Us Versus Them: The Roles of OrganizationalIdentification and DisidentiHcation in Social MarketingInitiatives

C.B. Bhattacharya and Kimberly D. Elsbach

Whereas organizational identification is defined as a cognitive connection between a personand an organization, disidentification is defined as a sense of separateness. The authorsconducted a mail survey to compare the attitudes and behaviors of people who identify ordisidentify with the National Rifle Association or view it in a neutral fashion. The resultsshow that whereas identification is related to people's personal experiences,disidentification is related to their values surrounding the organization. Moreover, althoughboth identifiers and disidentifiers talk, only identifiers take action.

In the organizational behavior literature, organizationalidentification is broadly defined as a cognitive connec-tion between a person and an organization (Dutton,

Dukerich, and Harquail 1994; Mael and Ashforth 1992),whereas disidentification is analogously defined as a senseof separateness (Elsbach and Bhattacharya 2001). To take afew real-world examples of these phenomena, through itsinnovative social responsibility practices, Ben and Jerry'sattempts to foster a sense of identification among many con-sumers (Bhattacharya, Rao, and Glynn 1995). Conversely,stakeholders' negative reactions to Exxon after the oil spill(Dutton, Dukerich, and Harquail 1994) or toward Nike afterallegations of sweatshop operations (McCall 1998) suggestthat people also disidentify with organizations. As the fol-lowing examples suggest, social marketers also attempt tobuild identification and/or disidentification as part of theirstrategies:

•In a campaign against tobacco giant Philip Morris, the Califor-nia Anti-tobacco Coalition has depicted the "Marlboro Man" inbillboards with the caption, "Bob, I've got emphysema."Through their television advertising and Web site links, organi-zations such as getoutraged.com are similarly trying to promoteseparation from Philip Morris. These examples illustrate howorganizations such as the California Anti-tobacco Coalition andgetoutraged.com are trying to influence consumers to disiden-tify with Philip Morris.

•People for the Ethical Treatment of Animals (PETA) has sev-eral initiatives in place (e.g., membership program, newsletter,online mall) to foster a sense of identification with the organi-zation. Simultaneously, PETA is also urging consumers todisidentify with The Gap for using questionable methods to

C.B. BHATTACHARYA is Associate Professor of Marketing, BostonUniversity. KIMBERLY D. ELSBACH is Associate Professor of Man-agement, Graduate School of Management, University of Califor-nia, Davis. The authors thank Sythan Voun, Ron Harris, Haya-greeva Rao, Paul Berger, and tbe three anonymous JPP&Mreviewers for their help with this article. A previous version of thiswork received the William Novelli Best Paper Award at the 1997Innovations in Social Marketing Conference.

import leather from India. This example illustrates how PETAis trying to induce consumers to identify with it on the one handand disidentify with The Gap on the other hand.

These examples are by no means exhaustive, but theyhelp illustrate the need for social marketers to better under-stand how both identification and disidentification work.Specifically, we use the term "social marketing" in this arti-cle to mean "the design, implementation, and control of pro-grams seeking to increase the acceptability of a social ideaor practice in a target group(s)" (Kotler 1975, p. 283). In linewith Bloom and Novelli (1981), we contend that social mar-keting is an endeavor that can be engaged in by profit-making organizations (e.g., Anheuser-Busch encouragingresponsible drinking), as well as by nonprofit and publicorganizations. In other words, regardless of the sponsor,encouraging stakeholders to engage in socially responsiblebehavior (e.g., quit smoking, not be cruel toward animals)and to avoid or act against socially irresponsible behavior(e.g., oppose sweatshop operations) are among the principalgoals of social marketing (Andreasen 1995). Identificationand disidentification can help achieve these respectivegoals.

Moreover, as Goldberg (1995) suggests in his plea forbroadening the scope ofthe extant social marketing domain,such goals of individual behavior change are complemen-tary to broader societal goals of change at the policy andsocial environment levels. Thus, beyond this study's interestto social marketers who work for individual behaviorchange, the potential to affect policy as a result of these indi-vidual behavior changes (e.g., the positive effects of reducedsmoking rates on health care costs leading to a ban on themarketing of tobacco products) increases the interest ofidentification and disidentification to policymakers as wellas legislators. A similar argument is put forth in the healthcommunications literature by Wallack (1990, p. 371), whoasserts that "the way a society thinks about cigarette smok-ing, in the long run, is certainly as important as, and may beeven more important than, getting small numbers of peopleto quit smoking." We contend that identification anddisidentification are two important cognitive constructs that

26 Journal of Public Policy & MarketingVol. 21 (I)

Spring 2002, 26-36

Journal of Public Policy & Marketing 27

have not been studied previously in the social marketing lit-erature and tbat not only shape people's thinking and lead toindividual-level behavior change (i.e., the social marketingobjective) but also could lead to related macro changes (i.e.,in policy and the social environment).

More important, in some contexts, many people have arelatively inert, "neutral" orientation toward the organiza-tion, which makes them possible targets for transition toeither positive or negative polarization. Knowledge of howto convert people from an apathetic state to a stronger posi-tive or negative relationship with the organization andthereby galvanize desired behavioral change would be help-ful for social marketers. Similarly, depending on his or hervantage point, a manager may be interested in not onlyproactively fostering identification (or disidentification) toinfluence bebavior but also reacting to and containing (neu-tralizing) such efforts when opposing organizations go onthe offensive.

To use identification and disidentification to their benefit,social marketers need an understanding of these constructs,their antecedents, and their consequences. More specifi-cally, are positive affiliations formed in the same way asnegative ones? How prevalent are such valenced affiliationscompared with more apathetic relationships with organiza-tions? Are there asymmetries in the behavior of people whoidentify with an organization versus those wbo disidentify?How do identifiers and disidentifiers differ from each otherand from those who view organizations neutrally? Ourobjective in this article is to address these issues.

The empirical context of this study is the National RifleAssociation (NRA). Several factors point to the suitabilityof the NRA for an exploratory study of this type. As Gold-berg (1995, p. 350) asserts, "tbe dark side of marketing,involving a much wider range of products and issues, alsoawaits research scrutiny. Consider, for example, how thelegal and illegal marketing of guns and automatic weaponscontributes to the disease of violence." Rightly or wrongly,the NRA has been viewed as a contributor to such violence(e.g., Erickson 1997), which makes it a worthwhile subjectof inquiry. Conversely, the opposing gun control organiza-tions are positioned on the social platform of increased"consumer safety," and they are actively marketing the plat-form of increased safety to the public (Pitts 1998). Disiden-tification with the NRA and subsequent behavior modiflca-tion in terms of reduced gun ownership, negative word ofmouth, and attendance at anti-NRA rallies may lead tostricter gun control measures, which would lower the rate ofunintentional shootings and violent crimes and benefit soci-ety as a whole. In contrast, identification with the NRA maylead to increased gun ownership and increased support forthe NRA through donations, attendance at rallies, and soforth, which may in turn impede the efforts of reduced vio-lence and increased safety.

