Journal of Personality and Social Psychology1998, Vol. 74, No- 3, 646-658

Copyright 1998 by the American Psychological Association, Inc.OO22-3514/98/$3.OO

Collective Identification and Social Movement Participation

Bernd SimonWestfalische Wilhelms-Universitat

Michael LoewySan Diego State University

Stefan SturmerUniversitat Leipzig

Ulrike Weber, Peter Freytag, Corinna Habig,Claudia Kampmeier, and Peter Spahlinger

Westfalische Wilhelms-Universitat

Determinants of collective behavior, as suggested by the social identity or self-categorization approachand social movement research, were examined in 2 field studies. Study 1 was conducted in thecontext of the older people's movement in Germany and Study 2 in the context of the gay movementin the United States. Both studies yielded similar results pointing to 2 independent pathways towillingness to participate in collective action; one is based on cost-benefit calculations (includingnormative considerations), and the other is based on collective identification as an activist. Study 2included an experimental manipulation and provided evidence for the causal role of collective identi-fication as an activist. Directions for future research on the proposed dual-pathway model aresuggested.

Members of disadvantaged groups who do not want to pas-sively accept their lot have to find ways to improve their situa-tion. To do so, they can adopt a variety of strategies that canrange from individual strategies of social mobility to collectivestrategies of social change. The former rest on the belief thatone's own position can be improved by moving from one socialposition to another as an individual (Tajfel, 1981). These indi-vidual strategies thus involve leaving a disadvantaged groupphysically or at least psychologically. Collective strategies, onthe other hand, are adopted to the extent that a person believesthat "the only way for him to change these [disadvantageous]conditions . . . is together with his group as a whole" (Tajfel,1981, p. 247). Collective strategies include not only militantforms of intergroup behavior or collective action such as revoltsand strikes but also more moderate forms such as signing apetition or attending a group meeting.

Although individual social mobility strategies are often pre-

Bernd Simon, Ulrike Weber, Peter Freytag, Corinna Habig, CJaudiaKampmeier, and Peter Spahlinger, Psychologisches Institut IV, West-falische Wilhelms-Universitat, Minister, Germany; Michael Loewy, De-partment of Counseling and School Psychology, San Diego State Univer-sity; Stefan Sturmer, Fachbereich Psychologie, Universitat Leipzig, Leip-zig, Germany.

Stefan StUrmer is now at Psychologisches Institut IV, WestfalischeWilhelms-Universitat, Miinster, Germany. Peter Freytag is now at Psy-chologisches Institut, Universitat Heidelberg, Heidelberg, Germany.

This research was made possible by a grant from the DeutscheForschungsgemeinschaft (Si 428/4-1,2). We are grateful to PennyOakes, Claudia Dalbert, and Manfred Schmitt for their helpful commentson an earlier version of this article and to Dirk Wentura and ThomasKessler for statistical advice.

Correspondence concerning this article should be addressed toBernd Simon, Psychologisches Institut IV, Westfalische Wilhelms—Uni-versitat, 48149 Miinster, Germany. Electronic mail may be sent tosimon@psy.uni-muenster.de.

ferred over collective social change strategies (Lalonde & Sil-verman, 1994; Wright, Taylor, & Moghaddam, 1990), this isnot always the case. Under certain circumstances, members ofdisadvantaged groups do, in fact, engage in collective behaviorin order to improve their situation. Social movements are a casein point. They have been defined, from a psychological pointof view, as *'effort [s] by a large number of people to solvecollectively a problem that they feel they have in common"(Toch, 1965, p. 5). Questions of when and how people cometo join social movements, or form them in the first place, can beapproached from both a macrolevel and a microlevel perspective(McAdam, McCarthy, & Zald, 1988). These are not competingbut complementary perspectives. The former focuses on socio-structural, political, and organizational antecedents of socialmovements. For example, an important aim of macrolevel analy-ses is to identify the frictions and contradictions inherent inthe larger society that provide the structural basis of socialmovements (Smelser, 1962). Thus, the origins of the labormovement have been traced back to the contradictions of asociety structured by social class, whereas the roots of the femi-nist movement are seen in the integrative problems arising forwomen in modern, functionally differentiated societies (Neid-hardt & Rucht, 1993). However, we also need microlevel analy-ses in order to understand how sociostructural, political, andorganizational factors translate into concrete experiences, mo-tives, and actions of people, which may then feed back againon macrolevel factors. We therefore turn now to the microlevelperspective.

Collective, Social, and Reward Motives for SocialMovement Participation

On the basis of a comprehensive review of pertinent micro-level research, Klandermans (1997b; also Klandermans &Oegema, 1987) developed a model of social movement partici-pation, which includes four steps. These steps are (a) becoming

646

SOCIAL MOVEMENT PARTICIPATION 647

part of the mobilization potential, (b) becoming a target ofmobilization attempts, (c) becoming motivated to participate,and (d) overcoming barriers to participation. Klandermans andOegema (1987) have substantiated the relevance of these foursteps empirically in research on mobilization and participationin the Dutch peace movement. The research presented in thisarticle focuses on microlevel antecedents of group members'willingness to participate in collective actions of social move-ments. Accordingly, the third step in Klandermans' (1997b)model is of particular relevance here. In that model, motivationor willingness to participate in a specific collective action ofthe social movement is viewed as a function of the expectedcosts and benefits of participation. At this point it is importantto note that the goal of the movement, if achieved, is a collectivegood. That is, all people sympathetic to the movement's causewill benefit from goal achievement no matter whether they haveparticipated personally in collective action, and thus contributedto goal achievement, or not. For example, all workers of a fac-tory will benefit from improvement in working conditions suc-cessfully fought for in a strike no matter whether they have orhave not participated personally in the strike. It follows that thecollective good may be insufficient as a motivating force, be-cause people can hope for a "free ride." In addition to thecollective good, potential free riders may therefore need addi-tional selective Incentives to participate, which can involve bothsocial and material costs and benefits (Olson, 1977). Accordingto Klandermans (1984), three different motives for social move-ment participation can thus be distinguished, with each of thethree motives originating from a different type of expected costsor benefits. First, the collective motive derives from the collectivegoals of the movement. In line with other expectancy-valuemodels (Ajzen, 1991; Ajzen & Fishbein, 1980; Fishbein & Aj-zen, 1975), the collective motive is assumed to be directly pro-portional to the value of the collective goals for the potentialparticipator, weighted by the perceived likelihood that thesegoals will be reached. More formally, the collective motive isconceptualized as the multiplicative function of the subjectivevalue of the goals of the social movement and the subjectiveexpectation that these goals will be reached. Second, the socialmotive derives from the expected reactions of significant othersto one's own participation in collective action (e.g., ridicule oradmiration by friends or family). It is directly proportional tothe positivity of the expected reactions of others, weighted by thepersonal importance of these others' reactions to the potentialparticipator. More formally, it is conceptualized as the multipli-cative function of the subjective (positive or negative) qualityof the expected reactions of others and the personal importanceof these reactions. This approach corresponds to Ajzen andFishbein's (1980) conceptualization of a subjective norm as themultiplicative function of normative beliefs and the motivationto comply. Finally, the reward motive results from selectiveincentives having to do with other personal costs and benefitssuch as losing money or time, making new friends, or riskingone's health. As with the collective motive, the reward motiveis conceptualized as a multiplicative function of value and ex-pectancy components. All three motives are assumed to contrib-ute positively to the willingness to participate in collective actionorganized by the social movement. Willingness to participateshould increase with the expected value of the collective out-

come of participation (collective motive), with the positivity ofthe expected reactions of important others (social motive), andwith the expected value of the personal outcome of participation(reward motive). Put in terms of Ajzen and Fishbein's (1980)theory of reasoned action, the collective motive and the rewardmotive determine together the attitude toward participation incollective action, whereas the social motive corresponds to thesubjective norm component. Attitude and subjective norm thencodetermine the intention or willingness to participate in collec-tive action.

