group processes inter group relations 2005 louis 355 74

8/3/2019 Group Processes Inter Group Relations 2005 Louis 355 74

1/21

http://gpi.sagepub.com/ Relations

Group Processes & Intergroup

http://gpi.sagepub.com/content/8/4/355The online version of this article can be foun d at:

DOI: 10.1177/13684302050564652005 8: 355Group Processes Intergroup Relations

Winnifred R. Louis, Donald M. Taylor and Rebecca L. DouglasCost-Benefit Analyses for Intergroup Behavior

Normative Influence and Rational Conflict Decisions: Group Norms and

Published by:

http://www.sagepublications.com

can be found at:Group Processes & Intergroup Relations Additional services and information for

http://gpi.sagepub.com/cgi/alertsEmail Alerts:

http://gpi.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://gpi.sagepub.com/content/8/4/355.refs.htmlCitations:

What is This?

- Oct 18, 2005Version of Record>>

by Lia Oltean on October 26, 2011gpi.sagepub.comDownloaded from
http://gpi.sagepub.com/http://gpi.sagepub.com/http://gpi.sagepub.com/http://gpi.sagepub.com/content/8/4/355http://gpi.sagepub.com/content/8/4/355http://gpi.sagepub.com/content/8/4/355http://www.sagepublications.com/http://gpi.sagepub.com/cgi/alertshttp://gpi.sagepub.com/cgi/alertshttp://gpi.sagepub.com/subscriptionshttp://gpi.sagepub.com/subscriptionshttp://www.sagepub.com/journalsReprints.navhttp://www.sagepub.com/journalsReprints.navhttp://www.sagepub.com/journalsPermissions.navhttp://gpi.sagepub.com/content/8/4/355.refs.htmlhttp://gpi.sagepub.com/content/8/4/355.refs.htmlhttp://gpi.sagepub.com/content/8/4/355.refs.htmlhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://gpi.sagepub.com/content/8/4/355.full.pdfhttp://gpi.sagepub.com/content/8/4/355.full.pdfhttp://gpi.sagepub.com/http://gpi.sagepub.com/http://gpi.sagepub.com/http://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://gpi.sagepub.com/content/8/4/355.full.pdfhttp://gpi.sagepub.com/content/8/4/355.refs.htmlhttp://www.sagepub.com/journalsPermissions.navhttp://www.sagepub.com/journalsReprints.navhttp://gpi.sagepub.com/subscriptionshttp://gpi.sagepub.com/cgi/alertshttp://www.sagepublications.com/http://gpi.sagepub.com/content/8/4/355http://gpi.sagepub.com/


2/21

Normative Inuence andRational Conict Decisions: Group Normsand Cost-Benet Analyses

for Intergroup Behavior Winnifred R. LouisUniversity of Queensland

Donald M. TaylorMcGill University

Rebecca L. Douglas Dalhousie University

The present paper articulates a model in which ingroup and outgroup norms inform rationaldecision-making (cost-benet analysis) for conict behaviors. Norms inuence perceptions of the consequences of the behavior, and individuals may thus strategically conform to or violatenorms in order to acquire benets and avoid costs. Two studies demonstrate these processes inthe context of conict in Qubec. In the rst study, Anglophones perceptions of Francophoneand Anglophone norms for pro-English behaviors predicted evaluations of the benets andcosts of the behaviors, and these cost-benet evaluations in turn mediated the norm-intentionlinks for both group norms. In the second study, a manipulated focus on supportive versushostile ingroup and outgroup norms also predicted cost-benet evaluations, which mediatedthe norm-intention relationships. The studies support a model of strategic conict choices in

which group norms inform, rather than suppress, rational expectancy-value processes.Implications for theories of decision-making and normative inuence are discussed.

k e y w o r d s cost-benet analyses, expectancy-value processes, group processes,intergroup conict, norms, rational choice

Group Processes & Intergroup Relations

2005 Vol 8(4) 355374

WHE N individuals identify as group membersrather than as unique individuals, they aremotivated to engage in pro-group action andto conform to ingroup norms (Tajfel & Turner,1979; Turner, Wetherell, & Hogg, 1989). Therelationship between pro-group conict choices,

GPIR

Copyright 2005 SAGE Publications(London, Thousand Oaks, CA and New Delhi)8:4; 355374; DOI: 10.1177/1368430205056465

Authors note Address correspondence to: Winnifred R.Louis, School of Psychology, University of Queensland, McElwain Building, St. Lucia,QLD 4072, Australia[email: [email protected]]

http://gpi.sagepub.com/http://gpi.sagepub.com/http://gpi.sagepub.com/http://gpi.sagepub.com/


3/21

such as decisions to engage in collective action,and individuals evaluations of the costs andbenets of these actions for themselves person-ally has been the subject of much debate. Many theorists have argued that individual-levelincentives are a primary motivation for partici-pation in collective action (e.g. Olson, 1968).Recent research however suggests that activistsconict choices may be predicted by group-level factors independent of individual-levelcosts and benets (e.g. Kelly, 1993; Simon et al.,1998). In particular, intergroup research hasrepeatedly demonstrated non-instrumentalconformity to ingroup norms, based on self-stereotyping and esteem-enhancement pro-cesses (see e.g. Terry & Hogg, 1996).

It is argued here that in the context of intergroup conflict, group norms may alsoinuence decision-making because they denethe means to acquire benefits and to avoidcosts. A model of agentic normative inuence ispresented which predicts cost-benefit evalu-ations for conict choices from group norms.In this model, cost-benefit evaluations areproposed to mediate the relationship betweengroup norms and individuals behavioralintentions. If group norms and cost-benefit perceptions are evaluated as independent predictors of pro-group choices, it is proposed,the role of group norms is underestimatedbecause cost-benefit perceptions may them-selves be predicted by norms. Mediationalanalyses will then reveal that for conflict behaviors cost-benet analyses play a role inindividuals strategic reactions to group norms.

We argue that not only ingroup norms, but alsooutgroup norms, may be implicated in thisrational normative influence. Thus, thepresent paper attempts to address two still-contested theoretical positions: that cost-benefit calculations may be normative; andthat outgroup norms may be influential inintergroup decision-making.

Norms in decision-making Classic research in conformity confirms thestrong impact of norms on individualsbehavior, not only in cases of high judgmental

uncertainty (Sherif, 1936), but even whenconformity means ignoring strong perceptualcues (Asch, 1956), or performing ostensibly life-threatening actions (Milgram, 1974). Normsat the individual, interpersonal, and intergrouplevels have been argued to inuence decision-making. Needs to be authentic, or to afrmmoral commitments, are thought to motivateindividuals to act out personal or moral norms (Gorsuch & Ortberg, 1983; in politics, Gecas,2000). Individuals need for approval fromsignicant others motivates conformity to inter-personal norms (Ajzen, 1991; Fishbein & Ajzen,1975; in political behavior, Klandermans,1984). At the group level, referent group norms arethought to dene individuals behavioral reper-toires as a function of salient identities, inu-encing decision-making via self-stereotypingprocesses (Terry & Hogg, 1996; in politics,Kelly, 1993; Simon et al., 1998).

Psychological research has tended to study conformity to group norms in the context of destructive (e.g. Milgram, 1965), incorrect (e.g.

Asch, 1956) or arbitrary (e.g. Sherif, 1936)behaviors (Turner et al., 1989; see also Kelly,1993; Simon et al., 1998). Perhaps for thisreason, conformity is often presented as aprocess that overrides rationality, causing indi-

viduals to engage in personally costly actions inthe service of authenticity, or interpersonal orgroup afliation. However, group norms havealso been shown to motivate deliberative pro-cesses (e.g. Abrams, 1994; Mackie, Gastardo-Conaco, & Skelly, 1992; Mackie & Queller,2000; Moscovici, 1985): when the behavior isingroup normative, individuals may systemati-cally assess, rather than automatically endorse,the consequences of the actions. Similarly,decision-making research demonstrates that beliefs about the costs and benets of behaviorsmay be correlated with perceived norms, ratherthan independent of them (e.g. Vallerand,Deshaies, Cuerrier, Pelletier, & Mongeau, 1992;see Liska, 1984; Miniard & Cohen, 1981). Inter-active and mediational models of normativeinuence and cost-benet evaluations are not typically assessed, however, because dominant decision-making models for individual decision-making (Ajzen, 1991), intergroup decisions

Group Processes & Intergroup Relations 8(4)

356



4/21

(Turner et al., 1989), and social movement participation (Klandermans, 1984) proposeindependent paths.

