smoking&drinking

Upload: shannon-cassidy

Post on 09-Apr-2018

221 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/8/2019 smoking&drinking

    1/28

    Social Influence, Self-Referent Identity Labels, and BehaviorAuthor(s): Bruce J. Biddle, Barbara J. Bank, Don S. Anderson, Ragnar Hauge, Daphne M.Keats, John A. Keats, Marjorie M. Marlin, Simone ValantinSource: The Sociological Quarterly, Vol. 26, No. 2 (Summer, 1985), pp. 159-185Published by: Blackwell Publishing on behalf of the Midwest Sociological SocietyStable URL: http://www.jstor.org/stable/4106374

    Accessed: 18/04/2010 13:55

    Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

    http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

    you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

    may use content in the JSTOR archive only for your personal, non-commercial use.

    Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

    http://www.jstor.org/action/showPublisher?publisherCode=black.

    Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

    page of such transmission.

    JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

    content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

    of scholarship. For more information about JSTOR, please contact [email protected].

    Blackwell Publishing andMidwest Sociological Society are collaborating with JSTOR to digitize, preserve and

    extend access to The Sociological Quarterly.

    http://www.jstor.org

    http://www.jstor.org/stable/4106374?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/action/showPublisher?publisherCode=blackhttp://www.jstor.org/action/showPublisher?publisherCode=blackhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/4106374?origin=JSTOR-pdf
  • 8/8/2019 smoking&drinking

    2/28

    SOCIALINFLUENCE,SELF-REFERENTIDENTITYLABELS,AND BEHAVIORBruceJ. BiddleBarbara . Bank

    Universityof Missouri-ColumbiaDon S. Anderson

    AustralianNationalUniversityRagnarHaugeNationalInstitute orAlcohol Research Norway)

    Daphne M. KeatsJohnA. KeatsUniversityof NewcastleMarjorieM. Marlin

    Universityof Missouri-ColumbiaSimone ValantinUniversitede Paris

    Self-referentidentitylabels arefrequentlyargued o be a centralcomponentof the selfand to be important n the planning of conduct. Despite the attractivenessof thisargument,relativelylittle researchhas yet appeared hatsupports t, and studiesof the

    Directall communicationso: Dr. BruceJ. Biddle,Center or Researchn SocialBehavior,University fMissouri-Columbia,olumbia,MO 65211.TheSociologicalQuarterly,Volume26, Number2, pages159-185.Copyright? 1985by JAI Press, Inc.All rightsof reproductionn anyform reserved.ISSN:0038-0253

  • 8/8/2019 smoking&drinking

    3/28

    160 THE SOCIOLOGICALQUARTERLYVol. 26/No. 2/1985

    etiologyandeffectsof self-referentabelsoften fail to control ortheconfoundingeffectsofpreferences,orms, rother onceptionshatareknown o affect onduct. nthispaper,propositionsreargued oncerninghe causesand ndependentffectsofself-referentabels. Someof thesepropositionsre thentested n two studies,oneconcernedwithalcoholuseamongadolescentsn fourWestern ountries,heotherfocusedon alcoholand obaccouseamongAmerican igh-schoolndundergraduatestudents.Bothstudies oundstrong ies betweenreports f peermodelingandre-spondents' elf-referentabels andthat the latterhad significant ndindependenteffectson reported ehaviors.Theoreticalndpracticalmplicationsf findingsarediscussed.