The Neglected Role of Collective Identification

The conceptualization of the potential participator in socialmovements as someone who weighs costs and benefits has beenchallenged, however, as overly atomistic and individualistic.Specifically, the neglect of the role of collective-identificationprocesses has been criticized (Friedman & Me Adam, 1992;Gamson, 1992; Kelly, 1993; Rucht, 1995). This criticism is inaccord with the social identity or self-categorization approachto group processes (Tajfel & Turner, 1986; Turner, Hogg, Oakes,Reicher, & Wetherell, 1987), which has long argued for thepivotal role of collective-identification processes in group be-havior (see also Ellemers, Spears, & Doosje, 1997). In fact, itwas Tajfel who, more than 15 years ago, reformulated Toch's(1965) definition of social movements in order to add a collec-tive-identification component. Tajfel (1981) suggested that so-cial movements should be understood on the social psychologi-cal level as "efforts by large numbers of people, who definethemselves and are also often defined by others as a group[italics added], to solve collectively a problem they feel theyhave in common, and which is perceived to arise from theirrelations with other groups" (p. 244). In other words, socialmovement participation is behavior qua group member so thatcollective identification is the basic social psychological processunderlying that behavior (Turner et al., 1987). Unfortunately,this definition, as well as the social identity or self-categoriza-tion approach in general, has had little influence on social move-ment research so far (for exceptions, see Kelly, 1993; Kelly &Breinlinger, 1996; also Kawakami & Dion, 1993, 1995). Thisis a consequence of the fact that social movement research andresearch on group processes and intergroup relations have longcoexisted as two rather strictly separated domains, with theformer being oriented more toward sociology and the latterbeing oriented more toward psychology.

The aim of this article is to bring together these differentresearch traditions and thus contribute to empirical and theoreti-cal integration (see also Simon, 1998). Two field studies arepresented in which we examine social psychological determi-nants of group members' willingness to participate in collectiveactions organized by contemporary social movements. Study1 concerns the older people's movement (Gray Panthers) inGermany. The focus of Study 2 is on the gay movement in theUnited States. In both cases, we distributed questionnairesamong members of the respective mobilization potential in orderto measure and compare the influence of collective-identificationprocesses and of collective, social, and reward motives on will-ingness to participate in collective action. Whereas Study 1 isrestricted to correlational analyses, we included an experimental

648 SIMON ET AL.

manipulation of the salience of collective identity in Study 2 inorder to move toward a causal analysis of the role of collectiveidentification.

Study 1

As indicated above, social movement research and researchon group processes and intergroup relations have hitherto beenpursued quite separately with little theoretical and empiricalintegration. A notable exception is the work by Kelly and col-leagues (for reviews, see Kelly, 1993; Kelly & Breinlinger,1996). Drawing directly on the social identity or self-categoriza-tion approach, they explored the role of collective identificationin the process of becoming motivated to participate in collectiveaction. For example, Kelly and Kelly (reported in Kelly, 1993)hypothesized and found that, for members of a British tradeunion, strength of identification as unionists was a positive pre-dictor of the willingness to participate in trade union activism.In line with Klandennans (1984), collective, social, and rewardmotives were also measured, but these motives did not predictwillingness to participate. Similarly, in the context of the wom-en's movement, Kelly and Breinlinger (1995) found that wom-en's willingness to participate in collective action increased withcollective identification as women. They also measured collec-tive identification as feminist activists, and this variable turnedout to be an even stronger predictor of willingness to participate(and actual participation) in collective action. Unfortunately,this latter variable was measured only with a single item, andcollective, social, and reward motives were not considered sys-tematically in that study. Nevertheless, these results are encour-aging. They clearly point to the role of collective-identificationprocesses in social movement participation and thus warrantfurther investigation.

The purpose of this study was to further substantiate that,beyond the collective, social, and reward motives usually consid-ered in social movement research, collective identification is aunique predictor of group members' willingness to participatein collective action. More specifically, two levels of collectiveidentity are considered here. The first concerns the broader socialcategory from which the social movement under scrutiny typi-cally recruits its members. In the present case that is the socialcategory of older people. The second level targets the socialmovement itself, the Gray Panthers. Unlike the research by Kellyand Breinlinger (1995), identification at both levels is measuredin the present study with comparable multi-item scales. It isexpected that identification with the social movement itself (theGray Panthers) will be a better predictor than identification withthe broader recruitment category (older people). This shouldbe so because the former is more directly tied to an activistidentity, which typically has embedded within it specific impli-cations for action (Friedman & McAdam, 1992). This reasoningis also consistent with Ajzen and Fishbein's (1977) principleof compatibility, which states that attitudes predict behaviors tothe extent that predictor and criterion are matched in terms oflevel of specificity or generality. By the same token, identifica-tion with the specific group of Gray Panthers should be a betterpredictor of specific activist behavior, or of the willingness toengage in such behavior, than identification with the more inclu-sive social category of older people.

To summarize, our main hypotheses were as follows. Collec-tive identification, especially at the level of the specific socialmovement organization (the Gray Panthers), should contributepositively to the prediction of the willingness to participate incollective action. This relationship should remain intact evenafter the contributions of the three motives typically consideredin social movement research are partialed out. For these lattermotives, we generally expected positive relationships with will-ingness to participate. However, we did not venture any compar-ative predictions as to the relative contributions of collective,social, and reward motives in this specific social movementcontext. We refrained from doing so because the extant socialmovement literature, although it acknowledges possible varia-tion in the relevance of each motive across social movementcontexts, has little to say as to the exact determinants of thisvariation.

Method

Participants and Procedure

Ninety-five registered members of the Senior Protection League GrayPanthers {Senioren-Schutz-Bund Graue Panther) participated in thisquestionnaire study (mean age = 74 years, SD = 7.4 years, range = 6 0 -91 years). The sample consisted of 80 female and 14male participants; 1participant did not provide information regarding his or her gender. Asto (former) occupation, the sample was quite heterogeneous. About onehalf of the participants had been employed in the areas of craft andadministration or in the business sector. Approximately 20% of the parti-cipants had been housewives. The remaining participants had worked invarious other fields. Only a small minority (less than 4%) had a univer-sity degree. About two thirds of the participants were widowed, 16%were still married, 13% were separated from their former partners, and5% were single.

Participants were recruited at regular group meetings of the targetorganization. The study was introduced as part of a research projectconcerned with the living conditions of older people. A female and amale investigator distributed the questionnaires. On average, it tookparticipants about 30 min to complete the questionnaire. Care was takento ensure that none of the participants completed more than one question-naire and that questionnaires were completed individually andanonymously.

The Questionnaire

Overview. To ensure that the questionnaire included meaningfulitems (e.g., items concerning relevant costs and benefits), the investiga-tors consulted extensively with leading activists of the Gray Panthers.In addition, the questionnaire was further refined on the basis of a pretestwith 10 people from the relevant age group. The variables measured inthe final questionnaire fall into six main categories: (a) identificationwith the social category of older people, (b) identification with the olderpeople's movement (i.e., the Gray Panthers), (c) collective motive, (d)social motive, (e) reward motive, and (f) willingness to participate incollective action organized by the older people's movement. The specificitems are described below.

Moreover, as Klandennans (1997b) pointed out, people belong to themobilization potential of a given social movement if, in addition to acollective identity, they also share feelings of injustice and believe inthe possibility of social change. Two additional questions therefore mea-sured the extent to which participants thought that older people werediscriminated against in society and the extent to which they thoughtthat the older people's movement could be effective in combating age

SOCIAL MOVEMENT PARTICIPATION 649

discrimination. These questions were answered on 5-point scales, withhigher scores indicating a higher degree of perceived discrimination oreffectiveness. They were inserted, between the identification items andthose items concerning the collective motive. On the last page, partici-pants provided demographic data (i.e., age, occupation, marital status,and so on).

