ED 064 659

AUTHORTITLEINSTITUTICN

SPONS AGENCYREPORT NOBUREAU NOPUB DATECONTRACTNOTE

EDRS PRICEDESCRIPTORS

ABSTRACT

DOCUMENT RESUME

24 CG 007 508

Levine, John M.; Allen, Vernon L.Reactions to Attitudinal Deviancy.Wisconsin Univ., Madison. Research and DevelopmentCenter for Cognitive Learning.Office of Education (DHEW), Washington, D.C.TP-14BR-5-0216Nov 68OEC-5-10-15421p.

MF-$0.65 HC-$3.29*Attitudes; Behavior Patterns; *Group Dynamics;*Group Norms; *Individual Differences; *PsychologicalPatterns; Reactive Behavior; Rejection

This paper presents a critical review of empiricaland theoretical treatmente of group reaction to attitudinal deviancy.Inspired by Festingerls (1950) ideas on resolution of attitudinaldiscrepancies in groups, Schachter (1951) conducted an experimentthat has greatly influenced subsequent research and theory concerningreaction to attitudinal deviancy. Although Schachter's findings orsociometric rejection of the deviate have been frequently replicated,his data on communication to the deviate are partially inconsistentwith theory and other research. Additional evidence is presentedwhich suggests that individuals who deviate in nonattitudinal spheresare also subject to negative group sanction. Rejection may beemployed against persons who impede group goals, transgress groupnorms, or differ in personality from other group members. Finally,data are presented which indicate that deviates are not alwaysrejected; rather, deviation in some settings and by some individualsmay be tolerated or even encouraged. (Author)

REACTIONS TO ATTITUDINLDEVIANCY

a

U.S. DEP0111III41OF SEP.t.114.

IOUCA-110e1& VIIII.F

OE

OFFICEOF ouciolos

Itits DOCUMENT*

14P.SSEEN

REPRO

CODEDEXP.CILI

ASovE0ac

FROM

THE PERSONOR ORGPNOMON

WO -

04P.IINGit PONIS

OF NOWOR OIAN

IONSSI Alf.0

DO NO1NECESSARN.1

REPRESENT

OFOCANt.000. OF EDO.

CATONPOSiDON

OR POOC4

R.FE SEAR CH

ER FOR

ITIVE LEARNING

N D DFV F L N1FNT

let

E A

COGNITIVEo LEARNING

Ict-e o P

CLILr1

Theoretical Paper No. 14CD

REACTIONS TO ATTITUDINAL DEVIANCYuJ

John M. Levine and Vernon L. Allen

Report from the Peer Group Pressures on Learning ProjectVernon L. Allen, Principal Investigator

Wisconsin Research and DevelopmentCenter for Cognitive LearningThe University of Wisconsin

Madison, Wisconsin

November 1968

The research reported herein was performed pursuant to a contract with the United States Office ofEducation, Department of Health, Educatior., and Welfare, under the provisions of the CooperativeResearch Program.

Center No. C-03 / Contract OE 5-10-154

PREFACE

Contributing to an understanding of children's cognitive learning andimproving related educational practices is the goal of the Wisconsin R & DCenter. One of the Center's three major research and development pro-gramsConditions and Processes of Learningconsists of laboratory-typeresearch projects, each concentrating on certain basic organismic or situa-tional determinants of cognitive learning, but all united in the task of pro-viding knowledge which can be utilized in the construction of instructionalsystems.

Any complete study of the variables which influence human learningwhether in or out of the classroommust ultimately consider social influ-ences. Professor Allen and his associates are engaged in a research projectdirected toward the analysis of social determinants in the acquisition andretention of basic cognitive skills. In this Theoretical Paper, Mr. Levineand Professor Allen provide a critical review of the literature on groupreaction to attitudinal deviancy.

Herbert J. KlausmeierDirector

iii

CONTENTS

Abstract

I. Introduction

page

vii

II. Negative Reaction to Attitudinal Deviancy 2

Early Research 2

Schachter's and Related Contributions 3

Recent ExperimentsSummary 6

Additional Evidence 7

III. Negative Reaction to Nonattitudinal Deviancy 8

IV. Group Size and Mediation of Rejection 1 1

V. Alternative Reactions to Deviancy I 3

References 1 5

I

ABSTRACT

This paper presents a critical review of empirical and theoretical treat-ments of group reaction to attitudinal deviancy. Inspired by Festinger's(1950) ideas on resolution of attitudinal discrepancies in groups, Schachter(1951) conducted an experiment that has greatly influenced subsequentresearch and theory concerning reaction to attitudinal deviancy. AlthoughSchachter's findings on sociometric rejection of the deviate have beenfrequently replicated, his data on communication to the deviate are partiallyinconsistent with theory and other research. Additional evidence is presentedwhich suggests that individuals who deviate in nonattitudinal spheres are alsosubject to negative group sanction. Rejection may be employed against per-sons who impede group goals , transgress group norms, or differ in personalityfrom other group members . Finally, data are presented which indicate thatdeviates are not always rejected; rather, deviation in some settings and bysome individuals may be tolerated or even encouraged.

r.,vi i

I

INTRODUCTION

While reaction to attitudinal deviancy inthe small group setting has received relativelylittle systematic empirical investigation, thetheoretical attention devoted to this problemhas been even more limited. This paucity oftheoretical consideration is somewhat surpris-ing in view of the centrality of the devianceproblem to the basic relationship between theindividual and the group. Perhaps, most in-vestigators consider the problem solved, con-cluding, as do Kelley and Thibaut (1954, p. 7614,

It is common knowledge that when amember deviates markedly from a group3tandard, the remaining members of thegroup bring pressures to bear on thedeviate to return to conformity. If pres-sure is of no avail, the deviate is re-jected and cast out of the group. Theresearch on this point is consistentwith common sense.

However, is group reaction really this sim-ple and invariant? To answer this question,we shall attempt to review and integrate dataand theory relevant to the problem of groupreaction to attitudinal deviancy. We believethat careful consideration of the data will indi-cate that Kelley and Thibaut' s (1954) statementvastly oversimplifies the range and complexityof treatment accorded an individual who deviatesfrom group consensus.

Leon Festinger and his students in the early1950's provided the initial impetus for study ofgroup reaction to deviance; unfortunately, littlenew theoretical clarification has emerged since.In an influential paper, Festinger (1950) assertedthat pressures toward grovp uniformity arise

from two sources: (1) social reality, groupmembers seek consensus to validate opinionsnot anchored in physical reality; and (2) grouplocomotion, uniformity is sought to enableattainment of group goals. Uniformity prns-sures (from either source) produce communi-cations oriented to reducing discrepanciesamong group members. Such communicationcan resolve attitudinal discrepancies in threeways: (1) the deviate may change his opiniontoward the group, (2) the group may change itsmodal opinion toward the deviate, or (3) thegroup may redefine its boundaries by rejectingthe deviate.

