The effect of difference oriented communication on the subjective validity of an in-group norm: Doc can treat the group

The Effect of Difference Oriented Communication on the SubjectiveValidity of an In-Group Norm: DOC Can Treat the Group

Daniel FringsLondon South Bank University

Dominic AbramsUniversity of Kent

The subjective group dynamics model predicts that in-group deviants who violatein-group norms that differentiate between the in-group and the out-group threaten thein-group’s public image and its sense of validity. Previous research has shown that, toreduce this threat, group members attempt to symbolically marginalize in-group devi-ants through negative evaluation. In the current study (N � 107), the effect of anotherform of symbolic marginalization (difference oriented communication) is investigated.The findings support the subjective group dynamics model by showing that groupmembers whose communications to deviants highlighted differences experienced asubsequent increase in subjective validity of in-group norms.

Keywords: communication, deviance, social identity, in-group

Deviant in-group members are those whoseconduct departs from group-specific behavioraland attitudinal norms that sustain the subjectivevalidity of the in-group (Abrams, Marques,Bown, & Henson, 2000). Subjective validity,the conviction that the in-group is “right” (cf.Festinger, 1954) and hence sustain positive so-cial identity reality for group members. A vari-ety of responses to in-group deviance have beenobserved. Research on the black sheep effect(Marques, Yzerbyt, & Leyens, 1988) shows thatfailure to meet normative standards leads togreater derogation when actors are members ofan in-group, as opposed to an out-group.

Derogation of in-group deviants may protectthe in-group in various ways. Low evaluationsof in-group deviants may signal to both in-group and out-group members that the deviant’sbehavior does not represent that of a typicalgroup member (Marques, Abrams, Paez, &Martinez Taboada, 1998). Harsh punishment of

in-group deviants may also stop the spread ofdeviant behavior to other group members(Ouwerkerk, Kerr, Gallucci, & Van Lange,2005). Other responses to deviance have alsobeen documented—including declassifying thedeviant as a group member (Eidelman, Silvia, &Biernat, 2006) and increasing levels of commu-nication toward the deviant (Frings et al., 2010;Schachter, 1951). Although a variety of re-sponses to in-group deviance have been inves-tigated, there is relatively little research directlyexamining the role of communication betweengroup members and in-group deviants. Thepresent research does so and tests hypothesisfrom the Subjective Group Dynamics (SGD)model (Abrams et al., 2000; Marques, Abrams,& Serodio, 2001).

The SGD model (Abrams et al., 2000) arguesthat in-group deviants present a threat becausethey challenge the validity of positive in-group/out-group differentiation. To provide positivesocial identity, groups must provide a degree ofcertainty about how the self and others shouldbehave, and that group members agree withthese norms. The maintenance of such subjec-tive validity requires that in-group memberssustain the subjective reality of the group’snorms and standards by adhering to them, or bydifferentiation between normative and deviantmembers. Deviants may threaten subjective va-lidity by calling into question in-group normsand also potentially by reducing the overall

Daniel Frings, Department of Psychology, London SouthBank University; and Dominic Abrams, Department of Psy-chology, University of Kent.

This research was supported by ESRC Grant PTA-030-20003-01347 to the first author and complies with BPS andAPA ethical guidelines.

Correspondence concerning this article should be ad-dressed to Daniel Frings, Department of Psychology, Lon-don South Bank University, 102 Borough Road, London,UK SE1 6LN. E-mail: [email protected]

Group Dynamics: Theory, Research, and Practice © 2010 American Psychological Association2010, Vol. 14, No. 4, 281–291 1089-2699/10/$12.00 DOI: 10.1037/a0019162

281

difference between the in-group and out-group,generating a distinctiveness threat (see Jetten,Spears, & Postmes, 2006). Marques at al.(2001) argue that in many cases it is impossibleto remove the deviant from the group; thus,negative evaluation of the deviant serves tore-enforce group consensus by ‘symbolicallymarginalizing’ the deviant. We note that mar-ginalization can take various forms, for examplereducing the boundaries of what constitutes ac-ceptable behavior (Eidelman, Silvia, & Biernat,2006). Overall, therefore, highlighting the ab-normality of the deviant can strengthen groupnorms by reinforcing consensus among the ma-jority.

The present article tests a key assumption ofthe SGD model; namely that when deviants aresymbolically marginalized, the perceived valid-ity of positive in-group/out-group differentia-tion should increase. Additionally, it argues thatwhen group membership is fixed, forms of sym-bolic marginalization other than extreme negativeevaluation are likely to be used. Specifically, thisresearch considers the role of difference ori-ented communication.