Identification with older people. This variable was measured withthe following four items: "I feel strong ties with other old people," " Iidentify with the group of old people," "In many respects, I am likemost other old people,'' and ' 'In many respects, old people are similarto each other." Ratings were made on 5-point scales, ranging fromnever/not true (1) to very oftenlvery true (5) .

Identification with the older people's movement. Tb measure thisvariable, the same four items used to measure identification with olderpeople were reworded so that they referred to the Gray Panthers insteadof the social category of older people (e.g., "I feel strong ties withother Gray Panthers'').

Collective motive. In line with Klandermans (1984) and other ex-pectancy-value models (Ajzen, 1991; Ajzen & Fishbein, 1980; Fish-bein & Ajzen, 1975), the value and expectancy components of the collec-tive motive were measured separately and were dien combined multipli-catively. As to the value component, participants evaluated three goalsof the older people's movement: guaranteed minimum pension and medi-cal care, more self-determination in the homes for older people, and anti-age-discrimination legislation. These goals were extracted from relevantmovement publications (Unruh, 1990). Consultations with leadingmovement activists, pretesting, and responses to an additional questionin the main questionnaire confirmed that these were indeed relevantcollective goals. As to the expectancy component, participants indicatedthe extent to which they expected that a sufficient number of peoplecould be mobilized to achieve these goals. Ratings were again made on5-point scales, ranging from 1 to 5, with higher scores indicating a morefavorable evaluation of the collective goals or a stronger expectationthat a sufficient number of people would participate.

Social motive. Again, two components were measured separatelyand were then combined multiplicatively. First, the expected reaction ofothers to one's own participation in collective action was measured.This was done for five groups: family (spouse, partner, children), otherrelatives, neighbors, friends within the movement, and other friends.Ratings were made on 7-point bipolar scales ranging from very negativereactions ( - 3 ) to very positive reactions (3) , with a neutral midpoint(0). Second, the personal importance of the reactions of the five groupsof people was measured on 5-point scales ranging from not important(1) to very important (5). (Note that the expected reaction and impor-tance components measured here correspond, respectively, to the con-cepts of normative beliefs and motivation to comply as specified in Ajzenand Fishbein's, 1980, theory of reasoned action. Similarly, the socialmotive as the multiplicative combination of the expected reaction andimportance components corresponds to Ajzen and Fishbein's subjectivenorm.)

Reward motive. Value and expectancy components were measuredseparately and were then combined multiplicatively. As to the valuecomponent, participants evaluated three possible gains (actual improve-ment of own living conditions, social contact with other older people,and meaningful leisure time activity) and two possible losses (healthrisks and loss of time) that may result from their participation in collec-tive action. As reported above, consultation with leading movement ac-tivists and pretesting with 10 people from the relevant age group ensuredthat these outcomes were indeed perceived as relevant gains and losses.As to the expectancy component, participants rated the likelihood ofeach of the three gains and the two losses. Ratings were made on 5-point scales, ranging from 1 to 5, with higher scores indicating biggersubjective gains and losses or stronger expectations that the listed gainsand losses would ensue.

Willingness to participate in (future) collective action. Willingnessto participate was measured concerning each of the following four ac-tions: supporting other older people in dealing with public authorities,campaigning in public places and homes for older people, foundinghomes in which older people can live a self-determined life, and demon-strating publicly. These actions were selected on the basis of movementpublications, consultations with movement activists, and pretesting withpeople from the relevant age group. Ratings of one's own willingnessto participate were made on 5-point scales ranging from very low (1)to very high (5) .

Results

Preliminary Analyses

For each participant, we calculated an overall (mean) scorefor identification with older people and one for identificationwith the Gray Panthers (Cronbach's a = .71 and .76, respec-tively). Identification with the Gray Panthers was stronger thanidentification with older people (M = 3.7 vs. 3.3), F(\, 92)1

= 32.46, p < .001, but both grand means were significantlyabove the midpoint of the respective 5-point scales (alpha setat .05). The latter was also true for the ratings concerning per-ceived discrimination against older people and expected effec-tiveness of the older people's movement (Mdiscrim = 4.7, Metiea —4.5). Taken together, these data confirm that participants indeedbelonged to the mobilization potential (Klandermans, 1997b).

For each of the three motives, we then calculated an overallscore as follows. As to the collective motive, the value compo-nent was calculated by averaging the evaluations of the threecollective goals (Cronbach's a = .80). This score was thenmultiplied by the expectancy rating (i.e., the perceived likeli-hood that a sufficient number of people could be mobilized toachieve the collective goals). As to the social motive, we firstmultiplied, for each group of others (family, friends etc.), therating of the expected (positive or negative) reaction by thecorresponding personal importance rating. These product termswere then averaged (Cronbach's a = .86). As to the rewardmotive, product terms were also first calculated for each gainand each loss by multiplying the corresponding value and expec-tancy components. Product terms were then averaged acrossgains and losses, separately. Finally, the mean product term forlosses was subtracted from the mean product term for gains.Whereas the mean product term for gains possessed a highdegree of internal consistency (Cronbach's a = .78), that wasnot the case for the mean product term for the two loss items(r = .13, ns). At this point it should be noted, however, that ahigh degree of internal consistency is not a necessary require-ment for the reward motive to be included in the analysis. Thereis no compelling reason to postulate that the reward motivederives solely from closely related, almost mutually redundant,outcomes. Instead, it is very likely that people are motivated ordiscouraged by some outcomes but not by others. This selectiv-ity should also vary across people, thus a more heterogeneoussample of possible outcomes is quite appropriate to measurethe reward motive. Analyses with the reward motive that werebased on both gains and losses or on gains alone yielded very

1 Degrees of freedom may vary in both studies because of missingdata.

650 SIMON ET AL.

Table 1Intercorrelations, Means, and Standard Deviations for Motives, Identification, andWillingness to Participate (Study I)

Variable

1. Collective motive2. Social motive3. Reward motive4. Identification with older people5- Identification with Gray Panthers6. Willingness to participateM*SD

1

_

13.35.9

2

.32**—

3.94.8

3

31**57***-—

9.17.8

4

.19

.37***

.25*—

3.30.9

5

27**45***59***64***—

3.70.9

6

44***.59***72***31**57**_ .3.41.1

a Scale range varies between variables; see Method section for details. Higher means indicate strongermotives, stronger identification, and greater willingness to participate.

.05. .01. .001.

similar results. In fact, if anything, the predictive value of thereward motive was further strengthened when both gains andlosses were taken into account.

Finally, we calculated an overall (mean) score for willingnessto participate (Cronbach's a = .83). Grand means and standarddeviations for all relevant variables as well as their intercorrela-tions are presented in Table 1. Higher means indicate strongermotives, stronger identification, and greater willingness toparticipate.

Main Analyses: Motives, Collective Identification, andSocial Movement Participation

As expected, the collective, social, and reward motives, aswell as identification with older people and identification withthe Gray Panthers, were significantly and positively correlatedwith willingness to participate (see Table 1). Correlation coef-ficients ranged from .31 for identification with older people to.72 for the reward motive (on the basis of both gains and losses).A multiple regression analysis was then performed in which thethree motives and the two identification measures served aspredictor variables, and willingness to participate served as thecriterion. As can be seen in Table 2, the collective motive andthe reward motive remained significant predictors, whereas theunique contribution of the social motive was only marginal. Atthe same time, identification with the Gray Panthers remaineda significant predictor, whereas identification with older peoplelost its predictive value.

To summarize, the collective motive, the reward motive, andidentification with the Gray Panthers have independent effects

on the willingness to participate. A similar trend is also visiblefor the social motive, whereas identification with older peoplehas no unique contribution to make to the prediction of thewillingness to participate in collective action.