Festinger offered several hypotheses con-cerning variables which affect the relationshipbetween communication and uniformity. Regard-ing the magnitude of pressures to communicate,Festinger hypothesized that perceived discrep-ancy among group members , relevance of theissue, and group cohesiveness are all posi-tively related to communication pressure.Variables which determine communication to aparticular individual include perceived opiniondiscrepancy, perception of individual as agroup member and/or desire that individual bea group member, and perception that communi-cation will alter deviate's opinion. Factorspositively related to opinion change in a com-munication recipient include group pressurestoward uniformity, forces causing the indi-vidual to remain in the group, and lack ofissue anchorage in other group memberships.Finally, perceived discrepancy, relevance ofthe issue, and group cohesiveness are allpositively related to rejection of the deviateby the group.

13

1

NEGATIVE REACTION TO ATTITUDINAL DEVIANCY

EARLY RESEARCH

Several studies were conducted by restingerand his colleagues to investigate relationshipssuggested above. restinger, Schachter, andBack (1960) interviewed M.I.T. students, liv-ing in married-student housing developments,on attitudes toward a tenants' organization.Data indicated that cohesiveness (% of in-group sociometric choices) was negatively re-lated to the percentage of individuals whodeviated from modal group opinion on the ten-ents' organization. Thus, the greater theattractiveness (cohesiveness) of the group,the greater conformity (less deviation) to groupnorms. Regarding reaction to deviation, theinvestigators found that deviates received sig-nificantly fewer sociometric choices from groupmembers than did conformers. It appears, then,that deviates are rejected by group members.However, the correlational nature of the datamake causal inferences impossible; we do notknow whether deviation produced rejection orwheLher initial rejection freed individuals todeviate.

To resolve the direction of causality in thisrelatiqnship, we need either correlational datawith clear temporal sequencing or experimentaldata with one of the two variables independentlymanipulated. Two studies fulfill the former re-quirement. restinger and Thibaut (1951) hy-pothesized that (1) communication will bedirected primarily to extreme deviates: (2) ifthe group can subdivide or redefine boundaries ,communication to deviates will decrease overtime; and (3) greater pressure toward uniformityand/or lower possibility of group subdivisionwill produce more change in group members.In this experiment, groups of 6 to 14 Ss usednotes to discuss one of two issues, treatmentof a juvenile delinquent or football strategy.It was hypothesized that opinions on the formertopic would be more resistant to change thanopinions on the latter. Instructions were used

2

to manipulate pressure toward uniformity (High,Medium, Low) and group composition (Homo-geneous, Heterogeneous).

Results indicated that on both topics communi-cation varied directly with the extremeness ofthe recipient's opinion, i.e., extreme deviatesreceived the most communication. Moreover,communication to deviates varied directly withpressure to uniformity. Regarding changes incommunication to deviates over time, resultsdiffered in the Homogeneous and Heterogeneousconditions. In the Homogeneous conditionamount of communication to deviates did notchange significantly over time in any of thethree pressure conditions. In the Heterogeneouscondition, on the other hand, communication todeviates decreased significantly over time inthe Medium and Low pressure conditions, butnot in the High pressure condition. Thus, whenpressure to uniformity is high, communicationto deviates remains high, regardless of groupcomposition. However, when pressure to uni-formity is low and opportunity for subgroup for-mation exists, communication to deviates de-creases over time. Data concerning change inmembers' opinions are partially consistent withpredictions: both high pressure to uniformityand group homogeneity produce relatively highchange toward group uniformity. However,these results hold primarily for the footballstrategy issue; Ss' opinions on treatrrqnt of ajuvenile delinquent are highly resistant to groupinfluence under all conditions.

In a related experiment, Gerard (1953) soughtto test the effects of group composition end pres-sure to uniformity upon communication, usingtwo kinds of group disagreement. Subjects werebrought together in groups of 8 to 14 and askedto (1) indicate which of two issues relating tofederal aid to education was the more importantand (2) express their opinion on each issueusing seven-point scales. Instructions wereused to manipulate group composition (Homo-geneous, Heterogeneous) and pressure to

uniformity (High, Low). As in the previousexperiment, Ss used notes to communicatewith one another. Subjects were dichotomizedinto majority and minority members, dependingon the issue chosen as more important.

Results were partially consistent with thoseof restinger and Thibaut (1951), i.e., morecommunication was directed to extreme deviatesthan to conformers , and change toward groupuniformity, though relatively slight, occurredprimadly in the Homogeneous-High pressurecondition. However, Gerard found that com-munication to deviates decreased over timeonly in the Homogeneous condition; this resultis opposite to that reported by Festinger andThibaut. Gerard (1953) explained that sincefew opinion changes occurred in his Homoge-neous condition, Ss gave up trying to influencedeviates and decreased communication to themover time. In the Festinger and Thibaut study,however, opinion changes did occur, and,hence, successful communication did not de-crease. In Festinger and Thibaut's Hetero-geneous condition, uniformity pressures aroseand deviates were only rejected after opiniondiscrepancies became clear during discussion;thus, communication to deviates decreasedover time. On the other hand, Ss in Gerard'sHeterogeneous condition began rejecting devi-ates immediately because tendencies to agreewith perceived group "experts" reduced uni-formity pressures. Interestingly, Jones andGerard (1967) report that Gerard's (1953) resultssupport Festinger and Thibaut (1951): Jones andGerard (p. 351) state that Gerard (1953) foundthat communication "went to less extreme opin-ions only when subgroup formation was possi-ble, that is, in the heterogeneous condition."

Although results of the Festinger and Thibaut(1951) and Gerard (1953) experiments are notp6.fectly parallel, certain consistent and inter-esting relationships emerge. However, onemay still ask whether these data are directlyrelevant to the issue of group reaction to devi-ance. First, since deviance is not directlymanipulated, the number of deviates varies inan unknown fashion across studies and acrossgroups within a study. Although a weightingtechnique was employed in both experimentsto take account of the number of individuals ata particular opinion position, this method doesnot solve all problems. In essence, the num-ber of persons in the group who are "deviates"depends completely on S's own position; devia-tion is defined in terms of discrepancy from S'sposition rather than discrepancy from group con-sensus. It would seem to be quite different,psychologically, for S to confront one deviatevs. four deviates in a group of five. Thus, thebasic phenomenon of deviance from group con-

sensus is not directly tested in either the Fes-Unger and Thibaut (1951) or Gerard (1953) study.

Second, the a priori definitions of "communi-cation" (as attempt to influence) and of "decreasein communication" (as rejection) are problemati-cal. We have no independent evidence that Sstry to influence one another in their writtennotes; neither do we know that decreased com-munication indicates rejection. Data such ascontent analysis of notes and sociometric pref-erences would greatly clarify the mechanismsunderlying group reaction to deviance.