More generally, social movement theorists (e.g.,McAdam, McCarthy, and Zald 1988) suggest that organiza-tions such as the NRA also have "customers." The subjectsin our sample either already support the NRA's cause(through their donations or membership) or have the poten-tial to support the NRA's cause (e.g., by donating money,volunteering time, becoming a member). Conversely, everysubject also has the potential of similarly acting against the

NRA through a variety of forums. Organizations opposingthe NRA need to learn about the construct because theywant to engender such disidentiflcation among the public.The flip side is that, because widespread disidentification istypicaliy not to an organization's advantage, controversialorganizations (such as Planned Parenthood and HumanRights Campaign) need to understand the factors that arerelated to disidentification.

Not surprisingly, in the natural course of some focusgroups we conducted, the NRA emerged as the most dis-cussed organization—one that participants had feelingstoward and to which they could relate. It was also an orga-nization toward which participants felt both positive andnegative; tbat is, we had a good chance of tapping both iden-tifiers and disidentifiers through a random survey. Finally,with a membership base of more than 4 million and anannual budget of $150 million, tbe NRA is an establishedmarketing organization (www.nra.org; Davidson 1996). Butthe NRA's influence extends far beyond tbese numbers. In1999, the U.S. market consumed approximately $2.2 billionworth of firearms and ammunition—many ofwhich were nodoubt used for recreational hunting and target practice(Kesmodel 1999). However, the U.S. Department of Justicereports that there were more than 10,000 homicides and halfa million serious violent crimes committed with flrearms in1999. The NRA, therefore, seems to be a good starting pointfor better understanding the differences between the identi-fication and disidentification constructs as they relate toconsumers.

We also clarify at the outset that organizational member-ship and identification, though likely to be related, are notone and the same. Whereas identification is a cognitivestate, membership is a related bebavior. Therefore, peoplewho identify with an organization may take up membership(if that possibility exists) to manifest their identification;alternatively, people who join the organization as membersmay over time come to identify with it. However, all mem-bers need not identify with the focal organization (for an artmuseum example, see Bhattacharya, Rao, and Glynn 1995),and conversely, all those who identify with the organizationneed not be members (as in this study). For example,whereas 55 of the 405 respondents in this study identifywith the NRA, only 21 are members (of whom 10 are iden-tifiers and the remaining 11 view the organization neu-trally). Conversely, of the 50 who disidentify, only 6 aremembers of organizations that oppose the NRA.

This study contributes both to theory and to practice. Weadd to the literature in social marketing, corporate socialresponsibility, and organizational identification (e.g., Bhat-tacharya, Rao, and Glynn 1995; Drumwright 1996). More-over, extant research on identification has mostly focused onmembers or former members of an organization. In contrast,in recognition of the notion that these attitudes may affectthe behavior of both current and potential consumers of anorganization, our study focuses on the general population.Notably, altbough the primary focus of this article is socialcause organizations, our findings also have implications formarketers of products and services: Sen and Bhattacharya(2001) show that consumers' relationship with a companycan influence their reactions to the company's products.

28 Us Versus Them

Conceptual Framework andHypothesesOrganizational identiflcation is a well-established conceptin the organizational behavior literature (e.g., Mael and Ash-forth 1992; O'Reilly and Chatman 1986). Formally, Dutton,Dukerich, and Harquail (1994, p. 239) define identificationas the "degree to which a person defines him or herself ashaving the same attributes that he or she believes deflne theorganization." Note from the deflnition that identiflcation isbased on self-perception rather than the person's perceptionof an organization. Identifying with organizations is a wayto preserve (or enhance) the self-concept. With increasinginterest in relationship marketing strategies (see Morgan andHunt 1994), there has been growing interest in marketing inorganizational identiflcation and the way it relates to cus-tomer behavior (e.g., Bhattacharya, Rao, and Glynn 1995;Drumwright 1996; Sen and Bhattacharya 2001).

Similar to organizational identification, disidentifyingwith organizations is another mechanism, to preserve theself-concept (Elsbacb and Bhattacharya 2001). However, inthis case, the self-concept is preserved not by developing aconnection with the organization but through a sense of sep-aration. Formally, disidentification is a self-perceptionbased on (1) a cognitive separation between a person's iden-tity and his or her perception of the identity of an organiza-tion and (2) a negative relational categorization of the selfand the organization (Elsbach and Bhattacharya 2001). Theflrst part of this deflnition mirrors Dutton, Dukerich, andHarquail's (1994) deflnition by emphasizing separation (asopposed to overlap) of the person's self-concept from that ofthe organization. The second part of the deflnition suggeststhat in disidentiflcation, people afflrm their social identitiesby categorizing organizations into groups such as "rivals" or"enemies."

Overall, disidentification and identification are similar inthat botb are perceptual constructs or schemas that helpdefine a person's self-concept. Our views are consistentwith the message of schema research (Fiske and Taylor1991)—people simplify reality by storing knowledge at amolar, inclusive level. Specifically, in line with Ashforthand Mael (1989) and Bergami and Bagozzi (2000), we adopta "molecular" view of both identiflcation and disidentiflca-tion as summary or global perceptions of belongingness toor separation from an organization. This oneness or overlapis based not on a direct correspondence between a person'sown characteristics and the organization's but, instead, onan overall subjective sense of conformance between identi-ties. Both identification and disidentification are relevantonly in contexts in which a person has an opinion about theorganization—as opposed to a third, neutral state of cogni-tive apathy in which a person seeks neither a connectionwith nor separation from the organization.

The basis of identification and disidentification isgrounded in the general family of cognitive consistency the-ories. For example, Heider's (1958) balance theory—a the-oretical forebear of schema models (Fiske and Taylor1991)—suggests that people are motivated to maintain rela-tionships in which they agree with their friends and disagreewith their enemies; that is, both connections and disconnec-tions are important in people's self-concepts. When people

flnd themselves in situations in which they disagree with afriend, they are out of balance. To restore balance, they maychange their attitudes either about the issue or about theirfriend. When the link in question is between a person and anorganization, we propose that both identification anddisidentiflcation enable people to change their attitudesabout the organization and thereby enhance or maintaintheir self-concepts. Unlike in the extant consumer behaviorliterature, in the case of identiflcation and disidentiflcation,this attitude change is made not on the basis of consumercharacteristics or brand features and attributes but on thebasis of speciflc organizational ideologies, policies, andpractices.

To elaborate, ideologies, policies, and practices are inte-gral components of an organization's image (Barich andKotler 1991). In particular, social cause organizations areoften positioned on distinct platforms (e.g., Greenpeacehelps preserve the environment). In contrast, policies andpractices of goods and services providers typically involvethe conduct of corporate employees or human resources(e.g., racial and sexual fairness/discrimination in hiring,benefit policies, employee policies with regard to volunteer-ing), procurement (e.g., socially responsible buying), manu-facturing (e.g., using child labor), marketing (e.g., advertis-ing content and style, pricing practices), and business policy(e.g., the product lines of the company, corporate philan-thropy, alliances with nonprofits, pollution, toxic waste andother environmental policies). In some organizational con-texts, the practices are generally known to the public (e.g!,Phillip Morris sells cigarettes), and in other cases, unantici-pated events bring the practices to light (e.g., an employeetape-recorded conversations in a company meeting to revealracial discrimination at Texaco).

