Article 18. Relational Identity and Identification.pdf

RELATIONAL IDENTITY AND IDENTIFICATION:DEFINING OURSELVES THROUGH WORK

RELATIONSHIPS

DAVID M. SLUSSUniversity of South Carolina

BLAKE E. ASHFORTHArizona State University

We explore the meaning and significance of relational identity and relational iden-tification, predicated on the role-relationship between two individuals. We argue thatrelational identity integrates person- and role-based identities and thereby the indi-vidual, interpersonal, and collective levels of self; contrast relational identity andrelational identification with social identity and social identification; contend thatrelational identity and relational identification are each arranged in a cognitivehierarchy ranging from generalized to particularized schemas; and contrast relationalidentification with relational disidentification and ambivalent relational identifica-tion.

Identity is at its core psychosocial: self and other;inner and outer; being and doing; expression ofself for, with, against, or despite; but certainly inresponse to others. It is both those for whom oneworks and the work of loving (Josselson, 1994: 82).

The study of identity and identification in or-ganizations has focused almost entirely on theindividual vis-a`-vis a collective, such as a work-group, department, and the organization itself.This research has yielded a wealth of insightson how individuals define and locate them-selves within organizational contexts (Ashforth& Johnson, 2001; Ashforth & Mael, 1996; Barker &Tompkins, 1994; Dutton, Dukerich, & Harquail,1994; Elsbach, 1999; Haslam, van Knippenberg,Platow, & Ellemers, 2003; Hogg & Terry, 2001;Pratt, 1998; van Dick, 2004; van Knippenberg &van Schie, 2000). However, this focus on the in-dividual vis-a`-vis a collective has largely ig-nored the interpersonal level and its influenceon ones identity and identification in the work-place.To be sure, management scholars in areas

such as career development, leadership, socialnetworks, and positive organizational scholar-

ship have provided valuable insights on the im-pact of relationships on ones development, per-formance, and well-being (e.g., Baker, Cross, &Wooten, 2003; Dutton & Heaphy, 2003; Flum,2001b; Gibbons, 2004; Hall & Kahn, 2002; Kahn,1998; Lord & Brown, 2001; Morrison, 2002; Ragins,Cotton, & Miller, 2000; Sherony & Green, 2002;Wrzesniewski, Dutton, & Debebe, 2003). Whathas been missing, though, is a specific focus onhow ones definition of self might be influencedby interpersonal relationships and the conse-quent interplay of three levels of identity: in-dividual (or personal), interpersonal, and collec-tive (or group, social). A major breakthroughoccurred with Brewer and Gardners (1996) arti-cle, Who Is This We? in which they contrastedthe three levels of self (see also Brickson, 2000;Brickson & Brewer, 2001; Sedikides & Brewer,2001). According to these and other scholars(e.g., Kashima & Hardie, 2000; Lord, Brown, &Freiberg, 1999), the individual level focuses ononeself as a unique being, and self-esteem de-rives from interpersonal comparisons of traits,abilities, goals, performance, and the like. Thebasic motivation is self-interest, and the individ-ual is essentially independent and autonomous.The interpersonal level focuses on ones role-

related relationships (henceforth role-relation-ships), such as supervisor-subordinate and co-worker-coworker. Individuals are thereforeinterdependent, placing a premium on the na-

We thank Glen Kreiner, Loriann Roberson, former associ-ate editor Elizabeth Mannix, and three anonymous review-ers for their very helpful comments on earlier drafts. Anearlier version of the paper was presented at the 2003 an-nual meeting of the Academy of Management in Seattle.

Academy of Management Review2007, Vol. 32, No. 1, 932.

9

ture of interaction and the potential for personalconnection and intimacy. As Andersen andChen put it, the self is relationalor even en-tangledwith significant others (2002: 619). Thebasic motivation is the dyads welfare, and self-esteem derives from fulfilling ones role-rela-tionship obligations.Finally, the collective level is the province of

social identity theory (SIT; Tajfel & Turner, 1986).At this level, the focus is on oneself as a proto-typical member of a group, such as an organi-zation, or a social category, such as gender. Self-esteem derives from intergroup comparisons,and the basic motivation is the welfare of thecollective, placing a premium on common fate,cohesion, and group norms. (As an anonymousreviewer noted, a second, macro perspectiveon the collective level of identity focuses on thesocial entity per se, independent of individuals.For example, Albert and Whetten [1985] concep-tualize organizational identity as the central,distinctive, and enduring attributes of an orga-nization qua organization. These attributes mayor may not correspond to the social identity ofthe individual as a member of the organization.Given our interest in the individual, our focuswill be restricted to the SIT conception of thecollective level.)Thus, as Brewer and Gardner argue,

Changes in levels of self-categorization reflectnot only differences in views of the self but alsodifferent worldviews (1996: 91), including val-ues, goals, and norms. It should be noted thatindividuals retain a sense of self from all threelevels, albeit in somewhat compartmentalizedform (Ashforth & Johnson, 2001; Trafimow, Trian-dis, & Goto, 1991), but that cognitive shifts be-tween the levels are easily primed by varioussituational cues, such as names, rewards, uni-forms, and saying we rather than you or I(Baldwin, 1994; Gaertner, Sedikides, Vevea, &Iuzzini, 2002; Gardner, Gabriel, & Hochschild,2002; Kark & Shamir, 2002; Pratt & Rafaeli, 2001).Research on self-conceptions born of role-

relationships is particularly important for atleast two reasons. First, according to the struc-tural functionalist wing of role theory (Merton,1957; see also Stryker & Statham, 1985), roles aresets of behavioral expectations associated withgiven positions in the social structure (Ebaugh,1988: 18). As such, roles are a basic buildingblock of organizations, and the purpose andmeaning of a given role depends on the network

of complementary roles within which it is em-bedded (Biddle, 1979; Katz & Kahn, 1978). Therole of supervisor is meaningless without thecomplementary role of subordinate; the role ofteam member requires at least one other teammember. As Stryker and Statham state, To usethe term role is necessarily to refer to interac-tion (1985: 323). Thus, self-definition in organi-zational contexts is predicated at least partly onones network of interdependent roles. For ex-ample, Pratt and Rafaeli (1997) discovered thatnurses choose their type of dress (an identitymarker) partially based on the type of patientserved (role-relationship). Moreover, becauseroles tend to be differentiated by function, hier-archy, and status, the nature of the complemen-tarity between any two roles (and how role in-cumbents choose to enact them) tends to bemore or less unique (Ashforth, 2001). Group-levelanalyses tend to overlook differences betweenspecific role-relationships such as coworker-coworker.Second, the interpersonal perspective on role-

relationships and, thus, interdependence and in-teraction is very timely. In the face of environ-mental turbulence, the emphasis on traditionalbureaucratic structures and control systems isshifting toward more fluid team- and project-based work, where interaction and personalconnection provide relatively informal socialcontrols. In Flums words, To work is to relate(2001a: 262), and the identities and identifica-tions flowing from role-relationships may pro-vide a much-needed cognitive and affectiveglue for organic organizations.In this article we build on Brewer and Gard-

ners (1996) notion of a relational self. First, wedefine the concepts of relational identity andrelational identification and discuss how eachis arranged in a cognitive hierarchy rangingfrom generalized to particularized schemas (i.e.,systems of beliefs). Second, we demonstratehow our conception of relational identity andidentification converges and diverges with for-mulations derived from SITthe predominanttheory of identity and identification in organiza-tional contexts. Third, we develop a typologyinvolving relational identification, relationaldisidentification, and ambivalent relationalidentification, and we discuss the functions anddysfunctions of each type. Finally, we close withsuggestions for future research.

10 JanuaryAcademy of Management Review

The topics of roles, identities, and relation-ships have been studied extensively in varioussocial domains, including organizations. Whatthe present paper adds to such models as roletheory, SIT, and identity theory, and to researchon personal relationshipsall four of which aretouched on at various pointsis a focus on howthe interpersonal level draws on and helps in-tegrate the personal and collective levels suchthat interpersonal relationships are simulta-neously informed by person- and role-basedidentities. Additionally, organizational scholarshave generally sidestepped discussing person-alized relationshipswith few exceptions (e.g.,Dutton & Heaphy, 2003; Gersick, Bartunek, & Dut-ton, 2000). In this paper we integrate insightsfrom the personal relationship literature withinthe broader conceptual landscape of roles andidentity to provide a more personalized and ho-listic understanding of ones work experience.