Discussion

The results of this study clearly confirmed our predictions.First, we replicated for all three motives typically considered insocial movement research (Klandermans, 1984; Klandermans &Oegema, 1987) a significant correlation with willingness to par-ticipate in collective action. It was further demonstrated that thecollective motive and the reward motive made unique contribu-tions to the prediction of willingness to participate even whenall three motives and two collective-identification measures weretaken into account simultaneously. The contribution of the socialmotive was only marginal in that analysis.

Second, willingness to participate in collective action wasalso significantly related to collective identification. This wastrue both at the level of collective identity as a member of thesocial category of older people and at the level of collectiveidentity as a Gray Panther. Also as expected, identification atthe latter level was clearly the better predictor. It remained asignificant predictor of the willingness to participate in collec-tive action even when collective, social, and reward motivesand identification with older people were taken into account.Identification with older people retained no unique predictivevalue.

The finding that identification with the social movement (i.e.,the Gray Panthers) was a better predictor than identification

Table 2Regression Analysis With Willingness to Participate as Criterion (Study 1)

Predictor

0r(81)

Collective motive

.203.05**

Social motive

.141.737

Reward motive

.475.30***

Identification witholder people

.020.23

Identification withGray Panthers

.242.44*

Note. R2 = .70, F(5, 81) = 37.29, p < .001.ip < .10 (marginally significant). *p ^ .05. **p < .01. ***p *= .001.

SOCIAL MOVEMENT PARTICIPATION 651

with the broader recruitment category (i.e., older people) sup-ports the assumption that a specific activist identity is particu-larly conducive to social movement participation (Friedman &McAdam, 1992; Kelly & Breinlinger, 1995). Apparently, mem-bership in the Gray Panthers makes an activist identity available,which includes specific implications for activist behavior. Thewillingness to perform such behavior then increases with theextent to which the activist identity is embraced. Alternatively,one could also argue that identification with the Gray Pantherswas the better predictor simply because it was more pronouncedand thus above some necessary threshold. Identification withthe Gray Panthers was indeed stronger than identification witholder people in general, and this may have been so because theGray Panthers are the less inclusive group and are thereforebetter able to provide a sense of optimal distinctiveness (Brewer,1991). Note that this alternative explanation focuses on thequantitative difference between identification with older peopleand identification with the Gray Panthers and gives less weightto the notion of a qualitatively distinct activist identity. Unfortu-nately, the present data do not allow us to rule out this alternativeexplanation, but we return to this issue when we discuss theresults of Study 2.

In any case, the results of this study strongly suggest thatcollective-identification processes constitute a second pathwayto social movement participation, in addition to and independentof cost-benefit calculations (including normative considera-tions). However, there are at least two major limitations to thecurrent study. First, the findings may be limited to the specificsocial movement under scrutiny here. The Gray Panthers are aformal organization with a registered membership, which is nottrue for all social movements (McAdam et al., 1988). It there-fore remains to be seen whether the present findings also general-ize to a less formally organized social movement. Second, giventhe correlational nature of this study, no definite conclusionsabout cause-effect relationships can be drawn. As with all cor-relational research, the direction of causality remains unclear,and there is always the possibility that the observed correlationsare spurious because of the influence of a third variable. Wetherefore conducted a second study to address these limitations.

Study 2

We pursued mainly two aims in this study. First, we wantedto examine whether the findings from Study 1 also hold true ina different social movement context. For this purpose, we re-cruited research participants in the context of the U.S. gay move-ment. Unlike the Gray Panthers in Study 1, the gay movementis not a single formal organization but rather a more looselyknit network of groups, which nevertheless elicits a fair degreeof collective identification (Adam, 1987; Cruikshank, 1992).2

Converging evidence for the significance of collective-identifi-cation processes from these two different social movement con-texts will strengthen our confidence in the validity and generaliz-ability of the obtained findings.

Our second aim was to move toward a causal analysis of therole of collective-identification processes in social movementparticipation. We therefore included an experimental manipula-tion of the strength of identification with the gay movement inour research design. Following a similar approach by Simon,

Pantaleo, and Mummendey (1995), we manipulated collectiveidentification by varying the salience of the common fate of gaypeople as a threatened minority. We expected that making theircommon fate as members of a threatened minority highly salientwould strengthen gay respondents' collective identification withthe gay movement, which should, in turn, lead to increasedwillingness to participate in collective action organized by thegay movement (for a similar reasoning, see Tajfel, 1981).

At this point it should be noted that, parallel to Study 1, wemeasured not only identification with the gay movement butalso identification with the broader category of gay people ingeneral. We did not rule out that the experimental manipulationmight also affect identification with gay people in general. How-ever, in a threatening or common fate situation, the more politi-cized level of collective identity should be particularly relevant(Klandermans, 1997a; Melucci, 1996; Taylor & Whittier, 1992,1995); thus we expected an experimental effect especially onidentification with the gay movement.

Method

Participants and Procedure

One hundred seventeen gay men (mean age = 39 years, SD = 11.8years, range = 19-73 years) volunteered to complete a questionnairethat was introduced as being concerned with the living conditions ofgay people. Eighty-seven participants worked professionally in variousfields, including the business sector, social services, teaching institutions,the fine arts, and the media. Sixty-five of these 87 participants indicatedthat they had a university or college education. Thirty participants werecurrently unemployed, of whom 19 indicated that they had a universityor college education. Participants were recruited at meetings of variousgay groups or in gay coffee houses in San Diego or San Francisco. Afemale and a male investigator distributed the questionnaires. On aver-age, it took participants about 20 min to complete the questionnaire.Care was taken to ensure thai none of the participants completed morethan one questionnaire and that questionnaires were completed individu-ally and anonymously. Though most questionnaires were completed im-mediately after they had been distributed, others were taken away andwere returned later by mail.

The Questionnaire

Overview. To ensure that the questionnaire included meaningfulitems (e.g., items concerning relevant costs and benefits), the investiga-tors drew on earlier research on gay movement participation (Deppe &Elkemann, 1996) and consulted extensively with gay activists. Parallelto Study 1, we measured the following variables: (a) identification withthe social category of gay people, (b) identification with the gay move-ment, (c) collective motive, (d) social motive, (e) reward motive, and(f) willingness to participate in collective action organized by the gaymovement. The specific items are described below. However, before themeasurement of these variables, the salience of the common fate of gaypeople as a threatened minority was manipulated by written instructions(see Experimental manipulation section below).

Tb ensure that participants belonged to the mobilization potential ofthe gay movement, we also measured the extent to which participantsthought that gay people were discriminated against in society and theextent to which they thought that the gay movement could be effective

2 We are grateful to an anonymous reviewer for drawing our attentionto this difference.

652 SIMON ET AL.

in combating discrimination against gay people (Klandermans, 1997b).The first question was inserted immediately after the experimental ma-nipulation, whereas the other question was placed after the items measur-ing identification with the gay movement. Ratings were made on 5-point scales, with high ratings indicating a high degree of perceiveddiscrimination or effectiveness. On the last page, participants furnisheddemographic data (age, education, occupation, and so on).

Experimental manipulation. The salience of the common fate of gaypeople as a threatened minority was manipulated as follows. In the(high-salience) common fate condition, participants were first asked torecall "an incident that you experienced yourself or you have heardabout where gay people were threatened or treated violently by non-gay people (e.g., gay bashing)." Participants were requested to writedown a short description of the recalled incident and then rate howviolent and how threatening that incident was and how distressed theywere as a result of that incident (or how distressed they were when theyheard about it). In the (low-salience) control condition, participantswere simply asked to write down a short description of ' 'what you didtoday" and then rate how usual or unusual their day was, how muchtime they had spent with friends, and how much time they had spentworking. Ratings in both conditions were made on 5-point scales, withhigher scores indicating more violence, greater threat, and more distress,or, a more unusual day, more time spent with friends and more time spentworking. Participants were randomly assigned to either the common fatecondition or the control condition. After the experimental manipulationand the three follow-up questions, participants proceeded to the othermeasures.