In terms of organizational identity construction (Scott andLane 2000), in general, a person's beliefs and inferenceswitb regard to these policies and practices coupled with hisor her prior knowledge and experience with an organization,as well as the organizations' reputation, lead to certainoverall evaluations regarding the organization's identity.These overall evaluations, when assessed for overlap withthe person's identity and self-concept, lead to feelings ofconnectedness (i.e., identiflcation), separation (i.e., disiden-tiflcation), or indifference. Note that identiflcation and dis-identiflcation need not be tightly coupled; in other words, itis not necessary to identify (disidentify) with OrganizationA in order to disidentify (identify) with Organization B. Forexample, a person may disidentify with the NRA but maynot identify with any particular gun control organization.Moreover, even within the context of one organization, prioridentification is not a necessary condition for disidentiflca-tion at some later stage.

In support of these basic notions, using a combination ofliterature reviews, archival data, focus groups, and surveyresearch, Elsbach and Bhattacharya (2001) show that orga-nizational identification and disidentification are qualita-tively distinct constructs (versus two oppositional ends ofthe same construct). Overall, their study suggests thatthough organizational identiflcation is primarily a mecha-nism for identity enhancement, disidentiflcation is relatednot only to enhancement but also to people's motivations foridentity maintenance. In other words, under certain condi-

Journal of Public Policy & Marketing 29

tions, people's identities may be threatened by an attitude ofindifference toward an organization, and disidentificationmay be necessary to maintain their identities and self-concepts. Other studies in psychology have found similardistinctions between constructs. For example, positive affectand negative affect are different from each other (Dienerand Emmons 1984).

Whereas Elsbach and Bhattacharya (2001) focus onestablishing that identification and disidentiflcation are dif-ferent constructs and investigating the antecedents and con-sequences of disidentiflcation, we focus on distinguishingbetween the proflles of people who identify or disidentifywith an organization or view it neutrally. In other words, westart by categorizing people into identifier, disidentifier, orneutral and subsequently establish systematic differencesacross these groups. Thus, our study is more in the spirit ofBelch and Belch (1987), who find that boycotters and non-boycotters of a product base their attitudes on different cri-teria: Whereas corporate associations matter more for boy-cotters, nonboycotters are affected more strongly by productattributes.

We draw on the organizational bebavior and marketingliterature in this area to propose factors that not only seemto be systematically related to botb organizational identifl-cation and disidentification but also likely discriminateamong the proflles of identifiers, disidentifiers, and thosewho neither identify nor disidentify with a focal organiza-tion of interest (i.e., the neutral group). For example, giventbat tbese feelings of connection with and/or separationfrom organizations are often rooted in people's value sys-tems, prior research suggests that values and beliefs towardissues that are central to the focal organization's identity arerelated to identiflcation (Dutton, Dukerich, and Harquail1994; Scott and Lane 2000). Similarly, contact with anorganization (Dutton, Dukerich, and Harquail 1994; Maeland Ashforth 1992) is associated witb identification; there-fore, we examine the role played by personal experiencewith issues that are central to the focal organization's iden-tity. Because the external image of an organization is relatedto identification (Dutton, Dukerich, and Harquail 1994;Pratt 1998), we believe that it is important to examine therole the perceived reputation ofthe focal organization playsamong identiflers and disidentiflers. Finally, the linkbetween identiflcation (and disidentification) and behaviorssuch as public discourse about the focal organization andaction for (against) the focal organization has been docu-mented by several researchers (e.g., Bergami and Bagozzi2000; Mael and Asbforth 1992), so we examine how identi-flers, disidentiflers, and neutrals differ in this regard. In thefollowing sections, we propose how the relative associationof these variables differs across people who identify,disidentify, and feel neutral toward an organization.

Values and Beliefs Toward Issues Central to theOrganization's IdentityOrganizational behavior researcbers (e.g., Ashforth andMael 1989) have suggested that identification is related tothe consistency between people's self-concepts and theirperceptions of an organization's identity. Researchers haveeven defined organizational identification as a congruencebetween organizational and individual values (Hall and

Schneider 1972). Therefore, identiflcation with the focalorganization will be related to the values the person holdstoward the issues that are central to the organization's iden-tity. Similarly, disidentiflcation may be necessary as ameans to preserve a person's self-concept when there is aperceived incongruence or conflict between the person'svalues and those of the focal organization. This is particu-larly likely if the values on which the organization is posi-tioned are also central to the person.

In general, identiflers want to be known as being assimi-lated with the organization, whereas disidentifiers want tobe known as distinct. Notably, in similar situations in psy-chology, the motivations for distinctiveness have beenshown to be stronger than the motivations for assimilation(Brown and Williams 1984; Oakes 1987). This suggests thatdisidentiflers who view themselves to be in an "us-them"relationship with the organization want to be differentiatedfrom the organization more than identifiers who view them-selves to be in an "us-us" relationship with the organizationwant to be considered similar. Thus, compared with theeffect on identiflers, disidentiflers' perceptions of the focalorganization will be more strongly associated with their val-ues and beliefs toward issues tbat are central to the organi-zation's identity. More formally.

Hi: On values central to the organization's identity, disidenti-fier > identifier > neutral.

Personal Experience in Contexts Central to theFocal Organization's IdentityPeople selectively expose themselves to particular informa-tion, tasks, and other people that permit the maintenance andstrengthening of desired identities (Schienker 1985). Studiesof organizational identification have shown that contactwith an organization, through participation in its activitiesand/or consumption of its products, is positively related toidentification (Bhattacharya, Rao, and Glynn 1995; Dutton,Dukerich, and Harquail 1994). Increasing contact leads togreater salience of the shared values between the organiza-tion and the person and therefore to identification.

In contrast, familiarity often leads to greater differentia-tion and perceived variability and thus to more moderatejudgments (Fiske and Taylor 1991). Bringing people withpositive and negative feelings together is one possible wayto break down their mutual misperceptions. Typically,disidentiflers are less likely to encounter additional, novel,or unusual information about the focal organization—andthis biases their encoding of all subsequent information inways that are consistent with their existing negative atti-tudes (Schienker 1985). This line of reasoning suggests thatpeople who have greater information, knowledge, familiar-ity, or experience with an organization and its traits will bemore moderate in tbeir feelings about the organization andless likely to disidentify.

H2: On personal experience in contexts relevant to focal organi-zation's identity, identifier > neutral > disidentifier.