SCHACHTER'S AND RELATED CONTRIBUTIONS

Let us now turn to experiments in which devi-ance is directly manipulated as dissent fromgroup consensus end in which less ambiguousmeasures of rejection are utilized. In perhapsthe most influential and misrepresented (Berkowitz,1967) study of group reaction to deviance,Schachter (1951) experimentally manipulateddiscrepancy between deviate and group, rele-vance of issue to group, and group cohesiveness.Subjects were recruited as members of clubsorganized around four different topics. One-half of the Ss were assigned to clubs they foundattractive (High Cohesive), while the remainingone-half were assigned to clubs they found rela-tively unattractive (Low Cohesive). In addition,the topic discussed in both High Cohesive andLow Cohesive conditions Ws either Relevantor Irrelevant to the group purpose. Thus, fourconditions were created: High Co-Relevant,High Co-Irrelevant, Low Co-Relevant, Low Go.Irrelevant. Each group contained five to se.rennaive Ss and three confederates (Mode, Slider,Deviate). The Mode agreed with Ss' modaiposition throughout the discussion. The Widerstarted as an extreme deviate, but graduallycame to agree with the modal position. TheDeviate maintained an extremely unpopularopinion throughout the discussion. The topicof discussion in all conditions was treetment ofa juvenile delinquent; as verbally commk:nicatedon the issue for 45 minutes.

Following the discussion, Ss filled out asociometric questionnaire and nominated groupmembers to committee positions varying in de-gree of attractiveness. Sociometric data indi-cated that in all conditions the Deviate wasrejected significantly more than either the Modeor Slider, who did not differ significantly fromone another. Moreover, while the Deviate wasrejected significantly more in the High Cohesivethan in the Low Cohesive condition, Relevancehad no differential effect on rejection. Regard-ing committee nominations, in three of the fourcoriditions (except Low Co-Irrelevant) the Devi-ate was overnominated for the least desirable

3

committee and undernominated for the mostdesirable committee; the Mode and Slider werenot systematically rejected. In addition, theDeviate was rejected aignificantly more in theRelevant than in the Irrelevant condition. How-ever, it should be noted that the differentialeffect of Relevance on rejection was primarilydue to a startlingly high degree of Deviateacceptance in the Low Co-Irrelevant condition.Finally, Cohesiveness was not significantlyrelated to rejection on the committee assign-ment measure.

Data on communication to the three confed-erates were also obtained; however, sinceSchachter's (1951) presentation of these resultswas somewhat unclear, our discussion will bebased partially on Mills' (1962) and Berkowitz(1967) summaries of Schachter's (1951) data.Regarding overall communication, the Deviatereceived approximately 4 times the number ofcommunications directed to the Slider, and 7times the number directed to the mode (Mills,1962). Thus, amount of communication varieddirectly with w:tremeness of deviancy fromgroup norms. Concerning trends over time,communication to the Mode remained uniform,communication to the Slider decreased, andcommunication to the Deviate increased steadilyin all conditions except High Co-Relevant,where a final decrease occurred. While theseresults generally support the Festinger (1950)and Schachter (1951) hypothesis that communi-cation is used to achieve group uniformity,other communisation data are not consistentwith the hypothesis. Regarding only communi-cation to the Deviate, the greatest amountoccurred in the Low Co-Irrelevant condition,where, theoretically, pressure to uniformityshould be lowest.

Let us review a partial replication of theSchachter study (Emerson, 1954) before dis-cussing the implications of Schachter's results.Except for the use of high school studentsrather than college students as Ss, and theomission of the Relevance manipulation,Emerson's experimental procedures closelyreplicated those of Schachter. Emerson'ssociornetric data showed that the Deviate wasrejected more than the Mode or Slider, who didnot differ from one another. Also, the Deviatewas rejected significantly more in the HighCohesive than in the Law Cohesive condition.On committee nominations, the Deviate wasovernominated for the least attractive commit-tee, but there was no significant differencebetween the High Cohesive and Low Cohesiveconditions on this measure of rejection. Morecommunication was directed to the Deviate inthe Low than in the High Cohesive condition,and communication to the Deviate consistently

4

rose over time with no final decline in any con-dition. No data on communication to the Modeor Slider were presented.

Rejection of the Deviate on all three meas-ures was lower in the replication than inSchachter's (1951) study. Emerson explainedthe relatively low rejection of the Deviate inhis experiment in terms of subject factors.That is, Emerson's Ss were younger than thoseof Schachter and, hence, tended to change theirown opinions rather than the Deviate's in orderto achieve uniformity. Emerson presented dataindicating that significantly more Ss changedtoward the Deviate in his study than in Schachter'sexperiment.

Several questions can be raised about theconclusiveness of Schachter's (1951) andEmerson's (1954) results on group reaction todeviance. First, communication data in bothexperiments inuicate that most, communicationis directed to the Deviate in conditions wheretheory predicts least pressure to uniformity.Thus, the validity of communication as an indexof "pressure to uniformity" is questionable.Moreover, since neither Schachter nor Emersonpresented content analyses of communications,we have no direct evidence that Ss try to influ-ence one another at all. This ommission seemsstrange since Schachter (1951) reported, in hismethods section, collecting data on the impliedapproval and disapproval contained in communi-cations. The questionable validity of communi-cation as an index of pressure to uniformity alsocasts doubt on the use of decreased communica-tion as an index of rejection.

We submit, therefore, that the communicationdata presented by Schachter and Emerson are,like earlier data of this type, inadequate to testhypotheses concerning pressure to uniformityand rejection of deviates. Fortunately, however,postexperimental evaluations of deviates pro-vid ed by Schachter (1951) and Emerson (1954)can be taken as fairly reliable indices of groupreaction to deviance.

Mills (1962) also argues that Schachter's(1951) communication data may not reflect at-tempts to achieve group uniformity, but ratherindicate aggression displaced from a powerfuland frustrating E to a helpless deviate. Millsasserts that Schachter angered Ss by violatingnorms of the experimenter-subject role relation-ship. That is, Schachter frustrated and angeredas by placing them in unattractive groups andimposing irrelevant tasks. By this reasoning,the runk order of conditions, in terms of frus-tration, is (from high to low): Low Co-Irrelevant,Low Co-Relevant or High Co-Irrelevant, andHigh Co-Relevant. Mills argues that, becauseIs cannot aggress against E, egression is dis-placed to the deviate. Communication, then,

should be rank ordered in the same manner asfrustration; the data approximate this relation-ship.

Mills' analysis , while interesting, is opento attack on two points. First, we have noevidence that communication to the deviateexpressed hostility displaced from E.; as men-tioned before, no data on communication con-tent were presented. Second, Schachter's(1951) results show that the Deviate receivedleast rejection (on the committee nominationmeasure) in the Low Co-Irrelevant condition,where Mills would have to predict greatestrejection.

Aderman (1968) has speculated regardingthe dramatic acceptance of the Deviate in theLow Co-Irrelevant condition on the committeenomination measure. Aderman suggests thatin the Low Co-Irrelevant condition acceptanceis positively related to an individual's com-munications to others , regardless of content.Thus, if we can assume that the Deviate spokemore in the Low Co-Irrelevant condition thanin any other condition, rejection should belowest in the Low Co-Irrelevant condition, asfound. However, the Deviate was strongly,rejected in the Low Co-Irrelevant condition onthe sociometric index. Unfortunately, sinceSchachter presented no data regarding volumeof communication initiated by any of the threeconfederates, Aderman's hypothesis cannot betested with existing data. Emerson (1954),however, provides evidence both supportiveand nonsupportive of the hypothesis. On theone hand, group members who received evenmore sociometric rejection than the Deviatecontributed least to group discussion; on theother hand, the Deviate, who received morerejection than the Mode or Slider, talked morethan any other single member. Clearly, judg-ment must be withheld on the communicationvolume hypothesis until adequate empiricalevidence is available.