    Muchof social psychologicalresearchmay be said to constitutea searchfor the central,conceptualprocesses throughwhich social behavior is generated. A good deal of thisresearchhas concentrated on two conceptual forms:preferences (or "attitudes") andnorms for conduct. Studies of preferencesand norms have numbered n the thousands,and althoughthese two forms are still sometimesconfused in research, a good deal isknowntoday aboutthe etiology and effects of these forms of conception.In contrast,much less research has appearedconcerningself-referent identitylabels.This is surprising,since Symbolic Interactionistsand othershave long arguedthat suchbeliefs are a centralcomponentof the self (BurkeandTully, 1977;McCallandSimmons,1966; Stryker, 1968, 1980) and are important n the planningof conduct (Foote, 1951;Gecas, 1982; Rosenberg, 1979, 1981; Stone, 1962; Wells, 1978; Wicklund and Goll-witzer, 1982). The basis of this argument s that the self-referent dentitylabel (or "self-concept") is thoughtto be formed largely through nteractionwith others in social con-texts, thus it is associatedwith situatedexpectationsfor conduct and has the capacityforaffectingbehaviorby providingthe personwith a frameof reference for interpretinghesituationand planningactions in it.Although this argumentis attractive,relatively little researchhas yet appearedthatsupports t, and some of the presumablysupportiveresearchmay also be interpretednpreferential r normative erms.To illustrate,altercastingis a form of influencein whicha persondeliversmessagesthataredesignedto changetheother'sself-concept(Weinsteinand Deutschberger, 1963). Altercastinghas been found to provide sharp changes inbehavior,andmost theoristsmake the assumption hatthesechangesare due to successfulinduction of changes in self-referent labels (Jensen and Moore, 1977; Kraut, 1973;McArthuret al., 1969; Miller et al., 1975; Steele, 1975; Strentaand DeJong, 1981).Unfortunately, vidence for this interpretations weak since few altercasting tudieshaveyet asked subjectsto reporttheir identity labels, and alternative nterpretationsor theeffect have been proposedthat are based on presumedshifts in subjectmood, preferencefor the influencer,or normsfor conduct (Gabrenya,1979;Jones, 1973; Shrauger,1975;Shraugerand Schoeneman, 1979).A second, relevant researchtraditionhas focused on the Foot-In-The-DoorTechniquein whichresearchsubjectsare induced to makea majorchangein behaviorby firstgettingthem to engage in a related, minor action. Early investigatorsalso argued that thisprocedure works because subjects change their self-concepts (Freedman and Fraser,1966), butagainstudies within the Foot-In-The-Doorraditionhave rarelyaskedsubjectsabout their self-concepts. Recent reviewers have suggestedthat the techniquemay pro-duce other conceptual changes as well as adjustmentof self-referentlabels, and theselattermay account for changes in behavior(DeJong, 1979, 1981; Rittle, 1981).

  • 8/8/2019 smoking&drinking

    4/28

    SocialInfluence,Self-ReferentdentityLabels, ndBehavior 161

    Yet anotherresearch traditionhas found significantrelationshipsbetween self-esteemlabels and many issues in conduct: e.g., academic achievement(Purkey, 1970; Wylie,1979);occupationalchoice (Dipboye, 1977;Korman, 1966);politicalparticipationCar-mines, 1978;Rosenberg, 1954-55; Sniderman,1975);or adolescentrejectionof deviance(Kaplan,1976, 1980;RosenbergandRosenberg, 1978). Such findingsareconundrums,however, since self-esteem labels are not usually phrasedso thatthey applyto a specifictopic in behavior, thus their ability to guide conduct is questionable.In addition,highlevels of self-esteemandcriterionbehaviorsarelikely to be preferred verlow levels, andthe formermay even be associated with positive norms, so the apparenteffects of self-esteem labels may in fact be due to other, associated, conceptual processes.In short, researchto date on such issues as altercasting,the Foot-In-The-DoorTech-nique, and self-esteem tends to support he thrustof self-referent dentitytheorybut canhardlybe considereddefinitive supportfor it. In contrast,only a small groupof studieshas yet tackledthe effects of experience on identityformation(for a review, see Alex-anderandWiley, 1981) whereasstudies of the effects of self-referent dentitieshave beenvanishingly few (Backman and Secord, 1968; Korman, 1967; Leonard et al., 1973;Reckless et al., 1956). It is difficult to find studies in either tradition hat have involvedcontrols for the presence of preferences, norms, or otherconceptionswhose appearanceand impact might be confused with that of self-referent abeling.The purposesof this paper, then, are to provideboth theory and controlled evidenceconcerning heetiology andeffects of self-referentdentity abels. We begin by reviewingarguments hat have been advanced for the origin and impact of the self-concept. Evi-dence is thenpresentedconcerningcausesand effects of self-referentabelingin decision-making by young persons.