Difference Oriented Communication asSymbolic Marginalization

Communications can take various forms.Some can be oriented toward differences be-tween the self and other individuals and groups(such as insults, persuasions or the use of eth-nonyms; see Gabriel, 1998; Mullen, Calegero,& Leader, 2007). Others can avoid highlightingsuch differences (by focusing on neutral, com-mon ground and agreed topics, for instance, inthe case of the British, a discussion of theweather). Small group research has shown thatconsiderable effort is expended communicatingwith in-group deviants (Festinger, 1950, 1954,1957; Schachter, 1951). Recently, Frings et al.(2010) showed that group members are pre-pared to attempt to persuade deviants to becomenormative. When fellow group members agreedto contact the deviant and the issue under con-sideration was considered important, intentionsto persuade and actual persuasive communica-tion also increased. However, the research didnot investigate whether persuasive (or othercommunications) aimed at deviants were re-lated to subjective validity, as would be pre-dicted by SGD.

To address this, the present research investi-gates ‘difference oriented communication’(DOC), which we consider a form of symbolicmarginalization. DOC is specifically defined asany communication which signals to the devi-ant, the self, and/or others that the deviant’sbehavior or attitude is non-normative. DOCmay operate by decreasing the possibility thatthe group’s position is miscast as the same asthe deviant’s. Potential miscasting has been pro-posed as an underlying threat posed by in-groupdeviants (see Eidelman & Biernat, 2003). Notethat to achieve this, the DOC need not change theattitude/behavior of the individual. The act ofcommunicating should itself suffice. AlthoughDOC could be directed either toward in-groupdeviants or out-group members, it is likely theformer will be more prevalent. Group members,particularly those showing in-group bias, typi-cally prefer contact with in-group members (seeBinder et al., 2009). Furthermore, addressingthe deviant’s continuing norm violation may beperceived as more effective and important thansimply limiting negative impact of deviancewhich has already occurred through communi-cating with out-group members (e.g., by tryingto dismiss or explain away the deviant’s behav-ior).

Engaging in DOC should increase feelings ofsubjective validity. The selection and evaluationof evidence that deviant opinions/behaviors arenon-normative necessitates an assessment of thein-group’s position relative to the deviant’s.This is likely to involve favorable comparisonsbetween other group members and the deviantand, simultaneously, comparisons that establishthe similarity and consensus among the rest ofthe group. Both comparisons will decrease un-certainty about the validity of the in-group’sattitudinal/behavioral position (Festinger,1954). Moreover, conditions which inducehighly negative evaluations of in-group devi-ants also invoke high levels of DOC, furthersuggesting both serve a similar psychologicalpurpose. For instance, in-group members statethat they are more willing to attempt to persuadein-group deviants when the status of the in-group is unstable or when the norm violated isperceived to be important (e.g., Marques et al.,2001).

DOC can take a variety of forms. Most di-rectly, group members may seek to activelypersuade deviants or out-group members. Such

282 FRINGS AND ABRAMS

communication is reflected by task oriented lan-guage which outlines the reason(s) that the per-suader’s point of view is correct. This can bebased on rational arguments (systematic pro-cessing) or by activating cognitive short-cuts(heuristic processing, see Chaiken, Liberman, &Eagly, 1989). The latter form of persuasion mayinclude reference to group membership. Lessdirectly, rather than attempting actively to per-suade others, group members can inquire intowhy the group member holds an opinion. Thismore passive form of communication focuseson information gathering rather than attitudechange. However, it still highlights differencesbetween the normative and deviant positions,and should still act as a form of symbolic mar-ginalization. Other difference highlighting re-sponses are also possible. Group members maytaunt and insult in-group deviants, or attempt toforce them to leave the group as a form ofsanction (Gabriel, 1998). These responses con-tain aggressive, insulting, or exclusionary lan-guage. They may also involve direct criticismand may hint at potential ostracism, by pointingout the differences between the deviant and thenormative in-group position. Although varioustypes of communication may have differingcontent, all attempt at some level attempt toquestion the deviant’s opinion and de-legitimizeit by highlighting differences between norma-tive and deviant members; hence, all arecounted as DOC in the present research. Incontrast, group members may engage in nondif-ference oriented communication. For instance,they may simply accept the difference in opin-ion between themselves and the deviant, andneither attempt to persuade or challenge thedeviant’s position. Such non-DOC responsesshould be less likely to increase subjective va-lidity.

Desire to engage in DOC is likely to varyaccording to situational and individual differ-ences. Frings et al. (2010) demonstrated thatattempts to persuade in-group deviants weremore likely when the norm violated was per-ceived as important. Violation of norms that areperceived to be important may have a greaterpotential impact on subjective validity and soshould also be associated with more DOCwhich in turn will be associated with increasedlevels of subjective validity.

To summarize, we contend that DOC is aform of symbolic marginalization. To investi-

gate the effects of DOC on subjective validity,three specific hypotheses are tested. Negativeevaluation of in-group deviants should be asso-ciated with a higher proportion of DOC relativeto non-DOC. In addition, feelings of subjectivevalidity of the in-group norm (relative to theout-group norm) should be greater after DOC,but not after non-DOC. DOC should be morelikely when the issue is perceived to be impor-tant.