RECENT EXPERIMENTS

Approximately a decade after the initialflurry or work on reaction to deviancy, Sampsonand Brandon (1964) sought to investigate groupreaction to opinion vs. role deviancy. Theseinvestigators hypothesized that both opiniondiscrepancies and role expectation discrepan-cies produce increased communication to theDeviate and/or redefinition of group boundariesto exclude the Deviate. It was suggested thatcommunication is employed when the deviationis specific and potentially modifiable and whenexternal demands for continued interaction arepresent.

Female Ss were brought together in groupsD$. tive to discuss (under no external uniformitypressure) treatment of a Negro juvenile delin-quent. A confederate initially presented herselfas either bigoted (Role Deviate) or raciallyliberal (Role Conformant). During the verbaldiscussion of the delinquent's treatment, theconfederate either agreed with group consensus(Opinion Conformant) or strongly and consistentlydisagreed (Opinion Deviate). Four experimentalconditions were thus ei-Pated: Role Deviate-Opinion Conformant, Role Deviate-OpinionDeviate, Role Conformant-Opinion Conformant,and Role Conformant-Opinion Deviate.

Results showed that Ss reacted very differ-ently to Role and Opinion Deviates. Communi-cation data indicated that the Opinion Deviate,as compared to the Opinion Conforinant, receivedsignificantly more overall communication, ex-pressions of hostility, and requests for informa-tion, and significantly fewer expressions ofsolidarity. In contrast, the Role Conformant,as compared to the Role Deviate, received sig-nificantly more overall communications, expres-sions of hostilitY, and information. The investi-gators interpret their results as indicating thatSs attempt to change the Opinion Deviate butwithdraw completely from the Role Deviate.Data on perception and evaluation of the Deviateindicated that (1) Ss saw themselves as signifi-cantly more similar to the Role Conformant thanthe Role Deviate, and (2) Ss expressed signifi-cantly more liking for the Role Conformant thanthe Role Deviate. The investigators suggestedthat, if E had applied external pressures to uni-formity, perhaps the Opinion Deviate wouldhave been liked significantly less than theOpinion Conformant. Indeed, on another socio-metric question the Opinion Deviate was likedsignificantly less than the Opinion Conformant.

The communication data in this experimentinteresting for several reasons. First,

Sampson and Brandon present the most adequatedata yet encountered,. Second, these dataclearly demonstrate that communication is nota unitary phenomenon; various communicationindices do not invariably correlate highly withone another or with overall communication.Third, several communication indices (e.g. ,expressions of solidarity and hostility, requestsfor information) do possess face validity asmeasures of pressure to uniformity. Finally,for the first time, extremeness of deviation isinversely related to amount of overall communi-cation directed toward the deviate, i.e. , theRole Deviate received relatively less communi-cation than did the Opinion Deviate.

Our interpretation of role deviancy as moreextreme than opinion deviancy requires someexplanation. We feel that Sampson and Brandon's

0

use of the term "role" to describe the confed-erate's bigotry or liberalism is questionable:it seems that extremeness of confederate'sopi:lion was actually manipulated here. (Inter-

Ss' perceptions of the confederate'sa .;:wie were used to validate the Role manipu-lation.) It appears reasonable that a bigotedconfederate would appear both more extremeand less amenable to change than a confederatewho merely deviates from group consensus ontreatment of a Negro juvenile delinquent. Thusthe Role Deviate is a more extreme opiniondeviate than the Opinion Deviate. Hence, wewould predict, as did Sampson and Brandon,more rejection of the Role Deviate than theOpinion Deviate. Sociometric data supportedthis prediction. However, overall communica-tion was inversely related to extremeness ofdeviancy, rather than directly related as previ-ous investigators have reported.

Sampson and Brandon assert that the RoleDeviate was so completely rejected that nopressures were exerted on her. Perhaps , asFestinger (1950) suggested, the relationshipof communication to deviate's extremenessdepends on Ss' perceptions of the deviate'swillingness to alter his opinion. If so, com-munication may be positively related to ex-tremeness when the deviate is malleable andinversely related when the deviate is rigid.Perhaps, then, only in the Sampson and Brandonexperiment was the extreme opinion deviate,i.e. Role Deviate, perceived as totally uname-nable to persuasion.

Katz, Libby, and Strodtbeck (1964) hypothe-sized that deviance may not be punished whenlack of devianc.e (conformity) threatens theexisting group status hierarchy. That is, sincehigher status group members are threatened bythe conformity of low status individuals whoare upwardly mobile, donformity may be pun-ished more than continued deviancy. In thisstudy, Is were female sales trainees who dis-cussed a tape recorded salespersoncustomerinteraction. Two confederates (Deviate andSlider) deviated from group consensus in amanner similar to Schachter's (1951) confed-erates. Subjects were toll that group statuspositions were either permanent throughout"group training" sessions (Stable status) orwould probably soon change (Mobile status).Results indicated that both the Deviate andSlider were ranked significantly lower on anindex of liking than were conforming groupmembers. Moreover, the Deviate was rejectedmore than the Slider, regardless of group statusmobtlity. Thus, the positive relationship be-tween extremeness of deviance and rejectionoverrides influence of other variables such asstatus threat.

6

Streufert (1965) investigated the influence ofa communicator's "importance" on the evaluationof his conformity or deviance. "Importance" washypothesized to vary inversely with interactiondistance (a composite of spatial closeness andtemporal length of interaction). Subjects listenedto two tape recorded communicators, one whoagreed with Ss' opinions on rock-and-roll music(Conformant) and one who disagreed (Deviate).Then, Ss were asked to rate the two communi-cators in three hypothetical situations varyingin communicator-S interaction distance. Resultsshowed that Ss rated the Deviate significantlyless favorably than the Conformant at all threeinteraction distances. Favorableness of ratingstoward the Conformant varied inversely with in-teraction distance, i.e. , the more important theConformant, the more Ss liked him. However,favorableness of ratings toward the Deviatevaried directly with interaction distance, i.e. ,the more important the Deviate, the less Ss likedhim. Streufert hypothesized that the Deviate'simportance was directly related to the amountof dissonance that he produced; this dissonance,in turn, was directly related to rejection.