    ETIOLOGYWheredo self-referentabelscome from? Underwhat conditionswill one form oracceptaself-referent dentitylabel? First, label acceptancemay be generatedwhen others applythatlabel to the personin a persuasivefashion(Backmanet al., 1963;K. Erikson, 1962;Gergen, 1965; Gove, 1975;MiyamotoandDornbusch,1956;Reeder et al., 1960;Scheff,1966; Sherwood, 1965, 1967; Videbeck, 1960). Strategiesfor persuasive labeling mayinclude altercasting, hypnotic suggestion, alertingthe person to "facts" about him orherself that were previouslyunknown, ceremonies in which identitylabels are publiclyappliedto the person, and some types of political demagoguery. These strategiesareassumedto workbecausethepersonis encouraged o believe thatmanyothers(orat leastthose otherswho are "important")share the belief that the labelappliesto theperson.Todenythevalidityof sucha consensual udgmentwouldpresumably akeconsiderable elf-confidence on the partof the person.This does not mean that all strategies for persuasionlead the person to form self-referent abels. To illustrate,normativeadvocacyby anothergenerally eads thepersontoattributenorms for his or her own conductto that other. Moreover,the personmay alsoconform in his or her behavior to the norm attributed f the other is deemed likely tosupportadvocacy with sanctions or if the personinternalizesadvocacy as his or her ownnorm or preference(Kelman, 1958; Parsons, 1951). But normativeadvocacy by itselfprovidesneithera vocabulary orlabelingnor motivation or thepersonto applythat abelto him orherself andis unlikelyto generateself-referentabels. Indeed,at least one study

  • 8/8/2019 smoking&drinking

    5/28

    162 THESOCIOLOGICALUARTERLYol. 26/No. 2/1985

    (Milleret al., 1975) has explored persuasivelabeling and normativeadvocacy as alter-native and opposed ways of generatingconformingbehavior.Second, the modeling of behavior by others may lead to self-referent abeling if theperson identifies with those others, the behaviors modeled are associated with mem-bershipin a social position those othersrepresent,and the person believes thathe or shehas the ability to performthem (Heiss, 1968; Latand, 1966; Morse and Gergen, 1970;Pettigrew, 1967). One example of this condition appears in anticipatorysocialization(Merton, 1949) whereinthe personapes the mannerismsof othersto whose position theperson aspires. Another occurs among adolescentswho may be convinced that specifichabits of behaviorare requisitefor acceptanceamongpeer groups. Still anotherappearsamong personswho label themselves positively in the hope of breakinga bad habit. Wehavepreviously argued hat themodelingof behaviorby others can also leadthepersonto

    form preferences for the behavior displayed (Biddle et al., 1980a). This means thatmodeling may cause the person to form preferences and self-referentbeliefs that arealigned. It does not mean that the person will always form consonantpreferences andbeliefs, of course. One may develop strongpreferencesfor a skilled behaviorthatonecannothope to imitate, and personssometimesadoptbehaviorpatterns hey thoroughlydislike in orderto obtainacceptanceby others in a group.Third, it is possible that persons may also form self-referentidentitiesby observingtheirown behaviors(Bandura, 1978; Bem, 1972; Ryle, 1949). To arguethat one's ownbehaviormayaffect one's self-conceptposes a conundrum.Most theoristshaveurgedthatexternal stimuli lead the person to form conceptions and that the latter then serve asgeneratorsof behavior.Nevertheless, a few authorshave arguedthatself-referent abelswill form afterthe person observes him or herself behaving, and Daryl Bem (1972), inparticular,has arguedthat the findings of many studiesof cognitive dissonancemay beexplainedparsimoniously f we assume that behaviorsproducealigned self-concepts. Ifthis were not sufficientlyconfusing, other authorshave arguedthat beliefs about the selfandbehaviors are interactiveandtend to supportone another(BurkeandReitzes, 1981;Wells, 1978) or that their apparentcovariation results from the causative influence ofother factors.