DEFINING RELATIONAL IDENTITY ANDRELATIONAL IDENTIFICATION

Relational Identity

Brewer and Gardner (1996), like many identitytheorists, conflate the terms identity and identi-fication, implicitly treating them as synonymsfor ones sense of self (who I am). However, thereis value in differentiating the terms. We define arelational identity as the nature of ones role-relationship, such as manager-subordinate andcoworker-coworker. It is how role occupants en-act their respective roles vis-a`-vis each other. Asnoted above, a role is fundamentally relational

and is largely understood with reference to thenetwork of interdependent roles. Indeed, it isrelational identities that knit the network ofroles and role incumbents together into a socialsystem.We define relational identification as the ex-

tent to which one defines oneself in terms of agiven role-relationship. Thus, one may have aclear sense of what it means to be a subordinatevis-a`-vis ones manager (i.e., a perceived rela-tional identity) but resist viewing that relationalidentity as self-defining (i.e., low relationalidentification). Note that both relational identity(What is the nature of our relationship?) andrelational identification (How much do I inter-nalize that identity as a partial definition ofself?) implicate the individual. We expand onthe concept of relational identity in this sectionand on relational identification in the next sec-tion.As shown in Figure 1, we argue that a rela-

tional identity consists of four parts: one individ-uals role- and person-based identities as theybear on the role-relationship, and another indi-viduals role- and person-based identities asthey bear on the role-relationship. A role-basedidentity is the goals, values, beliefs, norms, in-teraction styles, and time horizons typically as-sociated with the role (Ashforth, 2001)indepen-dent of who (what kind of person) may beenacting the role. For example, the role-basedidentity of a manager may include assigningtasks, monitoring performance, offering feed-back, liaising with other departments, and do-ing strategic planning. A relational identity,

FIGURE 1Relational Identity

Note: The dotted lines represent feedback loops.

2007 11Sluss and Ashforth

however, focuses on that portion of the role-based identity that is more or less directly rele-vant to ones role-relationship. Thus, liaisingand strategic planning may not be directly rel-evant to the managers relationship with a sub-ordinate and would therefore be excluded fromthe manager-subordinate relational identity.A person-based identity is the personal

qualities of the role occupant that bear on theenactment of the role-based identity. Roles areessentially abstractions until brought to lifeby flesh-and-blood individuals. Thus, barringvery strong situations, individuals have somelatitude or personal space in enacting therole-based identity according to their ownneeds and preferences. It is the individualwho must decide how to delegate an assign-ment or how to present negative feedback to acoworker. Because no two individuals areidentical, no two role enactments are identi-cal, except in extremely strong situations(Ashforth, 2001). Thus, the person-based iden-tity shapes the enactment of the role-basedidentity (and vice versa; cf. Stets, 1995). Forinstance, Witt, Burke, Barrick, and Mount (2002)found that conscientiousness, coupled withagreeableness, predicted higher job perfor-mance in jobs requiring cooperative interac-tions than jobs requiring little to no coopera-tion.Our model suggests that relational identities

involve all three levels of self articulated byBrewer and Gardner (1996). The role-based com-ponent draws on the collective level, focusing onprototypical role occupants. The person-basedcomponent draws on the individual level, focus-ing on the more or less unique ways an individ-ual may enact a given role-relationship. And therelational identity, of course, draws on the inter-personal level.However, we contend that these role- and per-

son-based identities interact such that a rela-tional identity is not simply an additive functionof the role- and person-based identities. The sec-ond wing of role theory, symbolic interaction-ism, holds that the meaning(s) of rolesandtherefore role-relationshipsand how they areenacted are socially constructed through inter-action, observation, negotiation, feedback, andother well-known social processes (Blumer, 1969;Stryker & Statham, 1985; e.g., Hosking, Dachler,& Gergen, 1995; cf. Emirbayer, 1997; cf. Weick,1979). Symbolic interactionism also informs

identity theory (Stets & Burke, 2003; Stryker,1980), which adds that roles are a central anchorfor the social construction of self. Individualsare typically hired into organizations to occupyand enact certain roles, and so social construc-tions of identities tend to crystallize aroundroles.Although the structural functionalist and sym-

bolic interactionist wings of role theory are oftenviewed as independent and even antithetical,following Sarbin (1954), Stryker and Statham(1985), and others, we view them as quite com-plementary. Briefly, structural functionalismemphasizes how roles are created to fulfill insti-tutional needs, whereas symbolic interaction-ism emphasizes the agency of individuals insocially (re)constructing the meaning and enact-ment of those roles. These processes result in anongoing, reciprocal interaction between systemand individual, structure and process, contextand interaction, and macro and micro (cf. Gid-dens, 1984). For example, although the term su-pervisor can be found in the dictionary, its livedmeaning emerges from the shared experiencesand sensemaking of unique but situated andinterdependent individuals. Thus, the literatureon leader-member exchange theory suggeststhat managers tend to modify the way they con-ceptualize their supervisory role as a function ofhow they view a given subordinate and enactthe role-relationship (Sparrowe & Liden, 1997).Similarly, research on upward influence indi-cates that subordinates personal characteris-tics and perceptions of their manager and theirrelationship with him or her affect these subor-dinates choice of influence tactics (Farmer &Maslyn, 1999).Moreover, as the individuals in a dyad gain

experience in their respective roles and witheach other, the nature of their relationship islikely to change. The literature on trust, for in-stance, suggests that as coworkers mutual ex-pectations are met over time, familiarity andempathy tend to develop, leading to a more nu-anced and trusting relationship (Lewicki & Bunk-er, 1996). A coworker may become less inclinedto monitor anothers behavior and more inclinedto give him or her the benefit of the doubt ifproblems arise. Similarly, situational leader-ship theory argues that as subordinates maturein their roles and become more capable, man-agers should become less directive and moreparticipative (Hersey & Blanchard, 1988).


This notion of interaction among role- andperson-based identities in the service of anemergent relational identity is not to say thatthe individuals who constitute the dyad willnecessarily agree on the nature of their rela-tional identity. However, research in a variety ofsocial settings indicates that individuals gener-ally strive for agreement, enact their agreed-upon selves, and attempt to have those selvessocially validated through interaction (Ashforth,2001; Hinde, 1997; Swann, 1999). Moreover, re-search on personal relationships suggests thatcomplementary perceptions tend to emerge andpredict relationship satisfaction and stability(Hardin & Conley, 2001; Holmes, 2000). Thus, theparties in a role-relationship tend to come to amutual understanding of their relational iden-tity, and this shared meaning facilitates the co-ordinated interaction that is the hallmark of ef-fective role-relationships.Finally, Figure 1 also indicates that a rela-

tional identity may, in turn, affect its constituentrole- and person-based identities. For instance,research suggests that transformational leadersmay enhance the self-efficacy of their followersand cause followers to see their roles in morevalue-laden terms (e.g., Dvir, Eden, Avolio, &Shamir, 2002).In sum, the power of the relational identity

construct is that it offers a conceptual tool forintegrating the individual (person-based), inter-personal, and collective (role-based) levels ofself. As such, relational identity may help an-swer calls for integrative constructs that canbridge multiple levels and, thus, more richlyand holistically describe ones work experiencewithin the organization. (As a reviewer noted, itis not accurate to use the term levels of analysis,because the interpersonal and collective levelsrefer here to an individuals conception of his orher relational and role-based identities ratherthan to identities apart from the individual.Thus, we use the term levels of self.) For exam-ple, the integration of person- and role-basedidentities may help explain how individuals areable to cognitively maintain and actively enacta seeming welter of identities. In enacting therole-relationship of, say, an employee vis-a`-visa manager, one simultaneously enacts person-and role-based identities in an interpersonalcontext: the relational identity is essentially ho-listic (Ashforth & Johnson, 2001; Thoits & Vir-shup, 1997). Further, the cognitive associations

between the nested levels of self likely make iteasier to seamlessly shift between the identitiesassociated with the levels.Finally, the notion of integration may help

explain a theoretical controversy in the litera-ture on relational identity. Whereas Brewer andher colleagues (Brewer & Gardner, 1996; Brick-son & Brewer, 2001) argue that the dyad consti-tutes a unique stand-alone level of self, Hogg(2001b) considers the dyad to be an epiphenom-enon of the individual and collective levels,where complementary individuals link upagainst the backdrop of their collective identi-ty(ies). The notion of integration suggests thatindividual and collective levels of self interactsuch that the resulting relational identity ismore than the sum of its parts.This discussion suggests the following sum-

mary proposition.

Proposition 1: The respective role-based and person-based identities oftwo individuals in a role-relationshipwill interactively influence the rela-tional identity such that the relationalidentity is more than the sum of itsparts.

