Identity Asymmetries: An Experimental Investigation of

r Academy of Management Journal2020, Vol. 63, No. 5, 1561–1590.https://doi.org/10.5465/amj.2018.0325

IDENTITY ASYMMETRIES: AN EXPERIMENTALINVESTIGATION OF SOCIAL IDENTITY AND

INFORMATION EXCHANGE IN MULTITEAM SYSTEMS

JULIJA N. MELLErasmus University Rotterdam

LESLIE A. DECHURCHNorthwestern University

ROGER TH. A. J. LEENDERSJheronimus Academy of Data Science and Tilburg University

NOSHIR CONTRACTORNorthwestern University

Many complex organizational tasks are performed by networks of teams, or “multiteamsystems.” A critical challenge in multiteam systems is how to promote informationexchange across teams. In three studies, we investigate how identity “asymmetries”—differences between teams in terms of whether the team or overarching system consti-tutes their primary focus of identification—affect interteam information sharing andperformance. In Study 1, we manipulate teams’ foci of identification (team vs. systemfocused) in a sample of 84 five-member teams working in one of 21 four-team multiteamsystems performing a computer strategy simulation. We find that, while system-focusedteams shared information equally with all teams, team-focused teams shared less in-formation with system-focused teams than they did with other team-focused teams.Interteam information sharing positively predicted interteam performance. In Study 2, we testthe assumptions underlying our theory in a vignette experiment, demonstrating that team-focused individuals adopt instrumentalmotives toward interteam interaction. Finally, in Study3,we investigate the implications of system composition in terms of team identity foci bymeansof a simulation study based on the empirical results of Study 1. The results of the simulationyield novel propositions about the nonlinear effects of social identity in multiteam systems.

Many complex organizational activities go beyondthe capabilities of single teams and require the

interdependent and coordinated action of multipleteams in the pursuit of collective goals (de Vries,Hollenbeck, Davison, Walter, & van der Vegt, 2016;Mathieu, Hollenbeck, van Knippenberg, & Ilgen,2017). For example, large-scale research or newproduct development projects often consist of multi-ple specialized teams that develop differentiatedmodules that need to be integrated into a coherentproduct (Aalbers, Dolfsma, & Leenders, 2016;Hoegl &Weinkauf, 2004;Leenders&Dolfsma,2016).Emergencymedical care requires interdependent collaborationof multiple teams—for example, a paramedic team,an emergency unit team, and a stationary careteam—in treating each patient (DiazGranados, Dow,Perry, & Palesis, 2014). Complex military operationsrequire the closely coordinated action of multipleteams—in the field aswell as in the “backoffice”—oftenspanning organizational and national boundaries

The authors would like to thank Professor PrithvirajChattopadhyay and three anonymous reviewers for theirthoughtful and developmental suggestions throughout thereview process. We also thank Daan van Knippenberg andSujin Jang, as well as seminar participants at ESSECBusiness School, INSEAD, and the Rotterdam School ofManagement, for insightful feedback. We also gratefullyacknowledge the funding received for this study throughthe Army Research Institute (W5J9CQ-12-C-0017), theNational Science Foundation (BCS-0940851), and theArmy Research Laboratory (W911NF-09-2-0053, NS-CTA). We also gratefully acknowledge the funding re-ceived for this study through the Army Research Institute(W5J9CQ-12-C-0017), the National Science Foundation(BCS-0940851), and the Army Research Laboratory(W911NF-09-2-0053, NS-CTA).

1561

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download, or email articles for individual use only.

https://doi.org/10.5465/amj.2018.0325

(Goodwin, Essens, & Smith, 2012). Space explora-tion missions rely on a network of ground teams toprepare,monitor, and support every step in theworkof the space crew (Mesmer-Magnus, Carter, Asencio,& DeChurch, 2016). All of these are examples of mul-titeam systems, “tightly coupled network[s] of teams”that “need to coordinate their efforts to achieve oneor more goals in addition to those of the componentteams” (Luciano, DeChurch, & Mathieu, 2018: 3;Mathieu, Marks, & Zaccaro, 2001).

The success of such multiteam systems criticallydepends on “interteam coordination”—organizingand aligning interdependent activities across teamboundaries (de Vries,Walter, van der Vegt, & Essens,2014; DeChurch&Marks, 2006)—and, especially, oninterteam information sharing, a core aspect of co-ordination (Marks, Mathieu, & Zaccaro, 2001). Thisis most evident when information sharing fails. Forinstance, insufficient interteam communication in newproduct development projects has been shown to com-promise the quality of the product and result in signifi-cant financial and reputational damages (Gokpinar,Hopp, & Iravani, 2010). Similarly, in health-caresettings, gaps in information sharing during patienthandoffs between medical teams have been shownto result in adverse clinical consequences for pa-tients (Horwitz, Moin, Krumholz, Wang, & Bradley,2008; Luciano, 2017).

A key factor affecting interteam coordination andinformation sharing is members’ social identity. So-cial identity theory (Tajfel & Turner, 1986) and itsextension, self-categorization theory (Turner, Hogg,Oakes, Reicher, & Wetherell, 1987), suggest that anindividual’s self-concept partly “derives from hisknowledge of his membership of a social group to-gether with the value and emotional significance at-tached to that membership” (Tajfel, 1978: 63). Suchself-concepts become behaviorally relevant as indi-viduals who strongly identify with a group showmorecommitmentandcooperation towardothermembersofthat group than toward out-group members (Ashforth,Harrison, & Corley, 2008). However, just as in anyfairly complex organization, amember of amultiteamsystem is simultaneously a member of multiple nes-ted groups—the more proximal component team andthe overarching multiteam system. These multiplememberships offer multiple “foci of identification”(van Knippenberg & van Schie, 2000).

Recentworkhighlights the importanceofmembers’identification with the superordinate focus—for ex-ample, the organization or themultiteam system—forinterteam coordination and effectiveness (Cuijpers,Uitdewilligen, & Guenter, 2016; de Vries et al., 2014;

Dokko, Kane, & Tortoriello, 2014; Lomi, Lusher,Pattison, & Robins, 2014; Richter, West, van Dick, &Dawson, 2006). While these insights advance ourunderstanding of the role of social identity for inter-teamprocesses and effectiveness, they share a criticalblind spot in ignoring that component teams withinthe same system can differ in the extent to which ei-ther the team or the multiteam system is their moresalient focus of identification.

Variation in identity foci can have different sour-ces. For example, in a new product developmentproject, some modules typically have more physicaland functional interfaces with other modules thanothers (Gokpinar et al., 2010). Teams working onthese modules might thus be more aware of theinterdependent nature of the team network as acollective system—the multiteam system member-ship becomes more salient for these teams than forteams working on modules with fewer interfaces(Connaughton, Williams, & Shuffler, 2012). As an-other example, in a space exploration mission, thephysical and social isolation of the space crew fromthe ground teams may make the team itself a moresalient focus for the space crew while the groundteams may identify most with the overarching sys-tem and its goals. In sum, variation in identity foci islikely prevalent in multiteam systems—and yet, todate, we have little theory and empirical insight intohow different configurations of identity foci affectinterteam collaboration.

The limited insight into the consequences of vari-ation in identity foci for interteam collaboration is themore striking in view of prior research that elucidatesantecedents that can lead to differences in identifi-cation within groups and examines consequencesthereof. For instance, drawing on social identity andself-categorization theories, the relational demogra-phy literature has shown how a team’s configura-tion of similarity and dissimilarity on demographic,occupational, or work status attributes can result ingroup members differentially identifying with differenttargets—that is, the team versus their demographic, oc-cupational,orworkstatuscategory (e.g.,Chattopadhyay,George, & Lawrence, 2004; George & Chattopadhyay,2005). These differences in identification, in turn, havebeen shown to result in asymmetrical individual-levelattitudes and behaviors toward the team, includinginteraction patterns, trust, organizational citizenship,and perceptions of conflict (e.g., Chattopadhyay, 1999;Chattopadhyay, George, & Shulman, 2008; George,Chattopadhyay, & Zhang, 2012).

The present paper, while building on the sametheoretical foundation as this prior work, goes

1562 OctoberAcademy of Management Journal

beyond it, not only in shifting the level of analysis,but also in spelling out the mechanisms governingthedyadic interaction between teams as a function ofthe specific configuration of identity foci in a teamdyad. More specifically, this paper explores the or-ganizational consequences of “identity asymme-tries” in multiteam systems—that is, situations inwhich interdependent component teams differ interms ofwhich entity (themore proximal componentteam or the more distal overarching system) consti-tutes their primary focus of identification. Drawingon social identity and self-categorization theories,we develop theory about the effect of identityasymmetries on interteam information sharing as animportant facet of interteam coordination (Markset al., 2001) and interteam performance (i.e., theachievement of team goals that require interdepen-dent work with other teams). We argue that differentidentity foci result in different motives underlyinginteraction with other component teams: while ateam focus elicits instrumental motives rooted in adesire to enhance team welfare by means of condi-tional cooperation with other teams, a system focuselicits more benevolent motives rooted in a desire toenhance shared welfare (Biel & Thøgersen, 2007;Chatman & Flynn, 2001; Goette, Huffman, & Meier,2006; Kerr, 1995). We further propose that, whileeither logic can sustain a productive collaborativerelationship between two teams when both teamsact based on similar motives, identity asymmetrieswill disrupt interteam coordination and, as a conse-quence, interteam performance. We test these pre-dictions in a laboratory experiment involving 84teams (252 team dyads) nested in 21 multiteam sys-tems. Ina secondstudy,we test the assumptions aboutthe underlying motivational mechanisms, positing thatteam-focused individuals adopt more instrumental mo-tives toward interteam interaction. Finally, building onour findings, we conduct a simulation study to considertheimplicationsof identityasymmetries forsystem-levelcoordination and performance given different compo-sitions of the system in terms of identity foci.

This study makes several contributions to the liter-ature. First, it provides causal evidence for the role ofidentity asymmetries for interteam information shar-ing andperformance inmultiteamsystems.Second, bycreating a better understanding of how identity asym-metries affect interteam collaboration and multiteamsystem functioning, it challenges the often-held as-sumption of a straightforwardly linear positive rela-tionship between superordinate identification andinterteam processes and system performance. Thirdand more broadly, the analysis of distinct identity

configurations and their effects extends fundamentaltheory on social identity and intergroup relations,highlighting identity composition as a characteristicthat, while having significant implications for multi-team system functioning, has hitherto been largelyoverlooked.

THEORETICAL BACKGROUNDAND HYPOTHESES

“Informational interdependence,”whichexistswhenone team (the “seeker”) requires information from an-other team(the“source”) for thepursuitof its goals, is animportant facet of interdependence among componentteams of a multiteam system. Collaboration betweenteams that are informationally interdependent requiresboundary-spanning communication and, in particular,“information sharing” as critical coordination pro-cesses. Interteam information sharing is a team-levelactivity emerging from individual behavior as teamscollectively organize their boundary-spanning inter-action (Marrone,2010). It canbe initiatedbyeither thesource or the seeker. In the first case, the source sharesinformation with the seeker without a request—thatis, proactively. Proactive information sharing consti-tutes a prototype of implicit coordination: the sourceanticipates the needs of the seeking team and actsupon them without a need for an explicit request(Fisher, Bell, Dierdorff, & Belohlav, 2012; Rico &Sanchez-Manzanares, 2008: 165). In the second case,the source shares information with the seeker in re-sponse to a request from the seeker—that is, reac-tively. Reactive information sharing constitutes anexplicit coordinationmechanism in the sense that theinformation-sharing activity itself is explicitly nego-tiated between the teams through the preceding re-quest. While both explicit and implicit coordinationgenerally have positive performance implications,their antecedents as well as their relative contribu-tion to collective performance may differ (Espinosa,Lerch, & Kraut, 2004), and we therefore considerthem side by side as we develop our theory.

Interteam information sharing has been recognized asa critical foundation of interteam effectiveness acrossdifferent fields of research (Best, 2011; Gokpinar et al.,2010; Horwitz et al., 2008). At the same time, it is aninherently challenging activity. Even within teams, in-formation isoftenexchanged less thanneeded (Mesmer-Magnus & DeChurch, 2009), and team boundaries onlyfurther limit interteam communication and informationexchange (Caimo & Lomi, 2015; Feld, 1981; Lomi et al.,2014). In the following sections, we examine the role of

2020 1563Mell, DeChurch, Leenders, and Contractor

social identity as a key factor that can help multiteamsystems to overcome this challenge.

Social Identity in Multiteam Systems

The social identity and self-categorization theories(Tajfel, 1978; Tajfel & Turner, 1986; Turner et al.,1987) suggest that an individual’s memberships inorganizational groups such as the organization itselfas well as workgroups, teams, divisions, or job cate-gories nested within the organization inform his orher self-concept (Hogg & Terry, 2000). To the extentthat the groupmembership is salient and valuable toindividuals—that is, to the extent that they stronglyidentify with the group—they perceive members ofthat group as their in-group and members of othergroups as out-group members. Behaviorally, this typi-cally results in in-groupmembers receivingpreferentialtreatment: individuals who strongly identify with agroup show more cooperation toward in-group mem-bers than toward out-group members (Ashforth et al.,2008; Brewer, 1979; Hewstone, Rubin, &Willis, 2002).However, in almost any organization and—veryprominently—in anymultiteam system, individuals aresimultaneouslymembers ofmultiple groups, whichprovidemultiple foci of identification (vanKnippenberg& van Schie, 2000). In a multiteam system, membershave two main foci—the component team and theoverarching system in which the teams are nested.These two foci can have different degrees of salienceto an individual and their relative salience shapeswhat an individual perceives as the primary bound-ary between in- and out-group (Gaertner, Dovidio,Anastasio, Bachman, & Rust, 1993; Hogg & Terry,2000). Below, we will refer to individuals who per-ceive theboundaryaround thecomponent teamas theprimary boundary separating in- and out-groups as“team focused” and to individuals who perceive theboundary around the multiteam system as the pri-mary boundary as “system focused.”