SUMMARY

Let us at this point attempt to summarize themore important findings reported so far. Theamount of communication directed to a givenindividue generally varies directly with extreme-ness of the individual's opinion vis-a-vis thecommunicator and/or the group (Festinger andThibaut, 1951; Gerard, 1953; Schachter, 1951).In addition, Festinger and Thibaut (1951) reportedthat increased pressure to uniformity producessignificantly more communication to deviates.These findings are consistent with Festinger's(1950) hypothesis that communication is em-ployed by group members in order to achievegroup uniformity. However, conflicting evi-dence also appears. Both Schachter (1951) andEmerson (1954) found that the deviate receivesthe greatest amount of communication in theconditions where uniformity pressurer. are low-est (low group cohesion; irrelevant issue).Moreover, Sarapson and Brandon (1964) showedthat communication varies inversely with ex-tremeness of the deviate's position.

Decrease in communication to deviates overtime (often considered rejection) occurs (1) whenuniformity pressures are relatively low and sub-group formation is possible (Festinger andThibaut, 1951), (2) when subgroup formation isnat possible (Gerard, 1953), and (3) when ahighly cohesive group discusses a relevant issue(Schachter, 1951). Note that Sampson andBrandon's (1964) communication data may alsobe relevant to rejection of deviates.

Several studies report that group membersreject opinion deviates in sociometric ratingsand/or committee nominations (Schachter, 1951;Emerson, 1954; Sampson and Brandon, 1964;Katz, Libby, and Strodtbeck, 1964; Streufert,1965). Rejection generally varies directly withextremeness of the deviate's opinion (Schachter,1951; Emerson, 1954; Sampson and Brandon,1964; Katz, Libby, and Strodtbeck , 1964).There is also some evidence that rejection variesdirectly with both group pohesiveness (Schachter1951; Emerson, 1954) and issue relevance(Schachter, 1951). Finally, Streufert (1965)found that rejection varies directly with intimacyof hypothetical deviate-subject interaction.

Regarding group change toward uniformity,Festinger and Thibaut (1951) and Gerard (1953)found the greatest changes in homogeneousgroups under high uniformity pressures. Astudy by Back (1951), not directly pertinent toreaction to deviancy, reported more change inhigh cohesive than in low cohesive dyads.However, Emerson (1954) presented data indi-cating that in neither his nor Schachter's (1951)study did variation in group cohesiveness sig-nificantly affect the percentage of as changingtoward the deviate's position. Emerson didshow, however, that significantly more Sschanged toward the deviate in his replicationthan in Schachter's original experiment; this

difference was attributed to variations in Ss'ages in the two studies.

ADDITIONAL EVIDENCE

The studies reviewed above, while dealingwith both modification of the deviate and groupchange as possitle resolutions of attitudinaldiscrepancy, clearly stress rejection of thedeviate. Does evidence from other research

, areas substantiate the hypothesis that rejectionof the deviate is a ubiquitous reaction to groupdisagreement? Two conformity experiments pro-vide data that deviates are often rejected bygroup members. Allen (1965) manipulated devi-ance by having one group member deviate fromgroup consensus on 12 ota of 20 items dealingwith personality traits , famous persons , anddrawn figures. Questionnaire responses indi-cated that approximately one-fourth of the Ssreacted "very unfavorably" to the deviate. Allenand Levine (19613b) asked Ss in a conformityexperiment to rate both a Social Supporter (whoagreed with a but disagreed with the erroneousgroup) and an Extreme Dissenter (who answeredeven more incorrectly than the group). Resultsshowed that Ss rated the Extreme Dissenter sig-nificantly lower than the Social Supporter on fourevaluative scales (Likeableness, Intelligence,Sincerity, and Adjustment).

dr 9

7

III

NEGATIVE REACTION TO NONATTITUDINAL DEVIANCY

Several additional studies, while not deal-ing directly with attitudinal "deviance, are rele-vant to the general notion that groups rejectdeviates. An attempt was made by Schachter,Nut lin, DeMencheux, Maucorps, Owner,Duijker, Rommetveit, and Israel (1954) to in-vestigate the relationship between threat to agroup goal and rejection of a deviate in a cross-cultural setting. Threat was defined as varyingdirectly with valence of the goal and inverselywith probability that the goal would be achieved.Sublects were school boys , living in Holland,Sweden, France, Norway, Belgium, Germany,or England, who volunteered to participate inan aviation club. The As were brought togetherin groups of six or seven and told that thegroups would compete on a model-airplanebuilding task.

Valence of goal was manipulated by telling_Ss that the competition would determine eithercontinuation of the club (High Valence) or re-ceipt of tickets to a dull movie (Low Valence).Probability of attaining the goal was definedas either High or Low. Thus, four conditionswere created: High Valence-High Probability,High Valence-Low Probability, Low Valence-High Probability, and Low Valence-Low Proba-bility. During discussion of which model tobuild for competition, a confederate deviatedfrom group consensus by choosing a dull glider.Since As were instructed to decide on a model,the Deviate clearly retarded group locomotionrather than merely confusing social reality.

Unfortunately, results are difficult to inter-p.:et because of the confusing manner in whichthey were presented. Three experiments (Eng-lish, Belgian, and German) will not be dis-cussed here since adequate perceptions of theexperimental situation were not created. Forthe remaining countries, sociometric data indi-cated that As rejected the Deviate significantlymore in the Low Probability than in the HighProbability condition; the Valence manipulationhad no significant effect on rejection. Generally,

8

greater rejection occurnd in the High Valence-Low NV:tab:My than in the High Valence-HighProbability condition. Data on the second re-jection measure (nomination for club president)indicated that neither the Valence nor the Proba-bility manipulation produced consistent differ-ences. Although the investigators stete that"in all countries and in all conditions the devi-ate is considered relatively undesirable as eithera working partner or a club president," no sig-nificance tests are offered to substantiate thisstatement.

The communication data are interesting pri-marily because an attempt was made to includeonly "imperative-minority," "aggressive," ant!"threatening" remarks to the deviate. In essr.ce,the communication index is the proportion of allrelevant comments initiated by Ss which are in-tense communications to the Deviate. Thisindex is certainly far superior to the gross com-munication measures used in some of the previ-ously discussed studies. Unfortunately, noteven the improved communication index canclarify results of this experiment. Significantlymore communication was directed to the Deviatein the High Valence-High Probability conditionthan in the High Valence-low Probability condi-tion; this result is both opposite to that foundon the sociometric measure and contrary to theo-retical predictions.

At the end of their paper, Schachter et al.(1954) offer theoretical interpretation of the dataand indulge a penchant for drawing theoreticalcurves. First, the investigators "demonstrate"that cohesiveness is independent of both val-ence and probability by showing there are nosignificant differences between conditions onthe question, "How frequently do you think thegroup should meet?" This, of course, is onlyone definition of cohesiveness (Aderman, 1968),and, even if the definition were adequate, thissingle postexperimental question seems inade-quate to test the important hypothesis that cohe-siveness is independent of valence and probability.