Relational Identity Hierarchy

Thus far, we have discussed relational iden-tity in the context of a relationship between twospecific individuals. That is, the relational iden-tity is highly particularistic (e.g., Susan the man-ager vis-a`-vis Bob the subordinate). In additionto having one or more particularistic relationalidentities regarding a given role (Susan vis-a`-vis each of her six subordinates), an individualmay abstract a more global or generalized rela-tional identity (how Susan sees herself as a su-pervisor of subordinates).The generalized relational identity is both in-

formed by and informs the particularized rela-tional identities. Based on research on intimaterelationships (Fletcher, Simpson, Thomas, &Giles, 1999; Hassebrauck, 1997), we speculatethat when a person is initially placed within anorganizational role, he or she has or soon formsa (possibly crude) prototype or generalized per-spective of a given relational identity. For exam-ple, Susan, a neophyte manager, may havesome expectations about what it means to su-pervise others, but little practical experience. As


she enters into specific role-relationships withothers, her generalized relational identity in-forms these particularized relational identities(e.g., what to expect, what to do). The abstractknowledge is translated into grounded action.As Susan gains experience with specific subor-dinates in specific contexts and learns the art ofsupervision, the particularized qualities of theserelationships may, in turn, inform the more gen-eralized relational identity (Pierce & Lydon,2001; Shaver, Collins, & Clark, 1996). As a resultof this fleshing out process, the generalized re-lational identity tends to become richer andmore nuanced with experience, and Susan willlikely approach subsequent particularized role-relationships with more skill and confidence.This discussion suggests the following sum-

mary proposition.

Proposition 2: The generalized rela-tional identity and particularized re-lational identity (or identities) con-cerning a given role-relationship (e.g.,manager-subordinate) will recipro-cally influence each other.

Of course, the fewer ones particularized rela-tionships, the greater the impact a given partic-ularized relationship will likely have on thegeneralized relationship. However, even if anindividual has had only one particularized rela-tionship, the generalized relational identity isunlikely to be identical to the particularized re-lational identity, because the latter is partlybased on information that may be seen as idio-syncratic to the relevant individual.Accordingly, individuals are usually able to

articulate not only a generalized relationalidentity (e.g., how Susan sees herself as a man-ager of subordinates) but also particularized re-lational identities at various levels of aggrega-tion (e.g., how Susan sees herself as a managerof first-shift subordinates or as a manager ofBob the subordinate). For example, research onpersonal relationships indicates that individu-als possess relational schemas that range fromgeneral beliefs concerning a role-relationship todetailed beliefs for specific relationships andinteractions (Baldwin, 1992; Fletcher & Fitness,1993; Hinde, 1997). Further, as the generalizedrelational identity becomes grounded in myriadparticularized experiences, it tends to becomemore stable and resistant to disconfirmation(Ashforth, 2001; Epstein, 1980). Although incon-

sistencies among the particularized role-rela-tionships may reduce the stability of the gener-alized relational identity, individuals tend tostrive for coherence in their generalized viewseven when particularized experiences varygreatly (for a review, see Van Rooy, Overwalle,Vanhoomisen, Labiouse, & French, 2003). For in-stance, self-defense mechanisms often amelio-rate the threat of disconfirming information togeneralized identities (Breakwell, 1986;Sedikides & Strube, 1997). A line worker whoacts contrary to her managers stereotype of lineworkers as lazy may be labeled an exception tothe rule.

Proposition 3: The greater the numberof particularized relational identitiesinvolving a given role-relationship,the more stable and resistant to dis-confirmation the generalized rela-tional identity will tend to be.

It should be noted that a generalized rela-tional identity is not the same as a collective-level (social) identity. Recall that the focus of thecollective level is on the individual as a proto-typical member of a social category, such asmanager. As discussed later when we comparerelational identity to social identity, the latterinvolves a depersonalized sense of self: whenthe social identity is salientthat is, situation-ally relevant and/or subjectively important (Ash-forth, 2001)one sees the individual as an inter-changeable exemplar of the group. In contrast, ageneralized relational identity remains in-formed by the individuals person-based identi-ties and so is necessarily personalized (Susanas a manager), and it focuses on a role-relationship (manager-subordinate) rather thanon the entire social category of manager.It should also be noted that a generalized re-

lational identity differs from the concept of roleas defined by the structural functionalist wingof role theory. Recall that structural functional-ists define roles as the expectations associatedwith a position. These expectations are deter-mined by the functional requirements of the sys-tem and, thus, have an institutionalized qualitythat transcends individual role occupants (Mer-ton, 1957; Sarbin & Allen, 1967). Conversely,again, a generalized relational identity is per-sonalized (and focuses on a role-relationshiprather than on the entire role). A generalizedrelational identity is more similar to the concept


of role as defined by the symbolic interactionistview of role theory, in that roles are said to besocially constructed, implicitly implicating indi-viduals in the process. However, symbolic inter-actionism focuses on social processes and doesnot explicitly examine the personal qualities ofthe individuals (Stryker & Statham, 1985). Fur-ther, a generalized relational identity, onceagain, focuses on a given role-relationshiprather than on the entire role.Finally, in theory, one can have a particular-

ized relational identity with every person withwhom one interacts. However, such a large num-ber of identities may become cognitively over-whelming. As a result, individuals are morelikely to maintain a moderate, and thus tenable,number of identities (e.g., Thoits, 1986). There-fore, we speculate that, in practice, one tends todevelop particularized relational identitieswhere a given role-relationship is salient. Forexample, a software developer is most likely todevelop particularized relational identities withhis supervisor, project members, and clients. Fora role-relationship that is not salient, ones per-ception of the other is likely to default to theothers collective-level identities (as per SIT)that is, to stereotypical impressions of the otherbased on his or her group memberships andsocial categories. This argument suggests thefollowing proposition.

Proposition 4: The more salient a spe-cific role-relationship is to an individ-ual, the more likely the individual willdevelop a particularized relationalidentity.

Relational Identification

The literature on social identification in orga-nizations focuses on how individuals partly de-fine themselves in terms of a collective, such asan occupation, workgroup, or organization(Haslam, 2001; Pratt, 1998). Similarly, as noted,we define relational identification as a (partial)definition of oneself in terms of a given role-relationshipwhat the relationship means tothe individual. For instance, an individual mayidentify with his or her role-relationship with acoworker because of the appealing role-basedidentity of mutual support and the coworkersdisplay of empathy and humor (person-basedidentity).

Note that relational identification differs fromwhat Kelman terms classical identification,where one identifies with another individualattempting to be like or actually to be the otherperson (1961: 63). The focus of classical identi-fication is the other person and, thus, involves asuppression of ones own individuality in favorof the other person when that person is salient.For example, Stone (1990) describes the cult ofpersonality that arose around Michael Milken atDrexel Burnham Lambert that enabled him toassemble a force of like-minded disciples and toengage in various unethical practices. AsKelman himself notes, identification with an in-dividual differs from identification with a re-ciprocal role relationship . . . in which the rolesof the two parties are defined with reference toone another (1961: 6364). The focus of rela-tional identification is the relationshipspecif-ically, the relational identity and the role- andperson-based identities that inform it.Relational identification involves what Aron

and Aron (2000; see also Aron & McLaughlin-Volpe, 2001) refer to as self-expansionan inclu-sion of the relationship in ones definition of selfand, thus, an extension of self. When the rela-tionship is salient, the individuals sense of selftranscends his or her personal qualities to in-clude another and what their association isthought to mean. Thus, rather than replacing hisor her extant identities, the individual broadenshis or her repertoire of identities to include therelationship. Psychologists generally regardthis extension of self as psychologically healthy(Aron & Aron, 2000; Josselson, 1992). However, aswe discuss later regarding overidentification,relational identification can be taken to an un-healthy extreme. Additionally, this extension ofself may quickly become unhealthy if it includesan inherently negative role-relationship, suchas an abusive supervisor-employee relationship(e.g., Tepper, 2000).Finally, just as relational identities can be

arrayed in a hierarchy from generalized to par-ticularized, so, too, can relational identifica-tions. A person may experience relational iden-tification as a generalized perceived onenesswith the role-relationship (e.g., with the co-worker relationship apart from any specific co-worker) and as a particularized perceived one-ness with a specific role-relationship (e.g., Peterthe coworker). And just as generalized and par-ticularized relational identities are mutually re-


inforcing, so, too, are relational identifications.An initial affinity for ones generalized notion ofcoworker (e.g., someone who is supportive) mayrender one amenable to identifying with a spe-cific role-relationship, and specific experienceswith particularized role-relationships may rein-force the generalized identification (assumingthe experiences are reasonably consistent withthe expectations; again, although inconsisten-cies may exist at the particularized level, indi-viduals will commonly seek and even imposecoherence at the generalized level).We can summarize this discussion with two

propositions that parallel our earlier discussionof the relational identity hierarchy.

Proposition 5: The magnitude of gen-eralized relational identification witha given role-relationship will recipro-cally influence the magnitude of par-ticularized relational identification(s)with that role-relationship.

Proposition 6: For a given role-rela-tionship, the greater the number ofparticularized role-relationships withwhich one identifies, the more stableand resistant to disconfirmation willthe magnitude of identification withthe generalized role-relationship tendto be.