In the present study, group members evalu-ated the in-group and out-group, then evaluateda member of each group before being given anopportunity to send a message to a person whoopposed the in-group normative opinion andwho was either an out-group member or anin-group (deviant) member. The effect of mes-sage type upon changes in relative subjectivevalidity before and after communication weremeasured.

Method

Participants

Participants were 107 psychology undergrad-uates from the University of Kent who com-pleted the study in exchange for course credits.The mean age of participants was 20.57 years(SD � 4.63). Ninety-three were female, 13 weremale, and 1 did not indicate their sex.

Design

A 2 (Time: 1; Precommunication, 2; Post-communication) � 2 (Message content: DOCvs. non-DOC) design was used. Time was awithin participants variable, and DOC conditionwas based on coding of messages sent betweenthe participant and other group members (seebelow). An a priori assumption was that partic-ipants would favor the in-group as a whole overthe out-group as a whole. In addition, the de-pendent variables were evaluation of both thein-group deviant and the out-group member,and the change in subjective validity before andafter communication. A manipulation checkwas included to ensure that the deviant wasperceived to hold a deviant attitudinal positionrelative to the in-group.

283DIFFERENCE ORIENTED COMMUNICATION

Procedure

Data were collected using an online question-naire. Participants were initially contacted viaan email asking them to participate in a briefpsychological survey. When participants loggedonto the study they were informed of their rightsas participants and gave their informed consent.Participants could only commence the studyonce they had indicated they had read and un-derstood their rights, were over 18 and provideda valid email address for the purpose of debrief-ing.

Using an established paradigm (Marques etal., 1998), the first screen ostensibly testedwhether participants were either “creative” or“methodological” thinkers by asking them fourquestions (e.g., ‘Do you make decisions quicklyor slowly?’) and asking them twice which oftwo similar images they preferred from twopairs of pictures. In reality, participants wereassigned randomly to a group. Feedback in-formed them they were either “methodological”or ‘creative thinkers’. Methodological thinkerswere then told: “Methodological thinkers areefficient and are more likely to come up withpractical solutions in complex situations. Meth-odological thinkers have an eye for detail. Cre-ative thinkers are good at looking at problems ina new light but they are sometimes impractical.”Creative thinkers were told: “Creative thinkersfind novel solutions to problems and they tendto think of the big picture. In contrast, Method-ological thinkers have an eye for detail, but theprocesses they use to arrive at solutions maylack flair and originality.” Subsequent analysisrevealed no differences in the responses of par-ticipants assigned to either of the two thinkinggroups and no main effects or interactions withother variables.

Participants were then presented with twoarguments in favor of the introduction of smok-ing restrictions on their university campus andtwo against. Participants were told they wouldbe placed in small discussion groups of peoplewith the same thinking style, and would berequired to come to a group decision as towhether they supported or were against suchrestrictions. It was stressed in the text that thediscussion group’s decision would only be re-corded if a unanimous decision was reached. Anew page then recorded the participant’s deci-sion (agree with the ban, disagree, or unsure)

and the importance the issue had for them. Thismethodology was adopted to ensure the devi-ant’s behavior had actual ramifications for thediscussion group and, in turn, the public imageof the thinking group as a whole (see Frings etal., 2010).

The next screen established contrasting in-group and out-group norms on the opinion issueand depicted the in-group deviant’s position inthat intergroup context. Participants were in-formed that 84.6% of members of previous dis-cussion groups with the in-group’s thinkingstyle had expressed the same opinion as theparticipant. In addition, an initially blank tablewas displayed that was gradually populated byinformation about the opinions of other mem-bers of the participant’s specific thinking group(creating a discussion group comprised of in-group members). Participants were shown a ta-ble displaying the opinions of each of the mem-bers (identifiable only by participant number),including the participant.1 Normative in-groupmembers expressed the same attitudinal posi-tion as the participant, consistent with the normheld by the in-group as a whole. One in-groupmember, identified as “participant #130” (thedeviant), disagreed with this normative posi-tion.

To set the scenario in an intergroup context,participants were simultaneously shown a tableshowing the opinions of out-group members(those from the thinking group to which theparticipant did not belong). All out-group mem-bers expressed the opposing view to the in-group. This procedure ensured that the evalua-tion of the deviant was an intragroup judgmenttaking place in an intergroup context, and alsoensured that the discussion groups were per-ceived to be synonymous with the thinkinggroups.

Once the participant indicated he or she hadread the information the measures of group fa-vorability and the Time 1 (precommunication)measures of subjective validity were presented.Participants were presented with a screen thatmade them wait for several seconds while twogroup members were ostensibly selected at ran-

1 Either 2 or 5 in-group and out-group members weredisplayed. This manipulation did not have a main effect on,or interactive effects upon, subjective validity when in-cluded in a separate ANOVA. Nor did it moderate theeffects of type of communication.

284 FRINGS AND ABRAMS

dom to be evaluated by the participant. Thedeviant group member and an out-group mem-ber were always selected (this was to ensureparticipants always evaluated both an out-groupmember and the deviant).