Identification with gay people. This variable was measured with thefollowing seven items: ' 'I see myself as a gay man,'' ' 'I feel strong tieswith other gay men," "I am glad to be gay," "I identify with the gaycommunity," "It is important to me to belong to the gay community,"" I feel ashamed to be gay," and " I try to hide that I am gay." Ratingswere made on 5-point scales, ranging from never {1) to very often (5).

Identification with the gay movement This variable was measuredwith the following five items: " I see myself as a member of the gaymovement," " I feel strong ties with the gay movement," " I identifywith the gay movement," "It is important to me to belong to the gaymovement," and "I consider myself an activist of the gay movement."Ratings were made on 5-point scales, ranging from never/not at allagree (1) to very often/strongly agree (5).

Collective motive. As in Study 1, the value and expectancy compo-nents of the collective motive were measured separately and were thencombined multiplicatively. As to the value component, participants eval-uated five goals of the gay movement: (a) legal recognition of gaymarriages, (b) equal employment opportunities, (c) allowing gay peopleto serve in the military, (d) abolishing laws that prohibit gay sexualrelationships, and (e) educating the public about gay issues and gaypeople. Ratings were made on 7-point bipolar scales, ranging fromstrongly negative (—3) to strongly positive (3), with a neutral midpoint(0). The five goals were extracted from relevant movement publications,and consultations with several gay activists and responses to an addi-tional question in the main questionnaire confirmed that these wereindeed relevant collective goals. As to the expectancy component, parti-cipants indicated the extent to which they expected that a sufficientnumber of people could be mobilized to achieve the above goals. Thisrating was made on a 5-point scale, with a higher score indicating astronger expectation that a sufficient number of people would participate.

Social motive. Again, two components were measured separatelyand were then combined multiplicatively. First, the expected reaction ofothers to one's own participation in collective action was measured.This was done again for five groups: family (e.g., parents, siblings,children), nongay friends, colleagues or classmates, supervisors or pro-fessors, and gay friends. Ratings were made on 7-point bipolar scales,ranging from strongly negative ( - 3 ) to strongly positive ( 3 ) , with a

neutral midpoint (0). Second, the personal importance of the reactionsof the five groups of people was measured on 5-point scales, rangingfrom not important (1) to very important (5).

Reward motive. Value and expectancy components were measuredseparately and were then combined multiplicatively. As to the valuecomponent, participants evaluated two possible gains (meeting othergay men and improvement of own personal life) and three possiblelosses (negative consequences at work or school, negative consequencesfor one's family, and reduction by others to one's sexual orientationwithout appreciation of the whole person) that may result from theirparticipation in collective action. As to the expectancy component, theyrated the likelihood of each of the two gains and the three losses. Ratingswere made on 5-point scales, with higher scores indicating bigger subjec-tive gains and losses or stronger expectations that the listed gains andlosses would ensue.

Willingness to participate in (future) collective action. Willingnessto participate was measured concerning each of the following four ac-tions: public demonstrations, gay pride activities, civil disobedience(e.g., sit-ins in government offices or businesses), and public discus-sions. Ratings were made on 5-point scales ranging from not at allwilling (1) to strongly willing (5). Finally, for each of these actions,participants indicated (yes or no) whether they had already participatedin that action during the past 3 years.

Results

Preliminary Analyses

For each participant, we calculated an overall (mean) scorefor identification with gay people and one for identification withthe gay movement (Cronbach's a = .83 and .94, respectively;all items were coded such that higher scores indicated strongeridentification). Identification with gay people in general wasstronger than identification with the gay movement (M = 4.0vs. 3.0), F ( l , 116) = 170.72, p < .001. The grand mean foridentification with gay people was significantly above the scalemidpoint. This was also true for the ratings concerning perceiveddiscrimination against gay people and expected effectiveness ofthe gay movement (Muscrim = 4.4, Mcf{eci = 4.0). Taken together,these data confirm that participants indeed belonged to the mobi-lization potential of the gay movement (Klandermans, 1997b).

Parallel to Study 1, we then calculated an overall score foreach of the three motives. Internal consistency was satisfactoryfor all three motives (Cronbach's alpha was between .61 and.71), although it was again lowest for the reward motive (Cron-bach's a = .61 and .66 for gains and losses, respectively).Finally, we calculated an overall (mean) score for willingnessto participate (Cronbach's a = .75). Grand means and standarddeviations for all relevant variables as well as their intercorrela-tions are presented in Table 3. Higher means indicate strongermotives, stronger identification, and greater willingness toparticipate.

Main Analyses

We first conducted similar correlation and multiple regressionanalyses as in Study 1 in order to examine whether the resultsobtained in the previous study could be replicated in the contextof the gay movement. We then analyzed the effects of the experi-mental variable, salience of common fate as a threatenedminority.

SOCIAL MOVEMENT PARTICIPATION 653

Table 3Intercorrelations, Means, and Standard Deviations for Motives, Identification, PreviousParticipation, and Willingness to Participate (Study 2)

1.2.3.4.5.6.7.Ma

SD

Variable

Collective motiveSocial motiveReward motiveIdentification with gay peopleIdentification with gay movementPrevious participationWillingness to participate

1

_

24.57.2

2

.14—

3.13.4

3

24**42***—

7.76.1

4

.1837*##45***—

4.00.7

5

.1638***49***

• . 5 9 * * *

—

3.01.0

6

.02

.20*

.34***

.23**

.48***—

1.81.0

7

.27**39***43***40***.64***.57***

—3.30.9

a Scale range varies between variables; See Method section for details. Higher means indicate strongermotives, stronger identification, and greater willingness to participate.* p £ .05. **p s .01. ***/? ^ .001.

Motives, Collective Identification, and Social MovementParticipation

As in Study 1, the collective, social, and reward motives, aswell as identification with gay people and identification withthe gay movement, were significantly and positively correlatedwith willingness to participate (see Table 3). Correlation coef-ficients ranged from .27 for the collective motive to .64 foridentification with the gay movement. A multiple regressionanalysis was then performed in which the three motives and thetwo identification measures served as predictor variables, andwillingness to participate served as the criterion (see Table 4) .The results were very similar to those of Study 1, with oneexception. As in Study 1, the collective motive retained a sig-nificant predictive value, and the social motive retained a mar-ginal predictive value. Moreover, identification with the gaymovement was clearly a significant predictor, whereas identifi-cation with gay people had no predictive value. Unlike in Study1, however, the reward motive had no unique contribution tomake to the prediction of the willingness to participate.

Study 2 also provided us with data concerning past behavior.More specifically, we counted in how many of the collectiveactions listed in the questionnaire participants had actually par-ticipated during the past 3 years. We entered these frequencydata as an additional predictor variable into the original multipleregression analysis, with willingness to participate as the crite-rion. The results are presented in Table 5. Although the new

variable received a significant regression weight, the resultsfor the original variables remained virtually unchanged. Thecollective motive retained a significant predictive value, and thesocial motive retained a marginal predictive value. Identificationwith the gay movement also remained a significant predictor,whereas identification with gay people had no predictive value.

Finally, we performed a multiple regression analysis, with thethree motives and the two identification measures as predictorvariables and the frequency of previous participation in collec-tive action as the criterion, R2 = .25, F(5, 109) = 7.24, p <.001. Only identification with the gay movement turned out tobe a significant predictor, /3 = .46, ((109) = 4.24, p < .001.In addition, the reward motive was a marginal predictor, 0 —.19, t(\09) = 1.80, p = .08. All other regression coefficientswere nonsignificant (ps > .23).