Indeed, conformity to ingroup norms may be motivating to individuals for many reasonsarticulated in previous theories that are inde-pendent of cost-benefit evaluations of thebehavior. For example, ingroup norms may motivate conformity through self-stereotyp-ing, as in referent informational influence (Terry & Hogg, 1996); through a need for approvalfrom significant others, as in the theory of planned behaviors subjective normative influence (Ajzen, 1986); or through the moralimperative created by internalized personal ormoral norms (Gecas, 2000). In these processes,performing the behavior intrinsically satisfiesthe motivation. In intergroup conflict,however, an additional process must surely also operate: conformity to ingroup normscoordinates group members to advanceingroup interest in the ongoing interaction.Thus performing the ingroup normativebehavior generates additional benefits for thegroup and thus the individual group member,

whereas violation of ingroup norms generatesadditional costs for the group and thus theindividual members of the group (Abrams,1994; Turner et al., 1989).

In intergroup contexts, individuals may perform expectancy-value calculations (seeFeather, 1982) at a group level of analysis as

well as an individual analysis, estimating thelikelihood (expectancy) and importance(value) of the costs and benets that will accrueto the group if particular behaviors are per-formed (see Louis & Taylor, 2002; Louis, Taylor,& Neil, 2005; Turner et al., 1989). The present research, however, focuses on the derivation of individual-level cost-benefit analyses fromgroup norms, proposing that in conict choicesgroup norms contribute to the denition of individuals self-interest, rather than overridingit or operating in parallel. Testing a model of agentic normative inuence in which group normspredict individuals cost-benefit analyses andthese analyses, in turn, signicantly mediate thenorm-intention relationship, is the rst goal of the present research.

The relevance of outgroup norms?

Outgroups, that is groups to which individualactors do not belong, or with whom individualsdo not identify, do not generally trigger needsfor approval, for afrmation, or understanding(Turner et al., 1989). Outgroup norms havebeen observed to inuence behavior only whenbacked by rewards and punishments, and com-pliance with coercive outgroup normative inu-ence is limited to outgroups with both powerand surveillance ability (see e.g. Reicher &Levine, 1994). In the absence of coercive sur-

veillance, then, outgroup norms are theoreti-cally irrelevant to individuals decision-making.For example, safe sex behavior may be guidedby perceptions of what is appropriate for indi-

viduals peer groups, or what signicant others want, or what students personally believe to bemorally appropriate; knowing that an outgroup(e.g. the Roman Catholic Church) frowns onthe use of condoms may simply have no impact (see e.g. Mackie et al., 1992; Terry & Hogg,1996).

For intergroup behavior, in contrast, it isargued here that outgroup norms should inu-ence decision-making. Coercive normative inuence is relevant to many intergroup decisions,particularly for low power groups who may beexposed to surveillance and the threat of reprisals if the status hierarchy is challenged.However, even without surveillance abilities tocreate coercion outgroups may exert agentic normative inuence , because outgroup reactionsthemselves constitute important consequencesor targets of action in intergroup conict. Inconict decisions, individuals may strategically conform to or violate outgroup norms becausethis conformity or violation is seen as helpful tothe group, with implications for the individualgroup member.

Consistent with this argument, the expectedbehavior of outgroup members has been shownto influence intergroup behavior in many contexts. For example, anticipated discrimi-nation on the part of the outgroup may produce retaliatory discrimination from equalor high power groups (e.g. Blanz, Mummendey,& Otten, 1995), and egalitarian or appeasement

Louis et al. n o r m s a n d r a t i o n a l c o n f l i c t

357



5/21

behaviors from low power groups (e.g. Ng,1982). Anticipated outgroup behavior denesthe consequences of inaction: if the ingroupdoes not act, the outgroup has the initiative,

which is an essential strategic consideration(see e.g. L. Gaertner & Insko, 2000). Forexample, escalating violence in eld studies hasbeen associated with the perception of intransi-gent outgroup determination to maintain thestatus quo in Israel (Rouhana & Bar-Tal, 1998)and in Ireland (Cairns & Darby, 1998). At themost serious level, belief in the threateningintentions of the outgroup has been observedto be a critical process in the escalation of intergroup violence, including riots (Drury &Reicher, 2000; Reicher, 1984) and genocide(Staub, 2001). Anticipated or past outgroupbehavior, in short, motivates behavior to changeor defend against outgroup actions. Impor-tantly, because attributions regarding intentionmoderate the link between outgroup past behavior and inferences for future behavior,perceptions of outgroup norms are critical inorder for outgroup behavior to motivateingroup action. If the outgroup behavior isoutgroup normative, it may be perceived asintentional and likely to be repeated (seeHunter, Stringer, & Watson, 1991). Thus, aterrorist act attributed to the situation (Warinevitably produces atrocities) or seen asoutgroup antinormative (Most guerrillas abhorterrorism) may have a far different effect indening the consequences of inaction than abehavior seen as outgroup normative (Theseguerillas endorse terrorism).

Ingroup members respond to the anticipatedbehavior of the outgroup if ingroup membersdo nothing, but they also react to the anticipatedresponses of the outgroup to their own action.That is, an analysis of intergroup decision-making must model the effects of ingroup andoutgroup norms on evaluations of the conse-quences of conict choices, from inaction toconciliatory or confrontational action (see also,Louis & Taylor, 2002; Louis et al., 2005). Forexample, if the outgroup is known to opposesit-ins but tolerate peaceful demonstrations, at least two effects of that knowledge might beexpected. Strategic conformity to the outgroup

normby demonstrating peacefully, in thecurrent examplemight be seen to benet thegroup because it will elicit positive reactionsfrom the outgroup: perceived normative simi-larity might (1) reduce symbolic threat andtherefore defensive behavior (Rokeach, 1960;see also De Ridder & Tripathi, 1992); (2) makesalient a shared superordinate identity (seeS. L. Gaertner, Dovidio, Anastasio, Bachman, &Rust, 1993) that delegitimizes the subgroupshorizontal conict behavior; and/or (3) signal

willingness to cooperate for mutual gain, trig-gering reciprocity dynamics (Osgood, 1962)and reducing realistic conict (Sherif, 1966).

Alternatively, strategic violation of outgroupnormsengaging in a sit-inmay be seen asbenecial if the action increases the costs of maintaining an unfavorable status quo for theadvantaged group, motivating attention or con-cessions (Dalton, 1988; Gurr, 1970; Tilly, 1975).For example, some research in political sciencehas suggested that violent riots may be particu-larly effective in attracting ameliorative govern-ment funding (Piven & Cloward, 1977),although others have challenged this analysis(see e.g. Burstein & Freudenberg, 1978). Theseempirical results of intergroup conflict researchsuggest that a theoretical model of conflict decisions should do more than consider groupprocesses of normative influence and cost-benefit analyses independently. The modelshould include the normative aspects of instru-mental decision-making, and the instrumentalaspects of normative influence. If groupmembers react dynamically to anticipatedbehaviors of the outgroup, explicitly modelingthe effects of ingroup and outgroup norms onactors cost-benet calculations will be useful inunderstanding and predicting actions inconict.

Agentic normative inuenceGroup norms exert agentic influence, wepropose, when they dene the consequences of intergroup behaviors: the means by whichgroups achieve benets or avoid costs. Groupnorms are understood as strategic, inuencingactive decision-makers as they choose among


358



6/21

intergroup behaviors which vary in theirconsequences for self and group. Agentic normative influence may be contrasted with othercommonly studied normative influence pro-cesses on three dimensions: the level of analysis,the degree of deliberation, and the motivationfor conformity.

In conict, agentic norms are measured at the group level of analysis. This is similar toreferent group norms (e.g. Terry & Hogg,1996), but in contrast to the individual leveltapped with measures of significant othersinterpersonal norms (e.g. Ajzen, 1991) orpersonal moral norms (e.g. Gecas, 2000).Group conict is likely to make social identitiescontextually salient, so that group-level vari-ables play a primary role in decision-making(Tajfel & Turner, 1979; Turner et al., 1989).

Whereas in social identity models ingroupsources alone exert referent group inuence,however, we propose here that outgroup as wellas ingroup norms can exert agentic inuence.

Agentic influence for conflict choicesoperates via deliberative cost-benet analyses(expectancy-value processes). This is similar tothe social incentives of the Klandermans(1984) social movement model, but in contrast to self-stereotyping or internalized self-afrma-tion in the referent informational influencemodel (Terry & Hogg, 1996). The agenticmodel seeks to integrate decision-makingtheories of rational choice and of intergroupsocial influence by hypothesizing that oneprocess by which groups exert influence inconict is by shaping members instrumentalperceptions of the costs and benefits of behavior (see also Louis et al., 2005).

Finally, agentic norms are injunctive, not descriptive; they are prescriptive, not behav-ioral (Cialdini, Kallgren, & Reno, 1991).

Agentic norms, like subjective norms (e.g. Ajzen, 1991), tap expected source approval ordisapproval, in contrast to norms that tap what the source will do themselves, like referent group norms. Referent informational inuencesuggests that if union members perceive adescriptive norm that strikes are what proto-typical union members do in conict situations,the union members will be more likely to strike.