In summary, a relational identity is ones def-inition of a role-relationship, whereas relationalidentification is the perceived oneness with therole-relationship. Relational identification in-volves a (usually) psychologically healthy ex-tension of self, resulting in a partial definition ofoneself in terms of the role-relationship. Bothrelational identity and relational identificationcan be generalized at the overall role-relation-ship level or particularized at the specific role-relationship level.

RELATIONAL IDENTITY AND IDENTIFICATIONVERSUS SOCIAL IDENTITY AND

IDENTIFICATION

Because much of the literature on identity andidentification in organizations is predicated onSIT (or its derivative, self-categorization theory),it is important to articulate the points of conver-gence and divergence between relational iden-tity and identification and social identity and

identification. Of course, the obvious differenceis that the target of relational identity and iden-tification is a role-relationship, whereas the tar-get of social identity and identification is a col-lective, such as an organization, or a socialcategory, such as gender. In this section we dis-cuss four social psychological concepts that fol-low from this distinction: personalization, inter-personal attraction, role transcendence, andgeneralization.

Personalization

SIT suggests that individuals interact basedon group prototypes rather than personal char-acteristics (Tajfel & Turner, 1986). Hogg andTerry define a group prototype as the cognitiverepresentation of features that describe and pre-scribe attributes of the group (2000: 123). Agroup prototype includes a perception of thegroup and how members should act. Two indi-viduals will tend to interact based on the proto-type of the most salient grouping, where sa-lience is influenced by contextual cues,historical precedent, purpose of the interaction,and so on (Ashforth & Humphrey, 1995; Hogg &Terry, 2000). In interacting based on a grouping,individuals depersonalize each otherthat is,they view the other as an interchangeable ex-emplar of [the] social category rather than as aunique person (Turner, Hogg, Oakes, Reicher, &Wetherell, 1987: 50).In contrast, our discussion of role-relation-

ships indicates that interactions between twoindividuals become personalized over time. Pet-tigrew (1986), commenting on the intergroup con-flict literature, argues that role-relationships re-quiring frequent contact become progressivelypersonalized, reducing stereotype-based inter-actions. Further, research, reviewed by Miller(2002), indicates that personalization tends to in-crease perceived similarity, interpersonalattraction, and positive affect. As a result, per-sonalization enhances the impact of the person-based identities on the nature of the role-relationship and, subsequently, on relationalidentity and identification. To be sure, groupprototypes are highly relevant to relationalidentities as they initially inform the partici-pants expectations of each others role-basedidentities. An engineer encountering a produc-tion supervisor for the first time will likely relyon prototypic expectations of production super-


visorswhat we referred to above as a general-ized role-based identityto inform his or herinitial behavior. However, these generalized ex-pectations will quickly give way to the particu-larized role- and person-based identities en-acted by the production supervisor.As this example indicates, the others role-

based identity provides the initial context forinterpreting his or her person-based identity.Organizations tend to be structured around dif-ferentiated roles, which are organized to accom-modate task interdependencies, and individualsare assigned to these roles. Accordingly, indi-viduals interact with one another in their capac-ity as role occupants (e.g., as coworkers, as man-ager and subordinate). Thus, an individualsperception of anothers person-based identity(how that person enacts the role-relationship) isheavily filtered through or conditioned by therole expectations on which the interaction ispredicated. For example, research on labelingprocesses in organizations suggests that a rolestereotype may continue to color interactionlong after a persons actual behavior shouldhave refuted the stereotype (Ashforth & Hum-phrey, 1995).In sum, what the notion of a relational identity

adds to the literature on social identity is anunderstanding of what happens between twoindividuals once interaction commences andthe group prototypes become softened by per-sonalization. SIT speaks only to interaction be-tween depersonalized and therefore relativelystatic entitiesentities that lack the color, nu-ances, and vagaries that define individuals andshape their interactions.

Interpersonal Attraction

Interpersonal attraction is the second elementthat distinguishes relational identity/identifica-tion from social identity/identification. Hogg(1992) has distinguished between being at-tracted to the qualities of the individual (inter-personal attraction) and the qualities of the col-lective of which the individual is a member(social attraction), and Markus and Kitayama(1991) have found that an interdependent viewof the selfwhich Brewer and Gardner (1996)suggest is analogous to the relational selfleads individuals to become interpersonally at-tracted to significant others. As one enacts therole-relationship, one becomes familiar with

specific role-relationship incumbents. Thus, asnoted, the generalized perspective of the role-relationship becomes particularized: one canput a face to the role. Given that most interde-pendent organizational roles require some coop-eration (Katz & Kahn, 1978), and given that hu-mans are social creatures and generally striveto be liked and to like others (Aronson, 1995),interpersonal attraction tends to occur. Addi-tionally, the value of resources generated fromthe role-relationship tends to influence onesevaluation of the relational identity (cf. Dutton &Heaphy, 2003; Kelley & Thibaut, 1978), predispos-ing one to feel an affinity for ones partner. Forinstance, a mentor-protege role-relationshipmay produce valued resources for the protege,in the form of psychosocial support and careeradvice, promoting a positive evaluation of therelational identity and liking for the mentor.Thus, interpersonal attraction tends to occur

as the relational identity becomes particular-ized and/or one associates ones partner withvalued resources. In sum, interpersonal attrac-tion is most closely related with relational iden-tity/identification, whereas social attraction ismost closely related with social identity/identification.

Role Transcendence

Because SIT focuses on depersonalized socialattraction, it provides no mechanism for friend-ships born of interpersonal attraction. Thus, ifone of the two individuals in a role-relationshipexits his or her role, the two will have no need ordesire for further interaction unless required bytheir new role-relationship. In short, the rela-tionship will not transcend their roles.Conversely, from the relational identity per-

spective, as a role-relationship becomes person-alized, interpersonal attraction tends to occur:the particularized role-relationship may becomea friendship such that the relationship can besaid to be multiplex (i.e., based on more thanone set of roles; e.g., Ashforth & Sluss, 2006;Valcour, 2002). Thus, the interpersonal relation-ship itself may create enough value for the in-dividuals to warrant continuance, even after oneof the individuals moves into another role oreven another organization (Markus & Kitayama,1991). For example, ex-coworkers may continueto interact (e.g., catching up over lunch, invitingeach other to parties), even when one has moved


to another organization. Additionally, the friend-ship ties may result, via job referrals, in a newcoworker-coworker role-relationship (in a neworganization) or a client-consultant role-rela-tionship. In other words, although the relationalidentity is situated within the respective roles,the relationship may transcend the bounds ofthe roles. To be sure, a social identity such asorganizational membership or gender can alsoprovide a basis for developing a friendship, butthe processes of personalization and interper-sonal attraction that friendship entails are be-yond the purview of SIT.

Generalization

Ashforth and Johnson (2001) argue that socialidentification with lower-order collectives (e.g.,work unit) may generalize to higher-order col-lectives (e.g., organization) because the latterare seen as vehicles for the former. By extension,we speculate that relational identification maygeneralize to identification with the salientgroups and social categories that the individualand the significant other share.How might this happen? First, in identifying

with a role-relationship, one may come to iden-tify with the collective that embodies and sus-tains the role-relationship. In effect, one sees thecollective as an extension of the role-relation-ship. For instance, Pratt (2000) describes howAmway distributors are encouraged to identifywith their sponsors (nominal supervisors) as ameans of identifying with the organization it-self. Additionally, in extending the self to in-clude the relationship and the others role, one islikely to develop a broader understanding of thewider organization (cf. Aron & Aron, 2000). Thus,relational identification may enhance ones un-derstanding and appreciation of the collective.Second, generalization occurs when one per-

ceives the other individual as representing orexemplifying the group or social category suchthat the relational identity and identificationspill over (Shamir, Zakay, Brainin, & Popper,2000: 615) onto the social identity and identifica-tion. For example, relational identification witha leader may strengthen ones identificationwith the relevant subunit and organization.Third, generalization may occur through inter-

personal attraction and personalization. As onecomes to know and value another individual,one is more likely to be influenced by the others

opinions. Thus, if the other person values thegroup or social category, one may also come tovalue it. Pratt (2000) also describes how Amwaydistributors are encouraged to socialize withtheir sponsors and often come to think of themas parents, faithfully replicating the sponsorspositive attitudes toward Amway.In sum, the process of generalization appears

to be common to both relational identity/identification and social identity/identification,whereas the processes of personalization, inter-personal attraction, and role transcendence dis-tinguish between the two.

Proposition 7: Relational identitiesand identifications are likely to be as-sociated with the processes of person-alization, interpersonal attraction,and role transcendence, whereas so-cial identities and identifications arelikely to be associated with the pro-cesses of depersonalization, social at-traction, and nontranscendence.

