (2000). Defining and Seibert, S. E., Silver, S. R., & Randolph, W. A. (2004). Taking empow-
measuring empowering leader behaviors: Development of an upward erment to the next level: A multiple-level model of empowerment,
feedback instrument. Educational and Psychological Measurement, 60, performance and satisfaction. Academy of Management Journal, 47,
301–313. doi:10.1177/00131640021970420 332–349. doi:10.2307/20159585
Kozlowski, S. W. J., & Bell, B. S. (2003). Work groups and teams in Seo, M.-G., Barrett, L. F., & Bartunek, J. M. (2004). The role of affective
organizations. In W. C. Borman & D. R. Ilgen (Eds.), Comprehensive experience in work motivation. Academy of Management Review, 29,
handbook of psychology: Industrial and organizational psychology (pp. 423– 439. doi:10.2307/20159052
333–375). New York, NY: Wiley. Seo, M.-G., & Ilies, R. (2009). The role of self-efficacy, goal, and affect in
LeBreton, J. M., & Senter, J. L. (2008). Answers to twenty questions about dynamic motivational self-regulation. Organizational Behavior and Human
interrater reliability and interrater agreement. Organizational Research Decision Processes, 109, 120 –133. doi:10.1016/j.obhdp.2009.03.001
Methods, 11, 815– 852. doi:10.1177/1094428106296642 Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis: An intro-
LePine, J. A., Podsakoff, N. P., & LePine, M. A. (2005). A meta-analytic duction to basic and advanced multilevel modeling. London, England:
test of the challenge stressor-hindrance stressor framework: An expla- Sage.
nation for inconsistent relationships among stressors and performance. Spreitzer, G. M. (1995). Psychological empowerment in the workplace:
Academy of Management Journal, 48, 764 –775. Dimensions, measurement, and validation. Academy of Management
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.