They will not be inuenced by the outgroupreferent normthe perception of whether ornot managers strike. In agentic normative inu-ence, it is expected that union members will beresponsive to managers (as well as unions)injunctive norms for striking, because tactics inunfolding conict interactions must be evalu-ated with an eye to the perceptions of theconict partner/opponent. It is both intuitively compelling and (as discussed above) consistent

with past research to argue that unionmembers perception that managers (as well asthe union) think the union ought or ought not to strike will change the perceived benets andcosts of striking, and thus drive behavior.

Empirically, the agentic normative inuencemodel may improve the power of models of decision-making in conict by drawing atten-tion to relevant outgroup norms (which couldbe overlooked in predictive models that assumethat outgroups are irrelevant). But the primary goal of the present paper is to model the medi-ating process whereby ingroup and outgroupinjunctive norms inuence behavior via cost-benet perceptions explicitly. If this mediationoccurs and models such as the theory of planned behavior (Ajzen, 1991) or the selectiveincentives model (Klandermans, 1984) areapplied to conict choices, the effect will be tounderrate the normative inuence that is rep-resented in the decision-making model. This isbecause the instrumental cost-benet analysesfor the behavior and behavioral attitudes willbe understood as independent predictors,

whereas if agentic normative inuence is occur-ring, cost-benefit analyses mediate indirect normative influence (see also, Liska, 1984;Miniard & Cohen, 1981). If agentic processesoccur and typical intergroup conict modelsare applied, measuring norms and behavior but not cost-benet perceptions for the actor, theeffect is to overlook the extent of rationaldecision-making and the mediating role that can be played by cost-benefit perceptions. Thus,assessing agentic inuence processes may allowresearchers to study the interplay of rationaldecision-making and group normsa con-tentious and important topic, both socially andtheoretically. In the present paper, two studies


359



7/21

analyze agentic influence processes in thecontext of EnglishFrench relations in Qubec.

Study 1The population of Qubec is predominantly Francophone, and coexists uneasily in federa-tion with 12 predominantly Anglophoneprovinces and territories. Since the 1960s,Francophone rights have become a dominant political issue within Qubec and a sovereign-tist movement has championed the politicalindependence of Qubec (see e.g. Linteau,Durocher, Robert, & Ricard, 1991). The sover-eigntist movement draws its support largely from Francophone Qubecers and is opposedby a federalist movement supported by amajority of Anglophone Qubecers as well as aminority of Francophones (e.g. Maclure, 2000).

In the rst study, young Anglophone partici-pants intentions to use English in consumercontexts, and their cost-benet evaluations of these behaviors, were assessed as a function of

Anglophone and Francophone norms. Young Anglophone Qubecers speak French relatively uently, and provincial legislation mandates that French should be the language of commerceand work in Qubec. Accordingly, using Englishin consumer contexts is not merely a matter of necessity or convenience: it is also a controver-sial intergroup behavior (see Chevrier, 1997;OMalley & Bowman, 2001). In the present study, we hypothesized in accordance with thesocial identity approach that ingroup norms

would be associated with (1) higher intentionsto engage in the behaviors and (2) more favor-able cost-benefit evaluations of the conse-quences of the behavior for the actor. In line

with past research concerning the dynamicimpact of anticipated outgroup reactions,outgroup norms for conict behaviors were alsoexpected to be associated (3) with intentionsand (4) with cost-benet evaluations. Finally, inaccordance with the agentic inuence model,(5) cost-benet evaluations were predicted tomediate the relationship between norms andintentions, demonstrating a process by whichdecision-making at the individual level may berationally informed by distal group-level factors.

Method

Participants Participants were required toself-identify as Anglophone Qubecers, and tobe born in Qubec or to have lived in theprovince for at least 10 years. Seventeen of the224 recruited were excluded on these grounds;of the remaining sample ( N = 207), women

were slightly in the majority ( n = 118, 57%) andparticipants were predominantly federalist ( n =172, 83%), with minorities who were neutral ( n = 23, 11%), undecided ( n = 10, 5%), or sover-eigntist ( n = 1, 1%).

Procedure Participants were recruited fromtwo English-language universities in Montral,Canada for a study on Anglophone Qubecersresponses to the language situation in Qubec.Those who were eligible to complete the20-minute study were given consent forms,tickets for two lotteries of C$100 (which wereawarded at the completion of the recruiting),and questionnaire booklets. After completingthe questionnaire, participants were thankedand debriefed verbally and in writing.

MaterialsIntroduction Under the title psychology ques-tionnaire participants read: In the Province of Qubec, ethnicity strongly inuences behaviorin everyday life. Language tensions between

Anglophones and Francophones provoke,among different ethnic groups, differingreactions. For example, Francophone Qube-cers might respond to the situation by callingfor more dialogue, or by involving themselvesin language protection groups. A wide variety of reactions are also available to AnglophoneQubecers. We have designed this question-naire to determine how you yourself will act inresponse to the language situation in Qubec.

Demographics An initial demographics sectionof the questionnaire included written reitera-tions of the screening questions, questionsregarding political affiliation, and French-language skills. As expected, all participantsindicated that they could both understand andspeak French, although two participants indi-cated they could read but not write in French.


360



8/21

Social identification Identification as Anglo-phones was assessed with a scale adapted fromPorter (1995). Three items measured import-ance of the Anglophone identity, commitment to expressing the identity, and perception of community on 11-point scales from 0 (Not at all) to 10 (Extremely), = .87.

Behavior evaluations Participants rated fivepro-English behaviors involving the use of English in commercial contexts: demandingthat business correspondence be in English,not shopping at a store because of monolingualFrench signs, avoiding certain stores because of being rudely treated as an Anglo, responding inEnglish to a store clerk who addressed you inFrench, and openly using English in French

work environments.Participants evaluation of the Anglophone

and Francophone group norms were establishedby two questions for each of the ve behaviors.The 11-point items were adapted from Ajzen(1991) (e.g. How would Francophones/other

Anglophones react to a person who did thisbehavior?; 5 (intensely negatively) to +5(intensely positively)). 1 Scales were created for

Anglophone norm, = .80, and Francophonenorm, = .75, by averaging across the fivebehaviors.

Cost-benefit evaluations regarding the conse-quences of the behavior were assessed onbipolar measures What is the effect of thisbehavior on the individual actor?, from 5

(very negative: costs overwhelmingly outweighbenets) to +5 (very positive: benets over-

whelmingly outweigh costs). Cost-benet scores were created by averaging across the fivebehaviors, = .64.

Finally, participants completed a measure of intentions to engage in each of the ve options:When the situation comes up, I react this way ____% of the time (0, Never to 100, Always).By averaging across the ve behaviors, a scalefor intentions was created, = .77.

Results

Following Baron and Kenny (1986), the model was tested by predicting intentions from norms(the distal independent variables); predictingthe mediator (cost-benefit evaluations) fromnorms; and using the Sobel test to evaluate thesignicance of the change in coefcients fornorms when the cost-benefit analyses wereentered in the hierarchical linear regression.Means, standard deviations, and intercorrela-tions are presented in Table 1. Two multivariateoutliers identied on the basis of Mahalanobisdistance were deleted.

Group norms predict intentions Anglophone(ingroup) and Francophone (outgroup) normsand expectancy-value scores were included ascentered continuous predictors of intentions toengage in pro-English consumer behaviors (seeTable 2). Group norms were signicantly associ-ated with intentions ( adj. R 2 = .10, F (2, 193) =


361

Table 1. Means, standard deviations (SDs), and intercorrelations for group norms, cost-benet evaluations,and intentions

Variables 1 2 3 4

1. Anglophone norm 1.32(1.42)

2. Francophone norm .17* 1.71(1.49)

3. Individual-level cost-benet evaluation .59*** .17* 0.05(1.47)

4. Behavioral intention .26*** .17* .41*** 46.00(26.04)

*** p < .001; ** p < .01; * p < .05.

Notes : Uncentered means (SD) in diagonal, and zero-order correlations below. Norms and cost-benet evaluations ranged from +5 to 5. Intentions ranged from 0 to 100.



9/21

12.24, p = .000). Perceptions that the behaviors were more ingroup normative were associated with higher intentions ( = .30, p = .000; seeTable 2). Independently, higher intentions

were associated with perceptions the behavior was outgroup normative ( = .22, p = .002).2

Group norms predict cost-benet evaluationsGroup norms were also associated with cost-benefit analyses: Anglophone and Franco-phone norms together predicted 42% of theadjusted variance in consequences for the indi-

vidual ( F (2, 195) = 71.98, p = .000). The morethat the ingroup was perceived to support abehavior, the more the behaviors were thought to benet the individual actor ( = .64, p =.000). Behaviors that the outgroup was per-ceived to support were also given more positivecost-benefit evaluations ( = .27, p = .000).Thus, rather than being independent of socialinfluence processes, cost-benefit evaluationscould be predicted from group-level norms.