Proposition 8: Relational identifica-tion will tend to generalize to identifi-cation with the salient collectives thatare shared by the individual and hisor her partner.

RELATIONAL IDENTIFICATION: A TYPOLOGY

In this section we develop a 2 2 typology ofrelational identification based on the earlierdistinction between role- and person-basedidentities, complemented with the distinctiondiscussed below between positive and negativevalence. We also delineate functions and dys-functions of each type of relational identifica-tion.

Positive and Negative Valence

The valence of a role-based, person-based, orrelational identity is its perceived attractive-ness or desirability, where positive valence re-flects desirability and negative valence reflectsundesirability. Valence results from an individ-uals evaluation of an identity. As the individualenters a role-relationship, he or she is likely toevaluate the way in which the other person en-actsor is expected to enactthe relationship.This evaluation is not performed in the abstract,with the other as the sole referent; rather, it is


done with regard to how the other meshes withoneself. The focus, in other words, is on how theothers role and personal qualities bearing onthe role-relationship affect the nature of the re-lationship.The initial evaluation is likely to be influ-

enced by expectations stemming from the gen-eralized and particularized perspectives held bythe individual. For instance, based on stereo-typic (generalized) beliefs about coworkers, onemay expect to like another as yet unmet co-worker and, therefore, approach the relationshipwith a positive frame of mind. Research on la-beling theory and the Pygmalion effect suggeststhat such expectations may prove self-fulfillingas ones frame of mind influences ones initialbehavior toward the other and, thus, the othersreciprocal behavior (Ashforth & Humphrey, 1995;Eden, 1990). Additionally, as noted, the value ofresources accruing from the role-relationshiptends to affect ones evaluation of the relationalidentity (cf. Dutton & Heaphy, 2003; Kelley &Thibaut, 1978).The more positive the evaluation of the role-

relationship, the more likely one will identifywith it (Aron & Aron, 2000). As relational identi-fication grows, one tends to monitor the costs ofsocial exchanges less closely and to take plea-sure in contributing to the welfare of the role-

relationship because of the inclusion of the re-lationship in ones own self-concept (Blau, 1964;Brewer & Gardner, 1996).Complicating the issue, however, is the dis-

tinction between role- and person-based identi-ties: one may view the valence of the othersrole-based identity differently than his or herperson-based identity. (For the sake of simplic-ity, we assume that one views ones own role-and person-based identities bearing on the role-relationship in positive terms.) For instance, asubordinate may evaluate his or her managersrole-based identity positively (the managerprovides important resources for doing my job)but the person-based identity negatively (thismanager is verbally abusive). Evaluating thevalence of both the role- and person-based iden-tities results in qualitatively different types ofrelational identification.As shown in Figure 2, the type of the others

identity (role-based, person-based) and identityvalence (positive, negative) combine to form a2 2 typology of relational identification. Asdiscussed below, the typology indicates that in-dividuals tend to fall into one of three types ofidentification: relational identification, rela-tional disidentification, or ambivalent relationalidentification. This typology is loosely based onthe expanded model of organizational identifi-

FIGURE 2A Typology of Relational Identification


cation (Dukerich, Kramer, & McLean Parks,1998; Elsbach, 1999; Kreiner & Ashforth, 2004; cf.Pratt, 2000), except that (1) the drivers of thetypology are not high and low magnitude ofidentification and disidentification but the pos-itive and negative valences of role- and person-based identities, and (2) the referent is relationalrather than organizational identification. The ty-pology facilitates predictions regarding the na-ture and outcomes of various forms of role-relationships, a topic not broached by theexpanded model of organizational identifica-tion.It should be noted that an individual will tend

to have multiple and more or less separate eval-uations of relational identification at the partic-ularized level (i.e., a different assessment foreach specific incumbent in a given role-relation-ship), as well as a global evaluation of rela-tional identification at the generalized level (i.e.,an assessment of the role-relationship apartfrom specific individuals). As noted, because thelatter is often abstracted from many specificrole-relationships, it tends to be more stable.It should also be noted that we present the two

dimensions (i.e., valence of role-based identity,valence of person-based identity) as dichoto-mous for pedagogical purposes, allowing us todescribe each type of relational identificationdiscreetly (i.e., relational identification, rela-tional disidentification, and ambivalent rela-tional identification). However, we recognizethat these two dimensions are actually continuasuch that there are fine gradations among thethree types of relational identification.

Relational Identification

As Figure 2 shows, we argue that relationalidentification occurs when the others role- andperson-based identities have positive valence.As defined above, relational identification is theperceived oneness with the role-relationship,and it can be either generalized or particular-ized.Functions. Social identity theorists have ar-

gued that identification with a group or socialcategory serves various functions, particularly(1) uncertainty reduction, by situating oneself insocial space, and (2) self-enhancement (amongother self-related motives), by becoming onewith a larger and perhaps distinctive and pres-tigious social entity (e.g., Dutton et al., 1994;

Hogg, 2001a). It seems likely that relational iden-tification similarly provides uncertainty reduc-tion via internalizing a role-relationship with itsattendant goals, norms, and so on, as well asself-enhancement insofar as the role-relation-ship is regarded as distinctive and prestigious.A key additional function of relational identi-

fication that has been largely overlooked by so-cial identity scholars (see Cheney, 1983, andPratt, 1998, for exceptions) is interpersonal con-nection and belongingness. Scholars have pro-posed that individuals have a fundamentalneed to identify with other individuals or socialentities, variously termed a need for identifica-tion (Glynn, 1998), self-expansion motive (Aron &Aron, 2000), need to belong (Baumeister & Leary,1995), desire for intimacy and interdependence(Brewer & Roccas, 2001), need for relatedness(Deci & Ryan, 1991), and so forth. What theseneeds suggest is that individuals are motivatedto identify with others as a means of attaining ahuman connection (in addition to the uncer-tainty reduction and self-enhancement laudedby SIT).In relational identification, as noted, the self

is expanded to include those facets of the com-plementary role and its incumbent(s) that bearon the role-relationship. Aron and McLaughlin-Volpe (2001; see also Cross, Morris, & Gore, 2002;Fiske & Haslam, 1996; Gardner et al., 2002) doc-ument associations between self-expansion andtendencies to discriminate less between oneselfand ones relationship partner when allocatingresources, to adopt the perspective of ones part-ner, to confuse what one thought or did withwhat ones partner thought or did, to confuseones traits with ones partners traits, and to feelbadly when ones partner does poorly (con-versely, one may bask in their reflected glory[Cialdini et al., 1976], feeling pride in a partnersaccomplishments). Such tendencies, in turn,may increase the likelihood the relational iden-tification will be reciprocated. Moreover, thecloser the relationship, the stronger these ten-dencies tend to be (Aron & Fraley, 1999). Indeed,given the expansion of self, helping ones rela-tional partner is tantamount to helping oneself;as Cialdini, Brown, Lewis, Luce, and Neubergsresearch demonstrates, It is the commonality,not the compassion, that generates helping(1997: 491).The upshot of these tendencies is that rela-

tional identification is likely to foster an array of


more distal interpersonal benefits, includingempathy, mutual understanding, loyalty, coop-eration, social support, altruism, and inrole per-formance (i.e., work behaviors that are pre-scribed by formal job roles; Hui, Law, & Chen,1999: 4). However, at extremely high degrees ofrelational identificationtermed overidentifica-tion belowinrole performance beyond thedyad may suffer if other important role-relation-ships are thereby starved of support.Kahn (1998) and Flum (2001b) add that, in the

context of organizational life, the interpersonalbenefits noted above are essential for psycho-logical growth. Relationships anchor the indi-vidual and help provide the confidence for ex-ploration (cf. Bowlby, 1969). Kark and Shamirnote that it is commonly believed that followergrowth and empowerment imply a greater fol-lower independence (2002: 84) from the leader,when, in fact, empowerment may flow fromtighter supportive and developmental relationalties with the leader. For example, the number ofindividuals managed by Southwest Airlinesfrontline supervisors is less than at other majorcarriers precisely because Southwest seeks tofoster positive subordinate-supervisor relation-ships and a sense of empowerment through ac-tive and ongoing coaching (Gittell, 2003). Inshort, autonomy and individuality, far from be-ing antagonistic to personal connections, actu-ally grow from them; as Ryan puts it, Individu-ation is . . . [not] something that happens fromothers but rather with them (1991: 223; see alsoFletcher, 1999, and Josselson, 1992).Dysfunctions. Conversely, as some scholars

have noted in the case of social identification(e.g., Ashforth & Mael, 1992; Dukerich et al., 1998),relational identification carries the risk of over-identification. That is, one may identify sostrongly with the role-relationship that either (1)the relational identity remains chronically acti-vated, even when it is not appropriate (e.g., acoworker who talks shop incessantly during abaseball game), or (2) when the relational iden-tity is relevant, other relevant aspects of onesrole(s) or individuality are suppressed (e.g., Katz& Genevay, 2002). In a real sense, one may loseoneself in the relationship, elevating it aboveother important considerations. For instance,Sias, Heath, Perry, Silva, and Fix (2004) describea supervisor who, to maintain a friendship witha subordinate, offered to change the subordi-nates performance evaluation.