Organizational ReputationOrganizational reputation refers to outsiders' beliefs aboutwhat distinguishes an organization (Dutton and Dukerich

30 Us Versus Them

1991). These reputations are constructed from a mix of sig-nals that include factors such as social responsiveness andmedia exposure (Fombrun and Shanley 1990). Organiza-tional identification researchers have established that identi-fication is positively related to the prestige of the organiza-tion's identity (Bhattacharya, Rao, and Glynn 1995) and asense that the organization is respected and admired bymeaningful referents (Bergami and Bagozzi 2000; Dutton,Dukerich, and Harquail 1994). In essence, when people con-strue the reputation as attractive, affiliating with the organi-zation creates an opportunity for a positive social identity.Thus, those who identify with an organization should bemore affected by their perceptions of its reputation than arethose who view it neutrally. The reverse arguments shouldhold true for disidentification: A perceived unattractive ornegative organizational reputation should lead to disidenti-fication. That is, people maintain their senses of self by dis-tinguishing themselves from salient but (perceived) unat-tractive organizational reputations.

Researchers have argued, however, that negative informa-tion is perceptually more salient than positively valencedinformation and given more weight than positive information(Peeters and Czapinski 1990; Taylor 1991). Therefore, nega-tive organizational information may have greater impact inshaping a person's identity than does positive information. Inother words, organizational reputation is likely to affectdisidentifiers to a greater extent than identifiers.

H3: On perceived organizational reputation, disidentifier> iden-tifier > neutral.

Public Discourse About the Focal OrganizationTalking about tbe focal organization is a way of increasingthe relevance of the organization as a source of self-defini-tion. In this case, publicly talking about the organization isakin to being a spokesperson for the organization, whichmay be considered a form of citizenship behavior (Bergamiand Bagozzi 2000). People who identify with organizationsperceive the organization's successes and failures as theirown (Ashforth and Mael 1989) and expend efforts that aredirected toward preserving, supporting, and improving theorganization (Dutton, Dukerich, and Harquail 1994).

Similarly, we believe that people who disidentify with anorganization will also be more prone to talking about theorganization than are those who view the organization neu-trally. In this case, saying negative things about the organi-zation is people's way of protecting their social identity andself-integrity. Notably, such criticism is often a more imme-diate and convenient form of self-affirmation than is takingaction against the focal organization (Taylor 1991).

One question is whether identifiers are more or less likelyto talk publicly about the organization than disidentifiersare. It is possible that identifiers are more exposed to socialcontexts that are more conducive to such conversation. Incontrast, negative feelings about organizations may be moresalient and more easily recalled than positive feelings; thissuggests that disidentifiers are more prone to talking pub-licly about the organization than identifiers are. On balance,therefore, we posit tbat there will not be any differencebetween these two groups in the extent to which they pub-licly talk about the focal organization.

H4: On public discourse related to organization, identifier =disidentifier > neutral.

Actions For or Against the Focal OrganizationWe posit that identification will lead to actions in support ofthe focal organization, whereas disidentification will lead toactions against the focal organization. Mael and Ashforth(1992) have shown that identification on the part of alumnileads to increased donations to the alma mater. Similarly,O'Reilly and Chatman (1986) show significant linkagesbetween extra-role behavior (for which the person receivesno immediate reward and which benefits the larger organi-zation) and identification. Finally, in the context of restau-rant employees, Bergami and Bagozzi (2000) show thatidentification leads to citizenship behaviors.

In the realm of disidentification, people will take actionsagainst the focal organization to maintain their sense of selfThese actions can range from personally boycotting a productor service, writing to tbe media and other relevant constituents(Day and Landon 1977; Singh 1988), or even supporting anopposing organization (e.g., donating to the American CancerSociety to maintain disidentification with cigarette compa-nies). These actions are motivated not only by people'sdesires to enhance their image but also by people's desires toavoid dissonance associated with acting inconsistently withtheir established beliefs and prior commitments.

Again, there may be arguments suggesting the relativeextent to wbich identifiers and disidentifiers will takeactions in favor of or against a focal organization. Someexploratory focus group data suggest that though doinggood does not help, doing bad hurts; that is, the negativeactions of disidentifiers may be greater in intensity than thepositive actions of identifiers (Barnard et al. 1993). How-ever, in reality, the extent to which action is taken maydepend on the relative ease of taking such action, the sizesof the focal organization and the opposition, and so on.Therefore, on balance, we posit no difference in the extentto which these two groups take action for or against the focalorganization.

H5: On action related to the organization, identifier = disidenti-fier > neutral.

In summary, people may feel positive, negative, or neu-tral toward an organization. We believe that these overallattitudes toward an organization are differentially related toa set of beliefs, attitudes, and behaviors. In the followingsection, we describe a large-scale empirical study that testsour assertions.

MethodsWe conducted a random survey of households in a largesoutheastern U.S. city that probed people's attitudes andbehaviors with specific reference to the NRA. In the survey,in addition to the scale items pertaining to the constructs ofinterest, we asked questions about the subjects' knowledgeand experience with the NRA and with organizations oppos-ing the NRA and questions on demographics such as sex,age, and occupation.

The execution ofthe survey closely followed the guidelinesprovided by Dillman (1977). We used a mailing list companytbat provided us witb names, addresses, and limited demo-

Journal of Public Policy & Marketing 3t

graphic information (e.g., age, income, sex) for a random sam-ple of lOCX) households. In addition to a cover letter outliningthe academic purpose ofthe survey, we attached postage-paidreturn envelopes and a $1 incentive. One week after the firstmailing, we mailed a follow-up postcard to all survey recipi-ents. Two weeks after the first mailing, we mailed a secondwave of surveys along with an updated cover letter to recip-ients who still had not responded. Of the 1000 surveys wemailed, 962 were received by respondents (38 were returnedby the Post Office). As a result of these efforts, of the 962received by respondents, 531 were completed and returnedto us, for a response rate of 55.2%. To ensure that our sam-ple was knowledgeable regarding the issue at hand, we elim-inated 126 respondents who professed that they were "not atall familiar with the NRA," which left a usable sample of405 respondents. Of these, 21 (5%) were NRA members,and to the extent that this is an overrepresentation of theNRA membersbip in the U.S. population (1.6% of individu-als and 4% of households), our results may have limitedgeneralizability.

In terms of demographics, respondents were 41% female,59% male. The age of respondents ranged from 24 to 78years, with an average of 48 years and a standard deviationof 9.9 years. Respondents self-categorized their occupationsas 49% professional, 21% managerial, 12% clerical/techni-cal, 6% labor/blue collar, 9% other, and 3% currently unem-ployed. The median income range was between $60,000 and$65,000. In terms of nonresponse bias tests, we investigatedhow the respondents compared with the overall sample towhich we mailed the questionnaires in terms of three keydemographics—age, income, and sex. Our mailing popula-tion was 50% women and 50% men, so it seems that ourrespondent sample has a slight overrepresentation of men.There is no age difference among respondents and nonre-spondents. At first blush, the median income ofthe respon-dents at $60,000-$65,000 seems at variance with the popu-lation statistics (median at $5O,OOO-$55,OOO). However,when we factor in the sex representation of the respondentsand the higher earnings of men than of women in the mail-ing population, the overall disparity between the respon-dents and nonrespondents ceases to be of much concern.