Schachter and his associates state next thatboth tendency to redefine the group (rejection)and tendency to restore harmony (communica-tion) vary positively with group cohesivenessand valence of the goal and negatively withprobability of achieving the goal. How, then,can one predict which action the group willtake to attain uniformity? Schachter et al. donot answer this question straightforwardly;instead, they employ unstated and undefendedassumptions to create theoretical curves whichare then purported to accurately fit the data.We are offered, however, illuminating formulaesuch as the following; Rcjectic.n = cohesive-ness + tendency to redefine group - tendencyto integrate; and Communication = cohesive-ness - tendency to redefine group + tendAncyto integrate. The investigators state that theirtheory, although not well substantiated, offersthe possibility of integrating communicationand rejection data into one conceptual frame-work. Such an integration would be welcome,indeed, but we are skeptical that the Schachteret al. formulation, as articulated, is powerfulenough to accomplish the task.

Two related studies have obtained evidenceconcerning reaction to deviation which retardsgroup locomotion toward a valued goal Berkowitzand Howard (1959) told groups of Ss that prizeswould be awarded on the basis of either groupperformance (High Interdependence) or individualperformance (Low Interdependence) on a judg-mental opinion task. Subjects were led to be-lieve that one group member differed stronglyfrom modal group opinion. Results showed thatmore overall communication was directed towardthe deviate in the High than Low Interdependencecondition; moreover, Ss in the former conditionhad less desire for future interaction with thedeviate than did Ss in the latter condition. Itappears, then, that rejection of the deviatevaries directly with his interference with goalattainment. A more direct test of this hypothe-sis was conducted recently by Wiggins, Dill,and Schwartz (1965). These investigators ledSs to believe that the probability of groupattainment of a $50 prize had been lowered byone member who did not follow directions.Results indicated that negative evaluation ofthe deviate increased as a function of the de-gree to which he hampered realization of thegroup goal.

Mudd (1968) wished to demonstrate that in-tensity of deviate rejection is a direct functionof the interaction between degree of deviationand importance of violated norm. Mudd devel-oped equal interval scales to measure (1) de-gree of deviation from norm, (2) importance ofnorm to group (speech, cleanliness , honesty),and (3) severity of group sanction administered

for deviation. Subjects were presented 27behavioral items , varying in degree of devia-tion and norm importance, and asked to assignone of 11 sanctions to each behavior. Signifi-cant main effects for norm relevance and de-gree of deviation were obtained: rejectionvaried positively with increasing leveis ofboth variables. Moreover, as norm relevanceincreased, severity of sanction per unit devia-tion increased. Results for norm relevancesupport Schachter's (1951) findings on issuerelevance; data on degree of deviation arecongruent with results obtained by severalinvestigators (Schachter, 1951; Emerson, 1954;Sampson and Brandon, 1964: Katz, Libby, andStrodtbeck, 1964).

Freedman and Doob (1968) manipulated devi-ancy by varying Ss' perceptions of their simi-larity to other group members on general per-sonality scores'. An initial experiment wasdesigned to investigate aggression directedtoward deviant and nondeviant frustrators bydeviant and nondeviant Ss , who had no choiceof victim. High school girls were brought to-gether in small groups; each S was told that(1) she was similar to or different from othergroup members, and (2) another S (confederate)was similar to or different from other groupmembers. The confederate then broke a ma-chine, eliminating Ss' chances to earn $5.00.Next, Ss were asked to rate the confederate'sperformance on a task; medium intensity shockswere hypothetically administered to the con-federate when a majority of Ss rated her nega-tively. Results indicated that nondeviant Ss(the group pertinent to our discussion) did notrate the deviant confederate more negativelythan the nondeviant confederate; thus, itwould seem that the "deviate" was not rejected.

However, note that two types of deviancywere actually manipulated in this experiment:(1) personality deviancy, and (2) unauthorizedtampering with equipment. Both the personality"deviate" and "nondeviate" were guilty oftampering with the equipment. Perhaps, Ssreacted very strongly to this tampering, cre-ating a ceiling on aggression; if so, we wouldexpect no difference in rejection of the person-ality "deviate". and "nondeviate." To test thisnotion, we need an experiment identical to theone above, with the exception that the confed-erate not break the equipment. Freedman andDoob conducted such a study; results againindicated no difference in treatment of the devi-ate and nondeviate confederates. Evidently,reported personality deviance does not elicitaggress ton .

In yet another experiment, Freedman andDoob investigated aggression by deviant andnondeviant Ss, when the Ss were allowed to

4

9

choose either deviant or nondeviant individualsas victims. Each S was told that (1) she wassimilar to or different from other group members,and (2) the remainder of the group was composedof three nondeviants and one deviant. Half theSs were asked to choose another group memberto fulfill an unattractive (shock) task, while theremaining Ss chose another person to participatein an attractive (money) task. Results indicatedthat nondeviant Ss tended to choose deviatesfor punishment but not for reward. Deviant Ss,on the other hand, chose deviates for rewardbut not for punishment. Thus, in this experi-ment, nondeviates did tend to reject deviates.

Unfortunately, Freedman and Doob do notprovide adequate theoretical integration of

10

their discrepant results. The investigatorsconclude that differential aggression to devi-ates occurs when the victim can be selectedby Ss but not when choice of victim is elimi-nated. Thus, deviancy affects selection oftarget for good or bad treatment, rather thanamount of aggression directed toward an alreadychosen target. This, of course, is description,not explanation. An attempt is made to relatethe "scapegoat" theory of aggression to rejec-tion of the deviate when choice is involved.The success of this attempt is questionablebecause Ss were not frustrated or angered inthe last experiment; hence, explanation of theresults in terms of aggression displaced to thedeviate is uncompelling.

IvGROUP SIZE AND MEDIATION OF REJECTION

Much of the data we have reviewed to thispoint strongly suggests that individuals whodeviate from group consensus are rejected.However, one relevant and theoretically cru-cial question about rejection has not been con-sidered. No one has yet investigated whetherrejection occurs because the deviate breaksgroup consensus or because the deviate dis-agrees personally with each member. In otherwords, would each group member in a dyadicinteraction with the deviate reject him asstrongly as when the deviate disagrees in agroup situation?

Recall Pestinger's (1950) assertion that lackof group uniformity may impede group locomo-tion and/or confuse social reality. Regardinggroup locomotion it is clear that in either adyadic or group situation where unanimity isessentiel to valued action, disagreement mightproduce strain eventuating in the rejection ofan unyielding deviate after influence attemptshave failed. Thus, no clear prediction can bemade regarding differential treatment of thedeviate in the dyad or group where unanimityis essential to locomotion. However, whereonly general consensus is necessary to loco-motion, we might predict more rejection of thedeviate in the group than in the dyad, sinceconsensus can be reached without the deviatein the group but not in the dyad. That is, thegroup can afford to reject the deviate and stillbe able to reach its goal, while the deviate isnecessary for locomotion in the dyad .

Turning now to social reality, the situationbecomes more complex. If individuals seekattitudinal congruence with one another in orderto have an unambiguous picture of the world(social reality), one might predict more rejec-tion of the deviate in the dyad than in the group.That is, the deviate in the group would presentlittle threat to social reality because severalpeople would agree on one opinion; hence,the deviate would not be rejected. In the dyad,on the other hand, S would have no social

support for his position and, thus, S wouldreject the deviate who challenges his percep-tion of social reality. We are asserting, then,that rejection is inversely related to E's con-fidence on the relevant issue. This hypothesisderives some support from a study by Worcheland McCormick (1963).