We speculate that overidentification is morelikely in role-relationships of unequal power,such as supplier-client and manager-subordi-nate, given the propensity of those with lesspower to defer to those with more power. Notethat overidentification within relationships ofunequal power may be problematic for both par-ties. For example, those with less power may bemore susceptible to maltreatment (e.g., Tepper,2000), whereas those with more power may re-ceive less critical evaluation from others, result-ing in lower-quality decisions (e.g., Jehn & Shah,1997). Further, overidentification is more likelywhere one is less certain (than ones partner)about role expectations and ones own role ca-pabilities. For example, newcomers to a roleoften defer to their more experienced peers andsupervisors for cues about not only what to dobut what to think and even feel (Ashforth, 2001;Ibarra, 1999).Overidentification with a particularized role-

relationship may also be harmful to other role-relationships within the individuals network.Research on personal relationships shows thatindividuals in close relationships tend to exag-gerate the person-based differences betweentheir current partner and alternative partners(e.g., Buunk & Van Yperen, 1991; Johnson & Rus-bult, 1989). Hinde argues that this devaluing ofalternative partners helps to develop and en-hance positive views of oneself, ones partner,and ones relationship (1997: 466), thus bolster-ing relational identification. Although bolster-ing certainly has its benefits, devaluing alterna-tive partners may undermine other importantrole-relationships, given that organizations arebuilt on networks of interdependent roles.Finally, although not addressed by social

identity theorists (to our knowledge), overidenti-ficationin the relational sensemay tip intocodependency. Springer, Britt, and Schlenkerhave defined codependency as a dependenceon anothers approval . . . designed to find asense of safety, identity, and self-worth (1998:141; see also Allcorn, 1992). The notion of code-pendency assumes that the partner exhibitschronic and harmful behaviors toward the indi-vidual. Paradoxically, codependency involves aneed to control the partner to ensure that self-worth continues to be derived from the relation-ship (Le Poire, Hallett, & Giles, 1998).Ones desire for approval leads to the subju-

gation of ones other identities to the relational


identity. In other words, the relational identitybecomes a surrogate for who one is, at least inthat setting. As a result, (1) ones other role- andperson-based identities become suppressed orforgotten, and (2) one internalizes portions ofthe others role- and person-based identitiesthat are not relevant to the role-relationship(e.g., an engineer internalizes a managers dis-dain for the marketing department, biasing theengineers own interactions with marketing).Because those involved in a particularized

role-relationship socially construct its livedmeaning, the resulting relational identity can beviewed as a shared reality (Hardin & Conley,2001). And because this shared reality is local-ized (i.e., specific to the partners), it may be onlyloosely coupled with the broader social con-structions that typify the organization. Thus,codependency may result in a pathological re-lationship where role-relevant information isconstrued in a way contrary to organizational orsocial norms. Examples of codependency in or-ganizational contexts include forms of abusivesupervision where the targets essentially col-lude in their victimization (Burris, 1999; Tepper,2000) and forms of corruption where individualsbecome willing accomplices and view corruptpractices as normal and acceptable (Ashforth &Anand, 2003).Our discussion can be summarized as follows.

Proposition 9: The greater ones rela-tional identification, (a) the more em-pathy, understanding, and loyalty onewill have regarding ones partner andthe more cooperation, support, and al-truism one will display toward onespartner, and (b) the greater ones in-role performance will be, providedother important role-relationships arenot denied support.

However:

Proposition 10: The less power one hasvis-a`-vis ones partner and/or the moreuncertain one is (relative to ones part-ner) about role expectations and onesown capabilities, the greater the like-lihood one will overidentify with therole-relationship and become code-pendent.

We should note, however, two countervailingtendencies to overidentification and codepen-

dence. First, optimal distinctiveness theoryholds that individuals strive to balance the ten-sion between assimilation in and separationfrom a relationship or group such that strongassimilationoveridentificationfosters acompeting desire for separation (Brewer, 1991;Brewer & Roccas, 2001; see also Mashek & Sher-man, 2004). Thus, for most individuals and situ-ations, there are natural checks and balances onthe magnitude of identification. Second, our ear-lier discussion of personalization suggests that,unlike social identification, relational identifi-cation is a meld of role- and person-based iden-tities. It is the resonance between ones self-in-role and other-in-role that fosters identificationsuch that the suppression of oneself or ones roleapart from the relationship is not likelyalthough nonetheless possible.

Relational Disidentification

Just as positive valence encourages one toidentify with a role-relationship, negative va-lence encourages one to disidentify with a rela-tionship (cf. Elsbach & Bhattacharya, 2001), andjust as identification facilitates self-definition(I am), so, too, does disidentification (Im not).For example, Levy (1996, in Elsbach, 1999) reportshow Scott McNealy, CEO of Sun Microsystems,described his very existence as an Anti-Bill(Bill Gates, then-CEO of Microsoft). This strongrelational disidentification likely impassionedMcNealy to continue in an aggressive andfiercely competitive battle for market share.Relational disidentification occurs when the

others role- and person-based identities havenegative valence. For instance, a manager mayrelationally disidentify with an internal auditorbecause of a negative role-based identity (au-diting slows down my work) and a negativeperson-based identity (this auditor is rude).Given the desire for cognitive consistency or

balance (Festinger, 1957; Heider, 1958), a nega-tively valent role-based identity may bias per-ceptions of the person. Because role-relation-ships are so consequential for individuals andorganizations alike, individuals often perceiveothers through the prism of their role-relation-ships. As Holmes states, Mental representa-tions of others are not simply organized aroundperson constructs, as has traditionally been as-sumed [in social cognition research], but in-stead, may reflect the type of relationship that


exists with the perceiver (2000: 473; see alsoFiske & Haslam, 1996). The nature of the role-relationship may be projected onto the individ-ual (e.g., a drill sergeant is seen as a nastyperson; Ashforth, 2001; Humphrey, 1985). Thus,individuals are predisposed to dislikeas peo-plethose who occupy roles that are perceivedas oppositional in some way to their own.Similarly, the valence of the person-based

identity may color the valence of the role-basedidentity. Kahn (1998) describes a social serviceagency where the director was perceived by thesocial workers as cold and manipulative suchthat an otherwise positive role-based identity(directors are supportive) came to be seen innegative terms (directors are unsupportive).However, given the greater stability of general-ized relational identities, an individual may per-ceive a particularized relational identity in neg-ative terms and yet retain a positive perceptionat the generalized level.Functions. Although the literature on social

disidentification in organizational contexts (pri-marily organizational disidentification) issparse, it does imply that disidentification mayalso address needs for uncertainty reduction (Iam not like X) and self-enhancement (I ambetter than X; Elsbach, 1999; Elsbach & Bhatta-charya, 2001). These functions similarly apply torelational disidentification. (However, we do notsee a clear analog to the third key function ofrelational identification noted above, interper-sonal connection; as discussed below, discon-nection is likely to be quite dysfunctional inmost organizational contexts.) In disidentifyingwith the others negatively valent role- and per-son-based identities, one effectively states, Imnot she or Im not he, and, possibly, Im betterthan she/he. Thus, relational disidentificationprovides a valuable foil for clarifying the bound-aries between who or what one is (or would liketo be) and who or what one is not (or would notlike to be). These functions are particularly im-portant for newcomers, who often find it easierto articulate what they do not like than whatthey do like (Elsbach, 1999). Indeed, as Ibarra(1999) found for neophyte consultants and in-vestment bankers, newcomers may test variousrole- and person-based identities and decide,via relational identification and disidentifica-tion, which ones best suit them.Moving beyond the literature on social disi-

dentification, relational disidentification may

also be more or less functional for roles that arestructured to be adversarial, akin to the court-room roles of prosecutor and defense attorney.Examples include groups that institute a gadflyrole to prevent groupthink (Janis, 1983) and man-agers who are encouraged to compete on behalfof their subunits for scarce resources. Individu-als may be encouraged to disidentify, at leasttemporarily, to sharpen the differences betweenthe adversaries, thereby allowing for a full (ifpolar) airing of the issues.Dysfunctions. That said, just as organizational

disidentification has been argued to be highlydivisive (Dukerich et al., 1998), relational disi-dentification is likely to be very dysfunctional inrole-relationships predicated on ongoing coop-erationthe vast majority of relationships inorganizations. If relational identification facili-tates mutual understanding, loyalty, coopera-tion, and altruism, disidentification facilitatesthe opposite. Moreover, negative expectationsmay prove self-fulfilling, giving rise to negativebehaviors that are likely to be reciprocated.Kahn (1998) describes the self-fulfilling prophe-cies that occurred in a social service agencywhen the director and social workers formedpejorative perceptions of each others role- andperson-based identities and, subsequently,withdrew from each other. The outcome of suchprocesses is often a downward spiral, where theunpleasant experience of conflict may soonovershadow the initial reasons for the conflict,inhibiting resolution (Andersson & Pearson,1999; Dukerich et al., 1998). Disidentification isnot a viable basis for a cooperative relationshipand, thus, is likely to provoke the exit of at leastone party.The discussion can be summarized in the fol-

lowing propositions.