Differences in what is perceived as in- and out-group boundaries result in differences in behaviortoward other component teams. Earlier research hasshown that, generally,moreproximate foci tend tobemore salient to individuals than more distal foci(Riketta & van Dick, 2005; van Knippenberg & vanSchie, 2000). As a result, individualsworkingwithina multiteam system tend to prioritize activities di-rected toward their own team over activities di-rected toward other teams. This, however, can bean impediment in the multiteam system contextwhere teams require intense interteam coordina-tion. A seemingly straightforward remedy, then, is to

foster a system focus in the component teams, thusextending the perceived in-group to include mem-bers of other component teams (Gaertner et al., 1993).Indeed, prior work highlights the importance ofmembers’ identification with a superordinate focusfor interteam coordination and effectiveness. For in-stance, prior researchhas found that identificationwitha superordinate focus increases the likelihood that in-dividuals will interact and cooperate with membersof other teams (de Vries et al., 2014; Dovidio, Gaertner,Validzic, Matoka, Johnson, & Frazier, 1997; Kramer &Brewer, 1984; Lomi et al., 2014; Richter et al., 2006;van Dick, van Knippenberg, Kerschreiter, Hertel, &Wieseke, 2008; Wit & Kerr, 2002). Furthermore, indi-viduals who identify with a superordinate focus havebeen shown to be more attentive to and to make moreuse of information they obtain from members of othergroups (Dokko et al., 2014; Kane, 2010; Kane, Argote, &Levine, 2005). On a system level, systems whose com-ponent teams share a superordinate identity focus havebeen found to collaborate more effectively (Cuijpersetal., 2016)—albeit, inaninterestingcounterpoint, recentwork found the opposite (Porck, Matta, Hollenbeck, Oh,Lanaj, & Lee, 2019).

Two critical assumptions are, to varying degrees,inherent in this line of research. The first is that whatconstitutes the primary identity focus varies acrossbut not within multiteam systems. This assumptionis most explicit in research in which measures ofsocial identification are aggregated to the systemlevel (Cuijpers et al., 2016; Porck et al., 2019). Yet,members of different component teams are embed-ded in different local contexts and subgroups and areexposed to different localized factors that can affectthe relative salience of the teamversus themultiteamsystem identity. For example, teams may have dif-ferent positions in the geographical arrangement orin the workflow of themultiteam system and, hence,havedifferent exposure to shared tasks andproblems(Bartel,Wrzesniewski, &Wiesenfeld, 2011; Davison,Hollenbeck, Barnes, Sleesman, & Ilgen, 2012; Hinds& Mortensen, 2005). Teams may also differ in status(Chattopadhyay, Tluchowska, &George, 2004; Tajfel&Turner, 1986), in the extent towhich teamgoals arecompatible with other teams’ and multiteam systemgoals (Rico, Hinsz, Burke, & Salas, 2017), or in teamleaders’ rhetoric and behavior (Shamir, Zakay,Breinin, & Popper, 1998). To the extent that theseantecedents alter the relative salience of team andsystemboundaries, they can result in asymmetries inwhat members of different component teams per-ceive as their primary foci of identification. Weargue that, given the many possible antecedents,


asymmetries in identity foci between teams are notonly possible but even probable.

The second implicit assumption in this line of re-search is that, even where differences in socialidentification between interacting parties exist, theyare not consequential to the interaction betweenthese parties. This assumption is implicit in researchthat considers a focal individual’s or group’s behav-ior toward another group as a function of theformer’s—but not the latter’s—social identification(de Vries et al., 2014; Dovidio et al., 1997; Hornsey &Hogg, 2000; Kane, 2010; Kane et al., 2005; Kramer &Brewer, 1984; Richter et al., 2006; van Dick, vanKnippenberg, Kerschreiter, et al., 2008; Wit & Kerr,2002). It remains present even in research thatconsiders social identification of both parties—forinstance, of both the information seeker and of theinformation source—but without recognizing thatthe effect of one party’s identity focus may be con-ditional on the identity focus of the other party(Dokko et al., 2014; Lomi et al., 2014). In contrast, weargue that identity asymmetries between componentteams have important and unique consequences forinterteam information sharing and performance.

Social Identity Asymmetries and ProactiveInformation Sharing between Teams

Cooperative interteambehavior—suchasproactiveinformation sharing—can be based on different mo-tives that range between self-interested instrumen-tality and other-interested benevolence. Followingprior work, we use the term “benevolence” in a broadsense, describing actions that are prosocial in thesense that they aim at enhancing the welfare of a rel-evant overarching collective that encompasses selfand other (Biel & Thøgersen, 2007: 102; Bolino &Grant, 2016). Thus, teams whose members are moti-vated by benevolence may share information withother teams because this contributes to the sharedwelfare of theoverarching system, even if itmaycomeat a cost to its intrateam-directed activities (Biel &Thøgersen, 2007; Chatman & Flynn, 2001). Informationsharing guided by a benevolence motive is not condi-tional on the behavior of the direct recipient but ratherfollows a logic of generalized reciprocity. It is based ontheassumption thatothers—whoarenotnecessarily thedirect recipients of their contribution—will equally co-operate in the future (Baker & Bulkley, 2014; Bearman,1997; Molm, Collett, & Schaefer, 2007). Teams whosemembers aremotivatedby instrumentality, on theotherhand, may share information with another team asa way to ensure that specific team’s reciprocal

cooperation. Such information sharing follows a logicof direct reciprocity, which may be viewed as “a formof ‘conditional kindness’ whereby advice is givenunder the expectation that itwill be received” (Caimo&Lomi, 2015: 671; Fehr & Gachter, 2000).

Prior research has shown that intra- and inter-group relations tend to be guided by different mo-tives. Because the perception of belonging to thesamegroup implies a concern for sharedwelfare and,thus, a motivation to ensure the success of not onlyself but also that of other group members, direct re-ciprocation is not necessary to motivate cooperativeaction toward an in-group member (Flynn, 2005).Correspondingly, empirical research has shown thatthe perception of belonging to the same group elicitsbenevolence toward in-group members and expec-tations of generalized reciprocity (Goette et al., 2006;Yamagishi &Kiyonari, 2000). The very same concernfor the welfare of the in-group, however, implies astronger focus on the instrumental value of interac-tions with those who are perceived as out-groupmembers. Correspondingly, advice and knowledgeexchange relationships between members of differ-ent organizational groups have been shown to begoverned more strongly by direct reciprocity thanintragroup relationships (Brennecke & Rank, 2016;Caimo & Lomi, 2015).

Because differences in identity focus imply differ-ences in where the subjective boundary between in-and out-group is drawn, multiteam system membersthat differ in identity fociwill likely differ inhow theyapproach relationswithmembers of other componentteams.While individuals with a team focus will viewrelations with members of other component teams asintergroup relations, system-focused individuals willperceive members of other component teams as in-group members and so they are likely to approachinterteam relations as they would intragroup relations.As a result, the behavior of team-focused teams towardother component teams is likely to be guided by moreinstrumental motives: cooperation with other compo-nent teams is a means to an end and conditional on itsinstrumental value. Conversely, the behavior of system-focused teams toward other component teams is likelyto be guided by more benevolent motives: cooperationwith other component teams is an end in itself and notconditional on its instrumental value. This differencehas multiple implications for interteam informationsharing.

First, both motives can, in principle, result in sus-tained cooperation. Members of team-focused teamswill invest resources in proactively sharing informa-tion with another team if they assume and observe


that the other team’s information sharing is condi-tional on their own behavior. Early work on individ-uals’ behavior in social dilemmas corroborates thisline of reasoning: when interacting with an opponentwho used a reciprocity-oriented tit-for-tat strategy,individuals primarily focused on maximizing theirown utility showed similar levels of cooperative be-havior as individuals focusedonmaximizing the sharedutility (Kuhlman & Marshello, 1975). Analogously, twoteamswhobothhave a team focus are likely to engage insustained proactive sharing, as both perceive the likeli-hood of receiving information from the other party asconditional on their own proactivity.

Second, when two teams differ in their primaryfocus of identity, we have no reason to expect thatmembers of a system-focused team would share in-formation differently with the team-focused teamthan they would with any system-focused team. Be-cause they perceive the superordinate membershipas more salient, their behavior toward other teamswill be more strongly guided by benevolence. Thus,we can expect that their information sharing withother teams would be as open, unconditional, andproactive as if they were members of the same team.

Third, and most importantly: the team-focusedteam in such an asymmetric dyad may behave quitedifferently. As described above, for a team-focusedteam, sharing informationwith another team ismoreof a means toward the end of obtaining informationnecessary for the pursuit of team goals rather than abehavior driven by concern for shared goals. Real-izing over the course of the interactionwith a system-focused team that the other party’s cooperation is notcontingent on their own behavior, the team-focusedteam is likely to shift its attention and resources to-ward other demands.While theymay still respond todirect requests, they will be less likely to invest theadditional effort of anticipating theother team’sneedsrequired by proactive information sharing. Again, wecan draw a parallel to individuals’ behavior in socialdilemmas: while individuals concerned with sharedwelfare show cooperative behavior both toward op-ponents who use a reciprocal tit-for-tat strategy andthose who consistently and unconditionally cooper-ate, individuals primarily concerned with their ownutility show considerably lower levels of cooperationtoward opponents who cooperate unconditionallythan toward opponentswho reciprocate bothpositiveandnegative behaviors (Kuhlman&Marshello, 1975).Correspondingly, we expect that, in the presence ofan identity asymmetry, a team-focused source teamwill reduce its level of proactive information sharing

toward a system-focused seeking team relative to ateam-focused seeking team. More formally:

Hypothesis 1a. There is an interaction between thesource team’s and the seeking team’s identity focussuch that team-focused source teams are less likely toproactively share information with system-focusedseeking teams thanwith team-focused seeking teams.

Social Identity Asymmetries and ReactiveInformation Sharing between Teams

While we expect that team-focused source teamsengage in less proactive information sharing towardsystem-focused seeking teams, we expect the oppo-site dynamic to arise with regard to reactive sharing.Our argument here rests on two assumptions. First,prior work has suggested that explicit and implicitcoordination are inversely related: when implicit co-ordination is established, the need for explicit coor-dination decreases (Espinosa et al., 2004; Rico &Sanchez-Manzanares, 2008). In the context of infor-mation sharing, this means that themore informationa source team shares with the seeking team proac-tively, the less the seeking team will have to ask thesource team for information. Conversely, this alsomeans that the less information a source team shareswith the seeking team proactively, the more theseeking teamwill need to ask the source team in orderto obtain the information they need. Thus, essentiallyas a side effect of team-focused source teams sharingless information with system-focused seeking teamsproactively, system-focused seeking teams will ex-tendmore information requests toward team-focusedsource teams. Furthermore, although there is, ofcourse, also a probability that a source team choosesnot to respond to an information request, work onknowledge hiding has shown that denying explic-itly requested information is a very rare behavior(Connelly, Zweig, & Webster, 2012). Thus, our sec-ond assumption is that most requests that are madeare also responded to. Therefore, we expect that theincrease in requests will be directly visible in an in-creased proportion of reactively shared informationby team-focused source teams toward system-focusedseeking teamsascomparedwith toward team-focusedseeking teams. In sum:

Hypothesis 1b. There is an interaction between thesource team’s and the seeking team’s identity focussuch that team-focused source teams are more likelyto reactively share information with system-focusedseeking teams thanwith team-focused seeking teams.


Social Identity Asymmetries, Information Sharing,and Interteam Performance

Interteam information sharing is consequential tomultiteam systems because, in the context of infor-mational interdependence among the componentteams, it directly impacts interteam performance. Wedefine “interteam performance” as a dyadic, directedconstruct that captures the extent to which a specificfocal teamsucceeds inachieving goals that require thecollaboration of a specific partner team. In our con-text, we consider the performance of a seeking team(this is the focal team) on tasks that require informa-tion from a specific source team (this is the partnerteam). Interteamperformance isdistinct fromwhatwemight call “intrateam performance” in that interteamperformance excludes from consideration the extenttowhich a focal teamachieves goals forwhich theydonot rely on other teams. Furthermore, interteam per-formance is a directed construct in the sense that, in adyad where both teams are mutually dependent oneachother, teamAmaybemore (or less) successful ongoals that require teamB’s collaboration thanteamBison goals that require team A’s collaboration.

The arguments in the preceding section imply that,where there is an identity asymmetry between theseeking and the source team, interteam coordinationshifts from implicit coordination based on proactiveinformation sharing to explicit coordination based onreactive information sharing. Both routes are, in prin-ciple, effective coordinationmechanisms—as long as ateamobtains the information it needs, it canproceed toutilize this information in its goal-directed activities.Thus, both proactive and reactive information sharingshould have positive implications for interteam per-formance. More formally:

Hypothesis 2a. Proactive information sharing has apositive effect on interteam performance.

Hypothesis 2b. Reactive information sharing has apositive effect on interteam performance.

While proactive as well as reactive informationsharing should contribute to interteam performance,the search and negotiation activities involved inreactive information sharing make this form ofexplicit coordination more costly (Rico & Sanchez-Manzanares, 2008). While a team is dedicating re-sources to searching for relevant information, theseresources are not available for putting the obtainedinformation into action. Thus, at least in a settingwhere it is relatively clear who may need to knowwhat (a boundary condition we examine at greater

detail in our Discussion section), proactive infor-mation sharing is arguably amore effective interteamcoordination mechanism than reactive informationsharing. Because of this, we expect that proactiveinformation sharing will have a stronger positiveimpact on interteam performance.

Hypothesis 3. Proactive information sharing has astronger positive effect on interteam performancethan reactive information sharing.

Together, the core logic underlying Hypotheses 1to 3 describes how the effect of the identity foci ofsource and seeker teams affects interteam perfor-mance. This logic suggests two mediators: proactiveand reactive sharing. Hypothesis 1 posits the inter-action between the source team’s and the seekingteam’s identity foci affects the probability that theyengage in proactive or reactive information sharing,respectively. Hypothesis 2 posits both types of in-formation sharing are positively related to interteamperformance, but that, per Hypothesis 3, the positiveeffect of proactive information sharing is strongerthan that of reactive information sharing. Taken to-gether, this implies that the configuration of theidentity foci between the source and seeker teamsindirectly affects interteam performance by influenc-ing the extent to which the teams engage in proactiveand reactive information sharing, and that the indirecteffect via proactive information sharing would bestronger than that via reactive information sharing.

Hypothesis 4a. There is an indirect effect of the in-teraction between the seeking team’s and the sourceteam’s identity foci on interteam performance, me-diated by proactive information sharing.

Hypothesis 4b. There is an indirect effect of the in-teraction between the seeking team’s and the sourceteam’s identity foci on interteam performance, me-diated by reactive information sharing.

Hypothesis 5. The indirect effect of the interactionbetween the seeking team’s and the source team’sidentity foci on interteam performance mediated byproactive information sharing is stronger than theindirect effect of the interaction between the seekingteam’s and the source team’s identity foci on interteamperformancemediated by reactive information sharing.

Our arguments thus far suggest that dyads withoutidentity asymmetries would achieve higher inter-team performance as a result of relying more onproactive rather than on reactive information shar-ing.Asprior research shows, however, identification


with the superordinate group (in our context, this isthe multiteam system) not only affects the sharing ofinformation but also makes a team more receptiveto external information, thus increasing the rate atwhich it will be utilized (Dokko et al., 2014; Kaneet al., 2005). That is, while symmetric team-focuseddyads may exchange information at a similar rate assymmetric system-focused dyads, the higher infor-mation utilization rate by system-focused teams thathas been established in prior work leads us to expectthat symmetric system-focused dyads will performat a higher level than dyads in which either one orboth of the parties have a team focus. In sum:

Hypothesis 6. Interteam performance is higher whenboth teams (seeking and source teams) have a systemfocus than when either the seeking, source, or bothteams have a team focus.