Measurement of VariablesThe two dependent variables of interest to us are identifica-tion and disidentification. Table 1 provides the scale itemsand reliabilities for all the constructs used in the study.Respondents indicated their level of (dis)agreement witheach statement on a five-point scale ("agree strongly" = 5).Elsbach and Bhattacharya (2001) used the same scale itemsand showed through confirmatory factor analysis that iden-tification, disidentification, and the other five variables ofinterest are all qualitatively distinct constructs. Specifically,the confirmatory factor analysis results show that the mea-surement model with all seven latent constructs had anacceptable fit: A,2 (131 degrees of freedom [d.f.]) = 627.08{p < .01), goodness-of-fit index = .87, normed fit index -.87, incremental fit index = .90, comparative fit index - .90,root mean square error of approximation = .09.

On the basis of the confirmatory factor analysis results,we used several tests to further assess discriminant and con-vergent validity for the measures. We constructed a 95%

Table 1. Scale Items and Reliabilities

Construct Scale Reliability

Organizational Identification .871. The NRA's successes are my successes.2. When someone praises the NRA, it feels like a personal

compliment.3. When someone criticizes the NRA, it feels like a personal

insult.Organizational Disidentification .79

1. The NRA's failures are my successes.2. When someone praises the NRA, it feels like a personal

insult.3. When someone criticizes the NRA, it feels like a personal

compliment.Perceptions of the NRA Based on Values and Beliefs AboutGuns and Control .81

1. I have values and beliefs about gun control that haveaffected my perceptions of the NRA.

2. I have values and beliefs about gun laws that have affectedmy perceptions of the NRA.

3. I have values and beliefs about gun ownership that haveaffected my perceptions of the NRA.

Perceptions of the NRA Based on the NRA's Reputation .701. The NRA's reputation in my community has affected my

perceptions of the NRA.2. The effect of joining the NRA on a person's reputation has

affected my perceptions of the NRA.Perceptions of the NRA Based on Personal Experiences .72

1. I have had personal experiences that have affected my per-ceptions of the NRA.

2. A friend or family member has had personal experiencesthat have affected my perceptions of the NRA.

Actions .831. I have done volunteer work (for or against the NRA).2. I have made charitable contributions (for or against the

NRA).3. I have joined organizations (for or against the NRA).

Public Discourse .901. I have publicly expressed my feelings (positive or negative)

about the NRA.2. I have publicly expressed my opinions (positive or negative)

about the NRA's goals and objectives.3. I have publicly expressed my feelings (positive or negative)

about supporters of the NRA.

"The reliability coefficient a is Cronbach's alpha.

confidence interval around the correlations between tbelatent constructs. None of tbe confidence intervals included1.0 or -1.0, which provided some evidence of discriminantvalidity. In addition, a series of nested model comparisonsassessed whether differences were present when correla-tions between the latent constructs were constrained to 1.0.Statistically significant differences between each model pairindicate discriminant validity. It is noteworthy that com-pared with tbe models in wbich the traits were allowed tocorrelate freely, the model fit worsened significantly eachtime we constrained tbe correlation between any pair of con-structs to 1 o r - 1 , implying again that all of these are differ-ent constructs. Finally, we conducted a more stringent test,recommended by Fornell and Larcker (1981), wbichdemonstrated discriminant validity by showing that the

32 Us Versus Them

Table 2. Within-Group Means

Variable Identifiers (n = 55) Disidentifiers (n = 57) Neutral (n = 293)

Values and beliefsPersonal experienceReputationPublic discourseAction

4.083.413.043.562.84

4.502.663.303.482.10

3.882.882.772.582.01

Tests of Equality of Group Means

Sum ofSquares

Wilks'Lambda F-Statistic d.f. 1 d.f. 2 Significance

Values and beliefsPersonal experienceReputationPublic discourseAction

25.32523.4948.334

77.22130.590

.906

.937

.974

.832

.893

20.89813.4115.422

40.62224.043

22222

402402402402402

.000

.000

.005

.000

.000

average variance extracted from each latent constructexceeds the squared correlation betv '̂een all pairs of con-structs. This series of tests provided evidence of discrimi-nant validity between all pairs of constructs. Together, theresults provide evidence that the measures have the soundpsychometric properties necessary for hypothesis testing.

For the purposes of this study, we classified identifiersand disidentifiers as those who scored higher than 3 on therespective scales (there were no overlaps), and we classifiedthe rest of the respondents as neutral (i.e., those who neitheridentify nor disidentify). Such classification into groupsbased on the midpoint ofthe scale is common in similar con-texts (e.g., Roberts and Berger 1999). This classificationscheme resulted in 55 identifiers, 57 disidentifiers, and 293people who neither identify nor disidentify (i.e., are neutral).The scale items and the internal reliabilities ofthe indepen-dent variables (i.e., discriminators) are also noted in Table 1.Note from the scale items that all participants (i.e., identi-fiers and disidentifiers alike) responded to the same scales;they only needed to know themselves whether they felt pos-itive or negative about the organization.

ResultsTo test the proposed hypotheses, we used multivariateanalysis of variance (MANOVA) and multiple discriminantanalysis (MDA), both well-accepted procedures for under-standing group differences. For the MANOVA, the scalevariables (i.e., values, personal experience, reputation, pub-lic discourse, and action) were the dependent variables, andthe classification variable (i.e., whether someone was anidentifier, a neutral respondent, or a disidentifier) was theindependent variable. Although the neutral category is muchlarger than the other two, these cell size differences per sedo not pose any problems with the analyses.'

All four multivariate tests (Wilks' lambda, Pillai's trace,Hotelling's trace, and Roy's largest root) suggest that theeffects in the model are highly significant (p < .001). Inother words, the mean vectors of the dependent variablesdiffer across the three groups. Table 2 provides the within-

' If we had a choice, we would have preferred to obtain more equal groupsizes (e.g.. 135 in each of the three groups) for higher overall power, butour survey results did not turn out that way.

Table 3. Test of Multiple Comparisons (Tamhane)

DependentVariable Comparison^

Mean Difference(Standard Error)

Values and beliefs

Personal experience

Reputation

Public discourse

Action

D-NN-l1-DD-NN-lI-DD-NN-lI-DD-NN-l1-DD-NN-lI-D

.62 (.12)***-.20 (.10)**-.42 (.15)***-.22 (.13)*-.53 (.14)***

.75 (.18)***

.53 (.13)***-.27 (.13)**-.26 (.17)

.90 (.15)***-.98 (.14)***

.08 (.19)

.09 (.12)-.83 (.12)***

.74 (.15)***

"D denotes disidentiflcation. N denotes neutral, and I denotes identiHcation.The corresponding tnean differences are shown in the next column.*p<.\0.**p< .05.***/>< .01.

group means and addresses the issue of whether the group-mean differences are significant for each dependent variableconsidered alone. As the results show, consistent with themultivariate results and in line with our expectations, all fivevariables show significant differences across groups.^

For a sense of the specific intergroup differences for eachvariable, we conducted multiple comparison tests thataccount for alpha inflation by focusing on the experiment-wise error rate. The results are consistent across various test-ing procedures. In Table 3, we report the results ofTamhane's test that accounts for unequal eovariance matri-

^Adtnittedly. because these tests are based on independent pairwise coin,parisons. there is a risk of alpha inflation. But given the strong significancelevels (the highest is .009), we expect the effects of each of these variablesto differ significantly across the three groups even after we account formulticollinearity.