    Which, then, are morelikely to inducechanges in the other:conceptionsor behaviors?In the case of preferences and norms we argue that conceptionsare more likely to beindependentvariables. Although conceptions in these modes may, occasionally, beformed romobservingone's own behavior,thistypeof reaction s probablynotcommon.Experiencesin which preferencesor normsare not matchedwith behaviorare frequent,andmost personsseem to react to these experiences by tryingto adjusttheir behaviors.Moreover,available evidence suggests thatpreferencesandnorms have more impactonbehaviorthan vice versa (Andrews and Kandel, 1979; Kahle and Berman, 1979).In contrast,beliefs are presumedto reflect reality, and it seems less likely that self-referentidentity labels will be embracedfor long if they are not matched, in part, byconduct. This means thatif a person is induced to change his or her habitualconductbyexternalpressures,norms, or preferences, it is also likely that he or she will presentlyadopta self-referent abel that reflects the new behavior. Adoptionof the label should beunlikely,however, if the new behavior is limited to a specific context, if the vocabularyfor that label is unfamiliar,or if the label has negative associationsfor the person.To illustrate the latterpoint, consider the heavy drinker who is beginning to sufferphysicalor social costs because of alcohol abuse. For some yearsthe personrejectsthe

  • 8/8/2019 smoking&drinking

    6/28

    SocialInfluence,Self-ReferentdentityLabels, nd Behavior 163

    label of "alcoholic" because it is opprobriousn the contextshe or she regularlyenters,and few pressuresareplaced on the personto matchbehaviorswith label. Then, perhapsfrightened by the escalating costs of abuse, the person is induced to join a chapterofAlcoholics Anonymous.Lo, the personnow regularlyenters a context in which the labelis no longerconsideredopprobrious,manyrewardsarenow offeredfor matching abel toconduct, and new and less-destructivebehaviors are encouragedthat are deemed label-appropriate.The personis now more likely to label him or herselfan "alcoholic" andisurgedto announcethis label in otherdrinkingcontextsas partof a strategyfor avoidingalcohol use (Turner, 1971). The entire enterprise s designed to provide environmentalforces that will enablethe personto overcome his or her admittedandcontinuingprefer-ence for heavy drinking,and clearly it works for some persons.To summarize, then, self-referentidentity labels are assumed to be generatedwhenotherslabel the personpersuasively,when performablebehaviorsare modeledby otherswith whom the person identifies, andwhen the personis inducedto behave habitually nsome way thatimplies a label which does not have strong, negativeassociationsfor theperson.Given ourearlierargument hat behaviormodelingby othersmay lead the personto form preferences for conduct, it follows that exposure to behavior modeling willsometimeslead to preferencesand self-referent abels that are aligned.

    EFFECTSWhat is the likely impactof self-referent abels on behavior?Under what conditionswillbeliefs aboutthe self lead to conformingbehavior,and when will those beliefs be ignoredor resisted?Probably he most common reasonfor conformingto self-referentbeliefs isthatthosebeliefs areassociated with preferences,norms,or otherexpectationsfor behav-iorthatalso induceconformity.To illustrate, hepersonwho identifieshimor herself as a"drinker" or "teetotaler" may exhibit drinkingbehaviorsappropriateo those labelsbecause he or she prefersorapproves hose behaviors.Althoughsuchassociationsmaybecommon, the answerthey suggest is misleadingif we are interested n the independentimpactof self-referentabels. Let us assume it possible to isolate self-referent abels frompreferences, norms, and other expectations for conduct. Having controlled for theseconfoundingvariables,what are the independenteffects of labels on behavior?