STUDY 1: EXPERIMENTAL INVESTIGATIONOF IDENTITY ASYMMETRIES IN

MULTITEAM SYSTEMS

In order to test our hypotheses, we conducteda laboratory experiment using a computer-based,team-based, dynamic strategy simulation, manipu-lating the focus of identification between componentteams nested in multiteam systems. Simulations ofthis kind are widely used in research on teams andmultiteam systems as they allow controlled experi-mentation, structured behavioral observation, andobjective measurement of process and performance(Beersma, Hollenbeck, Humphrey, Moon, Conlon, &Ilgen, 2003; DeChurch & Marks, 2006; Ellis, 2006;Homan, Hollenbeck, Humphrey, van Knippenberg,Ilgen, & van Kleef, 2008; Lanaj, Foulk, & Hollenbeck,2018; Lanaj, Hollenbeck, Ilgen, Barnes, & Harmon,2013; Marks, DeChurch, Mathieu, Panzer, & Alonso,2005; Mathieu, Heffner, Goodwin, Salas, & Cannon-Bowers, 2000; Pearsall & Venkataramani, 2015; Porcket al., 2019). Such simulations are alsowidely used toteach teamwork, coordination, and leadership; forexample, in military training (Beersma et al., 2003)and business education (Pearsall & Venkataramani,2015). While simulations naturally abstract from thehighly complex and specialized knowledge requiredin the field and use student rather than field samples,the teamprocesses that participants experience and theinterpersonal and intergroup behaviors they engage induring such simulations are generally deemed usefulanalogs to the processes and behaviors in the field.Correspondingly, meta-analytic evidence shows thatlab and field settings yield parallel findings with

respect to relationships relevant to our study, such asrelationships between team identity and team perfor-mance (Mesmer-Magnus, Asencio, Seely, & DeChurch,2018) as well as relationships between teamwork pro-cessesand teamperformance (LePine,Piccolo, Jackson,Mathieu, & Saul, 2008).

Sample

The initial sampleconsistedof440 individuals (188female, 252male)whowere recruited fromthecollegeand senior high school population within and in theneighborhood of a largeMidwestern university in theUnited States. Participants were between 16 and 35years old (M 5 21.32, SD 5 3.64). Fifty percent re-ported a Caucasian ethnic background, 20% Asian,12.6% African American, and 11.7% Hispanic. Par-ticipants were assigned to one of 22 multiteam sys-tems, each consisting of four component teams of fivemembers each. They received $35 for their partici-pation. Due to a computer error, one session’s recordof participants’ actions in the simulation was lost.This did not affect their experience nor the surveydata collection and hence we used the data from thefull sample for themanipulation checks. For ourmainanalyses involving data on actions within the simu-lation, however, weworkedwith the reduced sampleof 420 participants nested in 21 multiteam systems.

Experimental Task

In order to test our predictions experimentally, werequired a task with a number of specific character-istics. First, the task needed to contain goals at theteam level as well as at system level. Second, teamsmust be linked by informational interdependence(i.e., require information from other teams in thepursuit of their goals). Third, the task must allow usto capture rich data on all participants’ task-relatedactivity and communication. Based on these criteria,we developed a platform on the basis of a computer-based multiplayer strategy simulation.

The multiteam systems’ collective goal was tosafely direct a humanitarian aid convoy along apredefined route through a war-torn region repre-sented by a map divided into 100 cells. Seventy-fivethreats distributed across the map could damage theconvoy unless they were flagged and neutralizedprior to moving the convoy to the affected cell. Eachof the four component teams could only flag andneutralize threats located in their own district com-prising 25 cells on the map. Each district containedbetween 17 and 20 threats. Each participant


furthermorehada specific role allowinghimorher toperform particular actions in the simulation. “Re-connaissance officers”were responsible for flaggingthreats,while “field specialists”were responsible forneutralizing flagged threats and marking safe cells.Intelligence containing information necessary forflagging and neutralizing threats (i.e., type of threat,cell, and specific coordinates within the cell) wasdistributed among reconnaissance officers and fieldspecialists of all four component teams such thatonly about a quarter of the information required byany single team was given to members within thatteam. The remaining items of information were dis-tributed across members of the three other teams,such that each team required four or five items ofinformation fromeachother team. Finally, each teamfeatured one leader who was responsible for movingthe convoy in coordination with the leaders of theother teams. Given the special position and the dif-ferent task set of leaders, they did not receive anyintelligence items at the beginning of the mission.

In sum, in order to progress toward the system-levelgoal of safely moving the convoy, participants had to(a) exchange information such that relevant intelli-gence reached the teams and individualmemberswhoneeded it, (b) flag the threats based on the intelligence,(c) neutralize the flagged threats, and (d) move theconvoy once the next steps of the route were declaredsafe. In other words, multiteam system success wascritically dependent on effective collaboration andinformation exchange between the interdependentteams as well as on the teams’ successful utilizationof the received information.

Procedure

Upon arrival, participants were randomly assignedto one of the 20 roles in the multiteam system. Theywere seatedat individualworkstations, each teaminaseparate room, and viewed an instruction video aboutthe goals and the gameplay of the simulation. Thevideos were identical for each team up until a finalsegment,which contained the first part of the identitymanipulation. As a second part of the manipulation,following the instruction, participants engaged in avirtual banner-making exercise. We describe theseelements in the “Manipulation” section below. Next,participants filled in a brief survey, discussed strate-gies during a five-minute planning phase, and playeda practice mission of 15 minutes, during which theycould consult research assistants about the interfaceso as to ensure complete understanding of the game-play. The practice mission was followed by a brief

survey and a break. After the break, participants dis-cussed strategies during a seven-minute planningphase. Prior to beginning the main mission, theywatched another brief video in their rooms that aimedto recall and reinforce the identity manipulation.Then, they had 40 minutes for the main mission,which was followed by a final survey. The entireprocedure lasted about 3.5 hours. During both mis-sions, each participant could communicate with anyother participant using one on one Skype chat andcalls. We recorded and transcribed all communica-tionduring themissions aswell as all actions takenbyparticipants in the simulation. We use the data fromthe main mission to test our hypotheses.

Manipulation

Wemanipulated the identity focuswithinmultiteamsystems such that two teams of eachmultiteam systemwere placed in the “team focus” condition and twoteamswere placed in the “system focus” condition. Asa result, in eachmultiteam system,we obtained all fourpossible team-dyadic identity focus configurations: ofthe 12 directed ties in each four-team network, in twodirected dyads, both seeker and source had a team fo-cus; in twodirecteddyads,bothseekerandsourcehadasystem focus; in four directed dyads, the seeker wasteam focused while the source was system focused;and, in four directed dyads, the seeker was system fo-cused while the source was team focused.

Forourmanipulation,wecombinedmultipleelementsused in prior experimental research aiming to instill asense of shared identity with and attachment to a group(Cuijpers et al., 2016; De Cremer, van Knippenberg, vanDijke, & Bos, 2006; Eckel & Grossman, 2005; Gaertner,Mann,Murrell,&Dovidio,1989;Kaneetal.,2005;Kramer& Brewer, 1984). These elements included (a) video vi-gnettes emphasizing common fate with and emotionalattachment to the team or to the multiteam system; (b) abanner-making exercise, in which participants created abanner and a slogan for their team or for the multiteamsystem; and (c) symbols of intergroup comparison withother teams or with other multiteam systems.

Video vignettes. In the video vignettes, partici-pants were introduced to a background story abouttheir engagement, the emphasis of which differeddepending on the condition. In the system-focuscondition, the videos focused on the values andhistory of the greater region and participants’ sharedhistory of collaboration with the community in the re-gion. The videos emphasized to these participants thenotions of commitment, solidarity, and a sense of unitywith the “platoon” (i.e., the multiteam system) and


stressed that these shared experiences and achieve-ments distinguished their platoon from other platoonsoperating in other regions (i.e., other hypothetical mul-titeam systems). In the team-focus condition, partici-pants were shown an identical background story, butwith the difference being that it revolved around theirdistrict, emphasizing a sense of unity with the “squad”(i.e., their team) and contrasting this with other squads.

Banner-making exercise. Prior to the practicemission, participants designed a banner and a sloganfor their squad or their platoon using a virtual white-board app that allowed collaborative drawing andchatting using individual tablets. Participants inthe team-focus condition were only connected withmembers of their own team and designed a banner anda slogan for their ownsquad.Participants in the system-focus condition were also connected with members ofthe other system-focused team and designed a bannerand a slogan for the entire platoon. In order to sustainthe illusion that they were, in fact, connected withmembers of all teams rather than just one additionalteam,wesetupanonymousnumbersaschatnames.Thebanners continued to be displayed on large screens intheir respective rooms for the duration of bothmissions.

Symbols of intergroup comparison. To furtherreinforce a sense of distinctiveness of the team or themultiteam system, we placed a large poster in eachroom that displayed a fictitious ranking of the threebest-performing squads or platoons, depending onthe condition.

Measures

Manipulation checks. We conducted manipula-tion checks at three points in time. The first tookplace immediately after the instructions and ma-nipulation, prior to the practicemission. The secondcheck took place after the practicemission. The thirdcheck took place after the main mission. As manip-ulation checks, we asked participants to rank their“squad” (i.e., team), their “platoon” (i.e., multiteamsystem), and a fictitious superordinate “battalion”(that would include other platoons) in terms of howstrongly they identified with each. This measuredirectly captures the relative salience of the differentidentity foci. We then constructed an indicator vari-able that took the value of 1when participants rankedthe multiteam system more highly than their team(i.e., displaying a system focus) and 0 otherwise.

Interteam performance. In the simulation, infor-mational dependence arose from threats located in aseeking team’s district about which another source teamreceived information. We operationalized interteam

performance as the successful neutralization of suchthreats. That is, for each threat located in a seeking team’sdistrict and initially known to another source team, werecorded 1 if the seeking team successfully neutralized itand 0 otherwise.

Reactive information sharing. First, we identifiedall messages in the communication transcripts thatcontained an item of intelligence. We then coded allinstances in which the item was provided to the otherparticipant in response to an immediately precedingrequest.Asourmeasureof reactive information sharing,for each threat located in a seeking team’s district andinitially known to another source team,we recorded1 iftheinformationregarding this threathadbeenreactivelyprovided by the source team to the seeking team, and0 otherwise.

Proactive information sharing. We coded theremaining messages in which one participant pro-vided intelligence to another participant without animmediately preceding request as instances of proac-tive information sharing, and recorded1 for each threatabout which information was proactively provided bythe source team to the seeking team and 0 otherwise.1

Analytical Approach

Our set of observations consisted of 1,176 threatsnested in 252 directed dyads of source and seekingteams, which, in turn, were nested in 21 multiteamsystems. Because all of our dependent variables (proac-tive and reactive information sharing and interteamperformance) were binary variables, we estimated gen-eralized linear mixed models (with a probit link), in-cludingrandomeffects forseeker, source,andmultiteamsystem inorder to account for interdependencebetween

1 A small percentage of interdependent threats (5.8%)were neutralized even though we did not record thetransfer of the related information from source to seekingteam. A reexamination of the research logs suggested thatthis was primarily due to participants’ broadcasting in-formation through the status function of the software.Thus, a small part of the information was shared throughan unrecorded channel. A perusal of the communicationlogs suggested that this behavior emerged through imita-tion of other participants’ information-sharing behaviorrather than through explicitly coordinated requests to usethe status function in thismanner. Thus,we coded cases inwhich a record of the information transfer was missingdespite evidence of a transfer having taken place as pro-actively shared. Robustness checks in which we (a) codedproactive information sharing without this imputation or(b) treated these cases as missing observations yieldedidentical conclusions to the analyses reported below.


observations. We carried out these analyses using thelme4 package (Bates, Machler, Bolker, & Walker, 2015)in R (R Core Team, 2016).

Results

Manipulation checks. The results of three gener-alized linear mixedmodels accounting for nesting ofparticipants within teams and multiteam systemsshowed that participants in the system focus condi-tion were consistently more likely to report a systemfocus than participants in the team focus condition(Time 1: b 5 0.85, SE 5 0.28, p 5 .001; Time 2:b 5 0.38, SE5 0.23, p 5 .047; Time 3: b5 0.61, SE50.24, p 5 .005; all one-tailed2 tests of the directionalhypothesis that systemfocus [systemfocuscondition].system focus [team focus condition]).3 That is, our sys-tem focus manipulation successfully increased the rel-ative salience of themultiteam system identity vis-a-visthe team focus manipulation.4

Hypothesis tests. Table 1 presents the study’s de-scriptive statistics while Table 2 shows the results ofthe hypotheses tests. In the regressions, we usedcontrast coding for the identity focus conditions(system focus 5 10.5, team focus 5 20.5), as thisallows for the straightforward interpretation of theregressionparameters asmaineffects and interactionrather than conditional effects.

Model 1 tests the effects of seeker and source fociof identification on proactive information sharing.We found a statistically significant interaction be-tween seeker and source teams’ identity focus. To testHypothesis 1a, we computed a linear contrast betweentwo cells: team-focused seeker & team-focused sourcevs. system-focused seeker & team-focused source. Aspredicted, team-focused sources shared less informa-tion proactively with system-focused seekers than theydidwith team-focused seekers (b5 0.33,SE5 0.13,p5.015, one-tailed). In Model 2, we examined the effect ofseeker and source foci of identification on reactive in-formation sharing. In contrast to Hypothesis 1b, therewas no interaction between seeker and source foci ofidentification. We therefore did not proceed to probethe linear contrast.

Models 3and4examineHypotheses2 to5,whicharerelated to interteam performance. Consistent with Hy-potheses 2a and 2b, Model 4 shows a positive effect ofboth proactive (b 5 2.52, SE 5 0.16) and reactive (b 51.80, SE5 0.20) information sharing on interteam per-formance. To test Hypothesis 3 regarding the relativeimpact of proactive versus reactive information sharingon interteam performance, we tested the equality of thetwo regression coefficients through a linear hypothesistest. In linewithHypothesis3,wefoundthat theeffectofproactive sharing on interteam performance was sig-nificantlystronger thantheeffectof reactive informationsharingon interteamperformance (x25 15.29,p, .001,one-tailed).