Journal of Public Policy & Marketing 33

ces across the groups.^ Examining the /7-values that corre-spond to the multiple comparisons in conjunction with thegroup means, we find that H] is supported: Values andheliefs of both identifiers and disidentifiers more stronglyrelate to their respective perceptions ofthe NRA than do thevalues and beliefs ofthe neutral group. Moreover, disidenti-fiers' perceptions of the organization are more stronglyassociated with their values and beliefs than are those oftheidentifier group (p < .05). The results with regard to per-sonal experience also support our expectations. We findthat, compared with both the neutral group and disidenti-fiers, the role of personal experience is stronger for identi-fiers. Moreover, disidentifiers' perceptions of the focalorganization are less related to their personal experiencethan are those of the neutral group (p < .\). In other words,lack of personal experience is associated with disidentifica-tion. H2 is supported.

The results with regard to reputation (H3) suggest that, asexpected, both identifiers' and disidentifiers' perceptions ofthe focal organization are more strongly related to the per-ceived reputation of the organization than are the percep-tions of the neutral group. However, contrary to our expec-tation, disidentifiers' perceptions are not more stronglyrelated to the organization's reputation than are identifiers'perceptions.

As predicted in H4, identifiers and disidentifiers engage ingreater public discourse about the organization than do peo-ple who neither identify nor disidentify. Moreover, theredoes not seem to be any difference between the identifierand disidentifier groups on this score. With regard to actionfor or against the focal organization (H5), we find that, con-sistent with our expectations, identifiers act in support ofthefocal organization to a greater extent than the neutral groupdoes. Disidentifiers do not differ from the neutral group inthe extent to which they act against the organization.Finally, although we expected no difference between identi-fiers and disidentifiers in terms of support for or against thefocal organization, we fmd that identifiers "walk the talk"by supporting the NRA to a greater extent."*

The discriminant analysis adds a few more insights to theMANOVA results. Specifically, we estimated a three-groupMDA in which the groups composed of identifiers, disiden-tifiers, and neutral respondents and the five dependent vari-ables ofthe MANOVA now constitute the independent vari-able list. Table 4 summarizes the results of the discriminantanalyses. Again, various multivariate statistics such asWilks' lambda and Pillai's trace and the corresponding F-ratios show that the model is significant (p < .0001). More-over, as suggested by the Wilks' lambda and the associatedchi-square statistic, both canonical discriminant functionsare significant (j> < .001).

It is also instructive to examine the group centroids aswell as the standardized coefficients of the canonical dis-criminant function to learn more about the differencesamong the groups (Table 4). The group centroids suggestthat whereas identifiers and disidentifiers differ from each

-'A Box test for the equality of eovariance matriees rejects the hypothe-sis of equality.

"•It could be argued that the 11 NRA members in the identifier group aredriving the results for H,. However, the results hold good even after weremove these members from the identifier sample.

Table 4. Discriminant Analysis Results

Standardized Canonical Discriminant Function Coefficients

Function 1 Function 2

Values and beliefsPersonal experienceReputationPublic discourseAction

.449

.008

.225

.725

.123

-.308.645

-.447-.303

.725

Functions at Group Centroids

Function 1 Function 2

DisidentifiersNeutralIdentifiers

1.018-.308

.754

-.605.025.687

other mostly in terms of Function 2, the difference betweenboth identifiers and disidentifiers compared with the neutralgroup is the main source of discrimination for Function 1.Drawing on the discrimination function coefficients, wefind that identifiers and disidentifiers differ from the neutralrespondents mostly in terms of public discourse and values(the two strongest coefficients for Function I). Similarly,the biggest differences between identifiers and disidentifiersare in terms of actions and personal experience (the twostrongest coefficients for Function 2). Finally, although ourprimary objective is to understand intergroup differencesrather than to classify objects into groups, we computed a hitratio using the "leave one out" principle. The classificatorypower ofthe model (76.3%) is better than chance, as judgedby the proportional chance criteria (56.1%), the maximumchance criteria (73.3%), and the Press's Q statistic (p < .01).

DiscussionDo people affiliate with organizations both positively andnegatively? Are such affiliations (systematically) differen-tially related to a set of beliefs, attitudes, and behaviors?The results of our research suggest a positive response toboth questions and point to theoretical and practical impli-cations for strengthening identification and containingdisidentification.

Researchers have long emphasized that identification is away for people to preserve and enhance their self-concepts.This study suggests that, along with organizational identifi-cation, disidentitlcation is also part of the strategy peopleuse to sustain and enhance positive social identities. Thus,this study enlarges both the organizational identification lit-erature and the view ofthe "extended self (Belk 1988). Asour results suggest, the extended self seems to stem not onlyfrom material possessions or even memberships (Bhat-tacharya, Rao, and Glynn 1995) but also from people'spositive and negative psychological connections withorganizations.

In the context of organizations (such as the NRA) that areviewed both positively and negatively by constituents, ourcentral finding is that these affiliations are related differen-tially to the set of beliefs, attitudes, and behaviors investi-gated in this study. Thus, compared with an apathetic state,whereas positive affiliations with organizations are nurtured

34 Us Versus Them

through personal experience with the organization and/orthe products and issues surrounding it, disidentification isrelated to a sense of value incongruence between the personand the organization and simplified images of the organiza-tion developed through its perceived reputation. Our find-ings also show that, when compared with the neutral group,both identifiers and disidentifiers exhibit distinct behaviorpatterns such as talking about the focal organization. How-ever, identifiers go beyond talking and act on their beliefs,whereas disidentifiers stop at discourse. This distinction is avalidation of and addition to the identification literature thathas touted action—beyond words—as one of the benefits ofidentification (e.g., Bergami and Bagozzi 2000; Mael andAshforth 1992).

Along with Elsbach and Bhattacharya (2001), we alsoshow that formal organizational membership—either in theemployee or customer arena—is not a prerequisite for form-ing such positive or negative connections. As emphasizedpreviously, most of our identifiers (disidentiflers) were notNRA members (members of gun control organizations). Inother words, organizations that are not so strongly member-ship oriented are also likely targets for identification anddisidentiflcation. In general, organizations that are narrowlydefined and strongly associated with a particular value orissue (i.e., social cause organizations) are more likely tohave identifiers and disidentiflers specifically because ofthedistinctive value and issue. Examples would be organiza-tions that oppose smoking or domestic violence, pro-life andpro-choice organizations, gay rights organizations, animalrights organizations, and for-profit organizations that arepositioned or reputed for specific socially responsible orirresponsible activities (e.g., sweatshop operations, racialdiscrimination). Conversely, organizations that have diffusepositions and multiple identities in the marketplace (e.g..United Way) are less likely to be identified and/or disiden-tifled with.