    We shallarguethatself-referent abels affectbehaviorfor two, quitedifferentreasons.First, personsmay have internalneeds to conformto labels. Sometimes such needs areassumed o be universal. Forexample, Festinger's(1957) cognitivedissonance heoryandHeider's(1958) propositionsaboutcognitivebalancebothapparently rgued oruniversalneeds to make expectations and behaviors consistent. Although these theories have at-tracteda good deal of attention,empiricalsupportfor such universal needs is weak. Toillustrate,reviewers of cognitive dissonanceresearchnow suggest that this effect appearsonly whenthe issue is importanto the person(Aronson, 1969;FazioandCooper, 1983).If not universal,needs for conformitymightappear n variousdegreesamong persons,thus be considered a personalitycharacteristic. Saltzer (1981) argues the case for thisapproach.Several years ago Rotter(1966) proposedthatpersonsdiffer in whethertheybelieve they can determineoutcomes throughtheir own efforts or that outcomes aredeterminedby external factorssuch as luck, fate, or powerfulothers. Rotter'sconceptoflocusof control has been foundrelatedto individualdifferences n achievement Lefcourt,1976; Phares, 1976). Saltzerargues that it applies also to self-referentconformity, and

  • 8/8/2019 smoking&drinking

    7/28

    164 THESOCIOLOGICALUARTERLYol. 26/No. 2/1985

    thatpersonswho believe in their own efficacy will be moremotivatedto match behaviorsto self-referent abels than those who view outcomes as externally caused, presumablybecause lack of conformitywould imply thatbeliefs aboutcontrolare in error(Bandura,1977; deCharms, 1968). Similar argumentsabout conformity to labels have also beenmade for such motives as the need to maintainconsistency of the self-concept (Green-wald, 1980), and it seems unlikely that need for conformitywould be high for personswho were unintelligentor afflicted with schizophrenia.Second, persons may also conform to a self-referent abel if they value membership nthe social position that the label designatesand view behaviorsas a way of confirmingmembership n thatposition (Foote, 1951). To illustrate,professionalathletesmay prac-tice long hours and exertgreateffort if they are convincedthatsuch actionsarenecessaryto retaintheir desiredprofessionalstatus. The behaviorsthey exhibitareobviously thoseassociatedwith the label, but thosebehaviorsmaynotappear f the athleteno longercaresabout his or her professionalcareeror believes that careerto be guaranteedby talent orother means. This argumentsuggests that persons will conform only to "some" self-referent abels;other labelsthey applyto themselveswill not generateconformingbehav-iors. This pointis arguedby bothKelley (1955) andBem andAllen (1974), and the latterauthorsoffer datasuggestingthatpersonsare awareof the labels which generateconfor-mityfor them. Stryker 1980) andStrykerandSerpe(1982) urgethatconformity s morelikely when labels are salient for the person.To summarize, hen, self-referent abelsmayinduceconformingconducteitherbecausethe personhas internalneeds for conformityor becausethe personrespondsto externalforces that reward t. Interestingly,these reasonsfor conformityareindependentof thosenormallycited for conformityto preferencesand norms. Personsarethoughtto conformto preferences or consummatoryreasons whereasconformityto normsis assumedto beinduced hroughguiltor fearof consequences.Since self-referentabels, preferences,andnormsall have independentreasons for inducingconformity, personsmayhave to chooseamongthem in planningbehavior.Thus, in some contextsa person may have to decidewhether to do what he or she prefers to do, thinks "ought" to be done, or believesnecessary to do to confirm his or her identity. Such experiences of discordanceareresolved when the person figures out which conceptionsare stronger n those contexts(Biddle, 1979). (To illustrate,personsmaybe led to initiatea behaviorbecausetheyvaluemembershipn a social groupassociatedwitha given identity,but lateron-when trappedby habit or addiction-their behavior may be controlledmore by preferencesthan byidentity labels.) Thus norms, preferences, and self-referent labels are all capable ofgeneratingvariance n decisions made about behavior.Eachmaybe anindependent auseof conduct.