Cost-benefit evaluations mediate the norm-intention relationships The mediating role of cost-benet analyses in ingroup and outgroupnorm-intention relationships was thenexamined. When cost-benet analyses for theindividual actor were entered into the equationthe model t improved signicantly ( R 2change =.07, F (1, 192) = 16.49, p = .000) (see Table 2,block 2). Individuals had stronger intentions toengage in behaviors that they perceived wouldbenet them ( = .30, p = .000). Moreover, the

positive inuence of ingroup norms on inten-tions was fully mediated by individual-levelexpectancy-value processes, decreasing from

= .30 to = .08 (z = 3.86, p = .000), as was theinuence of outgroup norms, decreasing from

= .22 to = .12 (z = 3.14, p = .001).

Alternative causal models In the current analyses, the causal path is assumed to flowfrom norms to behavior, with cost-benefit analyses as a mediator. Since the data are cor-relational, alternative causal paths are possible,and it is possible that better model t may beprovided by (for example) assuming that behavioral intentions drive norm perceptions,rather than the reverse. 3 Table 3 provides t statistics from path analyses for the present model and three alternative theoretical models:(1) an economic model, in which cost-benet analyses drive behavior, with norms as amediator; and two reverse path models in

which behavioral intentions drive (2) cost-benet analyses with norms as a mediator, or(3) norms with cost-benefit analyses as amediator. Akaikes information criterion (AIC)is especially useful for comparing non-nestedmodels, with the model with the smallest AICbeing taken as the better tting. In the present data, however, all of the measures point in thesame direction: t statistics (GFI, AGFI) arehigher and error statistics (RMSEA, AIC) arelower for alternative causal models than for thepresent model. Although experimental manip-ulations and longitudinal data would comp-lement the present results, the interpretationthat norms are motivating behavior via cost-benet analyses is supported by the poorer t of alternative and reverse path causal models.

Summary In the present results, ingroupnorms were associated with intentions toengage in pro-group behaviors (hypothesis 1)and more favorable cost-benet evaluations of the consequences (hypothesis 2). In addition,the results conrm the hypothesized role of outgroup norms, which were associated withintentions (hypothesis 3) and cost-benefit evaluations (hypothesis 4). Finally, cost-benet evaluations were observed to mediate the


362

Table 2. Study 1: Prediction of behavioral intentionsas a function of Anglophone and Francophonegroup norms and cost-benet evaluations

Block 1 2Adj. R 2 .10*** .17***R 2 change .11*** .07***

Predictor Standardized coefcients

Anglophone norm .30*** .08 ***Francophone norm .22** * .12***Cost-benet evaluations .35***

* p < .05; ** p < .01; *** p < .001.



10/21

relationships between norms and intentions(hypothesis 5), providing evidence that decision-making at the individual level may berationally informed by distal group-levelfactors.

Discussion

The present results provide support for thecontentions that normative inuence can bedeliberative, expectancy-value processes may be normative, and that outgroup norms may beinfluential in intergroup decision-making.Specically, in the present results, Anglophonesupport for intergroup behaviors predictedintentions and cost-benefit calculations, con-sistent with a social identity model (e.g. Terry &Hogg, 1996) although inconsistent with a strict distinction between norms and cost-benefit analyses (e.g. Azjen, 1991; Klandermans, 1984).Thus, perceptions of what beneted the actor

were informed by ingroup norms, and thesecost-benet evaluations mediated the ingroupnorm-intention relationships. In addition,outgroup norms were associated with morefavorable cost-benet evaluations of intergroupbehaviors, as well as stronger intentions toengage in the behaviors. These normative cost-benefit evaluations mediated the outgroupnorm-intention relationship. Not only ingroupnorms, then, but also outgroup norms, may inuence decisions in intergroup conict by guiding strategic expectancy-value processes: aprocess of agentic normative inuence .

The present study, however, is open to criti-cism on several methodological grounds.Importantly, group norms were measuredcorrelationally for the same common set of

behaviors, leaving the causal path open toquestion. Reverse causal models produce worset statistics, in the present results, than a model

whereby norms drive behavior via cost-benet analyses. However, it seems likely that therelationship between norms and cost-benefit perceptions can be a feedback loop in somecircumstances: where it is difcult to obtainfeedback about other peoples views andactions, benets to the individual may motivateintentions and independently motivate partici-pants to see behaviors as descriptively and pre-scriptively normative (everyone does what Ido, a false consensus effect: Miller & Prentice,1994; everyone thinks whats good for me ismorally right, a legitimizing myth; Chen &Tyler, 2001). Conict contexts, which are char-acterized by repeated, consequential inter-actions with ingroup and outgroup members,may both motivate and inform accurate per-ceptions of ingroup and outgroup norms,increasing the effects of perceived norms oncost-benet analyses and reducing the feedbackfrom benefits or costs to norm perceptions(Louis & Taylor, 2002). However, in the present study, the two effects are not disentangled. Theindependent variables are highly intercorre-lated (see Table 1), and all participants are

judging the same behaviors as more or lessnormative. If indeed there are empirical Anglo-phone and Francophone norms for the behav-iors, the variation in norm perceptions implies(motivated?) inaccuracy in at least some par-ticipants, inflating the relationships amongmeasured variables.

Conversely, the power of the study wouldhave been weakened by the restricted range of


363

Table 3. Study 1: Fit statistics for the present theoretical model (1) and three alternative path models

Model Chi-square GFI AGFI RMSEA AIC

1. Norms -> CBA -> Beh 5.46*** .99 .87 .15 23.462. CBA -> Norms -> Beh 39.88*** .92 .59 .31 55.883. Beh -> CBA -> Norms 26.50*** .94 .81 .20 40.504. Beh -> Norms -> CBA 26.17*** .94 .71 .24 42.17

* p < .05; ** p < .01; *** p < .001.Note : GFI = goodness-of-t index; AGFI = adjusted GFI; RMSEA = AIC = Akaikes information criterion. CBA =cost-benet analyses; Beh = behavioral intentions.



11/21

the norm measures, in that all of the behaviorsinvolved the use of English in commercial/consumer contexts, and were evaluated by participants on average as ingroup normative(M = 1.32) and outgroup antinormative ( M =1.71; see Table 1). A more powerful analysis

would contrast ingroup and outgroup norma-tive and antinormative behaviors. Depressedestimates of the relationship could also havebeen produced by the modest reliability for thecritical mediator, cost-benet evaluations ( =.64). The total variance accounted for in inten-tions in the nal model was in fact quite low(adj. R 2 = .17; Table 2). Thus, although each of the predicted norm-evaluation-intentionrelationships was signicant, a more powerfultest of the model is desirable. In addition,although the use of English in consumercontexts is politically controversial and (as theresults suggest) informed by consideration of intergroup norms and consequences, extensionof the focus to an explicitly political context may be more compelling to political and inter-group psychologists. These issues wereaddressed in a second study.

Study 2In Study 2, Anglophone participants generatedand evaluated a range of intergroup behaviorsas a function of an experimental focus on sup-portive or hostile ingroup and outgroup norms.Specically, participants were asked to generateand evaluate lists of behaviors that Anglo-phones and Francophones would eithersupport or oppose. As in Study 1, it was hypoth-esized that (1) ingroup norms for conflict behaviors would predict intentions and (2)expectancy-value processes for the individualactors; that outgroup norms would also predict (3) intentions and expectancy-value processes;and that (4) expectancy-value processes wouldmediate the relationship between norms andintentions, replicating the agentic normative inuence process in which decision-making at the individual level is rationally informed by distal group-level factors.

Method

Participants As in Study 1, participants wererequired to self-identify as Anglophone, and tobe born in Qubec or to have lived in theprovince for at least 10 years. Of 101 partici-pants completing the study, 10 were excludedon these grounds. Ages for the 91 participantsremaining ranged from 18 to 29 with a mean of 20; most were women ( n = 70, 78%). Politically,most of the participants were federalists ( n = 76,84%) with 12 neutral participants (13%) andthree undecided (3%).

Procedure Forty students participated forcourse credit in lower-level psychology courses,choosing from over 20 available experiments.The remaining 51 participants were compen-sated with C$8 for their time; these had beenrecruited from undergraduate courses at thebeginning of the term after signing up for apaid psychology subject pool. During the45-minute session, participants rst completed aconsent form and a demographic informationquestionnaire, in which their identity as Anglo-phone Qubecers was checked. Using stratiedrandom number table, participants were thenassigned to one of four conditions, Anglophonenorm support/opposition by Francophonenorm support/opposition. Participants com-pleted the behavior generation task and thebehavioral evaluation measures, were debriefed

verbally, given a written description of therationale, and thanked for their participation.

MaterialsAnglophone identity Participants first self-identied as Anglophone (in comparison toFrancophone, or Other); then, identicationas Anglophones was assessed with a 5-item scaleadapted from Porter (1995). The measures(e.g. How much do you feel a part of a largergroup/community of people who share thisidentity?) were completed on 11-point scalesfrom 0 (Not at all) to 10 (Extremely) and

were averaged to form an index of identifi-cation, = .73.