Worchel and McCormick state that certaintyof one's opinion is inversely related to thedissonance produced by contradiction. Cer-tainty is also inversely related to tension.Therefore, an uncertain person reacts moreunfavorably to a contradictor (who increasestension) and more favorably to a supporter (whoreduces tension) than does a certain person.In this experiment, each S gave his opinionand degree of confidence on handling a hypo-thetical interpersonal problem, and then heardhis "partner" either agree or disagree. Post-experimental ratings on several evaluativescales indicated that uncertain Ss liked thesupportive "partner" more and contradictory"partner" le than did certain Ss. Here, then,rejection of the deviate was inversely relatedto S's initial confidence.

However, other data suggest that rejectionmay vary directly with confidence in somesituations. Mob (1951) found that singleindividuals who dissented from group consensuson simple visual judgments received "disdain"and "derision" from the majority. Strickland,Jones, and Smith (1960) told male Ss in groupsof three that they agreed on the issue of "bigtime" athletics and that one of ti would bechosen to attempt to influence a fourth indi-vidual. Subjects were asked to rank severalarguments in order of potential influence utility,and each was led to believe he had beenchosen as the communicator. The Ss were toldthat the other two group members either agreed(Support) or disacreed (No Support) with theirselection of arguments. Then, the argumentswere supposedly taken to the fourth individual,who derogated the arguments and S. On free

11

response personality sketches, significantlygreater hostility was directed to the derogatorby Ss who had received group support thanby those who had not. Thus, in this experi-ment, rejection was directly related to as'confidence.

Although the issue is clearly not resolved,let us assume for the moment that confidenceis inversely related to rejection. Can we thenpredict that rejection of deviates will generallybe greater in the dyad than in the larger group?Perhaps not, for several reasons. First, theperceived discrepancy between S and the deviate

12

tr.

may be larger in the group situation thanin the dyad. Second, the group may disinhibitaggression. Third, S may seek to gain statusin the group by zealously defending groupnorms. Any one or a combination of thesefactors might overcome the hypothesized ten-dency for a to reject the deviate more in thedyad than in the group. At this point, how-ever, we have little or no empirical data ade-quate to test these hypotheses. Clearly,further research is needed to provide adequatetheoretical conceptualization of the relation-ship between group size and deviate rejection.

t 116 g

GP* 610-876-3

sh

V

ALTERNATIVE REACTIONS TO DEVIANCY

As noted earlier, this paper has been pri-marily devoted to discussing rejection of thedeviate as a strategy for achieving group uni-formity. This stress seemed reasonable be-cause most research has been devoted to thisparticular resolution of deviance. We havealso mentioned two additional resolutions ofattitudinal discrepancy proposed by restinger(1950): attempt to modify the deviate andchange in modal group opinion. There is,however, a fourth solution to the problemdifferentiating or ignoring the deviate's re-sponse. In a paper of high potential, Dent lerand Erikson (1959) proposed that groups "in-duce, sustain and permit" deviant behaviorbecause of its functional eroperties. Essen-tially, deviant behavior is necessary becauseit delineates boundaries of acceptable groupbehavior and, in so doing, defines the groupand its members. Therefore, the group providesspecial roles for deviates (e.g., low socio-metric rank) and strongly resists alienation ofthese members unless deviant behavior becomescritically dangerous to group solidarity.

Dent ler and Erikson provide evidence thatdeviates in Quaker work projects and preschizo-phrenic trainees in U.S. Army squads are pro-tected and retained by the group. Relativelyweak pressures to conform are exerted on thesedeviates, and there are no attempts to expelthem from the group. While dramatic, theseexamples of deviate acceptance do not unequivo-cably validate Dent ler and Erikson's hypothesis.For, in both Quaker projects and Army squads,specific norms exist prescribing protection andacceptance of group members. Thus, to clearlysubstantiate that deviance is generally permit-ted, we need data from groups which do notpossess such clear-cut protection norms , e.g.,political groups, business men's associations.

Coser (1962) discusses in detail three po-tential reactions to deviancy and concomitantgroup consequences. First, the deviate canbe opposed and the group strengthened: bound-

sties are defined, common sentiments are re-vived, and normals feel righteous. Second,the deviate can be tolerated and the groupstrengthened: group norms are affirmed bytolerance (Quaker work groups). Third, thedeviate can be rejected and the group weak-ened: membership decreases, competing sectsemerge, etc. The first and second reactionsto deviancy (opposition and tolerance) areinteresting because both are based on strength-ening group norms. That is, some groupsstrengthen norms by opposing deviates (JohnBirch Society), while other groups strengthennorms by sheltering deviates (Salvation Army,Quakers). This line of thought supports ourdiscussion in the last paragraph concerningthe generality of deviate acceptance; more-over, it suggests that specific group norms ,and not merely variables such as extremenessof deviancy, may determine the manner in whicha group reacts to deviancy.

The ideas of Dentler and Erikson (1959) andCoser (1962) concerning the functions of devi-ancy have relevance to Festinger's (1954) no-tion of social comparison. Both formulationsimply that individuals utilize comparison withothers in order to evaluate and define them-selves. Festinger asserts that people seekout similar others as referents. However,Dentler and Erikson, and Coser, imply thatpersons utilize dev)ates who stand at thegroup boundary as referents. Thus, Festingerimplies an attempt to decrease the distance tothe comparison person (assimilation), whilethe other theorists imply an attempt to maxi-mize the distance from deviates (contrast).

Permissive, accepting behavior towarddeviates has also been reported in therapygroups (Stock, Whitman, and Lieberman, 1958).Patients were observed to alter their percep-tions of deviate behavior so that the behaviorwas acceptable. For example, remarks thatgroup therapy was "useless" were perceivedas merely superficial verbalizationr that masked

1 Q

13

actual acireen:ent with group consensus. Thisevidence for acceptance of deviation may clsobe explained in terms specific to the groupsobserved, That is, absence of deviate rejec-tion may have been due to (1) fear of doctors'reactions, (2) fear that a rejection precedentmight backfire on any group member, or (3) ad-herence to norms that idiosyncratic behaviorshould be tolerated in the therapy setting.While the latter two factors are mediated withinthe group, the first factor (fear of doctors) isclearly imposed by external forces. Thus, asCoser (1962) asserted, consideration of theexternal environment can clarify our under-standing of a group's internal response todeviancy.

A final example of deviate acceptance, e.g. ,the permissive treatment accorded a high statusnonconformer, has been discussed by Hollander(1958, 1960, 1964). Hollander suggested that"idiosyncrasy credits" accrue to individuals indirect proportion to conformity to group normsand demonstrated competence. These creditsare tangible indicants of status and are directlyrelated to the amount of nonconformity allowed.However, when an individual deviates fromgroup norms, he used up idiosyncrasy credits;complete exhaustion of credits produces rejec-tion.