Proposition 11: The greater ones rela-tional disidentification, the less empa-thy, understanding, and loyalty onewill tend to have regarding ones part-ner and the less cooperation, support,and altruism one will tend to displaytoward ones partner.

Proposition 12: Valence tends to beself-fulfilling; that is, the valence of aperson-based, role-based, or rela-tional identity may influence (a) thevalence of the other two and (b) sub-sequent behavior vis-a`-vis ones part-


ner such that (c) the partner recipro-cates both the valence and behavior.

Note that inrole performance is not includedin Proposition 11. The impact of disidentificationon inrole performance depends on the purposeof the role-relationship. For example, for rolespredicated on adversarial relationships, disi-dentification may at times actually promote in-role performance.

Ambivalent Relational Identification

The remaining two cells of Figure 2 aretermed ambivalent relational identification. Ac-cording to the expanded model of organization-al identification (Dukerich et al., 1998; Elsbach,1999; Kreiner & Ashforth, 2004), ambivalent iden-tification refers to simultaneously identifyingand disidentifying with a target. Ambivalent re-lational identification thus refers to the state inwhich the valences of the others role- and per-son-based identities are mixed, whether at thegeneralized (e.g., clients) or particularized (e.g.,Ms. Smith) levels of the relational hierarchy. Forexample, a subordinate may ambivalently iden-tify with the manager-subordinate role-relation-ship because of (1) evaluating the managersperson-based identity as positive (e.g., thismanager is respectful of who I am as an engi-neer) and the role-based identity as negative(e.g., I dont like having a nontechnical man-ager evaluate my contributions), (2) evaluatingthe person-based identity as negative (e.g., thismanager is belittling) and the role-based iden-tity as positive (e.g., I like having a managerbuffer client complaints, allowing me to focus onbeing an analyst), or (3) evaluating certain el-ements of the person- and/or role-based identi-ties as positive and others as negative (notshown in Figure 2).Ambivalent relational identification is likely

a common form of relational identification. Role-and person-based identities are multifaceted,and an individual is unlikely to perceive all ofthese facets as either positive or negative. Thus,mixed feelings may characterize many role-relationships. Research on personal relation-ships (Thompson & Holmes, 1996) and organiza-tion-based relationships (Kondo, 1990) suggeststhat individuals can indefinitely maintain astate of (moderate) ambivalence: Indeed, amoderate level of ambivalence may actually in-

dicate a balanced, realistic assessment of apartner (Thompson & Holmes, 1996: 502).However, the more salient a role-relationship

is to the individual, and the more pronouncedthe differences in valence between importantidentity facets, the greater the dissonance thatwill be arousedand the greater the disso-nance, the greater the motivation to resolve theambivalence (Festinger, 1957). Following Laza-rus and Folkman (1984), resolution efforts mayvary from problem focused to emotion or symp-tom focused. For example, problem-focused ef-forts may involve realigning the identities, suchas through the renegotiation of role expecta-tions (Schlenker, Britt, & Pennington, 1996;Swann, 1999), whereas symptom-focused effortsmay involve deferring to the role- or person-based identity that is most salient, vacillatingbetween the identities if they are equally sa-lient, avoiding problematic issues, and embel-lish[ing] a partners virtues and minimiz[ing]faults (Murray, Holmes, & Griffin, 1996: 80; Pratt& Doucet, 2000). In the absence of effective cop-ing, it may become very difficult to endure arelationship marked by strong dissonance.Functions. Because ambivalence represents

mixed valences regarding a given role-relation-ship, its impact on uncertainty reduction, self-enhancement, and interpersonal connection arelikely to be muted. For example, the mixedmessage of ambivalence creates uncertaintyabout whether to approach or avoid the relation-ship. However, ambivalence may nonethelessserve an important function. Given the potentialfor overidentification and the damage that disi-dentification can do to a relationship, ambiva-lent relational identification may at times behealthy for the individualand the relationship.Ambivalence may hold at bay an overwhelmingrelational identity, thereby allowing cognitiveand emotional space for ones personal- androle-based identities to survive (Ashforth &Mael, 1998; Fleming & Spicer, 2003). For instance,the head physicianmedical intern relationshipmay be very important to an intern because ofits impact on her medical career. Ambivalencemay enable the intern to maintain perspec-tivea certain ironic distance (Ashforth, 2001:81)when subjected to strong feedback,whether positive or negative. Moreover, much asconflict may promote constructive change, am-bivalence might signal divisive issues and trig-ger a constructive renegotiation of the role-


relationship (Thompson & Holmes, 1996); whilethe negatively viewed aspects motivate changeof some kind, the positively viewed aspects mo-tivate the parties to make changes that ulti-mately support the role-relationship.Dysfunctions. Conversely, based on our ear-

lier discussion, we speculate that ambivalenceis likely to be dysfunctional if it provokes strongdissonance that inhibits individuals from fullypartaking in important role-relationships. Forexample, an individual may wish to be fullyengaged with his or her coworker but may beput off by a personality clash. We suspect thatsuch dysfunctional effects will be more pro-nounced when the ambivalence is fueled bynegative person-based rather than role-basedidentities. When the negativity is attributed tothe character or intent of the person, one be-comes less inclined to display empathy, mutualsupport, loyalty, and other behaviors that sug-gest positive regard for the person in question(cf. Feather, 1999). Conversely, when the nega-tivity is attributed to impersonal role demandsthat impel the person to act as he or she does,one may continue to feel some personalized res-onance if not sympathy for the role incumbent.An experiment by Shaban and Welling (in

Glass & Singer, 1972) provides an illustration.Subjects were required to redo a long form, os-tensibly before engaging in the actual experi-ment. Subjects were induced to blame the redoon either the fastidiousness of the clerk (person-based identity) or on strict regulations that theclerk was required to enforce (role-based iden-tity). Subjects in the former condition were morelikely to cheat on a subsequent laboratory taskand to be less acquiescing in a bargaininggame. It should be remembered, however, thatbecause of the earlier-noted tendency to projecta persons role onto the person (and vice versa),the valences of role- and person-based identi-ties may blur somewhat.This discussion suggests our final set of prop-

ositions.

Proposition 13: The more salient arole-relationship is, and the strongerthe differences in valence betweenimportant identity facets, the morelikely ambivalent relational identifi-cation will foster problem-focusedand/or symptom-focused coping tac-tics.

Proposition 14: Ambivalent relationalidentification is more likely to impaira role-relationship if the ambivalenceis attributed to a negative person-based identity rather than to a nega-tive role-based identity.

Proposition 15: Ambivalent relationalidentification (a) reduces the likeli-hood of overidentification and (b) maytrigger constructive conflict resolu-tion.

In summary, relational identification occurs ifthe valence of the partners role- and person-based identities is positive, relational disiden-tification occurs if the valence is negative, andambivalent relational identification occurs ifthe valence is mixed. The form of identificationhas profound consequences for the nature of therelationship.

DISCUSSION

We have argued that a focus on the role-relationship between two individuals (e.g., co-workers, manager-subordinate) provides uniqueinsights into workplace identity and identifica-tion, beyond the conventional focus on individ-uals and collectives, that relational identitiesintegrate person-based (individual) and role-based (collective) identities, and that integrat-ing the personal relationship literature with theidentity literature provides a holistic under-standing of how one experiences work. Whereasthe dominant perspective on identity and iden-tification in organizationsSITsuggests thatrelationships are depersonalized and based onmutual attraction to the collective, a relationalperspective suggests that relationships are of-ten personalized and based on interpersonal at-traction. We have argued that relational identi-ties and identifications exist in a cognitivehierarchy ranging from generalized schemas(e.g., manager-subordinate) to particularizedschemas (e.g., Susan the managerBob the sub-ordinate). We also have argued that role-relationships can spawn not only relationalidentification but relational disidentificationand ambivalent relational identification, andwe have discussed the functions and dysfunc-tions of each for the individual and organiza-tion.