Hypotheses 4a and4bposited an indirect effect of theinteractionbetweenseeker andsource foci on interteamperformance, mediated by proactive and reactive in-formation sharing, respectively. In Model 3, we find asignificant total effect of the interaction between seekerand source foci of identification on interteam perfor-mance. To test themediation hypotheses, we estimateda path model in Mplus (Muthen & Muthen, 2017), astatistical software capable of estimating and testingindirect effects in complex multilevel data. We useda cross-classified probit model accounting for theclustering of observations in seeker and source teamssimultaneously.5 By default, Mplus uses Bayesian esti-mation for cross-classifiedmodels. Figure1presents the

2 Throughout ourmanuscript, we use one-tailed tests fordirectional hypotheses. This is consistent with recom-mendations put forward in earlier researchnoting that one-tailed tests provide amore precise, logical correspondencebetween a directional research hypothesis and its statisti-cal test (Cho & Abe, 2013; Gravetter & Wallnau, 2017;Schwab, 2005). Conversely, wherever we did not have apriori directional hypotheses—for example, in supple-mentary analyses—we used two-tailed tests.

3 We note that the secondmanipulation check showed aconsiderably smaller effect size than the first and the third.One explanation for this is that the second manipulationcheck took place after the practice mission during whichmastering the interface and experimenting with initialstrategies took the forefront over the mission and theidentity-relevant background story and context. Antici-pating that this could weaken the manipulation in the ab-sence of additional reinforcement, we had included therefresher video preceding the main mission, and, indeed,at Time 3, we again observed a stronger effect of themanipulation.

4 In supplementary analyses not reported here, we alsoexamined whether our manipulation affected percep-tions of task interdependence, goal interdependence, andinterteam competition. We found no significant differ-ences between conditions on any of these other variables.

5 In contrast to our main analyses, we could not accountfor clustering in multiteam systems simultaneously withthe cross-classified affiliation with seekers and sources inthis software. However, supplementary analyses not re-ported here indicated that the conclusions of our mainmodels presented in Table 2were robust to the omission ofthe multiteam system clustering variable, and we have noreason to expect any different in the path model.


results of the pathmodel, replicating our prior analyses.The estimate for the indirect effect of the interactionbetween seeker and source identity foci on interteamperformance via proactive information sharing was0.74, its 95%Bayesian credibility interval [0.28, 1.37]not including zero, thus supporting Hypothesis 4a.The estimate for the indirect effect of the interactionbetween seeker and source identity foci on interteamperformance via reactive information sharing, inturn, was 20.18, with its 95% Bayesian credibility in-terval [20.77, 0.35] including zero, thus not supportingHypothesis4b.To test thedirectedHypothesis5 that theindirect effect via proactive information sharing would

be stronger than the indirect effect via reactive infor-mation sharing,wecomputed90%Bayesian credibilityintervals around both indirect effects and examinedtheir overlap in the expected direction. The 90% cred-ibility interval around the indirect effect via proactivesharing [0.35, 1.25] did not overlap with the 90% cred-ibility interval around the indirect effect via reactiveinformation sharing [20.66, 0.25], thus supportingHypothesis 5.

Finally, to test Hypothesis 6, we computed threelinear contrasts comparing the seeker system focus–source system focus configuration with each othercombination of conditions based on the total effects

TABLE 2Results of Generalized Linear Mixed Models (Study 1)

Model 1 Model 2 Model 3 Model 4 Model 5

Proactiveinformation

sharing

Reactiveinformation

sharingInterteam

performanceInterteam

performance

Interteamperformance (shared

items only)

Random effects Variance Variance Variance Variance VarianceSeeker 0 0.69 0.09 0.26 0.30Source 0.33 0.14 0.11 0.05 0.06Multiteam system 0.04 0.31 0.06 0.03 0.02

Fixed effectsIntercept 0.69 (0.09)** 21.91 (0.22) 0.14 (0.08) 21.95 (0.16)** 0.72 (0.09)Seeker focus 20.06 (0.09) 20.15 (0.25) 0.10 (0.10) 0.24 (0.15) 0.28 (0.16)†

Source focus 0.12 (0.16) 0.28 (0.17)† 0.18 (0.11) 0.11 (0.11) 0.13 (0.12)Seeker focus3 Source focus 0.53 (0.18)** 20.28 (0.27) 0.38 (0.16)* 0.24 (0.20) 0.30 (0.21)Proactive information sharing 2.52 (0.16)**Reactive information sharing 1.80 (0.20)**

Log-likelihood 2658.00 2345.15 2777.69 2543.90 2503.80n 1176 1176 1176 1176 893

Notes: Conditions are contrast coded:20.55 teamfocus,10.55 systemfocus. Standarderrors inparentheses.Tests areone-tailed for tests ofdirectional hypotheses (effects of proactive and reactive information sharing), and two-tailed for all other coefficients.

†p , .10*p , .05

**p , .01

TABLE 1Descriptive Statistics (Study 1)

Variable

Source focus: Team Source focus: System Correlations

Seeker focus:Team

Seeker focus:System

Seeker focus:Team

Seeker focus:System 2. 3.

1. Proactive information sharing 0.75 [3.4] 0.66 [2.8] 0.70 [3.5] 0.77 [3.9]2. Reactive information sharing 0.09 [0.4] 0.10 [0.4] 0.14 [0.7] 0.11 [0.5] 2.083. Interteam performance 0.54 [2.4] 0.51 [2.2] 0.53 [2.7] 0.63 [3.2] .41** .18**

Notes: Values indicate the proportion of shared or neutralized threats in each seeker–source condition. Values in square brackets indicateaverage number of items shared or neutralized in each seeker–source condition. n5 1,176 threats in 252 dyads. Correlations were calculatedon the level of the dyad (i.e., setting in relation the proportion of shared or neutralized threats in each dyad).

** p , .01, two-tailed


presented in Model 3. Consistent with Hypothesis 6,interteam performance was higher when both teamshad a system focus thanwhen either seeker (b5 0.37,SE5 0.14, p5 .012), source (b5 0.30, SE5 0.13, p5

.023), or both had a team focus (b 5 0.28, SE 5 0.17,p 5 .046; all three comparisons were one-tailed testsusing the Holm, 1979, correction for multiple com-parisons). A single linear contrast comparing the

FIGURE 1Interaction of Seeker and Source Identity Focus on Information Sharing and Interteam Performance (Study 1)

Proactive Information Sharing

0.8

0.6

0.4

0.2

0.0

0.8

0.6

0.4

0.2

0.0

0.8

0.6

0.4

0.2

0.0

Team System

Seeker focus

Team System

Seeker focus

Team

Team System

System

Seeker focus

Source focus

Reactive Information Sharing Interteam Performance

Notes: Bars represent predicted values based on the estimated models (Models 1, 2, and 3 in Table 2, respectively). Lines represent 95%confidence intervals of the predicted values.

FIGURE 2Path Model Results (Study 1)

Seeking team’ssystem focus

Source team’ssystem focus

Seeking × Sourceteams’ foci

0.56 –0.19

0.09

–0.14

0.27

–0.20

–0.08

ProactiveInformation Sharing

ReactiveInformation Sharing

0.15

1.01

1.35

InterteamPerformance

Notes: Bold paths and coefficients indicate that the 95% Bayesian credibility interval around the parameter did not contain zero.


system–system configuration with all other config-urations combined further yielded consistent evi-dence for higher performance of the system–systemconfiguration as comparedwith all other configurations(b 5 0.31, SE 5 0.12, p 5 .005, one-tailed). Figure 2presents the predicted means per condition based onthe fitted models.

Supplementary analyses. To gain further insightinto the role of social identity for information sharing,we conducted several supplementary analyses. First, asa check of one of the assumptions underlying Hypoth-esis 5—namely, that system-focused teams are morelikely to utilize externally obtained information (Dokkoet al., 2014; Kane et al., 2005)—we estimated the effectsof our identity foci conditions on interteam perfor-mance, conditional on information having been shared(Model 5 inTable 2). Consistentwith prior research,wefound that system-focused teamsweremarginallymorelikely to proceed to neutralize the threats whose loca-tion they obtained from other teams than were team-focused teams (b5 0.28,SE5 0.16,p5 .08, two-tailed).

In a second supplementary analysis, we sought tounderstand to what extent a team’s primary identityfocus affected their intrateam performance—that is,the successful completion of tasks in which they didnot depend on other teams. The argument here couldbe that system-focused teams’ indiscriminate invest-ment in interteam cooperation may reduce perfor-mance on intrateam tasks—for instance, as a result ofdepletion (Porck et al., 2019). In our study, intrateamperformancewas operationalized as the proportion ofthose threats about which information was given tothe team that needed it from the start (N5 399) thatwassuccessfully neutralized. Team-focused teams neutral-ized 62.38% of the threats initially known to them,while system-focused teams neutralized 65.08%. Weestimated a generalized linear mixed model predictingneutralization of a threat as a function of six conditions:intrateam knowledge of the threat in combination withteam focus of the focal team, intrateam knowledge incombination with system focus of the focal team, andthe four conditions capturing interteam knowledge to-gether with the four different combinations of seekerand source foci. In contrast analyses, we found thatperformance did not differ between the two intrateamconditions (b520.07, SE5 0.17, p5 .97). Combiningboth intrateam conditions on the one hand, and allfour interteam conditions on the other hand, we foundthat—as could be expected—performance was higherin intrateam conditions than in interteam conditions(b5 0.25,SE5 0.08,p5 .01). Finally,wedifferentiatedbetween the interteam condition in which seeker andsource had system focus and the three remaining

interteam conditions. We found no difference be-tween the combined intrateam conditions and theinterteam condition inwhich both seeker and sourcehad system focus (b 5 20.01, SE 5 0.13, p 5 1.0),while the contrast between the two intrateam con-ditions and the remaining three interteamconditionswas significant (b 5 0.32, SE 5 0.08, p , .001; alltwo-tailed tests using the Holm, 1979, correction formultiple comparisons). In sum, system focus didnot constrain intrateam performance. Furthermore,while interteam collaboration was more challengingthan intrateam collaboration for most team dyads,those team dyads in which both partners had a sys-tem focus collaborated as effectively as if there hadbeen no team boundary between them.

STUDY 2: IDENTITY FOCI AND MOTIVESTOWARD INTERTEAM INTERACTION

The theory underlying our key hypothesis abouthow identity focus affects proactive informationsharing is based on the assumption that identity fo-cus changes the way in which team members ap-proach collaborationwith other teams.Weargue thatteam-focused teams approach interteam collabora-tion based on instrumental motives—cooperation asa means to an end and conditional on its instru-mental value. Conversely, we argue that system-focused teams approach interteam collaborationbased on benevolent motives—cooperation as an endin itself and not conditional on its instrumental value.It is this mechanism, we argue, that underlies differ-ences in information-sharing behavior between team-and system-focused teams: because team-focusedteams approach interteam cooperation more as ameans to an end, theywill orient their information-sharing behavior more strongly on direct reciprocityconsiderations.

Although the behavioral differences we observed inStudy 1 support these theoretical arguments, the studydid not directly test this underlying assumption. To fillthis gapand test ourworkingassumption,wedesignedascenario experiment inspired by the experimental taskin Study 1. In this scenario, participants took the role ofan intelligence officer operating as part of a multiteamsystem securing a city. We manipulated identity focusand then measured the impact of the manipulation onparticipants’ conceptualization of interteam relations asmoreor less instrumental, and their information-sharingintentions (see also Appendix A).


Sample

We recruited 308 participants in the UK and USAthrough the online research platform Prolific Aca-demic. Participants were paid £0.70 for a seven-minute study. Because prior research has raisedconcerns about response quality in online research,we included two comprehension checks and oneinstructional manipulation check at different pointsin our study (Berinsky, Margolis, & Sances, 2014;Fleischer, Mead, & Huang, 2015). Forty-six partici-pants failed either both comprehension checks or theinstructional manipulation check. Four other par-ticipants provided responses not conforming to therules set out by the scenario. We excluded thesefrom the analyses. The resulting sample contained258 individuals (134 men, 122 women, and twoindividualswhodidnot self-identify). Participationwas restricted to individuals reporting full-timeemployment. Participants were between 18 and 62years old (M 5 34.40, SD 5 9.70).

Procedure

Participants read a scenario in which their taskwas described as gathering intelligence about po-tential threats to a city and redirecting this infor-mation to field specialists within the task force. Thefull scenario and the measures are reproduced inAppendixA. The citywas described as consisting offive districts with a different component team op-erating in each district. Each participant was toldthat they were part of Team Center operating in theCenter District, but they could encounter intelli-gence about threats in any district. Following thegeneral introduction to the situation, participantsread the identity manipulation, which we adaptedfrom prior research (De Cremer et al., 2006). Next,participants read the information-sharing scenarioin which they were told that they had obtained twopieces of information that were relevant to field spe-cialists in two different teams, North and South. Theyfurthermore learned that, in thenear future,TeamNorthwas very likely to obtain information relevant to TeamCenter (i.e., the participant’s team) while Team Southwould most likely not obtain any information relevantto Team Center. Thus, from an instrumentality point ofview, Team North appears as a more relevant target forinformationsharing thanTeamSouth,assecuringTeamNorth’s future reciprocal cooperation is more valuablefor Team Center’s own performance. From a benevo-lence point of view, on the other hand, there is no suchdifference, as both teams’ performances equally

contribute to thesharedwelfareof thetaskforce.Finally,participants responded to a questionnaire containingthe measures of the dependent variables.

Measures

Manipulation check. As in Study 1, we measuredparticipants’primary identity focusdirectly byaskingthem to rank the team and the multiteam system interms of how strongly they identified with each.

Reciprocity-oriented information sharing. Afterreading the information-sharing scenario, partici-pantswere asked to decide how to allocate their timebetweenpreparingmemos for both teams. Theyweretold that the higher percentage of time allocated to amemo, themore useful it would be to the other team.Given that Team North was presented as havingmore relevant information to offer to Team Center inthe near future, higher time allocation to TeamNorthat the expense of Team South could be interpretedas favoring reciprocity-oriented information sharing,and was our dependent variable.

Instrumentality motive in interteam interactions.To measure the extent to which participants per-ceived collaboration with other teams as a means to anend, we adapted six items of Gruenfeld, Inesi, Magee,and Galinsky’s (2008) objectification scale. A sampleitem is “The main reason why relationships with otherteamswouldbeimportant tomeisbecause theyhelpmeaccomplish my team’s goals.” Participants respondedthrough reference toa5-pointLikert scale (15 “stronglydisagree” to 5 5 “strongly agree”). The adapted scaleshowed acceptable internal consistency (a 5 .72).

Results

Manipulation check.As intended, participants inthe system-focus condition were found to be morelikely to report the system as their primary identityfocus (78.4%) than participants in the team-focuscondition (16.1%, t5 12.77, p5, .001, one-tailed).