Our findings are important for social marketing becausethey suggest a broader scope for social marketing strategy.To elaborate, our study suggests that social marketingefforts need not necessarily be confined to changing behav-ior within an organization's own membership, attractingmore members, or even inducing the target audience only tothink favorably in terms of the mission of the organization.Identifying with the focal organization or disidentifyingwith an opposing organization while belonging to the gen-eral public and subsequently engaging in word of mouth andindividual-level action are also legitimate ways of support-ing the focal organization's social change efforts. For exam-ple, in certain contexts, the social marketer may decide thatit is more efficient to influence people to distance them-selves from opposing organizations (e.g., PETA's effortswith The Gap) than to affiliate with the focal organization.At a minimum, such disidentification will be associated withnegative word of mouth toward the opposing organization,which in turn may have more macro consequences such asshaping societal thinking on a particular issue (Wallack1990).

These findings have several practical implications thatcan be used to strengthen and/or contain identification anddisidentiflcation as is deemed necessary. The main practicalimplication is that people do identify and disidentify with

organizations, and this leads to certain systematic behaviorpatterns; therefore, social marketers should think strategi-cally about managing the patterns. Moreover, given that wefind a sizable segment that is apathetic toward the NRA,depending on the context, social marketers should also con-sider inducing disidentification among the neutral segment.Before delving into the specifics, it is worth emphasizing,however, that promoting identification and/or containingdisidentification may, in certain cases, involve repositioningthe organization in people's minds (e.g., the recent effortsby Phillip Morris to position itself as a socially responsiblecompany). This is a difficult task, particularly for organiza-tions that have an indelible, negative image to begin with.Notably, depending on the target audience's stance towardvarious issues surrounding the focal organization, in theircommunication strategies, managers may decide to reposi-tion either the focal organization itself or the competition.However, bear in mind that particularly in the context ofideological organizations or ones that deal with controver-sial causes, such repositioning and the business of fosteringidentification (or disidentification) in general is a double-edged sword. In promoting identification among one seg-ment, social marketers may inadvertently promote disiden-tification among another.

In terms of the practical implications, our findings speakto controversial organizations such as the NRA that maythemselves be the targets of disidentification. In this case,we believe that because perceived value incongruence, lackof personal experience, and (perhaps because of selectiveperception; Schienker 1985) the formation of simplifiedimages based on media reputation all relate to disidentifica-tion, fact-based information dissemination about the focalorganization and its actions can help quell disidentiflcation.Thus, many of our focus group participants agreed that theywould find it difficult to disidentify with the NRA if theyknew that the NRA was involved in funding rape crisis cen-ters or battered women's shelters. Similarly, Nike con-sumers who disidentify with the company because of itsoverseas sweatshop operations often do not know that thoseoverseas employees earn much more than do employeeswho work in local organizations. Thus, beyond strengthen-ing identification, ongoing information-oriented communi-cation can also help mitigate disidentification.

Our research also has implications for anti-NRA social-cause organizations such as Handgun Control Inc. Such anorganization is interested in bolstering identification withitself, influencing people to disidentify with the NRA, andconverting the neutraJ segment to identify with it and/ordisidentify with the NRA. First, given the role of personalexperience in identification. Handgun Control should focusits own identification-bolstering efforts mostly on peoplewho, either directly or indirectly, have suffered from gun-related incidents. Second, communication strategies height-ening the salience of people's values are appropriate forboth groups. However, the messages for garnering identifi-cation and disidentification should be different—whereasthe former would emphasize the similarity between the per-son's and Handgun Control's values, the latter would harpon the incongruence between the person's values and thoseof the NRA. Third, given that personal experience stronglydiscriminates between identifiers and disidentiflers, it may

Journal of Puhlic Policy & Marketing 35

be easier for an organization such as Handgun Control to geta larger body of people to disidentify witb tbe NRA than toidentify with itself. However, although the disidentifierswill benefit Handgun Control through their negative word ofmouth about the NRA, our results suggest that they areunlikely to undertake other proactive behaviors (e.g., writ-ing senators, participating in anti-NRA rallies).

Limitations and Future Research DirectionsThe generalizability ofthe findings in this study is limited intwo respects: First, we cannot be sure that the special placethe NRA has in the public's mind because of the fundamen-tal importance of the issues it deals with did not influencethe responses in some systematic fashion. Second, perhapsbecause the empirical investigation for this study wasconducted in the southeastern United States, the NRAmembersbip was slightly overrepresented in this study (21of 405, or 5%) compared with the national average (1.6% ofall individuals or 4% of all households). Given that themembers either identitled witb or were neutral toward theNRA, in particular, some of the responses of the identifierand neutral groups may have been influenced by thisoverrepresentation.

This study raises several research issues. Theories ofgroup identity and intergroup behavior introduced in thisarticle can also be used to understand the formation of brandcommunities (e.g., Muniz and O'Guinn 2001) and otbercause-based communities that are becoming increasinglypopular on the Internet. With wbat do tbese people iden-tify—is it the brand, tbe cause, or the opportunity to com-municate witb other people through this medium? Anotherpertinent research issue for botb theory and practice iswhether a distinctive, ideological corporate positioning isbetter than adopting a "middle ground." This issue involvestbe interdependency between identification and disidentifi-cation. Could strategies directed at building identificationresult in creating disidentification instead? A possible exam-ple wortby of empirical research is advertising that has asocial dimension (Drumwright 1996): Is it cause related(that could build identification) or cause exploitative (thatcould lead to disidentification)? More generally, are certainidentity-building dimensions more vulnerable to being mis-interpreted than others? We hope that this study will inspireinquiry along these lines.

ReferencesAndreasen, Alan (1995), Marketing Social Change. San Francisco:

Jossey-Bass Publishers.

Ashforth, Blake E. and Fred Mael (1989), "Social Identity Theoryand the Organization," Academy of Management Review, 14(1),20-39.

Barich, Howard and Philip Kotler (1991), "A Framework for Mar-keting Image Management," Sloan Management Review, 32(Winter), 94-104.

Barnard. Richard, Teresa Chin, Sharmila Naidu, and Rob Rode(1993), "A Research Study on the Impact of Social Responsibil-ity on Purchase Behavior," study conducted for Ben & Jerry'sIce Cream by MBA Students, Emory University.

Belch, George E. and Michael A. Belch (1987), "The Applicationof an Expectancy Value Operationaiization of Function Theory

to Examine Attitudes of Boycotters and Nonboycotters of aConsumer Product," in Advances in Consumer Research, Vol.14, Melanie Wallendorf and Paul F. Anderson, eds. Provo, UT:Association for Consumer Research, 232-36.

Belk, Russell W. (1988), "Possessions and the Extended Self,"Journal of Consumer Research, 15 (September), 139-68.

Bergami, Massimo and Richard P. Bagozzi (2000), "Self Catego-rization, Affective Commitment, and Croup Self-Esteem as Dis-tinct Aspects of Social Identity in the Organization," BritishJournal of Social Psychology, 39 (4), 555-77.

Bhattacharya, C.B., Hayagreeva Rao, and Mary Ann Clynn(1995), "Understanding the Bond of Identification: An Investi-gation of Its Correlates Among Art Museum Members," Journalof Marketing, 59 (October), 46-57.