Our analysis suggests at least four directionsfor future research. First, our contention thatrelational identities integrate the relevant per-son- and role-based identities of self and otherwarrants examination. While measures of roleexpectations can approximate the construct ofrole-based identity, a variety of individual-difference measures should be considered asproxies for person-based identity (e.g., self-monitoring, agreeableness, conscientiousness,gender). However, the construct of relationalidentity presents a measurement challenge be-cause it implicates the individual and, thus, isidiosyncratic, whether generalized (Susan vis-a`-vis her subordinates) or particularized (Susanvis-a`-vis Bob).There are at least three increasingly complex

possibilities for measurement (with the last be-ing most consistent with the construct): (1) treatrelational identity as an unmeasured blackbox variable, and examine how role- and per-son-based identities interact to affect variousoutcomes (such as relational identification andperformance); (2) treat relational identity as anomothetic or etic construct, and examine com-mon ways in which a given role-relationship(e.g., manager-subordinate) is enacted acrossdyads; and (3) treat relational identity as anideographic or emic construct, and examine howindividuals uniquely enact their respective role-relationships. In any event, given our relatedcontention that role- and person-based identi-ties interact over time in possibly unpredictableways, it is important that researchers use meth-odologies that allow for dynamic interactionism(Hattrup & Jackson, 1996)that is, that allow thevariables to mutually influence one another(both within-self and between partners) as wellas the unfolding relational identity (see also thesocial relations model; Kenny, Mohr, &Levesque, 2001). An example would be a longi-tudinal study of mentor-protege relationshipsusing data collection methods that provide thickdescriptions (e.g., surveys, interviews, observa-tions, diaries).A second direction for research is the typology

of relational identification shown in Figure 2. Thetypes of identification can be measured by chang-ing the referent of existing measures of social/or-ganizational identification (see Haslam, 2001, for areview) and of organizational disidentificationand ambivalent identification (Kreiner & Ash-forth, 2004) to the role-relationship. To what ex-

tent do the valences of role- and person-basedidentities predict the type of identification?What are the individual, relational, and orga-nizational antecedents and consequences ofeach type of identification?An example of a promising antecedent that

researchers might study are traitlike predispo-sitions to define oneself at least partly throughclose relationships, as measured by the Inclu-sion of Other in the Self Scale (Aron, Aron, &Smollan, 1992), the Relational-InterdependentSelf-Construal Scale (Cross, Bacon, & Morris,2000), or the Relational, Individual, and Collec-tive Self-Aspects Scale (Kashima & Hardie,2000). Examples of promising consequences ofrelational identification are motivation, self-regulation, and self-evaluation. For instance, inidentifying with a role-relationship, one will in-ternalize the performance standards (whetherhigh or low) and norms that define that relation-ship and evaluate oneself accordingly (cf. rela-tional schemas; Baldwin, 1992, 1994); a perfec-tionist mentor may impart unrealistic standardsthat become entrenched in ones generalizedmanager-subordinate relational identity, lead-ing to a career of personal frustration. Also, wedid not consider shifts between the types of re-lational identification. Researchers shouldexamine factors that may tip, say, relationalidentification into ambivalence or even disiden-tification (e.g., poor performance, trust viola-tions).Third, the concept of a cognitive hierarchy

raises intriguing research issues. How exactlydo hierarchies form as individuals enter newroles and role-relationships? Under what cir-cumstances do generalized relational identi-ties/identifications have a strong influence onparticularized relational identities/identifica-tionsand vice versa? How are particularizedrelational identities/identifications aggre-gated to influence generalized identities/identifications? For example, perhaps highlysalient particularized relational identities ex-ercise disproportionate influence (e.g., allow-ing ones protege to strongly shape expecta-tions of all subordinates). To what extent arerelational identities/identifications compart-mentalized both within levels (e.g., particular-ized) and between levels (generalized-partic-ularized)? How do particularization andgeneralization influence the newcomer adjust-ment process? For instance, particularization


may facilitate adjustment by providing di-verse relational models for emulation (e.g.,Ibarra, 1999), and generalization may facilitateadjustment by enabling a newcomer to place aparticularized relationship in perspective(e.g., realizing that a partners authoritarian-ism is not typical of partners).Fourth, given that organizations are networks

of complementary roles, individuals usuallyhave multiple relational identities (e.g., with co-workers, subordinates, clients). Are there limitsto how many role-relationships individuals canidentify with? Are there synergies among therole-relationships? For example, perhaps aproblematic relationship with a coworker maysensitize one to subordinates experiencing sim-ilar problems. Social network theory (e.g.,Granovetter, 1985) may shed light on the dynam-ics of multiple relational identities and identifi-cations. For instance, network centrality mayhelp delineate how central role-relationships in-fluence the initiation and maintenance of pe-ripheral role-relationships within the individu-als social network. And based on the concept oftransference, Andersen and her colleagues(Andersen & Berenson, 2001; Andersen & Chen,2002) argue that role-relationships with signifi-cant others create expectations that strongly in-fluence new relationships, adding a dynamicelement to multiple role-relationships. To whatextent do individuals transfer relational identi-ties inappropriately (e.g., treating subordinatesas if they are ones children), and how mightindividuals balance the benefits of transferencewith those of openness to learning (Ashforth &Sluss, 2006)?In conclusion, the constructs of relational

identity and relational identification offer in-sights into generalized and particularized role-relationships that are largely overlooked by ex-tant notions of individual identity and collectiveidentity and identification. As such, relationalidentity and relational identification offer afruitful perspective for understanding how indi-viduals define themselves within organization-al contexts.

REFERENCES

Albert, S., & Whetten, D. A. 1985. Organizational identity.Research in Organizational Behavior, 7: 263295.

Allcorn, S. 1992. Codependency in the workplace: A guide for

employee assistance and human resource professionals.Westport, CT: Quorum Books.

Andersen, S. M., & Berenson, K. R. 2001. Perceiving, feeling,and wanting: Experiencing prior relationships inpresent-day interpersonal relations. In J. P. Forgas, K. D.Williams, & L. Wheeler (Eds.), The social mind: Cognitiveand motivational aspects of interpersonal behavior: 231256. Cambridge: Cambridge University Press.

Andersen, S. M., & Chen, S. 2002. The relational self: Aninterpersonal social-cognitive theory. Psychological Re-view, 109: 619645.

Andersson, L. M., & Pearson, C. M. 1999. Tit for tat? Thespiraling effect of incivility in the workplace. Academyof Management Review, 24: 452471.

Aron, A., & Aron, E. 2000. Self-expansion motivation andincluding other in the self. In W. Ickes & S. Duck (Eds.),The social psychology of personal relationships: 109128.Chichester, UK: Wiley.

Aron, A., Aron, E. N., & Smollan, D. 1992. Inclusion of other inthe self scale and the structure of interpersonal close-ness. Journal of Personality and Social Psychology, 63:596612.

Aron, A., & Fraley, B. 1999. Relationship closeness as includ-ing other in the self: Cognitive underpinnings and mea-sures. Social Cognition, 17: 140160.

Aron, A., & McLaughlin-Volpe, T. 2001. Including others in theself: Extensions to own and partners group member-ship. In C. Sedikides & M. B. Brewer (Eds.), Individualself, relational self, collective self: 89108. Philadelphia:Psychology Press.

Aronson, E. 1995. The social animal (7th ed.). New York: Free-man.

Ashforth, B. E. 2001. Role transitions in organizational life: Anidentity-based perspective. Mahwah, NJ: Lawrence Erl-baum Associates.

Ashforth, B. E., & Anand, V. 2003. The normalization of cor-ruption in organizations. Research in Organizational Be-havior, 25: 152.

Ashforth, B. E., & Humphrey, R. H. 1995. Labeling processes inthe organization: Constructing the individual. Researchin Organizational Behavior, 17: 413461.

Ashforth, B. E., & Johnson, S. A. 2001. Which hat to wear? Therelative salience of multiple identities in organizationalcontexts. In M. A. Hogg & D. J. Terry (Eds.), Social identityprocesses in organizational contexts: 3148. Philadel-phia: Psychology Press.

Ashforth, B. E., & Mael, F. A. 1992. The dark side of organiza-tional identification. Paper presented at the annualmeeting of the Academy of Management, Las Vegas.

Ashforth, B. E., & Mael, F. A. 1996. Organizational identityand strategy as a context for the individual. Advances inStrategic Management, 13: 1964.

Ashforth, B. E., & Mael, F. A. 1998. The power of resistance:Sustaining valued identities. In R. M. Kramer & M. A.Neale (Eds.), Power and influence in organizations: 89119. Thousand Oaks, CA: Sage.

Ashforth, B. E., & Sluss, D. M. 2006. Relational identities in


Documents

Article 18. Relational Identity and Identification.pdf