Main results. Table 3 provides a summary of themainresults.Asexpected,participants in the team-focuscondition reported higher instrumentality of interteamrelationships (M5 3.40, SD5 0.66) than participants inthe system-focus condition (M 5 2.85, SD 5 0.69; t 56.63, p , .001, one-tailed). Furthermore, participantsin the team-focused condition showed higher levelsof reciprocity-oriented information sharing (M 558.27, SD 5 19.23) than participants in the system-focused condition (M 5 53.54, SD 5 19.99, t 5 1.93,p 5 .03, one-tailed). Finally, we conducted a media-tion analysis using Hayes’s PROCESS routine (Hayes,


2013). As predicted, instrumentality of interteam in-teractions mediated the effect of identity focus onreciprocity-oriented information sharing (b 5 22.48,with the 95%confidence interval of the indirect effect[24.63,20.58] not including 0).

STUDY 3: IDENTITY CONFIGURATIONSIN MULTITEAM SYSTEMS

Thus far, we have focused our investigation on theteam-dyadic level, arguing that identity asymmetrieswill disrupt implicit coordination between teamsand harm interteam performance. The results ofStudy 1 corroborate our line of reasoning, showingimpaired information sharing and lower interteamperformance in teamdyads consisting of a team-focusedsourceandasystem-focusedseeker.This insight, in turn,allows us to consider the effect of social identity onsystem-level coordination and performance in a moreprecise manner than prior research by considering theimplications of different identity configurations of mul-titeam systems.

A multiteam system’s “identity configuration” is thecomposition of the system in terms of its teams’primaryfoci of identification. It can be captured, for instance, asthe proportion of component teams whose primaryidentity focus is the multiteam system. Most prior re-search on the role of social identity in intergroup col-laboration broadly suggests that identification with theoverarching collective would have a (linearly) positiverelationship with collective performance, as it leads tomore (Lomi et al., 2014) and more effective (Cuijperset al., 2016; Dokko et al., 2014; Richter et al., 2006) in-teractions at the teamboundaries.That is, basedonpriorwork, we should expect that the larger the proportion ofsystem-focusedteamsinthesystemasawhole, thebetterthis system should perform. However, if—as we foundabove—thebenefit ofmultiteamsystem identificationofa component team is conditional on the identity focusof

the team it interacts with, then we may need to qualifythis claim: if identity asymmetries disrupt dyadic coor-dination and performance, then configurations with ahigher number of asymmetric dyads bear a disadvant-age that can counteract the positive effect of higher sys-tem focus in a system.

To gain a better understanding of these interactions,we conducted a third study in which we extrapolatedfrom our empirical results on the team-dyadic levelto develop propositions about team- and system-levelcoordination and performance by means of computa-tional simulation. The simulationmethod is particularlyuseful tounderstandtheimplicationsofdifferent identityconfigurations in multiteam systems as it enabled us toconduct virtual experiments manipulating the propor-tionof team-andsystem-focusedteamsinalargenumberof simulated multiteam systems. Thus, we were able togaininsightsnoteasilyobtainableinthelaborinthefield.

Simulation Procedure

In order to extrapolate from our results to the impli-cations of different identity configurations inmultiteamsystems, we used the expected values obtained in ourempirical models in Study 1 in a computational simu-lation mimicking multiteam systems engaged in a simi-lar task. That is, we simulated multiteam systems inwhich teams have tasks (e.g., neutralize threats) for thecompletion of which they require information fromother teams.

System setup. First, we generated synthetic mul-titeamsystems. In keepingwithStudy1,wemodeledfour-teammultiteamsystems.GoingbeyondStudy1,in Study 3, we varied identity configurations to createfive multiteam system configurations: 4T:0S, 3T:1S, 2T:2S, 1T:3S, and0T:4S,wherein the firstnumber indicatesthe number of team-focused and the second numberindicates the number of system-focused teams in eachmultiteam system. We generated 10,000 systems for

TABLE 3Results of Mediation Analysis (Study 2)

Instrumentality motiveReciprocity-orientedinformation sharing

Reciprocity-orientedinformation sharing

Regression models B SE t p B SE t p B SE t p

Intercept 3.40 0.06 55.94 , .01 58.27 1.76 33.06 , .01 42.95 6.34 6.77 , .01System identity focus 20.55 0.08 26.53 , .01 24.73 2.45 21.93 .03 22.25 2.61 20.86 .39Instrumentality motive 4.50 1.79 2.51 .01R2 0.38 0.01 0.04

Indirect effect Effect SE LCI UCIIdentity focus via instrumentality motive 22.48 1.05 24.63 20.58

Notes: All t tests are one-tailed. LCI and HCI5 lower and higher bounds, respectively, of 95% bootstrapped confidence interval.


each configuration. Next, in each system, we generated100 items of information and “distributed” these amongthe teams by randomly assigning a seeker (i.e., the teamthat needs this item) and a source (i.e., the team thatoriginally has this item) to each item.

Information sharing. In our simulation, each itemof information has the potential to be shared proac-tively and the potential to be shared reactively withthe seeking team. We assumed that the probabilityof an item being shared in either manner was de-pendent on the combination of seeker and sourceidentity foci. Each item was recorded as sharedproactively with an item-specific probability ppsi

and each itemwas recordedas shared reactivelywithan item-specific probability prsi. The probabilitieswere drawn from the distributions of expectedvalues generated byModel 1 (proactive sharing) andModel 2 (reactive sharing) for the correspondingcombinations of seeker and source identity foci. Wethen combined these two events in a single recordindicatingwhether or not an itemhadbeen shared bythe source with the seeker. Finally, we calculatedwhatproportion of information relevant to each teamwas actually obtained by that team (constituting a team-level outcome) and we calculated what proportion ofall information that could have been shared actuallywas shared (constituting a system-level outcome).

Performance. Next, each item that had been suc-cessfully shared had an opportunity to be neutralized—that is, the corresponding task could be completed.Amongthe itemsthathadbeenshared,werecordedeachitem as successfully neutralized with an item-specificprobability pni, which we draw from the distribution ofexpected values generated by Model 5 for the corre-sponding combinations of seeker and source identityfoci. We used Model 5 rather than Model 3 because itprovided us with expected values conditional on infor-mation having been shared, whichwas a better fit to thesequential nature of the simulation. We then calculatedwhat proportion of threats that could have been neu-tralized by each team were actually neutralized by thatteam, as a measure of team-level performance. For sys-tem performance, we made the simplifying assumptionthat each successfully completed task on the team levelequally andpositively contributed to the achievement ofthe system-level goal. Based on this assumption, wecomputed system-level performance as the total pro-portionof threats thatweresuccessfullyneutralized.Thisassumption is a simplification of reality, given that team-level goals may have different levels of compatibilitywith the system-level goal and with each other (Ricoet al., 2017). At a basic level, however, the assumptionthat completing team goals contributes to goals at the

higher level of the goal hierarchy is engrained in thedefinition ofmultiteam systems (Mathieu et al., 2001). Inaddition,whilegoalcompatibilitymayvaryinmultiteamsystems in the field, in our experiments, we held thisfactor constant. As our simulation was built on our em-piricaldata,wedeemedmaking this sameassumption inthe simulation reasonable.

Results

Figure 3 presents the results of the simulations asthe average proportions of items having been sharedand neutralized. The results can be interpreted asprecise point estimates, as standard errors convergeto 0 with sufficient simulation runs. Several insightsemerge from these analyses. First, we consider team-level outcomes (Panels A and B in Figure 3). Panel Ashows that the amount of information obtained by asystem-focused team depends on the identity config-uration of the system: a system-focused team sur-rounded by team-focused teams (3T:1S) obtains about10%less information thandoes a system-focused teamsurrounded by other system-focused teams (0T:4S). Incomparison, a team-focused team receives only 3%less information when being the only team-focusedteam (1T:3S), compared to being surrounded exclu-sively by other team-focused teams (4T:0S). Panel Bpresents a similar picture with regard to team perfor-mance: with each shift toward system focus in thesystem, a system-focused team succeeds in neutral-izing an additional 4.3% of its threats—resulting in a12.9% difference between the extreme scenarios—while the performance of a team-focused team ishardly affected by the system’s identity configuration(2% difference between the extreme scenarios).

These results indicate that team-level informationretrieval and performance are, to a considerable ex-tent, more dependent on the system-level identityconfiguration for system-focused teams than they arefor team-focused teams. That is,while a system-focusedteam can bemore successful on interdependent tasksthan a team-focused team, the composition of the restof the system in terms of identity focus is a criticalboundarycondition for thispositiveeffect. Ina systemthat predominantly consists of team-focused teams,on the otherhand, a system focusmay even turn into adisadvantage. More formally:

Proposition 1. The effect of system focus on a focalteam’s performance on interdependent tasks is mod-erated by the identity configuration of the multiteamsystem in which it is embedded.


Second, let us consider the system-level results(Panels C and D in Figure 3). The simulation resultssuggest that the relationships between an increasingproportion of system-focused teams and system-level coordination (i.e., information sharing) and

performance are convex rather than linear. This iseasily explained by the fact that the proportion ofasymmetric team-dyads in a system is higher, thecloser a system’s identity configuration is to 50:50. Thespecific shape of the function depends on the outcome

FIGURE 3Expected Effects of Different Multiteam System Identity Configurations on Multiteam System Information

Sharing and Performance (Study 3)

Team information retrieval Team performance

MTS information sharing MTS performance

0.8

0.7

0.6

0.5

0.8

0.7

0.6

0.5

0.8

0.7

0.6

0.5

0.8

0.7

0.6

0.5

Identity Configuration Identity Configuration

Identity ConfigurationIdentity Configuration

Average received by system-focused teamsAverage received by team-focused teams

Average neutralized by system-focused teamsAverage neutralized by team-focused teams

Exp

ecte

d p

rop

orti

on o

f sh

ared

ite

ms

Exp

ecte

d p

rop

orti

on o

f sh

ared

ite

ms

Exp

ecte

d p

rop

orti

on o

f n

eutr

aliz

ed i

tem

sE

xpec

ted

pro

por

tion

of

neu

tral

ized

ite

ms

Total shared Total neutralized

A B

C D

Notes: Squares represent system-focused teams, triangles represent team-focused teams. MTS 5multiteam system.


in question. For information sharing, our results sug-gest a U-shaped curve without a positive linear trend(R2

linear 5 0.04; R2quadratic 5 1.00). That is, compared

to a four-team multiteam system in which all com-ponent teams primarily identify with their team,shifting the identity focus of one or of two teams to themultiteam system can be counterproductive for in-formation sharing. Past the threshold of 50%, in-creasing theproportionof system-focusedcomponentteams is beneficial, however. For system-level per-formance, our results suggest aU-shaped curvewith apositive linear trend (R2

linear 5 0.70; R2quadratic 5

1.00). More specifically, while increasing theproportion of system-focused teams hardly affectsmultiteamsystemperformanceup to the thresholdof50%, beyond this threshold, the proportion of sys-tem focus has an increasingly positive effect. Viewedfrom the opposite direction, it is the smallest devi-ation from a 100% system-focus configuration thatis associated with the largest drop in multiteamsystem performance. Based on this, we put forwarda final proposition:

Proposition 2. There is a convex relationship betweenthe proportion of system-focused teams in a multi-team system and system-level information sharingand performance.

DISCUSSION

Multiteam systems tackle many complex orga-nizational tasks, in settings as varied as scien-tific innovation, new product development, healthcare, themilitary, and space exploration. In each ofthese settings, there is an increasing realizationthat success hinges on both “intra-” and “inter-”team processes. While sharing unique informationis a challenge evenwithin a team (Mesmer-Magnus& DeChurch, 2009), the “us-versus-them” socialcategorizations prevalent in “teams of teams” fur-ther compound the challenges of sharing uniqueinformation across teams. In the present work, weconducted three studies examining how the com-position of multiteam systems in terms of compo-nent teams’ primary foci of identification affectsinformation sharing and performance. Our find-ings provide causal evidence for the role of socialidentity in these processes, and highlight the dis-ruptive role of identity asymmetries—arisingwhencomponent teams differ inwhat they consider to betheir primary group.

Theoretical Implications

Multiteam system composition. Our study high-lights the importance of considering both sides of therelationship when considering interteam collabora-tion. With few exceptions (e.g., Bresman, 2013), thebroader research on team boundary spanning orinterteam coordination and collaboration takes theperspective of one focal team and examines the in-fluence of individual, team, or contextual factors onthis team’s interaction with external constituencies(e.g., deVries et al., 2014; Joshi, Pandey, &Han, 2009;Marrone, 2010; Marrone, Tesluk, & Carson, 2007;Richter et al., 2006). Yet, such interaction is of afundamentally dyadic nature—collaboration cannothappen if the other side does not cooperate. Thus,compositional factors of both teams as well as (asevident from our results) the interaction betweenthese variables on seeker and on source side play animportant role in shaping intergroup collaboration. Thisnotion, while often absent in the broader boundaryspanning and interteam collaboration literature, is nat-urally embedded in the multiteam systems literatureand, especially, in work onmultiteam system composi-tion (Lanaj et al., 2018; Luciano et al., 2018; Shuffler,Jimenez-Rodrıguez, & Kramer, 2015). Luciano andcolleagues (2018), for instance, discussed several com-positional factors that induce differentiation betweencomponent teams, such as goals, competencies, norms,workprocesses, and information, suggesting that greaterlevels of differentiation will result in processes that un-dermine collaborative interactions between teams. Weaddtwoimportantnuances to thisclaim: first,within thesame system, some teams may perceive componentteams as more differentiated than others; second, suchasymmetries have implications for interteam informa-tion sharing and performance.

It is tempting to interpret our results in homophilyterms: teams share more information when theirprimary focusof identity coincides thanwhen it doesnot. However, homophily would imply that the coor-dination breakdownwould affect both teams in a dyadina symmetric fashion: if itwas amatter of homophily,weshouldsee that system-focused teamswouldbe lesslikely to share information with team-focused teamsjust as team-focused teams are less likely to share in-formationwith system-focused teams. Conversely, ourfirst study found a clear difference between the be-havior of team-focused teams toward system-focusedteams, on the one hand, and the behavior of system-focused teams toward team-focused teams on theother.


In sum, this implies that the specific configurationof differences and similarities may be as important asthe overall level of differentiation to multiteam systemfunctioning. Just as research on team composition andprocesses increasingly adopts configural perspectivesrather thanmain-effects approaches (Crawford&LePine,2013;Humphrey&Aime,2014;vanKnippenberg&Mell,2016), we can achieve a deeper understanding of multi-team system functioning by considering how differentconfigurations of team attributes, processes, and emer-gent states result in different patterns of interteam inter-action and, consequently, influence system outcomes.