    MODELAND HYPOTHESESThe results we reportbelow were generatedby two studies concernedwith social influ-ence on young persons. Considerable researchhas investigatedthis issue, much of itconcerned with the impact of parentaland peer modeling and normativeadvocacy onadolescentconduct. In previous papers(Biddle et al., 1980a, b), we have arguedthatagood deal of this influence occurs becauseyoung personsform their own preferencesandnorms in response to parental and peer pressures. In this paper we assert that thesepressuresalso lead young personsto form self-referent abels, andthatthose labels have

  • 8/8/2019 smoking&drinking

    8/28

    SocialInfluence,Self-ReferentdentityLabels, nd Behavior 165

    substantialandindependentmpacton youthconduct. The argumentwe make is a generalone, althoughthe datawith which we test it concern the use of alcohol and tobaccobyyoung persons. (For reviews of researchon these topics see Blane and Hewitt [1977],Chassin [forthcoming],Flay et al. [1982], and Kandel [1980].) In summary, then, themodel to be tested is one in which youngpersons'drinkingor smokingarepredicted romfour, exogenousvariables(parentalandpeer drinkingor smoking, andparentalandpeernormativeadvocacy concerning these topics) and three, endogenous variables (youngpersons' self-referent abels, preferences, and normsfor drinkingand smoking).We state two hypotheses for self-referentidentity labels within such a model, oneconcernedwith etiology, the other with effects. First, we have arguedthatpreferences,norms,andself-referent abels have somewhat differentetiologies. In particular,we havesuggestedthatself-referent abels aregeneratedwhenperformablebehaviorsare modeledby otherswith whom the person identifies. Drinkingand smokingare forms of behaviorthatareeasy to perform,andit is reasonable o believe thatmany young persons identifywiththeirparentsor peers. Therefore,parentalandpeermodelingshouldlead, in part,toformationof self-referent abels for drinkingandsmoking. In contrast,it seems unlikelythatnormativeadvocacyby parentsandpeerswould lead to identityformation or theseorothertopics in youth behavior.

    Hypothesis 1. Behavior modeling by parents and peers have significant effects onyoungpersons' self-referent abelingfor drinkingand smokingthat are independentof one anotherand of normativeadvocacy by the two referent groups. In contrast,normativeadvocacy by parents and peers have weak or nill impacton young per-sons' self-referent abeling for these issues.Second, followingthe leadof E. Erikson(1968), manywritershave suggestedthat self-concept and the formingof acceptableidentities are importantconcerns for young per-sons. According to this argument, youth is a time when persons tend to discard thebehavioral standardsof others which have previously controlled conduct and seek toestablishhabits that are more closely aligned with acceptableideas about the self. Thissuggests that young persons have strong internal needs for conforming to some self-referentidentitylabels. Reflectionsalso suggest thatdrinkingandsmokingare issues forwhich external forces appearfavoring youth conformity. Both forms of behavior areproblematic or youth. Both are prohibitedwhen one is a child, and both are associatedwith adulthoodandconviviality. In addition,both are often used as behaviors that sym-bolize membership n groupsto whichyoungpersons aspire.(Thus, theyouthwho avoidstobacco or alcoholmay, in effect, be confirming membership n a religiousor residentialgroupthatstresseshealthyconduct,whereasthe young personwho drinksor smokes maydo so tojoin a social groupwherethese behaviorsarede rigeur.)Thesereflectionssuggestthat self-referentidentity labeling for drinkingand smoking have an impact on youthconduct that is strong and independentof preferences, norms, and other conceptualgeneratorsof behavior.