A screen of instructions was then displayedexplaining that the participant would be requiredto send a message to one of these two individuals.Participants were also told that failure to reach aconsensus within their in-group would mean thatno decision about smoking could be recorded.Participants were then asked to compose a shortmessage either to the deviant or the out-groupmember (their choice) and submit it for sending.After this was done, the postcommunication sub-jective validity measures were presented, and thenthe manipulation check of the deviant’s attitudinalposition was completed. This check was includedat the end of the study to ensure the participant didnot prematurely discover the hypotheses. Full de-briefing followed via email after the study hadbeen completed.

Measures

Importance. Participants were asked“How important to you is the issue of the smok-ing ban?” (1 � Very unimportant, 7 � Veryimportant).

Group preference. Group preference wasmeasured by asking participants how much theyagreed with the same two items for both the in-group and out-group; ‘I feel favorable toward[in-group/out-group] thinkers as a group’ and ‘[in-group/out-group] thinkers are good at solvingproblems’ (1 � ‘Not at all agree’, 7 � ‘Verymuch agree’). The in-group measure was pre-sented first, followed by the out-group measure.In-group favorability had a reasonable internalreliability (Cronbach’s � � .76), reliability ofout-group favorability was acceptable (Cron-bach’s � � .64).

Decision. Participants were asked to indi-cate their decision on proposals to increasesmoking restrictions on their campus. Partici-pants could say they agreed, disagreed, or wereunsure of their position on the issue. Fifty-nineparticipants were in favor of the restrictions, 40were against, and 8 were unsure. Data fromparticipants who were unsure were subse-quently discarded (during the study these par-ticipants were told they would be included asthe more popular “in favor” position, and the

rest of the study treated them as such). Thesedata were excluded to ensure that group membersincluded in the study clearly endorsedthe in-group normative position, and rejected theout-group position (to the extent to which theissue was viewed as important. Inclusion ofthese participants in all subsequent analysis didnot affect the significance of results, with theexception of one comparison.2 Subsequent anal-ysis revealed no effect of own opinion on theresponses of participants, and no interactionswith other variables.

Favorability. Favorability toward the de-viant in-group was measured using four itemsasking how likable, friendly, pleasant and intel-ligent they thought the group member was. Allitems were 7-point Likert-type scale anchoredat 1 (e.g., ‘Very unlikeable’) and 7 (e.g., ‘Verylikable’). A composite mean score was calcu-lated (Cronbach’s � � .87).

Subjective validity. Based on Randsley deMoura (2003) a 4-item scale was used to mea-sure relative subjective group validity. Theseitems asked how fair, valid, reasonable, andrational were the attitudes to smoking restric-tions held by both the in-group and out-group(1 � ‘Not at all,’ 7 � ‘Completely’). Randsleyde Moura (2003) found that overall reliabilityfor this scale was high (Cronbach’s �s � .80).Based on split half means from previous re-search the scale was split into two. At Time 1(precommunication) the “fair” and “valid”items were used. Scale reliability was good forthe in-group and out-group, (Cronbach’s �s �.77 and .83). At Time 2 (postcommunication)the items used were “reasonable” and “ratio-nal.” Reliabilities for these scales were good forthe in-group and the out-group (Cronbach’s�s � .89 and .79) so composite means werecalculated. At each time, mean out-group valid-ity was subtracted from mean in-group validityto create a relative subjective validity index.Positive scores on this scale indicate the partic-ipant felt their group’s attitude was more validthan that of the out-group.

Deviant’s attitudinal position. The devi-ant member’s attitude relative to the in-groupwas measured using the question ‘How typicalof (in-group) thinkers was participant #130’s

2 The difference between inquiring messages and non-DOC fell from p � .05 to p � .09.

285DIFFERENCE ORIENTED COMMUNICATION

attitude toward smoking restrictions on campus’(1 � ‘Not at all typical’, 7 � ‘Very typical’).

Results

Manipulation Checks

Group preference. In line with the manip-ulation, the in-group was preferred (M � 4.50,SD � 1.05) to the out-group (M � 3.97, SD �.96), t(102) � 4.21, p � .001, confirming thatgroup membership was valued by the participants.

Favorability. The SGD model holds thatpeople who more strongly differentiate betweenout-group and in-group norms should also morestrongly derogate in-group deviants. This pat-tern was observed in the present research. Con-trolling for evaluations of the in-group, themore negatively the out-group was evaluatedthe more negatively the in-group deviant wasevaluated, pr � .30, p � .003.

Deviant’s attitudinal position. Four par-ticipants did not complete this scale. The simi-larity between the deviant’s attitude and that ofthe group (M � 2.81, SD � 1.32) fell signifi-cantly below the midpoint of the scale (4),t(99) � 9.04, p � .001. This confirms that theattitude held by the deviant was perceived asbeing deviant.