Experimental Effects

The incidents recalled in the common fate condition wereindeed experienced as violent (A/ = 3.6), threatening (Af =4.3), and distressing (Af = 4.1). All means were significantlyabove the scale midpoint (3). One-way analyses of variance(ANOVAs) revealed that the experimental manipulation had thepredicted effect on identification with the gay movement (Af =2.8 vs. 3.2 for the control and common fate conditions, respec-tively), F ( l , 115) = 4.54, p < .05, and on willingness to

Table 4Regression Analysis With

Predictor Collective motive

0/(109)

.152.11*

Willingness to

Social motive

.141.74t

Participate as Criterion (Study 2)

Identification withReward motive gay people

.080.87

-0.04-0.46

Identification withgay movement

.545.84***

Note. R2 = .45, F(5, 109) = 18.12, p < .001.t p ^ .10 (marginally significant). * p =s .05. ***p ^ ,001,

654 SIMON ET AL.

Table 5Regression Analysis With Willingness to Participate as Criterion and Frequencyof Previous Participation as Additional Predictor Variable

Predictor

0f(108)

Note. R2

Previousparticipation

.354.62***

= .54, F(6, 108)

Collectivemotive

.182.68**

= 21.47, p <

Socialmotive

.141.90f

.001.

Rewardmotive

.010,15

Identification withgay people

.000.02

Identification withgay movement

.384.15***

t p s .10 (marginally significant). .01. .001.

participate (At = 3.1 vs. 3.5), F ( l , 115) = 8.50, p < .01. 'Both identification with the gay movement and willingness toparticipate increased significantly when common fate as athreatened minority was made salient (see Figure 1). No effectwas obtained for identification with gay people (M = 3.9 forthe control condition, M = 4.0 for the common fate condition;F < 1), nor for any of the three motives (Fs < 2.11, ps >. 14). Finally, there was a marginal effect on perceived discrimi-nation (Af = 4.2 for control condition, M - 4.5 for commonfate condition), F ( l , 115) = 2.85, p < .10; for expected effec-tiveness of the gay movement, F ( l , 115) = 2.03, p > .15.

We also examined whether identification with the gay move-ment functioned as a mediator of the experimental effect onwillingness to participate in collective action. According toBaron and Kenny (1986), three conditions must hold to establishmediation: First, the experimental variable must affect identifi-cation with the gay movement (i.e., the mediator); second, theexperimental variable must affect willingness to participate; andthird, identification with the gay movement must affect willing-ness to participate even when the experimental effect on willing-

I controlcondition

I commonfatecondition

Identificationwith the gaymovement

Willingness toparticipate in

collective action

ness to participate is controlled. The first two ANOVAs reportedabove confirm that Conditions 1 and 2 hold. (Note that theseANOVAs are equivalent to regression analyses with the experi-mental variable as the predictor and either identification withthe gay movement or willingness to participate as the criterion.)To check Condition 3, we performed a multiple regression analy-sis with both the experimental variable and identification withthe gay movement as predictors and willingness to participateas the criterion, R2 = .43, F(2, 114) = 42.34, p < .001. Identi-fication with the gay movement received a significant regressioncoefficient in this analysis, 0 - .61, r(114) = 8.43, p < .001,demonstrating that Condition 3 holds as well. As expected, thedirect experimental effect on willingness to participate obtainedin the multiple regression analysis was clearly weaker than thetotal effect in the original analysis (2.0% vs. 6.9% explainedvariance). Moreover, using the test procedure suggested by So-bel (1982; see Baron & Kenny, 1986), we confirmed that theindirect effect of the experimental variable by way of identifica-tion with the gay movement was indeed significant, r(114) =2.08, p < .05. The relevant path diagram is depicted in Figure2. Note that, although the effect of the experimental variableon willingness to participate was clearly attenuated when themediator was controlled, its direct effect was still significant(see Figure 2) . However, perfect mediation is rare in our fieldbecause most (social) psychological phenomena have multiplecauses, thus the search for partial mediation is usually the morerealistic goal (Baron & Kenny, 1986).

Finally, we performed separate multiple regression analysesfor the control and common fate conditions, with the three mo-tives and the two identification measures as predictors and will-ingness to participate as the criterion. Identification with thegay movement remained a significant predictor in both analyses,P = .51 and .48, ts > 3.37, ps s .001, for the control andcommon fate conditions, respectively. In addition, the socialmotive was a significant predictor, 0 = .23, f(53) = 1.98, p <.05, and the collective motive was a marginal predictor, /J =.19, r(53) = 1.79, p = .08, in the control condition, R2 = .52,F(5, 53) = 11.49, p < .001. In the common fate condition,identification with the gay movement was the only significantpredictor, R2 = .36, F(5, 50) - 5.58, p < .001; for all otherregression coefficients in both conditions, /JS > .39.

Figure 1. Identification with the gay movement and willingness toparticipate in collective action organized by the gay movement as afunction of the salience of common fate.

3 The experimental effect remained intact even after the frequency ofprevious participation was taken into account as a covariate, F(l, 114)= 3.95, p < .05.

SOCIAL MOVEMENT PARTICIPATION 655

.14"*

Figure 2. Path diagram (with standardized regression coefficients) ofthe mediational role of identification with the gay movement. 'Standardregression coefficients from multiple regression analysis. *p s .05.***p < .001.

Discussion

This study replicated and extended the findings from Study1 concerning the role of collective-identification processes insocial movement participation. Even though Study 2 focused ona less formally organized social movement than Study 1, thefindings are remarkably similar. As in Study 1, willingness toparticipate in collective action was significantly related to col-lective identification, both at the level of the broader recruitmentcategory (i.e., gay people) and at the level of the specific socialmovement (i.e., the gay movement). Also, identification at thelatter level was again the better predictor and retained its pre-dictive value even when collective, social, and reward motiveswere taken into account. Moreover, frequency of previous partic-ipation in collective action was measured and used as an addi-tional criterion in the present study, and the same relationshipswere observed as for willingness to participate in (future)actions.

The present data speak also to an issue raised in the Discus-sion of Study 1. There, it was speculated that identification withthe movement (i.e., the Gray Panthers) may have been a betterpredictor of willingness to participate than identification withthe broader recruitment category (i.e., older people) simplybecause it was more pronounced and thus above some necessarythreshold. However, this quantitative argument can be ruled outfor the present study, because this time, identification with themovement was less pronounced than identification with thebroader recruitment category (i.e., gay people). Instead, it ap-pears that there exists a qualitatively distinct identity as a gayactivist that is particularly conducive to participation in collec-tive action.

As to the motives typically considered in social movementresearch, the present results are also very similar to those ofStudy 1, with one exception. Whereas the collective motive wasagain a significant predictor and the social motive a marginalpredictor of willingness to participate in collective action, thereward motive played no significant role in the present context.Thus, despite the generally high degree of similarity betweenthe results of both studies, there was also some variation acrosssocial movement contexts in the relative importance of these

three motives. If replicated in future research, this finding mayrequire not only theoretical elaboration but it may also haveimportant practical implications for successful mobilizationcampaigns, for it would suggest that, in order to tailor successfulmobilization campaigns, social movement "entrepreneurs"need to analyze very carefully which motives are particularlyrelevant to their "clientele."

So far, we have discussed only correlational data. As indicatedearlier, the quality of such data may be compromised, amongothers, by the third-variable problem. For example, in the presentcase, one could possibly argue that the relationship betweenidentification with the gay movement and willingness to partici-pate in collective action was spurious because both variableswere actually driven by a common underlying individual-differ-ences variable such as general action orientation. People highon general action orientation may strongly identify with almostany social movement whose goal they find acceptable. At thesame time, they may be very willing to participate in all kinds ofactions associated with acceptable goals. However, such generalaction orientation should also be reflected in past behavior sothat people high on general action orientation should reportmore instances of previous participation in collective action thanpeople low on general action orientation. Fortunately, we hadcollected data concerning previous participation from our parti-cipants and could use them as an (indirect) control variable.Because the original results were replicated, it seems very un-likely that the observed relationship between identification andwillingness to participate was artificially produced by a con-found with general action orientation. This specific control anal-ysis also suggests that knowledge of previous participation isby no means sufficient to predict willingness to participate inthe future. In other words, identification with the gay movementand the (collective and social) motives measured more than justa general orientation or habit to participate in collective action.