    Hypothesis 2. Self-referent dentity abelsfor drinkingand smokinghave significanteffects on young persons' behaviors that are independent of the effects of otherconceptionsheld about these issues.

  • 8/8/2019 smoking&drinking

    9/28

    166 THESOCIOLOGICALQUARTERLYVol. 26/No. 2/1985

    METHODSWe turnnow to two studies that offer evidence bearing on these hypotheses. Study 1concerned the use of alcohol among adolescents in four, Westerncountries, and somefindings from it have alreadybeen presented(Bank et al., forthcoming;Biddle et al.,1980a, b; Biddle et al., 1981; Hauge, 1977, 1978; Keats et al., 1983). Details of itsmethodshave alreadybeen publishedandareonly summarizedhere. Study2 focused onboth alcohol and tobacco use among Americanhigh-school and college studentsand isdescribed in more detail.

    Study1DataforStudyI were collectedin 1975-76 frommatched,quotasamplesof adolescent

    respondentsby means of structured nterviews.Respondents nd Sample Design

    Respondents orthe studywere429 unmarried dolescentscomprising149Americans,96 Australians, 100 Norwegians, and 84 French youths. All four samples were con-structed n a common quota-sampledesign thatwas balanced for sex, age, and parentalsocial class (and, in the United Statesonly, for race). Each samplewas selected to fill a2x 3x 2 grid in which equal numbersof boys and girls (those aged 12, 15, and 18) andthose from middle-or-upperversus working-or-lowerclass homes were to appear. Thefoursampleswereurban n compositionandwere to consist of respondentswho were thenenrolledin state (public) schools in theirown countries.Instrumentsnd Procedures

    Data for the study were collected by individualinterviewsfor which a standardizedschedulewas developedandextensivelypretested n all fourcountries.The scheduledealtwithvarious issues in the lives of adolescents, includingthe use of alcohol, andprovidedboth lead questions and appropriateprobes to be used if respondentshad difficultyanswering.In all, interviews averagedan hour and a half in length, andresponsesweretape-recordedorsubsequentcoding, for which a standardizedmanualwas also developedandpretested n each country.Interviewingandcodingwere conductedby nationals n thenative languageof respondents n all cases.Respondentswere asked to report heir self-referent abels, preferences,and normsforalcoholuse, andfive-category, Likert-typescales were used for all threemeasurements.In the case of self-referent abeling, respondentswere asked to choose which of a set ofdrinking abels they would applyto themselves. Variouslabels for alcoholuse have beensuggested in prior studies (Liccione, 1977), and drawing from these we offered re-spondentsa choice of: ABSTAINER, TASTER, DRINKER,ALCOHOLIC,andALCO-HOL ADDICT. In coding responses it was assumed that these five labels fell into anordered cale ranging rom "abstainer"(coded 1) to "alcohol addict"' coded 5), and thisassumptionwas confirmed during pretesting. (Other ways of measuringself-referentlabelingare discussedbelow.) Questionsused formeasuringrespondents'preferencesandnorms were phrasedin traditional,Likert-typeformat. In the case of preference, re-spondentswere asked to indicate how muchthey enjoyeddrinking.In the case of norms,respondentswereaskedtheir evel of approvalordisapproval ortheirown use of alcohol.

  • 8/8/2019 smoking&drinking

    10/28

    SocialInfluence,Self-ReferentdentityLabels, nd Behavior 167

    In addition, respondentswere also asked to reporttheir own drinkingbehavior, thedrinkingbehaviorsof parentsand peers, andthe norms advocatedby these two referentgroupsfordrinkingby the respondent.Several, differentquestionswere used to generatereports orrespondents'own drinking,but we reportdatahereforonly one of these: HOWOFTEN DO YOU USUALLY DRINK? Separate reports were given concerning thedrinkingof each parent,andthese were summedto provideajoint index. A similarindexwas also prepared orpeer drinking. Finally, respondentswere askedto indicate the levelof approvalor disapprovalthat parentsand peers advocatedfor respondents'drinkingconduct. Responsesto all of these questionswere, again, coded as Likert-typescales.