Main Analysis

Message content. Group members sentmessages to their choice of either an in-groupdeviant (n � 75) or an out-group member (n �22). Content analysis of the messages was un-dertaken by two coders blind to condition andthe hypothesis of the study (Cronbach’s � �.79). Each message was classed as one of thefollowing: Nondifference oriented (n � 25;classed as explicitly accepting the target’s rightto hold their opinion or ignoring the difference;for instance by sending a friendly greeting, per-suasion (n � 43; e.g., “you should vote for aban because smoking is bad for everyone’shealth”), inquiry (n � 23; e.g., “what are yourreasons for your decision? how did you come tothis decision?”), or sending a blank message(n � 6). As the meaning of a “blank” messageis theoretically unclear, possibly signaling ei-ther acceptance or ostracism, these participantswere excluded from the main analysis. Frequen-cies of messages according to message target

can be seen in Table 1. Chi-square analysesrevealed nonequal frequencies between in-group and out-group targets, with more mes-sages being directed at the in-group than theout-group, �2(1) � 28.52, p � .001. In addition,persuasion was more frequent than other typesof communication, �2(2) � 8.00, p � .018.Chi-square on the complete design (represent-ing the interaction) revealed no significanteffect, �2(2) � 3.10, p � .21. Collapsing in-quiring and persuasive messages into a singlecategory (DOC) and comparing it to non-DOC,across message target showed that DOC wasmore frequent than non-DOC, �2(1) � 18.47,p � .001. Frequencies for the full 2 � 2 tablewere also nonequal, �2(2) � 31.66, p � .001,suggesting that DOCs targeted at in-groupmembers were most frequent.

Subjective validity. The small number ofeach type of message sent to out-group mem-bers precluded analysis of a fully crossedANCOVA design (Message target; in-group de-viant vs. out-group member X Message type;persuasion, inquiry, non-DOC, blank message)because of small cell sizes.3 Hence, the follow-ing analysis focused only on messages sent tothe in-group deviant to test the effects of differ-ent types of DOCs compared to Non-DOCs;Sixteen participants sent inquiring messages toin-group members, 37 sent persuasive messagesand 18 sent non-DOC. Two participants wereoutliers on the Time 1 subjective validity mea-sures (falling more than three standard devia-tions from the mean) and 1 of these was also anoutlier on the Time 2 measure. Following pro-cedures laid out by Tabachnick and Fidell

3 Although insufficient numbers of participants sent mes-sages to out-group members to provide detailed analysiscomparing differing types of DOC, sufficient were presentto compare non-DOC versus aggregated DOCs (mean ag-gregate of persuasion and inquiry conditions). ANCOVA onTime 2 subjective validity was conducted with Time 1subjective validity as a covariate. Message type (DOC vs.Non-DOC) and message target (in-group deviant vs. out-group member) were independent variables. The covariatewas significant, F(1, 86) � 28.83, p � .001, �2 � .25. Therewas a main effect of message type F(1, 86) � 6.48, p �.013, �2 � .07, reflecting higher levels of Time 2 subjectivevalidity amongst group members who sent DOCs(M � 1.11, SD � 1.43) than amongst those that sentnon-DOCs (M � .08, SD � 1.13). The main effect ofmessage target and the interaction term were both nonsig-nificant, Fs � .13, ps � .72, �2s � .001.

286 FRINGS AND ABRAMS

(2001) these scores were winnowized to 3 SDsfrom the mean.

ANCOVA was conducted to test the effectsof differing types of communication sent to thedeviant, with Time 2 subjective validity as thedependent variable, message type (Persuasion,inquiry, non-DOC) as the independent variableand Time 1 subjective validity as a covariate.The effect of the covariate showed that Time 1and Time 2 validity were significantly related,F(1, 67) � 28.02, p � .001, �2 � .30. Moreimportantly, message type had a significant ef-fect on Time 2 subjective validity (once it hadbeen adjusted for Time 1 validity), F(3,67) � 3.49, p � .036, �2 � .09. An a prioricontrast revealed Time 2 subjective validity waslower when a non-DOC was sent (M � .11,SD � 1.01), than when a DOC (a persuasive orinquiry message) was sent (M � 1.14,SD � 1.30), F(1, 67) � 12.81, p � .001, �2 �.16. To explore this effect further, the maineffect of message type was unpacked. Time 2subjective validity was significantly lower whennon-DOC (M � .11, SD � 1.01) was sent thanwhen the message was inquiring (M � 1.34,SD � 1.08, p � .013) or persuasive (M � .95,SD � 1.53, p � .045). There was no differencein Time 2 subjective validity following persua-sion and inquiry DOCs ( p � .34).

To test the effects of DOC controlling forevaluation of the deviant, a multiple regressionwas undertaken. Time 1 subjective validity wasincluded in the first step of the regression, sig-nificantly predicting Time 2 subjective validity, � .55, t(69) � 5.48, p � .001. Step 2 of themodel included favorability to the deviant anduse of DOC (coded as 1) versus Non-DOC(coded as 0) as an additional predictor. Favor-ability of the deviant did not predict Time 2validity, � .037, t(67) � .36, p � .71. DOCwas significantly related to increased Time 2validity, � .25, t(67) � 2.46, p � .017. An

interaction term for DOC and favorability wascalculated and included in the third step of themodel, but was nonsignificant, � .25, t(66) �7.5, p � .18. Finally, a correlation betweenfavorability to the in-group deviant and DOCrevealed that lower levels of favorability towardthe deviant were related to a higher proportionof DOC messages, r(71) � .26, p � .027.Favorability did not directly correlate withTime 1 or 2 validity, rs(71) � .18, ps � .14.