Social identity theory.Beyond the contribution tothemultiteamsystems literature, thiswork also feedsback to more fundamental social identity theory. Inthis paper, we break new ground by investigating theimplications of identity asymmetries for dyadic effec-tiveness. When it comes to the role of social identity ininterteam coordination and performance, the broadconsensus in the literature seems to be that a strongidentification with the overarching collective—the mul-titeam system or, in other contexts, the organization—isgenerally desirable (Cuijpers et al., 2016; de Vries et al.,2014; Dokko et al., 2014; Kane, 2010; Lomi et al., 2014;Richter et al., 2006—but cf. Porck et al., 2019). Whileour findings support the main corollary of thisproposition—that a system will be effective whenall its component teams have a system focus—ourproposition of a U-shaped relationship between thenumber of system-focused teams and system-level in-formation sharing and performance challenges simplis-tic assumptions.

The key proposition of this paper is that identityasymmetries have an influence on intergroup col-laboration. While our study focused on identityasymmetries between component teams within amultiteam system—and thus “intergroup” in ourcontext translates into interteam—arguably, similararguments may be made at other levels of analysis.For instance, within a team, individuals have mul-tiple foci of identity as they are simultaneously teammembers and representatives of demographic or pro-fessional social groups. Within-team differences in de-mographic or professional categories can be a strongfoundation for the formation of subgroups—and thusmay represent a situation in which intergroup relationsmust be managed within a team (Carton & Cummings,2012). Importantly, research on team diversity and re-lational demography has demonstrated that such dif-ferences can also result in identity asymmetries withinteams. For example, an individual’s dissimilarity toother members of the team can have different effectson the extent to which they identify with the team or

with the other social categories they belong to, depend-ing on status asymmetries (Chattopadhyay, George, &Ng, 2011; Chattopadhyay et al., 2008; Chattopadhyay,Tluchowska, & George, 2004). As another example,an individual’s perception of diversity in their teammayhavedifferent effects on the extent towhich theywill identify with their team, depending on whetherthey see a positive value in diversity (van Dick, vanKnippenberg, Hagele, Guillaume, & Brodbeck, 2008;van Knippenberg, Haslam, & Platow, 2007). Whilethese lines of research explain the existence ofidentity asymmetries also within teams, they do nottypically address the consequences of such asym-metries for dyadic interaction—that is, they do notexamine how the fact that two teammembers differ intheir identification with their team influences theircollaboration. Arguments we develop in the presentwork may contribute to future research on identityasymmetries across different levels of analysis.

In this study, we concentrated on the repercus-sions of differences in relative salience of the teamand the system as a focus of identification of multi-team systemmembers, and we were largely agnosticto differences in team and system identificationin absolute terms. While this binary distinction issuitable for a first investigation of identity asymme-tries in multiteam systems, undoubtedly we mayobtain a more differentiated understanding of iden-tity asymmetries by also considering similarities anddifferences in absolute levels of team and systemidentification on seeker and on source sides. In par-ticular, prior research has highlighted additive aswell as interactive effects of absolute proximate andoverarching identification on groups’ interactionwith other groups (van Dick, van Knippenberg,Kerschreiter, et al., 2008) and put an emphasis onthe role of dual identification—situations in whichindividuals have high absolute identification bothwith the component team and with the overarchingsystem (Brewer & Brown, 1998; Cuijpers et al., 2016;Hornsey&Hogg, 2000; Pettigrew, 1998; Richter et al.,2006). It is important to note here that neither ourconceptualization nor our operationalization of sys-tem focus imply that this is necessarily a situation ofhigh absolute system identification and low absoluteteam identification (and vice versa for team focus).Rather, system focus means that, at the margin, indi-viduals perceive the system rather than the teamboundary as the primary boundary. This can happenwhen system identification is high and team identi-fication is low—but this can also happen when bothsystem and team identification are high. Indeed, twopieces of meta-analytic evidence suggest that the


latter is amore likely occurrence underlying a systemfocus than the former. First, the levels of identificationwith the proximate and with the overarching groupstend to be highly correlated (Mesmer-Magnus et al.,2018). Second, where divergence does exist, teamidentification is typically higher than system identi-fication, because, for instance, the smaller size of theteam relative to the system allows for more intenseinteraction and results in greater familiarity (Riketta& van Dick, 2005). Thus, while this study does notdirectly speak to the dual identity hypothesis, itsfindings are not at odds with it.

Implicit and explicit coordination. Our findingsfurthermore contribute to a better understanding ofthe interplay between explicit and implicit coordi-nation in complex social systems. Theory on teamcoordination suggests that,while teams typically usea mix of explicit and implicit forms of coordination(Espinosa et al., 2004), there is some substitutabilitybetween explicit and implicit coordination, suchthat teams that can rely on implicit coordination to agreater extent engage in less explicit coordination(Rico & Sanchez-Manzanares, 2008). Equally, thisimplies that teams—or team dyads in our case—thatcannot rely on implicit coordination to the sameextent would compensate by increased explicit co-ordination. Yet, we did not find that teams whoobtained less information from other teams in an-ticipation of their needs compensated by obtainingmore information from those teams through makingtheir needs explicitly known. An explanation forthis may lie in the nature of information sharing asa coordination mechanism. For instance, work ontransactive memory systems—that is, team’s sharedcognitive systems for the division of cognitive labor(Hollingshead, 2001; Wegner, 1987)—suggests thatteam processes around sharing and retrieving infor-mation fromeachother benefit frommembershavingan understanding of who knows what (Mell, vanKnippenberg, & van Ginkel, 2014; van Ginkel & vanKnippenberg, 2009; Wegner, 1995). Expanding thisargument to interteam coordination within a multi-team system suggests that a seeking team that doesnot have an adequate representation of what infor-mation exists in the system and where it is locatedwould be less likely to attempt to retrieve it from theright source. This, in turn, implies that—under suchconditions, at least—the responsibility for ensuringthat information reaches the target in need of it pri-marily lieswith the source rather thanwith the seeker.Proactive, anticipatory interteam information sharingis key for multiteam system effectiveness.

Managerial Implications

In practice, this last insight finds exemplary applica-tion in policies developed by what we might call “in-formation professionals” in recent years. Following therecognition that information barriers between differentU.S. government agencies contributed to the failure toprevent the attacks on the World Trade Center (9/11Commission, 2004), the U.S. intelligence communityrevised its guidelines for interagency collaboration. Im-portantly, these guidelines include a shift from a“need to know” mindset, emphasizing access re-strictions, to a “responsibility to provide” mindset,emphasizing proactive information sharing withinthe community (Director of National Intelligence,2009). Adopting such guidelines, however, requiresa cultural shift in which collective identity plays akey role. In the example of the intelligence com-munity, information-sharing guidelines went handin hand with the establishment of superordinateentities charged with providing a focal point andsupporting coordination within the community(Best, 2011)—thereby increasing the salience of thesuperordinate community as a focus of identifica-tion. In other settings, such superordinate entitiesexist a priori—for example, the new product devel-opment team housing multiple interdependent sub-teams (Hoegl &Weinkauf, 2004)—and the question isone of managing identity in the multiteam system.

Our results have two implications with regard tothis question. First, our finding that identity asymme-tries can compromise interteam information sharingand performance suggests that organizations shouldpay attention to organizational arrangements that maymake such asymmetries particularly likely. These canbe situations in which some teams are more central inthe workflow than others, situations in which someteams are more physically or socially isolated fromthe rest of the system than others, or situations inwhich team leaders vary in their individual identityfoci and consequent rhetoric. Second, while our re-sults support the notion that interventions aimed atincreasing members’ identification with the systemcan improve system functioning, they highlight thatmanaging such a transition is not straightforward. Ourfinding that team-focused source teams withhold in-formation from system-focused seeking teams suggeststhat even having just one team-focused team on boardmaygodisproportionately far in spoiling theproverbialbarrel. Thus, an intervention aimed at shifting the pri-mary identity focus of only a part of the system holdslimited value. Furthermore, even if the intervention isaimed at the entire system but—perhaps for practical


reasons—it is staggered, such that some componentteams receive it later than others, the transition phaseitselfmaybe a source of vulnerability to the systemas itresults in temporary asymmetries, introducing the as-sociated coordination breakdowns. In sum, our studysuggests that, in order to be successful, interventionsaimed at shifting a system from team focus to systemfocus must be all encompassing and simultaneous.

Boundary Conditions

Our theory and results are subject to multipleboundary conditions. First, implicit in our theory isthe assumption that teams are reciprocally depen-dent on each other—in each dyad, both teams aresimultaneously seekers and sources and depend oneach other’s information to roughly the same extent.It is under these conditions that the instrumentalitymotive results in more reciprocity-oriented infor-mation sharing. If, on the other hand, dependence isasymmetric between teams—the extreme case beingone team depending on another team that does notdepend on the former—we may observe differentpatterns of interaction. The exact pattern would de-pend not only on the composition of the system interms of identity foci and information distribution,but also on the specific configuration and alignmentof the two aspects.

Second, we created a situation in which eachcomponent team’s main goal—eliminating threats—was equally instrumental to the systemgoal: it did notmatter in which district a possible attack would hap-pen; for a successful outcome, the entire region neededto be kept safe. While, at a basic level, a positive func-tional relationship between the achievement of teamgoals and the achievement of system goals is a definingelement of amultiteamsystem (Mathieu et al., 2001), inpractice, someteams’goalsmaybemoreclearlyalignedwith the systemgoal thanother teams’ goals (Rico et al.,2017).Arguably, strongerdifferentiationamong teamsin terms of goal compatibility could further exacer-bate the differences in information-sharing behavior,depending on how goal compatibility and identityfoci are aligned with each other.

Finally, as we note above, the importance of pro-active information sharing relative to reactive infor-mation sharing depends on the nature and structureof the task, and, with this, on the ability of multiteamsystem members to engage in proactive and reactiveinformation sharing effectively. In this study, memberslacked knowledge of who had what information—limiting their ability to effectively requestwhat theyneeded. Conversely, on tasks structured such that

members canmore easily develop anunderstandingof who knows what—for instance, in the presenceof clear expert roles—members have been shownto engage in more information retrieval, triggeringmore reactive information sharing (Mell et al., 2014).At the same time, in the present study, members hadthe knowledge of who needed what information—increasing their ability to sharewhat they knew. If thetaskwere structured such thatmemberswere less ableto anticipatewhowill needwhat information in orderto perform their part, proactive information sharingmay not only be less prevalent, but also less effective:pushing information to recipients for whom it is ir-relevant would increase counterproductive informa-tion overload (Ellwart, Happ, Gurtner, & Rack, 2015).Insum,effective information sharingdependsonbothmotivation and ability to seek and to share informa-tion (Reinholt, Pedersen, & Foss, 2011). While thefocus of our study lies on motivational antecedents,teams’ ability to seekand to share—inparticular, suchability as arises from features of the task—is an im-portant boundary condition.

Limitations

As discussed in the preceding section, the absenceof knowledge of who knows what in our setting mayhavemade itmore difficult for participants to engagein requesting information from other teams, result-ing in a relatively low base rate and low variability ofreactive information sharing. While this setup is notunrealistic—inmany situations, information seekersdo not know who has the information that theyneed—wecannot exclude thepossibility that the lowvariability may have limited our statistical power todetect differences in reactive information sharingbetween our conditions. Thus, our test of Hypothesis1b may have been underpowered.

In the present study, we examined the interplay be-tween proactive and reactive information sharing in ag-gregate form.While this allowed us to identify the effectof the identity manipulations on the total of a team’sinformation-sharing activity, examining the temporalpatternof this interplay is an intriguingavenue for futureresearch. For example, newmethods capable of captur-ing the complexity of group interaction over time couldallow researchers to examine hypotheses about tempo-ral sequences of proactive and reactive informationsharing (Leenders, Contractor, & DeChurch, 2016;Schecter, Pilny, Leung, Poole, & Contractor, 2018).

Although the laboratory setting of our main studypresents several advantages, it also poses limita-tions. There is certainly a difference in the intensity


of the identification that can be created in the lab ascompared with what exists in the field where teamscollaborate over long time spans and team interac-tions are settled in the context of power and statusdifferences, long-standing relationships, and orga-nizational politics. These factors, along with manyothers, can shape social identities as well as interteamcollaboration patterns independent of or in interactionwith identity concerns. Insofar as our study abstractsfrom this context, it is naturally a simplification of real-ity. On one hand, this ability to isolate a focal constructand investigate its implications while holding constantpotential confounding factors is a core strength of theexperimentalmethod. At the same time, future researchexamining the role of these factors as antecedents orpotential moderators of the effect of identity asymme-tries would be highly valuable.

A further limitation of our study inherent in thelaboratory setting is the relatively short duration ofthe task interaction. It isplausible that, over thecourseof prolonged interaction, component teams’ identityfoci may shift as a result of initial asymmetries andconsequent interaction patterns. The dynamic natureof identity asymmetries and their consequences re-mains a subject for future research.

Finally, as with any experimental study, there is thequestion of the extent to which its findings are general-izable to the field. As several decades of work on socialidentity have shown, identification can bemeaningfullymanipulated in the lab (Hornsey, 2008). As a recentmeta-analysis furthermore suggests, the effects of socialidentity foundinthe labgenerallyparallel those foundinthe field (Mesmer-Magnus et al., 2018). Thus, althoughthereareclear limitations to the labasa setting, given theevidenceprovidedby this streamof research as awhole,we have little reason to believe that the relationshipswefind are unique to this setting.

CONCLUSION

Managing interteam collaboration is a critical task inmultiteam systems and other complex organizationalarrangements. Social identity plays a key role in thisprocess. The present study not only contributes causalevidence for this claim, but also further extends our un-derstanding of the role of social identity in multiteamsystems by shedding first light on the implicationsof differences in identity foci between interdepen-dent teams for collaboration and performance.

REFERENCES

9/11 Commission. 2004. The 9/11 Commission report:Final report of the National Commission on Ter-rorist Attacks upon the United States. Washington,DC: U.S. Government Printing Office. Retrieved fromhttps://www.gpo.gov/fdsys/pkg/GPO-911REPORT/content-detail.html

Aalbers,R.,Dolfsma,W.A.,&Leenders,R.T.A. J.2016.Verticaland horizontal cross-ties: Benefits of cross-hierarchy andcross-unit ties for innovative projects. Journal of ProductInnovationManagement, 33: 141–153.

Ashforth, B. E., Harrison, S. H., & Corley, K. G. 2008. Identifi-cation in organizations: An examination of four funda-mental questions. Journal of Management, 34: 325–374.

Baker, W. E., & Bulkley, N. 2014. Paying it forward vs. re-warding reputation: Mechanisms of generalized reci-procity. Organization Science, 25: 1493–1510.

Bartel, C. A., Wrzesniewski, A., & Wiesenfeld, B. M. 2011.Knowing where you stand: Physical isolation, perceivedrespect, and organizational identification among virtualemployees.Organization Science, 23: 743–757.

Bates,D.,Machler,M.,Bolker, B., &Walker, S. 2015. Fittinglinear mixed-effects models using lme4. Journal ofStatistical Software, 67: 1–48.