Nevertheless, it is very difficult, if not impossible, to solvethe third-variable problem entirely with correlational methodsbecause the universe of such (third) variables is theoreticallyinfinite. Moreover, the directionality problem typically remainsunsolved in correlational research. Certainly a better way todeal with these issues is controlled experimentation, which iswhy we included an experimental manipulation in our researchdesign. More specifically, we experimentally varied the salienceof the common fate of gay people, and this manipulation af-fected, as predicted, both identification with the gay movementand willingness to participate in collective action organized bythe movement. Once their common fate as a threatened minoritywas made highly salient, research participants identified morestrongly with the gay movement and were also more willing toparticipate in collective action. Moreover, we could demonstrate,beyond these parallel effects, that collective identification withthe movement played at least a partial role in mediating theexperimental effect on willingness to participate.

General Discussion

The aim of this article was to bring together two differentresearch traditions concerned with collective behavior, namelythe social identity or self-categorization approach (Tajfel &Turner, 1986; Turner et al., 1987) and social movement research

656 SIMON ET AL.

(e.g., Klandermans, 1997b). These two traditions have typicallyfocused on quite different social psychological determinants ofcollective behavior; the former on collective-identification pro-cesses and the latter on collective, social, and reward motivesresulting from the perceived costs and benefits of participationin collective action. Two field studies were presented in thisarticle. Each study simultaneously considered both categoriesof postulated determinants but was conducted in the context ofa different social movement (the older people's movement andthe gay movement). Willingness to participate in collective ac-tion organized by the respective movement served as the maindependent variable.

The results concerning the role of collective-identificationprocesses are very similar for both studies. In both cases, identi-fication with the respective social movement turned out to be aunique predictor of willingness to participate in collective actioneven when collective, social, and reward motives were taken intoaccount at the same time. We also observed significantly positive(zero-order) correlations between willingness to participate andidentification with the broader social category from which themovement typically recruits its members. However, this form ofcollective identification had no independent or direct effect onwillingness to participate. In both studies, it lost its predictivevalue in multiple regression analyses in which identificationwith the relevant social movement as well as collective, social,and reward motives were considered as additional predictor vari-ables. Given the high correlation between both forms of collec-tive identification (r = .64 and .59, for Study 1 and Study 2,respectively; see Tables 1 and 3) and that identification with themovement retained its predictive value in the multiple regressionanalyses (see Tables 2 and 4) , it can be concluded that identifi-cation with the movement mediated the effect of identificationwith the recruitment category (Baron & Kenny, 1986). Indeed,the indirect effect of identification with the recruitment categoryon willingness to participate by way of identification with themovement was significant in both studies, t$ > 4.58, ps < .001.(Note that the collective, social, and reward motives can beruled out as similar mediators because identification with therecruitment category regained its predictive value as soon asidentification with the movement was excluded from the multi-ple regression analyses, even if the three motives were kept inthe analyses.4) It thus appears that identification with a broadersocial category increases one's willingness to participate in col-lective action only to the extent that it is transformed into amore politicized form of activist identification (Klandermans,1997a; Melucci, 1996; Taylor & Whittier, 1992, 1995).

The collective, social, and reward motives also made uniquecontributions to the prediction of willingness to participate, de-spite some variation between studies as to the relative impor-tance of these motives. Still, overall, the degree of similaritybetween the results of both studies is remarkable, consideringthat they were conducted in quite different political, social, orga-nizational, and even national contexts.

However, Study 2 did not only replicate important results ofStudy 1 but also extended them. This was accomplished byincluding an experimental manipulation in the design, thus com-bining the advantage of realism typical of field research withthe advantage of experimental control typically ascribed to labo-ratory research. Tn line with the social identity or self-categori-

zation approach, it was shown that collective identification var-ied as a function of the salient social psychological context,and this variation translated into more or less willingness toparticipate in collective action. In other words, Study 2 providedevidence for a causal, or better, mediational role of collective-identification processes in social movement participation. It isimportant to highlight again that it was identification as anactivist, and not identification with the broader recruitment cate-gory (i.e., identification with gay people in general), that func-tioned as a mediator. Apparently, a specific activist identity isparticularly conducive to social movement participation, and thismay be so because of specific implications for action inherentin such an identity (Friedman & McAdam, 1992; Kelly & Brein-linger, 1995). Unfortunately, we have no data available from thepresent studies as to the specific content of such activist identi-ties. Future research therefore ought to deal more directly withthe issue of identity content, and this is likely to necessitate thejoint use of both quantitative and qualitative methods (Rubin &Rubin, 1995).

In conclusion, on the basis of the present results, an integra-tion of the social identity or self-categorization approach intosocial movement research suggests the following preliminarymodel. There seem to exist two independent pathways to socialmovement participation, or at least to willingness to participate.One pathway appears to be calculation of the costs and benefitsof participation (including normative considerations). Thispathway was quantified in the present research by the measure-ment of the collective, social, and reward motives. The secondpathway seems to be identification with the movement or, inother words, adoption of a distinct activist identity. Here, calcu-lation processes should be of less importance than identificationor self-definition processes, meaning that ' 'if I know who I am,then I also know what to do, no matter what the consequencesare." This dual-pathway model is reminiscent of other dual-process models in social psychology, which distinguish betweencontrolled, strategic, or systematic processes, on the one hand,and automatic, spontaneous, or heuristic processes, on the otherhand (Chaiken, 1987; Devine, 1989; Petty & Cacioppo, 1986;see also Smith, 1994, for a review). Though it may be temptingto interpret the present calculation-identification distinctionalong these same lines, such an interpretation is beyond thescope of this article and certainly requires more direct empiricaltesting. In addition, future research should try to specify possi-ble sources for the differential use of these pathways. Par exam-ple, it is conceivable that people differ from each other as towhether they get motivated to participate in collective actioneither more by calculation or more by identification processes.Here, the individual-differences variable, idiocentrism versusallocentrism, may be a moderator variable of interest (Triandis,Bontempo, Villareal, Asai, & Lucca, 1988; also Kelly, 1993).The social context could be another source of variation. Study2, where cost-benefit calculations (including normative consid-erations) played a role in only the control condition and not inthe common fate condition, points in that direction. A thirdsource of variation may be the specific type of social movement.

4 In these analyses, the regression coefficients for identification withthe recruitment category were .15, f(83) - 2.09, p < .05, and .19,/(110) = 2.08, p < .05, for Study 1 and Study 2, respectively.

SOCIAL MOVEMENT PARTICIPATION 657

For example, it is very likely that identification processes aremore important in the context of religious or other fundamental-ist movements than in the context of movements with morecircumscribed agendas (e.g., for or against speed limits). Fi-nally, future research should also overcome another limitationof the present research. In this article, we were concerned pri-marily with the determinants of the willingness or intention toparticipate in collective action. What is needed next is a test ofthe predictive power of the model suggested here with respectto actual future behavior.

References

Adam, B. D. (1987). The rise of a gay and lesbian movement [TWayne'ssocial movements series]. Boston: Twayne.

Ajzen, I. (1991). The theory of planned behavior. Organizational Behav-ior and Human Decision Processes, 50, 179-211.

Ajzen, L, & Fishbein, M. (1977). Attitude-behavior relations: A theo-retical analysis and review of empirical research. Psychological Bulle-tin, 84, 888-918.

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and pre-dicting social behavior. Englewood Cliffs, NJ: Prentice Hall.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variabledistinction in social psychological research: Conceptual, strategic, andstatistical considerations. Journal of Personality and Social Psychol-ogy. 51, 1173-1182.