Meditational Analysis

To test the hypothesis that DOC would bemore prevalent when the normative positionwas perceived as important, meditational anal-ysis of norm importance, use of DOC/ non-Docand Time 2 validity was undertaken (see Figure1). All regressions predicting Time 2 validityincluded Time 1 validity as a covariate. Normimportance significantly predicted Time 2 va-lidity, � .22, t(88) � 2.42, p � .018. Afurther regression showed that norm importancealso predicted use of DOC, � .22,t(89) � 2.16, p � .033. Use of DOC predictedTime 2 validity, controlling for Time 1 validityand norm importance, � .19, t(87) � 2.1, p �.035. Once use of DOC was included in themodel, norm importance still predicted Time 2validity, � .18, t(87) � .199, p � .05. ASobel Z test revealed this change to be marginal(Z � 1.67, p � .09), suggesting partial media-tion.

Discussion

Research suggests that in-group deviantsthreaten the positive social identity of othermembers of the group (Turner et al., 1987). TheSGD model (Marques et al., 2001) predicts thatgroup members who symbolically marginalizedeviants should experience an increase in sub-

Table 1Frequency of Different Message Types According to Message Target

Message recipient

Message type

TotalPersuasion Inquiry Non-DOC Did not send

In-group deviant 37 (52%) 16 (23%) 18 (25%) 4 75Out-group normative 6 (30%) 7 (35%) 7 (35%) 2 22Total 43 (47%) 23 (25%) 25 (27%) 6 97

Note. Percentages are with rows, excluding group members who nominated a recipient but did not send a message.

287DIFFERENCE ORIENTED COMMUNICATION

jective validity. However, while previous re-search has examined evaluations of deviants, ithas not directly considered how other memberscommunicate with deviants, and whether thecontent of communication (rather than theamount) has implications for subjective valid-ity. The present study provides important newevidence addressing this gap.

As in previous research (e.g., Abrams, Rut-land, & Cameron, 2003) intergroup differentia-tion was related to negative evaluation of thein-group deviant, consistent with the idea thatparticipants aim to reject out-group attitudes,not just out-group members.

The present research showed that more neg-ative evaluations of the in-group deviant (a pro-cesses hypothesized to be a form of symbolicmarginalization; see Marques et al., 2001) arealso associated with greater use of DOC. Mostimportantly, we found that DOC is linked toincreased subjective in-group validity. This isan important finding for the SGD model as itsupports a key expectation, that communicatingin ways that reinforce differences between de-viant and other members is linked to feelings ofsubjective validity. The findings also increasethe scope of SGD. The SGD model developedin part from research into the “black sheep”effect which shows that derogation of groupmembers who fail to live up to a normativestandard is greater for in-group members thanout-group members (e.g., Marques & Yzerbyt,1988). The present findings suggest that desireto restore and protect subjective validity appliesin situations beyond those involving in theblack sheep effect; for instance, when one groupholds a norm in direct opposition with another.

In line with existing literature into in-groupdeviance, the present research confirms that re-

sponses to in-group deviance are likely to beparticularly pronounced when the violated normis perceived as being important; meditationalanalysis suggested that DOC was more likely tooccur under such conditions, and that DOC wasassociated with increased subjective validity ofthe in-group norm.

The present study also confirms a number ofprevious research findings. Previous researchhas shown that group members are highly mo-tivated to interact with other in-group members(e.g., Binder et al., 2009) and highly motivatedto persuade deviants (Frings et al., 2010). Thecurrent research supports these findings, 75% ofparticipants chose to talk to the in-group devi-ant, and of these, 75% engaged in DOC. Thisextends previous research suggesting in-groupdeviants are the likely to be the target of at-tempts to restore validity, and suggests thatDOC is one mechanism by which this can oc-cur.

Several limitations to the present study areacknowledged, and these suggest avenues forfurther empirical investigation. A key predic-tion of the SGD model is that in-group devi-ance, if left unchecked, should negatively affectsubjective validity, and reduce the potential forpositive differentiation between the in-groupand out-group. The present study does not di-rectly address this issue—rather it tests whethercommunicatively highlighting in-group devi-ance is linked to increased levels of subjectivevalidity. Future research should aim to measurelevels of subjective validity before deviants areencountered, and then again subsequently be-fore and after symbolic marginalization occurs.

The present study uses a correlational design.Future research could experimentally manipu-late whether it was possible to communicate

β = .18* (β = .22*),

β = .22*

Importance of

norm violated

Time 2 subjective

validity

Use of DOC /

non-DOC

β = .17*

Figure 1. The meditational role of difference orientation communication in the relationshipbetween issue importance and Time 2 subjective validity. Time 1 validity is included as acovariate. in parentheses indicates the nonmediated relationship between importance andsubjective validity. � p � .05.