Bearman, P. 1997. Generalized exchange. American Jour-nal of Sociology, 102: 1383–1415.

Beersma, B., Hollenbeck, J. R., Humphrey, S. E., Moon, H.,Conlon, D. E., & Ilgen, D. R. 2003. Cooperation, com-petition, and team performance: Toward a contin-gency approach. Academy of Management Journal,46: 572–590.

Berinsky, A. J., Margolis, M. F., & Sances, M. W. 2014. Sepa-rating the shirkers from the workers? Making sure re-spondents pay attention on self-administered surveys.American Journal of Political Science, 58: 739–753.

Best, R. A., Jr., 2011. Intelligence information: Need-to-knowvs. need-to-share. CRS report for Congress, no. R41848.Washington, DC: Congressional Research Service. Re-trieved from https://fas.org/sgp/crs/intel/R41848.pdf

Biel, A., & Thøgersen, J. 2007. Activation of social norms insocial dilemmas: A review of the evidence and reflec-tions on the implications for environmental behaviour.Journal of Economic Psychology, 28: 93–112.

Bolino, M. C., & Grant, A.M. 2016. The bright side of beingprosocial at work, and the dark side, too: A review andagenda for research on other-oriented motives, be-havior, and impact in organizations. Academy ofManagement Annals, 10: 599–670.

Brennecke, J., & Rank, O. N. 2016. The interplay betweenformal project memberships and informal adviceseeking in knowledge-intensive firms: A multilevelnetwork approach. Social Networks, 44: 307–318.


https://www.gpo.gov/fdsys/pkg/GPO-911REPORT/content-detail.html

https://www.gpo.gov/fdsys/pkg/GPO-911REPORT/content-detail.html

https://fas.org/sgp/crs/intel/R41848.pdf

Bresman, H. 2013. Changing routines: A process modelof vicarious group learning in pharmaceutical R&D.Academy of Management Journal, 56: 35–61.

Brewer, M. B. 1979. In-group bias in the minimal inter-group situation: A cognitive-motivational analysis.Psychological Bulletin, 86: 307–324.

Brewer, M. B., & Brown, R. J. 1998. Intergroup relations. InD. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), Thehandbook of social psychology, vols. 1–2 (4th ed.):554–594. New York, NY: McGraw-Hill.

Caimo, A., & Lomi, A. 2015. Knowledge sharing in orga-nizations: A Bayesian analysis of the role of reciproc-ity and formal structure. Journal of Management,41: 665–691.

Carton, A. M., & Cummings, J. N. 2012. A theory of sub-groups in work teams. Academy of ManagementReview, 37: 441–470.

Chatman, J.A.,&Flynn,F. J. 2001.The influenceofdemographicheterogeneity on the emergence and consequences of co-operative norms inwork teams.AcademyofManagementJournal, 44: 956–974.

Chattopadhyay,P.1999.Beyonddirectandsymmetricaleffects:The influence of demographic dissimilarity on organiza-tional citizenship behavior. Academy of ManagementJournal, 42: 273–287.

Chattopadhyay, P., George, E., & Lawrence, S. A. 2004. Whydoes dissimilarity matter? Exploring self-categorization,self-enhancement, and uncertainty reduction. Journal ofApplied Psychology, 89: 892–900.

Chattopadhyay, P.,George, E., &Ng,C. 2011.Anuncertaintyreduction model of relational demography. Researchin Personnel and Human Resources Management,30: 219–251.

Chattopadhyay, P., George, E., & Shulman, A. D. 2008. Theasymmetrical influence of sex dissimilarity in dis-tributive vs. colocated work groups. OrganizationScience, 19: 581–593.

Chattopadhyay, P., Tluchowska, M., & George, E. 2004.Identifying the ingroup:Acloser lookat the influenceofdemographicdissimilarityonemployee social identity.Academy of Management Review, 29: 180–202.

Cho, H.-C., & Abe, S. 2013. Is two-tailed testing for direc-tional research hypotheses tests legitimate? Journal ofBusiness Research, 66: 1261–1266.

Connaughton, S.A.,Williams, E. A., & Shuffler,M. L. 2012.Social identity issues in multiteam systems: Consid-erations for future research. In S. J. Zaccaro, M. A.Marks, & L. A. DeChurch (Eds.), Multiteam systems:An organization form for dynamic and complexenvironments: 109–140. New York, NY: Routledge.

Connelly, C., Zweig, D., & Webster, J. 2012. Knowledgehiding in organizations. Journal of OrganizationalBehavior, 33: 64–88.

Crawford, E. R., & LePine, J. A. 2013. A configural theory ofteam processes: Accounting for the structure of task-work and teamwork. Academy of ManagementReview, 38: 32–48.

Cuijpers, M., Uitdewilligen, S., & Guenter, H. 2016. Effectsof dual identification and interteam conflict on mul-titeam system performance. Journal of Occupationaland Organizational Psychology, 89: 141–171.

Davison, R. B., Hollenbeck, J., Barnes, C. M., Sleesman,D. J., & Ilgen, D. R. 2012. Coordinated action in mul-titeam systems. Journal of Applied Psychology, 97:808–824.

De Cremer, D., van Knippenberg, D., van Dijke, M., & Bos,A. E. R. 2006. Self-sacrificial leadership and followerself-esteem: When collective identification matters.Group Dynamics, 10: 233–245.

de Vries, T. A., Hollenbeck, J. R., Davison, R. B., Walter, F.,& van der Vegt, G. S. 2016. Managing coordination inmultiteam systems: Integrating micro and macro per-spectives. Academy of Management Journal, 59:1823–1844.

deVries,T.A.,Walter,F.,vanderVegt,G.S.,&Essens,P. J.M.D.2014.Antecedents of individuals’ interteamcoordination:Broad functional experiences as a mixed blessing. Acad-emy of Management Journal, 57: 1334–1359.

DeChurch, L. A., & Marks, M. A. 2006. Leadership inmultiteam systems. Journal of Applied Psychology,91: 311–329.

DiazGranados, D., Dow, A. W., Perry, S. J., & Palesis, J. A.2014. Understanding patient care as a multiteam sys-tem. In M. L. Shuffler, E. Salas, & R. Rico (Eds.),Pushing the boundaries: Multiteam systems in re-search and practice, vol. 16: 95–113. Bingley, U.K.:Emerald Group Publishing.

Director of National Intelligence. 2009. Intelligence com-munitydirective501:Discoveryanddisseminationorretrieval of information within the intelligence com-munity (ICD 501). Washington, DC: Office of the Di-rector of National Intelligence. Retrieved from https://www.dni.gov/files/documents/ICD/ICD_501.pdf

Dokko, G., Kane, A. A., & Tortoriello, M. 2014. One of us orone ofmy friends: How social identity and tie strengthshape the creative generativity of boundary-spanningties. Organization Studies, 35: 703–726.

Dovidio, J. F.,Gaertner,S.,Validzic,A.,Matoka,K., Johnson,B.,& Frazier, S. 1997. Extending the benefits of recategoriza-tion: Evaluations, self-disclosure, and helping. Journal ofExperimental Social Psychology, 33: 401–420.

Eckel, C. C., & Grossman, P. J. 2005. Managing diversity bycreating team identity. Journal of EconomicBehavior& Organization, 58: 371–392.

Ellis, A. P. J. 2006. System breakdown: The role of mentalmodels and transactive memory in the relationship


https://www.dni.gov/files/documents/ICD/ICD_501.pdf

https://www.dni.gov/files/documents/ICD/ICD_501.pdf

between acute stress and team performance. Acad-emy of Management Journal, 49: 576–589.

Ellwart, T., Happ, C., Gurtner, A., & Rack, O. 2015. Man-aging information overload in virtual teams: Effectsof a structured online team adaptation on cognitionand performance. European Journal of Work andOrganizational Psychology, 24: 1–15.

Espinosa, J. A., Lerch, J., & Kraut, R. E. 2004. Explicitversus implicit coordination mechanisms and taskdependencies: One size does not fit all. In E. Salas &S. M. Fiore (Eds.), Team cognition: Understandingthe factors that drive process and performance:107–129. Washington, DC: American PsychologicalAssociation.

Fehr, E., & Gachter, S. 2000. Fairness and retaliation: Theeconomics of reciprocity. Journal of Economic Per-spectives, 14: 159–181.

Feld, S. L. 1981. The focused organization of social ties.American Journal of Sociology, 86: 1015–1035.

Fisher, D. M., Bell, S. T., Dierdorff, E. C., & Belohlav, J. A.2012. Facet personality and surface-level diversity asteam mental model antecedents: Implications for im-plicit coordination. Journal of Applied Psychology,97: 825–841.

Fleischer, A., Mead, A. D., & Huang, J. 2015. Inattentiveresponding in MTurk and other online samples. In-dustrial and Organizational Psychology: Perspec-tives on Science and Practice, 8: 196–202.

Flynn, F. J. 2005. Identity orientations and forms of socialexchange in organizations.AcademyofManagementReview, 30: 737–750.

Gaertner, S. L., Dovidio, J. F., Anastasio, P. A., Bachman,B. A., & Rust, M. C. 1993. The common ingroupidentity model: Recategorization and the reduction ofintergroup bias. European Review of Social Psy-chology, 4: 1–26.

Gaertner, S. L., Mann, J., Murrell, A., & Dovidio, J. F. 1989.Reducing intergroup bias: The benefits of recategori-zation. Journal of Personality and Social Psychol-ogy, 57: 239–249.

George, E., & Chattopadhyay, P. 2005. One foot in eachcamp: The dual identification of contract workers.Administrative Science Quarterly, 50: 68–99.

George, E., Chattopadhyay, P., &Zhang, L. L. 2012.Helpinghand or competition? The moderating influence ofperceived upward mobility on the relationship be-tween blended workgroups and employee attitudesand behaviors. Organization Science, 23: 355–372.

Goette, L., Huffman, D., & Meier, S. 2006. The impact ofgroup membership on cooperation and norm enforce-ment: Evidence using randomassignment to real socialgroups. American Economic Review, 96: 212–216.

Gokpinar, B., Hopp, W. J., & Iravani, S. M. R. 2010. Theimpact ofmisalignment of organizational structure andproduct architecture on quality in complex productdevelopment.Management Science, 56: 468–484.

Goodwin, G. F., Essens, P. J. M. D., & Smith, D. 2012.Multiteam systems in the public sector. In S. J.Zaccaro, M. A. Marks, & L. A. DeChurch (Eds.),Multiteam systems: An organization form for dy-namic and complex environments: 53–78. NewYork, NY: Routledge.

Gravetter, F. J., & Wallnau, L. B. 2017. Statistics for thebehavioral sciences (10th ed.). Boston, MA: CengageLearning.

Gruenfeld, D.H., Inesi,M. E.,Magee, J. C., &Galinsky, A. D.2008. Power and the objectification of social targets.Journal of Personality and Social Psychology, 95:111–127.

Hayes,A. F. 2013. Introduction tomediation,moderation,and conditional process analysis a regression-basedapproach. New York, NY: The Guilford Press.

Hewstone, M., Rubin, M., & Willis, H. 2002. Intergroupbias. Annual Review of Psychology, 53: 575–604.

Hinds, P. J., &Mortensen,M. 2005. Understanding conflictin geographically distributed teams: The moderatingeffects of shared identity, shared context, and spon-taneous communication. Organization Science, 16:290–307.

Hoegl, M., & Weinkauf, K. 2004. Interteam coordination,project commitment, and teamwork in multiteamR&D projects: A longitudinal study. OrganizationScience, 15: 38–55.

Hogg, M. A., & Terry, D. J. 2000. Social identity and self-categorization: Processes in organizational contexts.Academy of Management Review, 25: 121–140.

Hollingshead, A. B. 2001. Cognitive interdependenceand convergent expectations in transactive mem-ory. Journal of Personality and Social Psychology,81: 1080–1089.

Holm, S. 1979. A simple sequentially rejective multipletest procedure. Scandinavian Journal of Statistics,6: 65–70.

Homan, A. C., Hollenbeck, J. R., Humphrey, S. E., vanKnippenberg, D., Ilgen, D. R., & van Kleef, G. A. 2008.Facing differences with an open mind: Openness toexperience, salience of intragroup differences, andperformance of diverse work groups. Academy ofManagement Journal, 51: 1204–1222.

Hornsey,M. J. 2008.Social identity theoryandself-categorizationtheory: A historical review. Social and Personality Psy-chology Compass, 2: 204–222.

Hornsey,M. J., &Hogg,M.A. 2000.Assimilationanddiversity:An integrative model of subgroup relations. Personalityand Social Psychology Review, 4: 143–156.


Horwitz, L. I., Moin, T., Krumholz, H. M., Wang, L., &Bradley, E. H. 2008. Consequences of inadequate sign-out for patient care. Archives of Internal Medicine,168: 1755–1760.

Humphrey, S. E., & Aime, F. 2014. Team microdynamics:Toward an organizing approach to teamwork. Acad-emy of Management Annals, 8: 443–503.

Joshi, A., Pandey, N., & Han, G. 2009. Bracketing teamboundary spanning: An examination of task-based,team-level, and contextual antecedents. Journal ofOrganizational Behavior, 30: 731–759.

Kane, A. A. 2010. Unlocking knowledge transfer potential:Knowledge demonstrability and superordinate socialidentity. Organization Science, 21: 643–660.

Kane, A. A., Argote, L., & Levine, J. M. 2005. Knowledgetransfer between groups via personnel rotation: Effects ofsocial identity and knowledge quality. OrganizationalBehavior and Human Decision Processes, 96: 56–71.

Kerr, N. L. 1995. Norms in social dilemmas. In D. Schroeder(Ed.), Social dilemmas: Perspectives on individualsand groups: 31–48. Westport, CT: Praeger.

Kramer, R. M., & Brewer, M. B. 1984. Effects of groupidentity on resource use in a simulated commons di-lemma. Journal of Personality and Social Psychol-ogy, 46: 1044–1057.

Kuhlman,M.,&Marshello,A.F.1975. Individualdifferences ingame motivation as moderators of preprogrammed strat-egy effects in prisoner’s dilemma. Journal of Personalityand Social Psychology, 32: 922–931.

Lanaj, K., Foulk, T. A., & Hollenbeck, J. R. 2018. The ben-efits of not seeing eye to eye with leadership: Diver-gence in risk preferences impacts multiteam systembehavior and performance. Academy of Manage-ment Journal, 61: 1554–1582.

Lanaj, K., Hollenbeck, J. R., Ilgen, D. R., Barnes, C. M., &Harmon, S. J. 2013. The double-edged sword of decen-tralized planning in multiteam systems. Academy ofManagement Journal, 56: 735–757.

Leenders, R. T. A. J., Contractor, N. S., & DeChurch, L. A.2016. Once upon a time Understanding team pro-cesses as relational event networks. OrganizationalPsychology Review, 6: 92–115.