Norm manipulation and behavior generations Thestudy comprised a 2 (Francophone normative/


364



12/21

antinormative) 2 (Anglophone normative/antinormative) between-subject design. Ingroupor outgroup normative behaviors were denedas behaviors that most group members approvedof or supported, whereas antinormative behav-iors were dened as behaviors that most groupmembers disapproved of or opposed. Partici-pants read a description of the task which began:

Whenever two groups are in conflict, groupmembers have to make decisions about how they are going to act. Sometimes people nd actionsthat both groups approve of. For example, in aunionmanagement context, both groups might agree to have contract negotiations . Sometimespeople use behaviors that one group approves of and the other disapproves of. For example, a strike might be called by the union, but disapproved of by management, whereas strike-breaking might besomething that the union disapproves of, and themanagement is promoting. Finally, there arebehaviors that both groups disapprove of. In alabor context, for example, both groups might disapprove of property destruction .

When Anglophones and Francophones interact in Qubec, there are many ways that Anglos can try to advance their own status or that of their group. We are going to ask you to list behaviors that help Anglophones advance either their own individualinterests or those of Anglophones as a whole, that most Francophones and most Anglophonesapprove of [that most Francophones and most Anglophones disapprove of/that most Anglo-phones approve of, but that most Francophonesdisapprove of/that most Anglophones disapproveof, but that most Francophones approve of].

When the participant signalled to the exper-imenter that the instructions had been read, asheet was provided repeating the informationcharacterizing the norm condition and the par-ticipant was given 15 minutes to generate vebehaviors that t the criteria.

Behavior evaluations Participants were thenprovided with a booklet in which they evaluatedthe rst three behaviors they had generated. Wedid not perform a qualitative analysis per se forthese behaviors, but the modal responses ineach condition could be classied as: violent acts, pro-Anglophone graffiti/vandalism of French signs, racist insults (ingroup and out-group antinormative); assimilation to French at

work, assimilation to French at school, support-ing separatism (ingroup antinormative, out-group normative); English at work, English at school, bilingual signs 4 (ingroup normative,outgroup antinormative); and bilingualism ineducation, bilingualism at work, friendliness andpoliteness (ingroup and outgroup normative).

Group norms regarding the behaviors wereestablished by two sets of three questions eachadapted from Ajzen (1991), e.g. How wouldFrancophones/Anglophones react to a person

who did this behavior? Responses were on11-point Likert type scales ranging from 5(antinormative) to zero (neutral) to +5 (norma-tive). Scales were created for Anglophonenorm, = .96, and Francophone norm, = .94,by averaging across the three behaviors.

Individual expectancy-value processes wereassessed with four separate measures of the sub-

jective probability and value of benets andcosts to the actor. For example, participants

were asked What is the probability that you asthe individual actor will benet if this behavioris performed? and How important do youthink that the probable costs are, for you as theindividual actor, if this behavior is performed?.In each case the scales were unipolar measures,from 0 (Unimportant/Zero probability) to 10(Extremely important) or 100 (Certainty).

Weighted bipolar expectancy-value scores werecreated by multiplying expectancy and value,subtracting costs from benets, and rescaling toa 21-point scale (from +10, certain and import-ant benets only to 10, certain and important costs only). Expectancy-value scores wereaveraged across the behaviors, = .72.

Two measures of intentions were included,namely When the situation comes up, what percent of the time do you react with thisbehavior? and How likely is it that you wouldengage in this behavior? (0, Never to 100,Always). These items were averaged for eachbehavior, and intentions for the three behaviors

were combined, = .88.

Results

Manipulation check A 2 Anglophone normcondition (Support/Opposition) 2 Franco-phone norm condition (Support/Opposition)


365



13/21

2 target group norm (perceived Anglo-phone/Francophone norm) mixed-modelanalysis of variance was performed on ratedgroup norms for the behaviors. As expected,interactions were observed between Anglo-phone norm condition and target group norm( F (1, 87) = 90.88, p = .000, 2 = .51), andbetween Francophone norm condition andtarget group norm ( F (1, 87) = 127.18, p = .000,

2 = .59), while the three-way interaction wasnot signicant ( F (1, 87) = 1.25, p = .267, 2 =.01). Post hoc tests revealed that Anglophones

were perceived to evaluate the behaviors gener-ated in the Anglophone support conditionmore favorably ( M = 2.52) than in the Anglo-phone opposition condition ( M = 2.11, p =.000), and Francophones were perceived toevaluate the behaviors more favorably in theFrancophone support condition ( M = 2.23)than in the Francophone opposition condition(M = 2.52, p = .000). The manipulated normconditions were accordingly used as variables insubsequent analyses.

Design As in Study 1, hierarchical regressionanalyses were conducted on mean behavioralintentions (the dependent variable), and par-ticipants expectancy-value scores (potentialmediators) to test the hypothesis that ingroupand outgroup norms would predict cost-benet evaluations and intentions, and the cost-benet evaluations would mediate the norm-intention

links. Means, standard deviations, and intercor-relations for the variables are presented inTable 4. Anglophone (ingroup) and Franco-phone (outgroup) norm conditions wereentered as dichotomous variables (effect codes)scored +1 (support) or 1 (opposition). 5Continuous expectancy-value scores werecentered, and one outlier was excluded.

Group norms predict intentions Anglophoneand Francophone group norms predicted 42%of the variance in behavioral intentions ( F (2,87) = 32.94, p = .000). Participants had strongerintentions to engage in the actions in the Anglo-phone support condition ( = .60, p = .000; seeTable 5). However, as in Study 1, outgroupnorms predicted intentions independent of theimpact of ingroup norms ( = .28, p = .001).6

Group norms predict expectancy-value scores Anglophone and Francophone norm manipu-lations together predicted 31% of the adjusted

variance in rated benefits and costs to theindividual ( F (2, 87) = 20.72, p = .000). Ingroupnormative behaviors were thought to benet the individual actors ( = 0.41, p = .000), andindependently, outgroup (Francophone) normsfor the behaviors also positively inuenced cost-benet analyses ( = .40, p = .000). As in Study 1, calculations of the benets and costs of inter-group behaviors for the actor were informed by group-level norms.


366

Table 4. Means, standard deviations (SDs), and intercorrelations for group norms, cost-benet expectancy-value scores, and intentions

Variables 1 2 3 4

1. Anglophone norm condition 0.04(1.00)

2. Francophone norm condition .02 0.00(1.01)

3. Cost-benet expectancy-value scores .40*** .39*** 0.19(3.88)

4. Behavioural intention .59*** .27** .50*** 43.65(30.21)

*** p < .001; ** p < .01; * p < .05.Notes : Uncentered means (SD) in diagonal, and zero-order correlations below. Norm conditions: +1 =

support, 1 = oppose. Cost-benet scores ranged from +10 (certain, valuable benets improbable,unimportant costs) to 10. Behavioral intentions ranged from 0 to 100.



14/21

Expectancy-value scores mediate norm-intentionrelationships The mediating role of expec-tancy-value processes in the ingroup andoutgroup norm-intention relationships was thenexamined (Table 5, Block 2). When cost-benet analyses were entered in a regression equationalong with norm conditions, model t signi-cantly improved ( R 2change = .03, F (1, 86) = 5.16,

p = .026). Cost-benet analyses signicantly pre-dicted intentions ( = .22, p = .026). Residualdirect effects of both ingroup and outgroupnorms were observed, when cost-benet evalu-ations were entered. However, the expectancy-

value processes had signicantly mediated theeffect of both Anglophone ingroup norms onintentions, which decreased from = .60 to =.51 ( z = 2.04, p = .041), and Francophoneoutgroup norms, which decreased from = .28to = .19 (z = 2.03, p = .042).

Alternative causal models As for Study 1,alternative causal models were constructed andevaluated using standard error of the mean.

Table 6 provides t statistics from path analysesfor the present model and the same threealternative theoretical models: cost-benefit analyses drive behavior, with norms as amediator; behavioral intentions drive cost-benet analyses with norms as a mediator; or(3) behavioral intentions drive norms with cost-benet analyses as a mediator. In the present data, as for Study 1, t statistics (GFI, AGFI) arehigher and error statistics (RMSEA, AIC) arelower for alternative causal models than for thepresent model. Only the present model, in fact,produces adequate t and error statistics forthe data, strengthening the interpretation that norms are motivating behavior via cost-benet analyses rather than the reverse/alternativepaths.

Summary These results replicate the correla-tional relationships observed in Study 1 using amanipulated focus on supportive or hostileingroup and outgroup norms. Ingroup support predicted intentions to engage in pro-groupbehaviors (hypothesis 1) and more favorablecost-benefit evaluations (hypothesis 2). Out-group norms were also associated with strongerintentions (hypothesis 3) and more favorablecost-benet evaluations (hypothesis 4). Finally,expectancy-value scores signicantly mediatedthe relationships between norms and intentions(hypothesis 5), although residual direct effectsof norms were also observed.