288 FRINGS AND ABRAMS

with the deviant or not, or manipulate whethercommunications were DOC or non-DOC, forinstance by instructing participants to commu-nicate about the discussion topic, or a separate,neutral topic. Despite this limitation, the statis-tical control of initial levels of subjective valid-ity via ANCOVA does suggest that it is DOC,and not simply initial levels of subjective valid-ity, that accounts for variance in post commu-nication validity in the present study.

Although the present research provides evi-dence that DOC does increase subjective valid-ity, it would be fruitful to investigate possibleaffective processes underlying the relationship.As DOC actively supports a normative positionit may trigger feelings of self- righteousness.Self-righteousness is thought to be a form ofidentity defense, as it transfers blame (seeHorowitz, 1981). The expression that one’s own(and by extension, the group’s) attitude/behavior is morally superior to that of the de-viant’s (and by extension any out-groups whoshare the deviant behavior) should, in parallelwith other processes, increase positive relativesubjective validity.

Furthermore, DOC may be effective in re-storing subjective validity by establishing attri-butions that account for why a deviant may holdor espouse an “incorrect” attitude. For example,people could make a dispositional attribution(to personality or lack of understanding) thatcould explain away the deviant attitude withoutchallenging the group’s validity. This might beparticularly likely if the deviant does not changefollowing initial persuasion attempts (seeOstrom, 1990, for a discussion). The presentstudy did not inform group members of theeffect their communication had upon the devi-ant. Future studies could measure the effect of adeviant’s behavioral change (or lack thereof)upon changes in subjective validity, attributionsand levels of derogation. It is likely that failureto change would lead to dispositional attribu-tions and increased derogation. If, on the otherhand, as the present paper suggests, the purposeof DOC is simply to highlight difference,changes in subjective validity may not alter as aresult of changes in the success of the commu-nication.

A further possibility is that the DOC is drivenby a sense of cognitive dissonance aroused byfeeling toward the deviants versus contrasting

feelings toward the deviant in-group (see Fest-inger, 1957; Hieder, 1958). A response to re-duce this dissonance could be to reinforce thebelief that the group maintains the same view asthe self. This new belief should lead to anincreased sense of subjective validity.

A final possibility, arising from recent re-search into the interaction between variousforms of identity defense, is that DOC is simplyone means by which people assert validity ofin-group norms or defend identity (see Eidel-man & Biernat, 2003; Eidelman, Silvia, & Bier-nat, 2006). This opens the possibility that othertypes of response could provide the same ben-efits, or indeed that if subjective validity isalready well defended in other ways, therewould be no need for DOC. In the present study,it is interesting that although evaluation of thedeviant appears to motivate DOC, it is DOCitself that predicts subjective validity. In futureresearch it will be useful to see whether evalu-ation or other responses predict subjective va-lidity when communication is prevented, andwhether encouraging DOC can increase subjec-tive validity regardless of other responses to thedeviant. When and how various forms of iden-tity defense are activated remains open ques-tion, although some research (e.g., Ostrom,1990) suggests that inclusive strategies may ini-tially be preferred to exclusive ones.

The present study used an online paradigmto investigate the effects of symbolic margin-alization. Online communication may besomewhat more distant and task focused thanface-to-face interaction, although because ofdeindividuation some group members may bemore expressive online than face-to-face (e.g.,Social Identity Model of DeindividuationEffects; Reicher, Spears, & Postmes, 1995; alsoBargh, McKenna, & Fitzsimmons, 2002). How-ever the now pervasive use of social profilewebsites containing interactions highly similarto those sent in our study (e.g., MySpace,Facebook, Instant messenger), blurs the bound-aries between online and offline communica-tion, particularly among the young populationsampled. While the content and effectiveness ofonline DOC may differ slightly from face-to-face communication, it seems probable that thesubsequent effects on subjective validity willremain similar.

In conclusion, the present research supportsthe idea that communications that highlight dif-

289DIFFERENCE ORIENTED COMMUNICATION

ferences with in-group deviants can function asa form of symbolic marginalization which in-crease relative in-group validity. In many con-texts deviants cannot be excluded or ignored,and rules of the situation (e.g., committee meet-ings) may preclude being rude to, or about, thedeviant. In such cases it seems likely thatgroups reassert the validity of their consensusby highlighting differences with the deviant.Thus, in the face of deviant members, differenceoriented communication may be an effectivestrategy by which groups can both allow differ-ences to exist but also sustain their norms.

References

Abrams, D., Marques, J. M., Bown, N., & Henson,M. (2000). Pro-norm and anti-norm deviancewithin and between groups. Journal of Personalityand Social Psychology, 78, 906–912.