Bloom, Paul N. and William D. Novelli (1981), "Problems andChallenges in Social Marketing," Journal of Marketing, 45(Spring), 79-88.

Brown, Rupert J. and Jennifer A. Williams (1984), "Group Identi-fication: The Same Thing to All People?" Human Relations, 37(7), 547-64.

Davidson, D. Kirk (1996), Selling Sin. Westport, CT: QuorumBooks.

Day, Ralph L. and E. Liard Landon Jr. (1977), 'Towards a Theoryof Consumer Complaining Behavior," in Consumer and Indus-trial Buying Behavior, Arch Woodside, Jagdish Sheth, and PeterBennet, eds. Amsterdam, The Netherlands: North-Holland Pub-lishing Company, 425-37.

Diener, Ed and Robert A. Emmons (1984), "The Independence ofPositive and Negative Affect," Journal of Personality andSocial Psychology, 47 (5), 1105-17.

Dillman, Don A. (1977), Mail and Telephone Surveys. New York:John Wiley & Sons.

Drumwright, Minette E. (1996). "Advertising with a SocialDimension: The Role of Noneconomie Criteria," Journal ofMarketing, 5^ {3), 1-19.

Dutton, Jane M. and Janet M. Dukerich (1991), "Keeping an Eyeon the Mirror: The Role of Image and Identity in OrganizationalAdaptation," Academy of Management Journal, 34 (3), 517-54.

, , and Celia V. Harquail (1994), "OrganizationalImages and Member Identification," Administrative ScienceQuarterly, 39 (June), 239-63.

Elsbach, Kimberly D. and C.B. Bhattacharya (2001), "DefiningWho You Are by What You're Not: A Study of OrganizationalDisidentification and the NRA," Organization Science, 12 (4),393-^13.

Erickson, Jan (1997), "NRA Continues Effort to Weaken DV Bill,"(May), (accessed March 19, 2002), [available at http://www.now.org/nnt/05-97/legis.html].

Fiske, Susan T. and Shelley E. Taylor (1991), Social Cognition, 2ded. New York: McGraw-Hill..

Fombrun, Charles and Mark Shanley (1990), "What's in a Name?Reputation Building and Corporate Strategy," Academy of Man-agement Journal, 33 (2), 233-58.

Fornell, Claes and David F. Larcker (1981), "Evaluating StructuralEquation Models with Unobservable Variables and Measure-ment Error," Journal of Marketing Research, 18 (February),39-50.

Goldberg, Marvin E. (1995), "Social Marketing: Are We FiddlingWhile Rome Burns?" Journal of Consumer Psychotogv, 4 (4).347-70.

36 Us Versus Them

Hall, Douglas T. and Benjamin Schneider (1972), "Correlates ofOrganizational Identification as a Function of Career Pattern andOrganizational Type," Administrative Science Quarterly, 17 (3),340-50.

Heider, Fritz (1958), The Psychology of Interpersonal Relations.New York: John Wiley & Sons.

Kesmodel, David (1999), "Target Market," Atlanta BusinessChronicle, 22 (August 13), IA.

Kleine, Susan Schultz, Robert E. Kleine III, and Chris T. Allen(1995), "How Is a Possession 'Me' or 'Not Me?" CharacterizingTypes and an Antecedent of Material Possession Attachment,"Journal of Consumer Research, 22 (December), 327-43.

Kotler, Philip (1975), Marketing for Nonprofit Organizations.Englewood Cliffs, NJ: Prentice Hall.

Mael, Fred and Blake E. Ashforth (1992), "Alumni and TheirAlma Mater: A Partial Test ofthe Reformulated Model of Orga-nizational Identification," Journal of Organizational Betiavior,13(2), 103-23.

McAdam. Doug. John D. McCarthy, and Mayer N. Zald (1988),"Social Movements," in Handbook of Sociology, Neil Smelser,ed. Thousand Oaks, CA: Sage Publications, 695-737.

McCall, William (1998), "Nike Battles Backlash from OverseasSweatshops," Marketing News, 32 (November 9), 14.

Morgan, Robert M. and Shelby D. Hunt (1994), 'The Commit-ment-Trust Theory of Relationship Marketing," Journal of Mar-keting, 58 (3), 30-38.

Muniz, Albert M. and Thomas C. O'Guinn (2001), "Brand Com-munity," Journal of Consumer Research, 27 (4), 412-32.

Oakes, Penelope J. (1987), 'The Salience of Social Categories," inRediscovering the Social Group: A Self-Categorization Theory,J.C. Turner, M.A. Hogg, P.J. Oakes, S.D. Reicher, and M.Wetherell, eds. Oxford and New York: Blackwell,

O'Reilly, Charles and Jennifer Chatman (1986), "OrganizationalCommitment and Psychological Attachment: The Effects ofCompliance, Identitlcation and Internalization on ProsocialBehaviour," Journal of Applied Psychology, 71 (3), 492-99.

Peeters, Guido and Janusz Czapinski (1990), "Positive-NegativeAsymmetry in Evaluations: The Distinction Between Affectiveand Informational Negativity Effects," European Review ofSocial Psychology, 1(1), 33-60.

Pitts, David (1998), 'The NRA Versus Handgun Control: A CaseStudy in Issue Advocacy," (June), [available at http://dns.usis-israel.org.il/publish/jpurnals/democracy/june98/guns.htm].

Pratt, Michael G. (1998), 'To Be or Not to Be? Central Questionsin Organizational Identification," in Identity in Organizations:Building Theory Through Conversations, D.A. Whetten andP.C. Godfrey, eds. Thousand Oaks, CA: Sage Publications,171-207.

Roberts, Mary Lou and Paul D. Berger (1999), Direct MarketingManagement. Upper Saddle River, NJ: Prentice Hall.

Schienker, Barry R. (1985), "Identity and Self Identification," inThe Self and Social Life, Barry R. Schienker eds. New York:McGraw-Hill, 65-99.

Scott, Susanne G. and Vicki R. Lane (2000), "A StakeholderApproach to Organizational Identity," Academy of ManagementReview, 25 0), 43-62.

Sen, Sankar and C.B. Bhattacharya (2001), "Does Doing GoodAlways Lead to Doing Better? Consumer Reactions to Corpo-rate Social Responsibility," Journal of Marketing Research, 38(May), 225^3.

Singh, Jagdip (1988), "Consumer Complaint Intentions andBehavior: Definitional and Taxonomical Issues," Journal ofMarketing, 52 (January), 93-107.

Taylor, Shelley E. (1991), "Asymmetrical Effects of Positive andNegative Events: The Mobilization-Minimization Hypothesis,"Psychological Bulletin, 110(1), 67-85.

Wallack, Lawrence (1990), "Media Advocacy: Promoting HealthThrough Mass Communication," in Health Behavior and HealthEducation, K. Glanz, F.M. Lewis, and B.K. Rimer, eds. SanFrancisco: Jossey-Bass Publishers, 370-86.