Discussion

In the present study, as in Study 1, ingroupnorms for conict behaviors predicted favor-able cost-benet perceptions and intentions forintergroup behaviors. Behaviors that ingroup


367

Table 5. Study 2: Prediction of behavioral intentionsas a function of Anglophone and Francophonenorms and cost-benet expectancy-value scores

Block 1 2Adj. R 2 .42*** .44***R 2 change .43*** .03*

Predictor Standardized coefcients

Anglophone norm .60*** .51***Francophone norm .28** * .19*Expectancy-value score .22*

* p < .05; ** p < .01; *** p < .001.

Table 6. Study 2: Fit statistics for the present theoretical model (1) and three alternative path models

Model Chi-square GFI AGFI RMSEA AIC

1. Norms -> CBA -> Beh 0.04*** 1.00 .99 .00 18.042. CBA -> Norms -> Beh 9.34*** .95 .77 .20 25.343. Beh -> CBA -> Norms 33.99*** .87 .57 .34 47.994. Beh -> Norms -> CBA 10.13*** .95 .74 .21 26.13

* p < .05; ** p < .01; *** p < .001.Note : See Table 3 for denitions.



15/21

members supported were seen as more bene-cial to the individual actor, and these cost-benefit analyses significantly mediated thenorm-intention relationships for ingroupnorms. The ndings suggest that the impact of ingroup norms and cost-benet evaluations arenot necessarily independent, and support asocial identity model in which ingroup normscan be important in deliberation and strategicchoices (e.g. Abrams, 1994; Terry & Hogg,1996). However, outgroup norms also influ-enced cost-benefit analyses, in the present study. Conformity to outgroup norms regard-ing ingroup members conflict choices wasseen as beneficial, and the influence of outgroup norms on intentions was signicantly mediated by evaluations. Thus, outgroupnorms predicted the perceived benets of thebehaviors, which in turn predicted intentionsto engage in conflict choices. The present study illustrates a process that we call agentic normative inuence in which both ingroup andoutgroup norms inform deliberation for inter-group behaviors.

Regarding the relative strength of the medi-ation of cost-benet evaluations in Studies 1(full) and 2 (partial): it may be the case that the intergroup context was more salient inStudy 2 than with the consumer choices of Study 1, triggering processes in which group-level variables motivate intentions indepen-dently from cost-benefit evaluations at theindividual level (see also, Kelly, 1993; Louiset al., 2005; Klandermans, 1984; Simon et al.,1998). The contingencies that govern therelative importance of agentic normativeinuence, as mediated by rational cost-benet calculations, versus other processes of referent or subjective normative inuence, seem likely to be a fruitful topic of future research. For thepurposes of this paper, however, the results of Study 2 are important in that they provide areplication of the agentic normative influence process: in intergroup decision-making, bothingroup and outgroup norms predicted inten-tions and the evaluation of costs and benets,and the norm-intention relationships were sig-nicantly mediated by these expectancy-valueprocesses.

Limitations It was a goal of Study 2 to providea stronger test of the agentic model by using amanipulation of focus on supportive or hostilegroup norms and by accessing a wider range of behavior. Study 1 analyzed correlational vari-ation in norms, cost-benefit analyses, andintentions among a common set of behaviors(which were perceived to be ingroup normative/outgroup anti-normative). However, in Study 2,participants rst generated and then evaluatedbehaviors as a function of ingroup and outgroupapproval or disapproval. The norm manipula-tions successfully induced participants to focuson qualitatively different behaviorse.g. racist insults vs. bilingualism at workand it was therandomly assigned norm condition, rather thanmeasured norm perceptions, that was observedin Study 2 to inuence intentions both directly and via cost-benet perceptions. These effectsare consistent with a causal role for normativeinuence, and this interpretation is strength-ened by consideration of alternative pathmodels, as described above. Nevertheless, amore stringent test would manipulate groupnorms for particular behaviors and demonstratesubsequent changing cost-benefit perceptionsand effects on intentions. Moreover, when par-ticipants evaluated the costs and benets of thebehavior for the individual actor, benets suchas the satisfaction of afliation needs were not ruled out explicitly. Indeed, order effects may have focused respondents attention on socialoutcomes for the behaviors, where the normmeasures or manipulation preceded measure-ment of cost-benefit perceptions. A betterdesign would include counterbalanced measuresof expressive motivation and agentic motivationand explicitly assess the mediation of ingroupnorms and outgroup norms by each. Futureresearch may thus provide a test of agenticnormative influence by using longitudinalresearch to assess the effects of changing normperceptions on subsequent expectancy-valueprocesses, intentions, and behavior in the eld.

General discussionIn two studies, ingroup and outgroup norms forconflict behaviors were positively related to


368



16/21

cost-benefit perceptions and intentions forintergroup behaviors. Behaviors that were sup-ported by the outgroup or the ingroup wereseen as benecial for the individual actor, and

were more likely to be endorsed by participants.Moreover, the cost-benet analyses signicantly mediated the norm-intention relationships forboth ingroup and outgroup norms. This wastrue when norms were assessed for the samebehaviors, in Study 1, and across behaviors, inStudy 2. Thus, for Anglophones consideringbehaviors relevant to intergroup conflict inQubec, Anglophone and Francophone normsare observed to predict rather than suppressrational processes (see also, Mackie et al.,1992).

Two aspects of these results are of theoreticalimportance. First, the present studies articulatea process whereby group norms in conflict decision-making direct deliberative expectancy-

value processes, rather than operating inparallel or in conict. It should be reiteratedthat in many contexts, group members may wellhave motives for conformity to ingroup normsthat are independent of cost-benet calcula-tions. For example, fans can wear team colorsto fulll a need for approval, or to afrm orexpress a valued identity. Where there are notactical issues of eliciting desired outgroupresponses in ongoing interaction, conformity may be exclusively to ingroup norms and moti-

vated primarily by expressive needs. Thus,many motives to conform may operate inparallel to cost-benet calculations (as indeed issuggested by residual group norm effects inStudy 2; but see also Ajzen, 1991; Fishbein &

Ajzen, 1975; in politics, Kelly, 1993; Klander-mans, 1984; Simon et al., 1998). However, formany choices in intergroup conict, norms forgroups in conict may also inuence decision-making because they define the means toacquire benefits and to avoid costs. In twostudies in the context of EnglishFrenchrelations in Qubec, this was apparently thecase. Mediational analyses of agentic normativeinuence processes in other intergroup con-icts may allow the role of group norms in cost-benefit analyses to be better understoodand/or predicted.

A second aspect of the present results of theoretical interest is evidence of an indepen-dent role of outgroup norms in decision-making for intergroup behavior. A range of research and theory in intergroup relationssuggests that outgroup norms are often irrele-

vant to individuals decision-making (e.g. Terry & Hogg, 1996; Turner et al., 1989). However,because outgroups motivations and reactionsare targeted in conflict behaviors, salient outgroup norms may motivate deliberate con-formity or violation even when ingroup identi-ties are salient. In the present study, outgroupnorms independently predicted intentions, andinfluenced the cost-benefit calculations of ingroup members. Allowing ingroup andoutgroup norms to be included in conflict decision-making models may also improve thepredictive or explanatory power of intergrouptheories.

In both studies, Anglophones perceived con-formity to outgroup norms to be benecial, but the positive relationship of conformity withoutgroup norms and group-level expectancy-

value processes needs to hold across time orintergroup contexts (see also, Louis & Taylor,2002; Louis et al., 2005). Future research isrequired to specify when outgroup norms areconsidered irrelevant, versus motivating strate-gic conformity versus strategic norm violation.

We may offer several predictions on that score,however. First, when a societal consensusexists, outgroup norms may add nothing to theprediction of behavior once ingroup norms arecontrolled. For example, if both the ingroupand outgroup favor individuals voting ratherthan rioting, outgroup norms are statistically redundant when ingroup norms are controlled.Moreover, where no conict exists, there is noreason to consider behavioral alternatives andthus no reason to deliberate. Implicitly, a super-ordinate category and norm govern thebehavior via referent informational inuence .

However, when real differences in groupnorms exist, the salience of alternatives may engage deliberative processing. Strategic con-formity to outgroup norms may be motivated by desires to signal common values, a commonsuperordinate identity, or common goals;


369



17/21

strategic violation may be motivated by attemptsto increase the costs of inaction for theoutgroup. In the present study, conformity withoutgroup Francophone norms was seen as ben-eting the group, but future research is neededto tease apart the specic rationales that wereimportant.