Abrams, D., Rutland, A., & Cameron, L. (2003). Thedevelopment of subjective group dynamics: Chil-dren’s judgments of normative and deviant in-group and out-group individuals. Child Develop-ment, 74, 1840–1856.

Bargh, J. A., McKenna, K. Y. A., & Fitzsimons,G. M. (2002). Can you see the real me? activationand expression of the “True Self” on the Internet.Journal of Social Issues, 58, 33–48.

Binder, J., Zagefka, H., Brown, R., Funke, F.,Kessler, T., Mummendey, A., . . . Leyens, J. P.(2009). Does contact reduce prejudice or does prej-udice reduce contact? A longitudinal test of thecontact hypothesis amongst majority and minoritygroups in three European countries. Journal ofPersonality and Social Psychology, 96, 843–856.

Chaiken, S., Liberman, A., & Eagly, A. H. (1989).Heuristic and systematic information processingwithin and beyond the persuasion context. In J. S.Uleman & J. A. Bargh. (Eds.), Unintended thought(pp. 212–252). New York: Guilford Press.

Eidelman, S., & Biernat, M. (2003). Derogatingblack sheep: Individual or group protection? Jour-nal of Experimental Social Psychology, 39, 602–609.

Eidelman, S., Silvia, P. J., & Biernat, M. (2006).Responding to deviance: Target exclusion and dif-ferential devaluation. Personality and Social Psy-chology Bulletin, 32, 1153–1164.

Festinger, L. (1950). Informational social communi-cation, Psychological Review, 57, 271–282.

Festinger, L. (1954). A theory of social comparisonprocesses. Human Relations, 7, 117–140.

Festinger, L. (1957). A theory of cognitive disso-nance. New York: Harper and Row.

Frings, D., Abrams, D., Randsley de Moura, G., &Marques, J. (2010). The effects of cost, normativesupport and issue importance on motivation topersuade in-group deviants. Group Dynamics:Theory, Research, and Practice, 14, 80–91.

Gabriel, Y. (1998). An introduction to the socialpsychology of insults in organizations. Human Re-lations, 51, 1329–1354.

Heider, F. (1958). The psychology of interpersonalrelations. New York: Wiley.

Horowitz, M. J. (1981). Self-righteous rage and theattribution of blame. Archives of General Psychi-atry, 38, 1233–1238.

Jetten, J., Spears, R., & Postmes, T. (2004). Inter-group distinctiveness and differentiation: A meta-analytical investigation. Journal of Personalityand Social Psychology, 86, 862–879.

Marques, J., Abrams, D., Paez, D., & MartinezTaboada, C. (1998). The role of categorization andin-group norms in judgments of groups and theirmembers. Journal of Personality and Social Psy-chology, 75, 976–988.

Marques, J., Abrams, D., & Serodio, R. G. (2001).Being better by being right: Subjective group dy-namics and derogation of in-group deviants whengeneric norms are undermined. Journal of Person-ality and Social Psychology, 81, 436–447.

Marques, J. M., & Yzerbyt, V. Y. (1988). The blacksheep effect: Judgmental extremity towards in-group members in inter- and intra-group situations.European Journal of Social Psychology, 18, 287–292.

Marques, J. M., Yzerbyt, V. Y., & Leyens, J. P.(1988). The “Black Sheep Effect”: Extremity ofjudgments towards in-group members as a func-tion of group identification. European Journal ofSocial Psychology, 18, 1–16.

Mullen, B., Calogero, R. M., & Leader, T. I. (2007).A social psychological study of ethnonyms: Cog-nitive representation of in-group and intergrouphostility. Journal of Personality and Social Psy-chology, 92, 612–630.

Ostrom, E. (1990). Governing the commons: Theevolution of institutions for collective action. Cam-bridge: Cambridge University Press.

Ouwerkerk, J. W., Kerr, N. L., Gallucci, M., & VanLange, P. A. M. (2005). Avoiding the social deathpenalty: Ostracism and cooperation in social di-lemmas. In K. D. Williams, J. P. Forgas, & W. vonHippel (Eds.). The social outcast: Ostracism, so-cial exclusion, rejection, and bullying (pp. 321–332). New York: Psychology Press.

Randsley de Moura, G. (2003). Subjective groupdynamics: Leadership and prototypicality. PhDThesis, Kent University.

Reicher, S., Spears, R., & Postmes, T. (1995). Asocial identity model of deindividuation phenom-

290 FRINGS AND ABRAMS

ena. European Review of Social Psychology, 6,161–197.

Schachter, S. (1951). Deviation, rejection and com-munication, Journal of Abnormal and Social Psy-chology, 41, 190–207.

Tabachnick, B. G., & Fidell, L. S. (2001). Usingmultivariate statistics (4th ed.). New York: Allyn& Bacon.

Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher,S. D., & Wetherell, M. S. (1987). Rediscoveringthe social group: A self-categorization theory. Ox-ford: Basil Blackwell.

Received August 17, 2009Revision received December 19, 2009

Accepted January 13, 2010 �

291DIFFERENCE ORIENTED COMMUNICATION