Note that Hollander suggested that statusis directly related to license for deviation.Dent ler and Erikson (1959), on the other hand,posited an inverse relationship. Homans (1961)suggested that a curvilinear relation exists be-tween status and deviation: middle status per-sons conform more than individuals of eitherhigh or low status. The previously mentionedexperiment by Wiggins et al. (1965) providesdata relevant to the conflicting predictionsmentioned above. These investigators foundthat high status deviates receive more pun-ishment for major interference with attainmentof group goals and less punishment for minorinterference than do lower status deviates.This interaction between status and degree of

14

interference may be due in part to specificexpectations regarding the role of the highstatus individual. Although relatively immunefrom group sanction in matters of little conse-quence, the high status person who retardsgroup locomotion in important matters (wherehe is expected to lead) will be severely criti-cized. Thus, the low status person is expectedto c.onform on small matters but is "allowed" tohinder goal attainment: the high status indi-vidual may deviate on unimportant matters butis expected to perform capably in pursuingimportant group goals. These speculations,if true, seriously limit the adequacy of idio-syncrasy credits in explaining the relationshipbetween status and deviance. It would appearthat idiosyncrasy credits may only have pre-dictive utility in matters of relatively littleimportance to the group.

Hollander (1964, p. 214) stated that non-conformity is "perceived" and waluated" inlight of the idiosyncrasy credits a person hasaccumulated. Though perception and evalua-tion seem to be used synonymously by Hollander,a distinction can be made. It would seem mean-ingful to oistinguish between the following twostatements about a heavy drinker: (1) "Joereally doesn't drink much," and (2) "Joe drinksa lot, but with hi4 wife and job, he deservesit." While Hollander, Dent ler and Erikson,Homans, and Wiggins et al. stress evaluativefactors in deviate acceptance, Stock, Whitman,and Lieberman suggest altered perception aswell. Thi c. listinction seems important toanalysis oi e mechanisms underlying groupreaction to .laviancy.

This paprir has attempted to review andmeaningfull? integrate empirical and theoreti-cal work on ...he issue of group reaction to atti-tudinal deviancy. Clearly, many issues arestill unresolved because adequate data are notavailable. Thus, in conclusion, we echo thehackneyed but so often inescapable conclusionthat more research is needed.

REFERENCES

Aderman, D. Determinants of a majority'sreaction to a disagreeing minority. Un-published manuscript, University ofWisconsin, 1968.

Allen, V. L. Conformity and the role of thedeviant. J. Pers., 1965, 33, 584-597. 7.

Allen, V. L., and Levine, J. M. Consensusand conformity. Technical Report from theWisconsin R & D Center for CognitiveLearning, University of Wisconsin, 1968,No. 41,

Asch, S. Effects of group pressure upon modi-fication and distortion of judgments. InH. Guetzkow (Ed.), Groups, leadership,and men. Pittsburgh: Carnegie Press,1951.

Back, K. W. Influence through social com-munication. J. abnor. soc. Psych., 1951,46 9-23.

Berkowitz, L. Reporting an experiment: acase study in leveling, sharpening andassimilation. Unpublished manuscript,University of Wisconsin, 1967.

Berkowitz, L., & Howard, R. Reactions toopinion deviates as affected by affiliationneed (n) and group member interdependence,Sociometry, 1959, 22, 81-91.

Coser, L. Some functions of deviant behaviorand normative flexibility. Amer. J. Sociol. ,1962, a, 172-181.

Dent ler, R., & Erikson, K. The functions ofdeviance in groups. Social Prob., 1959,7, 98-107.

Emerson, R. M. Deviation and rejection: anexperimental replication. Amer. soe. Rev.,1954, 19, 688-694.

Festinger, L. Informal social communication.Psych. Rev., 1950, 57, 271-282.

Festinger, L. Theory of social comparisonprocesses. Human. Re lat., 1954, 7,117-140.

Festinger, L., Schachter, S., & Back, K.The operation of group standards. In D.Cartwright and A. Zander (Eds.), Group

dynamics, research and theory. New York:Harper and Row, 1960. Pp. 241-259.

Festinger, L., & Thibaut, J. Interpersonalcommunication in small groups. J. abnor.soe. Psych., 1951, 46, 92-99.

Freedman, J., & Doob, A. Deviancy, theysy-chology of being different. New York:Academic Press, 1968.

Gerard, H. B. The effect of different dimen-sions of disagreement on the communicationprocess in small groups. Human Relate,1953, 6, 249-271.

Hollander, E. P. Corformity, status and idio-syncrasy credit. Psych. Rev., 1958, 65,117-127.

Hollander, E. P. Competence and conformityin the acceptance of influence. J. abnor.soc. Psych., 1960, 61, 365-369.

Hollander, E. P. Leaders, groups and influence.New York: Oxford University Press, 1964.

Homans, G. Social behavior: its elementaryforms. New York: Harcourt, Brace, & World,1961.

Jones, E., & Gerard, H. Foundations of socialpsychology. New York: Wiley, 1967.

Katz, E., Libby, W., & Strodtbeck, F. Statusmobility and reactions to deviance and sub-sequent conformity. Sociometry, 1964, 27,245-260.

Kelley, H., & Thibaut, J. Experimental studiesof group problem-solving and process. InG. Lindzey (Ed.), Handbook of social _psy-chology, II. Cambridge, Mass.: Addison-Wesley, 1954. Pp. 768-769.

Mills, T. A sleeper variable in small groupsresearch: the experimenter. Pacific Sociol.Rev., 1962, 5, 21-28.

Mudd, S. A. Group sanction severity as afunction of behavior deviation and relevanceof norm. I. pers. Soo. Psych., 1968, 8,258-260.

Sampson, E., & Brandon, A. The effects of roleand opinion deviation on small groups behavior.Sociometry, 1964, 27, 261-281.

20

15

Schachter, S. Deviation, rejection and com-munication. J. abnorrn. soc. Psych., 1951,46, 190-207. (In D. Cartwright and A.Zander (tds.), Group dynamics, researchand theory. New York: Harper and Row,1960. Pp. 260-285.)

Schachter, S., Nut lin, J., DeMencheux, C.,Maucorps, P., Osmer, D., Duijker, H.,Rommetveit, R., & Israel, j. Cross-cultural experiments on threat and rejec-tion. Human Re lat., 1954, 72, 403-409.

Stock, 13. , Whitman, R., & Lieberman, M.The deviant member in therapy groups.Human Relat., 1958, 11, 341-371.

16

Streufert, S. Communicator importance andinterpersonal attitudes toward conformingand deviant group members. J. pers. soc.Psych., 1961, 2, 242-246.

Strickland, L. 11., Jones, E. L., & smith,W. P. Effects of group support on theevaluation of an antagonist. J. abnorm,soc. Psych., 1960, 61, 73-81.

Wiggins, J., Dill, 1%, & Schwartz, R. On"status-liability," Sociometry, 1965, 28.197-209.

Worchel, P., & McCormick, 13 Self-conceptand dissonance reduction. J. pers., 1963,31, 588-599.

GPO J a!t..-2