Relative group power seems likely to be afactor moderating the relationship betweengroup norms and choices for conformity ornorm violation. Research in escalating inter-group violence suggests that perceptions of outgroup power may motivate norm violationbecause of stronger threat perceptions (e.g.Staub, 2001); however, research in discrimi-nation suggests that threatening outgrouppower may sometimes produce appeasement (Ng, 1982). We might predict that low relativepower creates stronger contingencies foroutgroup norms both for norm violation (themotivation for deance) and conformity (thecosts of deance). Similarly, low relative power

would presumably weaken both positive andnegative contingencies for ingroup norms.Further research may test these hypothesesempirically by manipulating power positions inad hoc groups, or by measuring the moderatingeffects of power perceptions on norm-evalu-ation relationships for real groups such as

Anglophones and Francophones.Finally, outgroup norms may drive attitudes

and behavior unstrategically when ingroupmembers ostentatiously reject the outgroupinfluence: a phenomenon called reactance(e.g. Brehm, 1966), or anticonformity (Nail,MacDonald, & Levy, 2000). Some research onstigmatized minorities suggests that defensiverejection of outgroup norms should be con-sidered in addition to, or even instead of,positive conformity to the ingroup (e.g.Falomir, Mugny, & Prez, 2000; in socialmovement research, e.g. Gurr, 1970; Kaplan& Liu, 2000). However, reactance appears toinvolve an esteem-motivated rejection of theoutgroup identity and we speculate that it doesnot implicate cost-benet calculations, whereasstrategic norm violation in the agentic model

would be motivated by expected beneficialconsequences to the ingroup and self.

In conclusion, the present studies draw onthe larger literature that deals with the dynamicof action and reaction in intergroup conict.Ingroup and outgroup norms for intergroupbehaviors predict cost-benet analyses that areproposed to mediate the relationship betweennorms and intentions to engage in conflict choices. The agentic normative inuence processseems likely to occur commonly in the decision-making of group members in intergroupconict. Assessment of this process may play apart in understanding the psychological pro-cesses that translate dynamics of intergroupaction and reaction into individuals choicesamong intergroup behaviors. Moreover,analysis of agentic inuence processes suggestsone means by which rational decision-makingis consistent with, and informed by, group-levelsocial inuence. In modeling the relationshipbetween cost-benet analyses and ingroup andoutgroup norms, intergroup researchers may begin to address two theoretically and socially important questions: How does group identity influence rational decision-making? andHow do individuals in intergroup conflict choose among inaction, conciliation, and con-frontation?

Notes1. Thus, as noted in the introduction, the group

norms here are not descriptive (whether Anglophones and Francophones would behavethis way themselves) but rather injunctive andrelational (whether Anglophone andFrancophone groups approve of the targetsbehavior). Just as with interpersonal sources (see

Ajzen, 1991), ingroup and outgroup membersmay react positively to behaviors that they are not likely to do themselves, and vice versa.

2. The impact of ingroup norms was not moderatedby outgroup norms. Inclusion of the interactiondid not increase model t ( R 2change = .00,

F (1, 192) = 2.64, p = .106), nor was the interactionterm signicant. Alternative models might include identication as an independent predictor (e.g. Kelly, 1993) or as a moderator of the effects of ingroup norms (e.g. Turner et al.,1989) or cost-benet perceptions (e.g. Terry &

Hogg, 1996) on intentions. However, these variables do not increase model t when group


370



18/21

norms and cost-benet perceptions arecontrolled ( F (3, 189) = 0.57, p = .637), nor areany coefcients signicant ( id = .06, p = .394;

idxnorm = .09, p = .290; idxcba = .07, p = .771).

3. We thank anonymous reviewers for thissuggestion. For a discussion of measures of modelt in structural equation modeling, see e.g. Huand Bentler (1995). The results in Table 3 suggest that although the alternative models tested aremuch worse than the present model, it itself haslarger than desirable error measures and smallerthan desirable adjusted goodness-of-t (AGFI).Modication indices suggest that the t wouldimprove to acceptable levels if ingroup andoutgroup norm measures were correlated, not modeled as independent. This intercorrelation is

also apparent in Table 1, as noted in thediscussion.4. Qubec legislation mandated that French

predominate on commercial signs and thusbilingual signs in Qubec can be subject to nesin some conditions (see Chevrier, 1997).

5. Effect codes (1/1) and dummy codes (1/0) forcategorical variables generate the same R 2change. However, Aiken and West (1991) suggest that where there are two or more categorical variables that may interact, effect coding ispreferable because dummy variables produce

correlated contrast vectors for the two maineffects and the interaction, whereas with equaln effect codes do not. As in Study 1, however, theinteraction of the variables was not signicant:the inclusion of the interaction term did not increase the t of the model ( R 2change = .02, F (3,82) = 0.93, p = .430), nor was the coefcient signicant.

6. As in Study 1, alternative models withidentication as an independent predictor, as amoderator of the effects of ingroup norms and of cost-benet perceptions on intentions were also

considered. Even when group norms andcost-benet perceptions were controlled, theentry of identication and the interaction termsincreased model t ( F (3, 83) = 4.04, p = .010). Analysis of the coefcients suggested that participants higher in identication were morelikely to engage in the conict behaviors in allconditions ( id = .23, p = .003). As in Study 1,however, strongly identied respondents were not more responsive to ingroup norms ( idxnorm = .11,

p = .312) or less inuenced by cost-benet calculations ( idxcba = .04, p = .410). It should be

noted that weak effects of identication, in bothStudies 1 and 2, may have occurred because the

identication measures preceded the normmeasures or manipulations. As participantssubsequently evaluated pro-group behaviorsand/or explicitly considered conicting norms

for conict behavior, the salience of the identity may have changed through the course of thestudy. Ingroup identity salience and identicationhave been robustly linked to greater conformity to ingroup norms in previous theory and research(e.g. Terry & Hogg, 1996). Thus, although it isnot a focus of the present paper, it would besurprising if moderating effects were not observed with a more powerful test, such asre-running the analyses with post-measures of identity salience.

AcknowledgmentsThis research was supported by a doctoral fellowshipto the rst author from the Social Sciences andHumanities Research Council of Canada. Theauthors wish to thank Deborah Terry, MatthewHornsey, and Kelly Fielding for their helpfulfeedback on an earlier draft of this manuscript.Results from Study 2 were presented in a paper at the bi-annual meeting of the Society for thePsychological Study of Social Issues, Toronto, ON,

June 2002.

References Abrams, D. (1994). Social self-regulation. Personality

and Social Psychology Bulletin, 20,473483. Aiken, L. S., & West, S. G. (1991). Multiple regression:

Testing and interpreting interactions . London: Sage. Ajzen, I. (1991). The theory of planned behavior.

Organizational Behavior and Human Decision Processes, 50,179211.

Asch, S. E. (1956). Studies of independence andconformity: A minority of one against aunanimous majority. Psychological Monographs, 70,170.

Baron, R. M., & Kenny, D. A. (1986). Themoderatormediator variable distinction in socialpsychological research: Conceptual, strategic, andstatistical considerations. Journal of Personality and Social Psychology, 51,11731182.

Blanz, M., Mummendey, A., & Otten, S. (1995).Perceptions of relative group size and groupstatus: Effects on intergroup discrimination in

negative evaluations. European Journal of Social Psychology, 25,231247.


371



19/21

Brehm, J. (1966). A theory of psychological reactance.New York: Academic Press.

Burstein, P., & Freudenberg, W. (1978). Changingpublic policy. American Journal of Sociology, 84,

99122.Cairns, E., & Darby, J. (1998). The conict in

Northern Ireland. American Psychologist, 53 (7),754760.

Chen, E. S., & Tyler, T. R. (2001). Cloaking power;Legitimizing myths and the psychology of theadvantaged. In A. Y. Lee-Chai & J. A. Bargh (Eds.),The use and abuse of power: Multiple perspectives on the causes of corruption (pp. 241261). Philadelphia:Psychology Press.

Chevrier, M. (1997). Laws and language in Qubec;The principles and means of Qubecs language

policy. Ministre des Relations internationals.Downloaded from http://www.mri.gouv.qc.ca/la_bibliotheque/langue/loi-langue_an.html January 9, 2003.

Dalton, R. J. (1988). Citizen politics in western democracies . Chatham, NJ: Chatham House.

De Ridder, R., & Tripathi, R. C. (Eds.). (1992). Norm violation and intergroup relations. New York:Clarendon Press.

Dion, K. L. (1979). Intergroup conict andintragroup cohesiveness. In W. G. Austin &S. Worchel (Eds.), The social psychology of intergroup

relations (pp. 211224). Monterey, CA:Brooks/Cole.Dion, K. L. (1986). Responses to perceived

discrimination and relative deprivation. In J. M.Olson, C. P. Herman, & M. P. Zanna (Eds.),Relative deprivation and social comparison: The Ontario symposium (Vol. 4, pp. 159179). Hillsdale,NJ: Erlbaum.

Drury, J., & Reicher, S. (2000). Collective action andpsychological change: The emergence of newsocial identities. British Journal of Social Psychology,39 , 579604.

Falomir, J. M., Mugny, G., & Prez, J. A. (2000).Social inuence and identity conict. In D. J.Terry & M. A. Hogg (Eds.), Attitudes, behavior, and social context: The role of norms and group membership (pp. 313337). London: Erlbaum.