Brewer, M. B. (1991). The social self: On being the same and differentat the same time. Personality and Social Psychology Bulletin, 17,475-482.

Chaiken, S. (1987). The heuristic model of persuasion. In M. P. Zanna,J. M. Olson, & C. P. Herman (Eds.), Social influence: The OntarioSymposium (Vol. 5, pp. 3-39) . Hillsdale, NJ: Erlbaum.

Cruikshank, M. (1992). The gay and lesbian liberation movement. NewYork: Routledge, Chapman & Hall.

Deppe, A., & Elkemann, U. (1996). Kollektive Identitdt und sozialeBewegungen: Untersuchung zum Zusammenhang von Minoritatensta-tus und Bewegungsbeteiligung am Beispiel schwuler Mdnner [Collec-tive identity and social movement: On the relationship between minor-ity status and social movement participation—The case of gay men].Unpublished master's thesis, Westfalische Wilhelms—Universitat,MUnster, Germany.

Devine, P. G. (1989). Stereotypes and prejustice: Their automatic andcontrolled components. Journal of Personality and Social Psychology,56, 5-18.

Ellemers, N., Spears, R., & Doosje, B. (1997). Sticking together orfalling apart: In-group identification as a psychological determinantof group commitment versus individual mobility. Journal of Personal-ity and Social Psychology, 72, 617-626.

Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behav-ior: An introduction to theory and research. Reading, MA: Addison-Wesley.

Friedman, D., & McAdam, D. (1992). Collective identity and activism:Networks, choices, and the life of a social movement, hi A. D. Mor-ris & C. M. Mueller (Eds.), Frontiers in social movement theory (pp.156-173). New Haven, CT Yale University Press.

Gamson, W. A. (1992). The social psychology of collective action. InA. D. Morris & C. M. Mueller (Eds.), Frontiers in social movementtheory (pp. 53-76) . New Haven, CT Yale University Press.

Kawakami, K., & Dion, K. L. (1993). The impact of salient self-identi-ties on relative deprivation and action intentions. European Journalof Social Psychology, 23, 525-540.

Kawakami, K., & Dion, K. L. (1995). Social identity and affect asdeterminants of collective action: Toward an integration of relative

deprivation and social psychology theories. Theory and Psychology,5, 551-577.

Kelly, C. (1993). Group identification, intergroup perceptions and col-lective action. In W. Stroebe & M. Hewstone (Eds.), European reviewof social psychology (Vol. 4, pp. 59-83) . Chichester, England: Wiley

Kelly, C , & Breinlinger, S. (1995). Identity and injustice: Exploringwomen's participation in collective action. Journal of Community &Applied Social Psychology, 5, 41-57.

Kelly, C , & Breinlinger, S. (1996). The social psychology of collectiveaction: Identity, injustice and gender. London: Taylor & Francis.

Klandermans, B. (1984). Mobilization and participation: Social psycho-logical expansions of resource mobilization theory. American Socio-logical Review, 49, 583-600.

Klandermans, B. (1997a). Identimt und Protest Ein sozialpsycho-logischer Ansatz [Identity and protest. A social psychological ap-proach.] Forschungsjournal Neue Soziale Bewegungen, 70(3), 4 1 -51.

Klandermans, B. (1997b). The social psychology of protest. Oxford,England: Basil Black well.

Klandermans, B., & Oegema, D. (1987). Potentials, networks, motiva-tions, and barriers: Steps towards participation in social movements.American Sociological Review, 52, 519-531.

Lalonde, R. N., & Silverman, R. A. (1994). Behavioral preferences inresponse to social injustice: The effects of permeability and socialidentity salience. Journal of Personality and Social Psychology, 66,78-85.

McAdam, D., McCarthy, J. D., & Zald, M.N. (1988). Social move-ments. In N. J. Smelser (Ed.), Handbook of sociology (pp. 695-737).Newbury Park, CA: Sage.

Melucci, A. (1996). Challenging codes: Collective action in the infor-mation age. Cambridge, England: Cambridge University Press.

Neidhardt, F., & Rucht, D. (1993). Auf dem Weg in die ' 'Bewegungsge-sellschaft"? Uber die Stabilisierbarkeit sozialer Bewegungen [On theway to a "movement society"?]. Soziale Welt, 44, 305-326.

Olson, M. (1977). The logic of collective action: Public goods and thetheory of groups. Cambridge, MA: Harvard University Press.

Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion:Central and peripheral routes to attitude change. New "fork: Springer-Verlag.

Rubin, H. J., & Rubin, I. S. (1995). Qualitative interviewing: The artof hearing data. London: Sage.

Rucht, D. (1995). Kollektive Identitat: Konzeptuelle Uberlegungen zueinem Desiderat der Bewegungsforschung [Collective identity: Someconceptual comments on a lacuna of social movement research].Forschungsjournal Neue Soziale Bewegungen, 5(1) , 9-23.

Simon, B. (1998). Individuals, groups and social change: On the rela-tionship between individual and collective self-interpretations and col-lective action. In C. Sedikides, J. Schopler, & C. A. Insko (Eds.),Intergroup cognition and intergroup behavior (pp. 257-282). Hills-dale, NJ: Erlbaum.

Simon, B., Pantaleo, G., & Mummendey, A. (1995). Unique individualor interchangeable group member? The accentuation of intragroupdifferences versus similarities as an indicator of the individual selfversus the collective self. Journal of Personality and Social Psychol-ogy, 69, 106-119.

Smelser, N. J. (1962). A theory of collective behavior. New %rk: FreePress.

Smith, E. R. (1994). Procedural knowledge and processing strategiesin social cognition. In R. S. Wyer & T. K. Srull (Eds.), Handbook ofsocial cognition (Vol. 1, pp. 99-151). Hillsdale, NJ: Erlbaum.

Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effectsin structural equations models. In S. Leinnart (Ed.), Sociologicalmethodology (pp. 290-312). San Francisco: Jossey-Bass.

658 SIMON ET AL.

Tajfel, H. (1981). Human groups and social categories: Studies insocial psychology. Cambridge, England: Cambridge University Press.

Tajfel, H., & Turner, I.C. (1986). The social identity theory of in-tergroup behavior. In S. Worchel & W. G. Austin (Eds.), Psychologyof intergroup relations (pp. 7 -24) . Chicago: Nelson-Hall.

Taylor, V., & Whittier, N. E. (1992). Collective identity in social move-ment communities. In A. D. Morris, & C. M. Mueller (Eds.), Frontiersin social movement theory (pp. 104-129). New Haven, CT YaleUniversity Press.

Taylor, V., & Whittier, N. E. (1995). Analytical approaches to socialmovement culture: The culture of the women's movement. In H. John-son & B. Klandermans (Eds.), Social movements and culture (pp.163-187). Minneapolis-London: University of Minnesota Press—Ucl Press.

Toch, H. (1965). The social psychology of social movements. Indianapo-lis, IN: Bobbs-Merrill.

Triandis, H. C , Bontempo, R., Villareal, M. J., Asai, M., & Lucca, N.(1988). Individualism and collectivism: Cross-cultural perspectiveson self-ingroup relationships. Journal of Personality and Social Psy-chology, 54, 323-338.

Turner, J. C , Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell,M. S. (1987). Rediscovering the social group: A self-categorizationtheory. Oxford, England: Basil Blackwell.

Unruh, T. (1990). Grau kommt—das ist die Zukunft [Gray is coming—That's the future]. Miinchen, Germany: Goldmann.

Wright, S. C , Taylor, D. M., & Moghaddam, F. M, (1990). Respondingto membership in a disadvantaged group: From acceptance to collec-tive protest. Journal of Personality and Social Psychology, 58, 994-1003.

Received October 23, 1997Revision received October 28, 1997

Accepted October 28, 1997 •