Leenders, R. T.A. J., &Dolfsma,W.A.2016. Social networksfor innovation and new product development. Journalof Product Innovation Management, 33: 123–131.

LePine, J. A., Piccolo, R., Jackson, C. L., Mathieu, J. E., &Saul, J. R. 2008. A meta-analysis of teamwork pro-cesses: Tests of a multidimensional model and rela-tionships with team effectiveness criteria. PersonnelPsychology, 61: 273–307.

Lomi, A., Lusher, D., Pattison, P. E., & Robins, G. 2014. Thefocused organization of advice relations: A study inboundary crossing.Organization Science, 25: 438–457.

Luciano, M. M. 2017. Coordination at the interface:Understanding and improving cross-unit patienthandoffs.Academy ofManagement Proceedings.Published online ahead of print. doi: 10.5465/AMBPP.2017.259

Luciano, M. M., DeChurch, L. A., & Mathieu, J. E. 2018.Multiteam systems:A structural framework andmeso-theory of system functioning. Journal of Manage-ment, 44: 1065–1096.

Marks, M. A., DeChurch, L. A., Mathieu, J. E., Panzer, F. J.,& Alonso, A. 2005. Teamwork in multiteam systems.Journal of Applied Psychology, 90: 964–971.

Marks, M. A., Mathieu, J. E., & Zaccaro, S. J. 2001. A temporallybased framework and taxonomy of team processes. Acad-emy ofManagement Review, 26: 356–376.

Marrone, J. 2010. Team boundary spanning: A multilevelreview of past research and proposals for the future.Journal of Management, 36: 911–940.

Marrone, J., Tesluk, P. E., & Carson, J. B. 2007. A multilevelinvestigation of antecedents and consequences of teammember boundary-spanning behavior. Academy ofManagement Journal, 50: 1423–1439.

Mathieu, J. E., Heffner, T. S., Goodwin, G. F., Salas, E., &Cannon-Bowers, J. A. 2000. The influence of sharedmental models on team process and performance.Journal of Applied Psychology, 85: 273–283.

Mathieu, J. E., Hollenbeck, J. R., van Knippenberg, D., &Ilgen, D. R. 2017. A century of work teams in theJournal of Applied Psychology. Journal of AppliedPsychology, 102: 452–467.

Mathieu, J. E., Marks, M. A., & Zaccaro, S. J. 2001. Multi-team systems. In N. Anderson, D. Ones, H. K. Sinangil,& C. Viswesvaran (Eds.), International handbook ofwork and organizational psychology: 289–313. Lon-don, U.K.: SAGE.

Mell, J. N., van Knippenberg, D., & van Ginkel, W. P. 2014.The catalyst effect: The impact of transactive memorysystem structure on team performance. Academy ofManagement Journal, 57: 1154–1173.

Mesmer-Magnus, J. R., Asencio, R., Seely, P. W., & DeChurch,L. A. 2018. How organizational identity affects team func-tioning: The identity instrumentality hypothesis. JournalofManagement, 44: 1530–1550.

Mesmer-Magnus, J. R., Carter, D. R., Asencio, R., & DeChurch,L.A. 2016. Space exploration illuminates the next frontierfor teams research.Group&OrganizationManagement,41: 595–628.

Mesmer-Magnus, J. R., & DeChurch, L. A. 2009. Informa-tion sharing and team performance: A meta-analysis.Journal of Applied Psychology, 94: 535–546.

Molm, L. D., Collett, J. L., & Schaefer, D. R. 2007. Buildingsolidarity through generalized exchange: A theory of


reciprocity. American Journal of Sociology, 113: 205–242.

Muthen, L. K., & Muthen, B. O. 2017. Mplus user’s guide(8th ed.). Los Angeles, CA: Muthen & Muthen.

Pearsall, M. J., & Venkataramani, V. 2015. Overcomingasymmetric goals in teams: The interactive roles ofteam learning orientation and team identification.Journal of Applied Psychology, 100: 735–748.

Pettigrew, T. F. 1998. Intergroup contact theory. AnnualReview of Psychology, 49: 65–85.

Porck, J. P.,Matta, F.K.,Hollenbeck, J. R.,Oh, J. K., Lanaj,K.,& Lee, S. M. 2019. Social identification in multiteamsystems: The role of depletion and task complexity.Academy of Management Journal, 62: 1137–1162.

R Core Team. 2016. R: A language and environment forstatistical computing. Vienna, Austria: R Foundation forStatistical Computing. Retrieved from https://www.R-project.org

Reinholt, M., Pedersen, T., & Foss, N. J. 2011. Why a centralnetwork position isn’t enough: The role of motivationand ability for knowledge sharing in employee networks.Academy ofManagement Journal, 54: 1277–1297.

Richter, A.W.,West,M., vanDick, R., &Dawson, J. F. 2006.Boundary spanners’ identification, intergroup con-tact, and effective intergroup relations. Academy ofManagement Journal, 49: 1252–1269.

Rico, R., Hinsz, V. B., Burke, S., & Salas, E. 2017. A multi-level model ofmultiteammotivation and performance.Organizational Psychology Review, 7: 197–226.

Rico, R., & Sanchez-Manzanares, M. 2008. Team implicit co-ordination processes: A team knowledge-based approach.Academy ofManagement Review, 33: 163–184.

Riketta, M., & van Dick, R. 2005. Foci of attachment in orga-nizations:Ameta-analytic comparisonof thestrengthandcorrelates of workgroup versus organizational identifica-tion and commitment. Journal of Vocational Behavior,67: 490–510.

Schecter,A., Pilny,A., Leung,A., Poole,M.S.,&Contractor,N.2018. Step by step: Capturing the dynamics of work teamprocess through relational event sequences. Journal ofOrganizational Behavior, 39: 1163–1181.

Schwab, D. P. 2005.Researchmethods for organizationalstudies (2nd ed.). Mahwah, NJ: Psychology Press.

Shamir, B., Zakay, E., Breinin, E., & Popper, M. 1998. Cor-relates of charismatic leader behavior in military units:Subordinates’ attitudes, unit characteristics, and supe-riors’ appraisals of leader performance. Academy ofManagement Journal, 41: 387–409.

Shuffler, M. L., Jimenez-Rodrıguez, M., & Kramer, W. S.2015. The science of multiteam systems A review andfuture research agenda. Small Group Research, 46:659–699.

Tajfel, H. 1978. Social categorization, social identity, andsocial comparison. In H. Tajfel (Ed.), Differentiationbetween social groups: Studies in the social psy-chologyof intergroup relations: 61–76.London,U.K.:Academic Press.

Tajfel, H., & Turner, J. C. 1986. The social identity theory ofintergroup behavior. In S. Worchel & W. G. Austin(Eds.), Psychology of intergroup relations: 7–24. Chi-cago, IL: Nelson Hall.

Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., &Wetherell,M. S. 1987.Rediscovering the social group:A self-categorization theory. Oxford, U.K.: Blackwell.

van Dick, R., van Knippenberg, D., Hagele, S., Guillaume,Y. R. F., & Brodbeck, F. C. 2008. Group diversity andgroup identification: The moderating role of diversitybeliefs. Human Relations, 61: 1463–1492.

van Dick, R., van Knippenberg, D., Kerschreiter, R., Hertel,G., & Wieseke, J. 2008. Interactive effects of workgroup and organizational identification on job satis-faction and extra-role behavior. Journal ofVocationalBehavior, 72: 388–399.

van Ginkel, W. P., & van Knippenberg, D. 2009. Knowledgeabout the distribution of information and group decisionmaking: When and why does it work? OrganizationalBehaviorandHumanDecisionProcesses,108:218–229.

van Knippenberg, D., Haslam, S. A., & Platow, M. J. 2007.Unity through diversity: Value-in-diversity beliefs,work groupdiversity, and group identification.GroupDynamics, 11: 207–222.

van Knippenberg, D., & Mell, J. N. 2016. Past, present, andpotential future of team diversity research: From com-positional diversity to emergent diversity. Organiza-tional Behavior and HumanDecision Processes, 136:135–145.

van Knippenberg, D., & van Schie, E. C. M. 2000. Foci andcorrelates of organizational identification. Journal ofOccupational and Organizational Psychology, 73:137–147.

Wegner,D.M. 1987.Transactivememory:Acontemporaryanalysis of the group mind. In B. Mullen & G. R.Goethals (Eds.),Theories of group behavior, vol. 185:185–208. New York, NY: Springer-Verlag.

Wegner, D.M. 1995. A computer networkmodel of humantransactive memory. Social Cognition, 13: 319–339.

Wit, A. P., & Kerr, N. L. 2002. “Me versus just us versus usall” categorization and cooperation in nested socialdilemmas. Journal of Personality and Social Psy-chology, 83: 616–637.

Yamagishi, T., & Kiyonari, T. 2000. The group as the con-tainer of generalized reciprocity. Social PsychologyQuarterly, 63: 116–132.


https://www.R-project.org

https://www.R-project.org

Julija N. Mell ([email protected]) is an assistant professor atthe RotterdamSchool ofManagement, ErasmusUniversityRotterdam. She received her PhD in management fromErasmus University Rotterdam. Her research focuses onboundary-spanning collaboration in diverse teams, geo-graphically distributed teams, multiteam systems, andconfigurations involving multiple team membership.

Leslie A. DeChurch ([email protected]) is pro-fessor and chair of communication studies, professorof psychology, and director of the ATLAS laboratory atNorthwestern University. She received her PhD in indus-trial and organizational psychology, is the recipient ofa National Science Foundation CAREER award, and isa fellow of APA, APS, and SIOP.

Roger Th. A. J. Leenders ([email protected]) is pro-fessor of Organization Studies at the Jheronimus Academyof Data Science, Den Bosch. He received his PhD in sociol-ogy from the University of Groningen. His research interestsinclude dynamic social network analysis, team performance,sports analytics, and social influence in networks.

Noshir Contractor ([email protected]) is the Jane S.& William J. White professor of Behavioral Sciences atNorthwestern University. He is a fellow of the InternationalCommunication Association, Association of ComputingMachinery, and American Association for the Advancementof Science. He received a PhD from the Annenberg School ofCommunication, University of Southern California.

APPENDIX A: MATERIALS FOR STUDY 2

1. INTRODUCTION TO SCENARIO

Please imagine the following situation:

You are working as an intelligence officer in Task-force Delta. Delta’s main mission is to ensure the se-curity of Kazbar, a city in a conflict region thatregularly faces terrorist threats. This is a map of Kaz-bar (Figure A1):

Because Kazbar is a fairly large city, your taskforceconsists of five teams: Team Center, Team North,Team East, Team South, and TeamWest.

Each team has intelligence officers and field special-ists. Intelligence officers gather information aboutpotential threats from different sources. Field spe-cialists use this information to neutralize thesethreats. Each team is primarily active in their owndistrict, but together your objective is to secure thecity of Kazbar.

As an intelligence officer, you regularly talk to yoursources. Sometimes, you learn information about threatsin your district and sometimes you learn informationabout threats in other districts. Similarly, intelligenceofficers in other districts sometimes learn informationabout threats in your district from their sources.

When you learn useful information from your sources,you write this information in a memo and send it to thefield specialist for whom it will be relevant. Becausethere is never enough time, you often need to prioritizeand choose between sending different memos to differ-ent taskforce members.

2. IDENTITY FOCUS MANIPULATION

Team Focus

You are part of Team Center. While both yourmembership in Team Center and your membershipin Taskforce Delta are important to you, when youthink of yourself, you usually see yourself as amember of the team first—and a member of thetaskforce second.

You have shared many experiences with the othermembers of the teamand, as a result, you feel a strongsense of attachment and unity with the team. You

FIGURE A1Map of the City of Kazbar

North

CenterWest East

South

Map of cityof Kazbar


mailto:[email protected]




often compare your team to other teams operating inthe other districts and you are proud of what youhave achieved together with your team so far. Youview the cooperationwith the othermembers of yourteam as something particularly special.

All in all, even though you feel connected to bothyour team and the overarching taskforce, you feelparticularly at home in your team. When you thinkabout your team, you think “we.” When you thinkabout the other teams, you think “they.” (See alsoFigure A2.)

System Focus

You are part of Team Center. While both your mem-bership in Team Center and your membership inTaskforce Delta are important to you, when you thinkof yourself, youusually see yourself as amember of thetaskforce first—and a member of the team second.

You have shared many experiences with the othermembers of the taskforce and, as a result, you feel astrong sense of attachment and unity with thetaskforce. You often compare your taskforce toother taskforces operating in other cities and youare proud of what you have achieved together as ataskforce so far. You view the cooperation with theother members of your taskforce as something par-ticularly special.

All in all, even though you feel connected to bothyour team and the overarching taskforce, youfeel particularly at home in your taskforce. When

you think about the other teams in the taskforce,you always think “we”—just the same as when youthink about your own team—never “they.” (See alsoFigure A3.)

3. INFORMATION SCENARIO

When you talked with your sources today, you havelearned about two potential threats: one in the Northdistrict and one in the South district.

Apart from the information about the threats, yoursources had some additional insights for you.

Theymentioned that TeamNorth has just establisheda connection to a new source with ties to the CenterDistrict. Thismeans that TeamNorth is likely to learna lot of information about the Center District in theforeseeable future.

They also mentioned that one of Team South’s keysources of information about the Center District hasjust gone underground. This means that Team Southis not likely to learn any information about the CenterDistrict in the foreseeable future.

4. RECIPROCITY-BASED INFORMATIONSHARING: TIME ALLOCATION

Because time is limited, you have to split your timebetweenwritingmemos. Themore time you spendon a memo, the more useful it will be for the fieldspecialist who receives it.

FIGURE A2Structure of Taskforce Delta

Taskforce Delta

TeamCenter

TeamEast

TeamNorth

TeamSouth

TeamWest

FIGURE A3Structure of Taskforce Delta

Taskforce Delta

TeamCenter

TeamEast

TeamNorth

TeamSouth

TeamWest


Howwill you split your time? You can give between0% and 100% of your time to any of these twomemos, but it has to add up to 100%.

5. INSTRUMENTALITY MOTIVE

How would you generally think about your team’srelationshipwith the other teams on the task force?

(1) I would think more about what other teamscan do for my team than what I can do forthem.

(2) I would tend to contact other teams only when Ineed something from them.

(3) Themain reasonwhy relationshipswith otherteams would be important to me is becausethey help me accomplish my team’s goals.

(4) My relationship with another team would bebasedonhowproductive it is, rather thanonhowmuch I enjoy it.

(5) If the nature of my team’s task changed and an-other team wasn’t helpful anymore, the rela-tionship probably wouldn’t continue.

(6) Iwould likea teamthat isnotuseful tomyteamlessthan I would like a team that is useful to my team.


Documents

Identity Asymmetries: An Experimental Investigation of