Feather, N. T. (Ed.). (1982) Expectations and actions: Expectancy-value models in psychology . Hillsdale, NJ:Erlbaum.

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

Gaertner, L., & Insko, C. A. (2000). Intergroup

discrimination in the minimal group paradigm:Categorization, reciprocation, or fear? Journal of Personality and Social Psychology, 79,7794.

Gaertner, S. L., Dovidio, J. F., Anastasio, P. A.,Bachman, B. A., & Rust, M. C. (1993). Thecommon ingroup identity model: Recategorizationand the reduction of intergroup bias. In W. Stroebe

& M. Hewstone (Eds.), European review of social psychology (Vol. 4, pp. 126). New York: Wiley.

Gecas, V. (2000). Value identities, self-motives, andsocial movements. In S. Stryker, T. J. Owens, &R. W. White (Eds.), Self, identity, and social movements (pp. 93109). Minneapolis, MN:University of Minnesota Press.

Gorsuch, R. L., & Ortberg, J. (1983). Moralobligations and attitudes: Their relation tobehavioral intentions. Journal of Personality and Social Psychology, 44,10251028.

Gurr, T. R. (1970). Why men rebel.Princeton, NJ:

Princeton University Press.Hu, L.-T., & Bentler, P. M. (1995). Evaluating modelt. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 7699). London: Sage.

Hunter, J. A., Stringer, M., & Watson, R. P. (1991).Intergroup violence and intergroup attributions.British Journal of Social Psychology, 30,261266.

Kaplan, H. P., & Liu, X. (2000). Social movements ascollective coping with spoiled personal identities:Intimations from a panel study of changes in thelife course between adolescence and adulthood.

In S. Stryker, T. J. Owens, & R. W. White (Eds.),Self, identity, and social movements (pp. 215238).Minneapolis, MN: University of Minnesota Press.

Kelly, C. (1993). Group identication, intergroupperceptions and collective action. In W. Stroebe &M. Hewstone (Eds.), European review of social

psychology (Vol. 4, pp. 5983). New York: Wiley.Klandermans, B. (1984). Mobilization and

participation: Social-psychological expansions of resource mobilization theory. American Sociological Review, 49, 583600.

Lemaine, G. (1974). Social differentiation and social

originality. European Journal of Social Pyschology , 4 ,1752.Linteau, P.-A., Durocher, R., Robert, J.-C., & Ricard,

F. (1991). Quebec since 1930 (R. Chodos & E.Garmaise, Trans.). Toronto, ON: Lorimer.(Original work published 1986)

Liska, A. (1984). A critical examination of the causalstructure of the Fishbein and Ajzen attitude-behavior model. Social Psychology Quarterly, 47 ,6174.

Louis, W. R., & Taylor, D. M. (2002). Understandingthe September 11th terrorist attack on America:

The role of intergroup theories of normativeinuence. Analyses of Social Issues and Public Policy,2 (1), 87100.


372



20/21

Louis, W. R., Taylor, D. M., & Neil, T. (2005).Cost-benet analyses for your group and yourself:The rationality of decision-making in conict.International Journal of Conict Management , 15 (2),

110143.Mackie, D. M., Gastardo-Conaco, M. C., & Skelly, J. J.(1992). Knowledge of the advocated position andthe processing of in-group and out-grouppersuasive messages. Personality and Social Psychology Bulletin, 18,145151.

Mackie, D. M., & Queller, S. (2000). The impact of group membership on persuasion: RevisitingWho says what to whom with what effect? InD. J. Terry & M. A. Hogg (Eds.), Attitudes, behavior,and social context: The role of norms and group membership (pp. 135155). London: Erlbaum.

Maclure, J. (2000). Le projet nationaliste entransition: la n des vaches sacres? [Nationalism intransition: The end of sacred cows?]. In R. Ct(Ed.), Qubec 2001: Annuaire politique, social,conomique et culturel (pp. 9196). Qubec: Fides.

Milgram, S. (1974). Obedience to authority: An experimental view.New York: Harper & Row.

Miller, D. T., & Prentice, D. A. (1994). Collectiveerrors and errors about the collective. Personality and Social Psychology Bulletin, 20 , 541550.

Miniard, P. W., & Cohen, J. B. (1981). Anexamination of the Fishbein behavioral intentions

models concept and measures. Journal of Experimental Social Psychology, 17,309329.Moscovici, S. (1985). Social inuence and

conformity. In G. Lindzey & E Aronson (Eds.), The handbook of social psychology (3rd ed., pp. 347412).Hillsdale, NJ: Erlbaum.

Nail, P. R., MacDonald, G., & Levy, D. A. (2000).Proposal of a four-dimensional model of socialresponse. Psychological Bulletin, 126 , 454470.

Ng, S.-H. (1982). Power and appeasement inintergroup discrimination. Australian Journal of Psychology, 34 , 3744.

Olson, M. (1968). The logic of collective action.New York: Schocken.OMalley, M., & Bowman, J. (2001). Language in

Quebec . Canadian Broadcasting Corporation.Downloaded from http://www.cbc.ca/news/indepth/language/ January 9, 2003.

Osgood, C. E. (1962). An alternative to war or surrender . Urbana, IL: University of Illinois Press.

Piven, F., & Cloward, R. (1977). Poor peoples movements . New York: Pantheon.

Porter, L. A. (1995). Managing intergroup conict in interpersonal relationships: How women maintain

intimate relationships with men . Unpublisheddoctoral dissertation, McGill University, Montreal,Qubec, Canada.

Reicher, S. D. (1984). Social inuence in the crowd: Attitudinal and behavioral effects of deindividuation in conditions of high and lowgroup salience. British Journal of Social Psychology,

23 , 341350.Reicher, S., & Levine, M. (1994). Deindividuation,power relations between groups and theexpression of social identity: The effects of visibility to the outgroup. British Journal of Social Psychology, 33 , 145163.

Rokeach, M. (1960). The open and closed mind: Investigations into the nature of belief systems and

personality systems . New York: Basic Books.Rouhana, N. N., & Bar-Tal, D. (1998). Psychological

dynamics of intractable ethnonational conicts:The IsraeliPalestinian case. American Psychologist,

53 , 761770.Sherif, M. (1936). The psychology of social norms . New York: Octagon.

Sherif, M. (1966). Group conict and co-operation: Their social psychology.London: Routledge & Kegan Paul.

Simon, B., Loewy, M., Sturmer, S., Weber, U.,Freytag, P., Habig, C. et al. (1998). Collectiveidentication and social movement participation.

Journal of Personality and Social Psychology, 74 :646658.

Staub, E. (2001). Genocide and prevention: Theirroots and prevention. In D. J. Christie, R. V.

Wagner, & D. D. N. Winter (Eds.), Peace, conict,and violence (pp. 7686). Upper Saddle River, NJ:Prentice-Hall.

Tajfel, H., & Turner, J. C. (1979). An integrativetheory of intergroup conict. In W. G. Austin &S. Worchel (Eds.), The social psychology of intergroup relations (pp. 3347). Monterey, CA: Brooks/Cole.

Terry, D. J., & Hogg, M. A. (1996). Group norms andthe attitudebehavior relationship: A role forgroup identication. Personality and Social Psychology Bulletin, 22 , 776793.

Tilly, C. (1975). The rebellious century.Cambridge,

MA: Harvard University Press.Tramow, D. (2000). A theory of attitudes, subjectivenorms, and private versus collective self-concepts.In D. J. Terry & M. A. Hogg (Eds.), Attitudes,behavior, and social context: The role of norms and group membership (pp. 4765). London: Erlbaum.

Turner, J. C., Wetherell, M. S., & Hogg, M. A. (1989).Referent informational inuence and grouppolarization. British Journal of Social Psychology, 28,135147.

Vallerand, R. J., Deshaies, P., Cuerrier, J.-P., Pelletier,L. G., & Mongeau, C. (1992). Ajzen and Fishbeins

theory of reasoned action as applied to moralbehavior: A conrmatory analysis. Journal of Personality and Social Psychology, 62 , 98109.


373



21/21

Wright, S. C., Taylor, D. M., & Moghaddam, F. M.(1990). Responding to membership in adisadvantaged group: From acceptance tocollective protest. Journal of Personality and Social

Psychology, 58 , 9941003.

Paper received 1 April 2004; revised version accepted 19 August 2004.

Biographical notes W I NN I FR E D R . LOUIS , PhD, is a lecturer in the

school of psychology at the University of Queensland. Her research interests include

decision-making in intergroup conict, socialinuence, and collective action.

D ON AL D M . TAYLOR , PhD, is a professor in the

department of psychology at McGill University,Canada. He has conducted research in diversesettings involving issues of intergroup conict andminority language development.

R EB EC CA L. DOUGLAS (BA, McGill University) is adoctoral student in clinical psychology at theUniversity of Dalhousie, Canada. Her researchinterests include emotion and memory,psychopathy, and violent behavior.


374

group processes inter group relations 2005 louis 355 74

Documents