    Study2Data for Study 2 were also collected from matched, quotasamplesof respondentsbymeans of structurednterviews. However, the latterstudyconcerneddrinkingand smok-ing amonghigh-schooland undergraduatetudentsin an Americancommunityand wasconducted in 1979-80.

    Respondents nd Sample DesignRespondentsfor the study were 120 unmarriedpersons comprising60 high-schoolstudentsand60 undergraduates.As in Study 1, samplescontainedequalnumbersof boysandgirls andrespondentswho were frommiddle-or-upper ersus those fromworking-or-lower class homes. The two sampleswere drawn,respectively,from a large, publichighschool and a major, state university that are located in a moderate-sized,Midwesterncommunity. Few non-whites or personsof Spanishorigin appear n this community, sothetwo sampleswere limited to white, Anglo respondents.High-school respondentsweresampled romthejuniorclass. Undergraduate espondentswere enrolled in an introducto-ry course in the social sciences.

    Instruments nd ProceduresThestandardizednterview scheduleused in Study2 drewquestionwordingfromStudy1, butquestionswere revised andpretestedwithrepresentatives f the two populations or

    which the schedulewas designed. The scheduledealt with various issues in the lives ofrespondents, ncludingbothdrinkingandsmoking, andprovidedbothlead questionsandappropriateprobes to be used if respondentshad difficulty answering. All questionsincluded were asked in closed format. Interviews averaged an hour in length, and re-sponses were markedby the intervieweron the scheduleusing precategorizedresponsealternatives.As in Study 1, respondentswere asked to indicatetheir self-referentidentity labels,norms, andpreferences or alcohol use. In addition,respondentswere to report heir owndrinkingbehaviorsas well as the drinkingbehaviorsand norms advocatedby referentothers. Similarquestionswere also asked concerningtobacco smoking.The questionused for generatingself-referent abels for drinkingread:AT THE PRE-SENTTIME DO YOU CONSIDERYOURSELFA NON-DRINKER,AN OCCASION-AL USER, A MODERATE USER, A HEAVY USER, AN ALCOHOLIC,OR ANALCOHOLADDICT?That used for smoking was phrased:AT THE PRESENT TIMEDO YOU CONSIDER YOURSELFA NON-SMOKER,AN OCCASIONALSMOKER,

  • 8/8/2019 smoking&drinking

    11/28

    168 THESOCIOLOGICALQUARTERLY ol. 26/No. 2/1985

    A MODERATESMOKER, A HEAVY SMOKER, A CHAIN SMOKER, OR A TO-BACCO ADDICT? Questions designed to assess respondents'preferencesread: ALLTHINGSCONSIDERED, HOW OFTEN DO (or WOULD) YOU LIKETO DRINK?,and ALL THINGS CONSIDERED, HOW OFTEN DO (or WOULD) YOU LIKE TOSMOKE?Those for norms were phrased:DO YOU HOLD A SPECIFICOPINIONABOUT HOW FREQUENTLYYOU SHOULD DRINK (and if so) WHAT IS THATOPINION?,and DO YOU HOLD A SPECIFICOPINIONABOUT HOW MUCHYOUSHOULD SMOKE (and if so) WHAT IS THAT OPINION?Questions asking aboutdrinkingbehaviorwere phrased:DO YOU DRINK ALCOHOLICBEVERAGES(andifso) DURING THE PAST YEAR HOW OFTEN DID YOU DRINK ON AVERAGE?Thosefor smokingincluded:DO YOU SMOKECIGARETTES andif so) HOWMUCHDO YOU REGULARLYSMOKE?Responsesto all of thesequestionswerecodedas six-category, Likert-type cales with low scores indicatingdislike, disapproval,or low use ofalcohol or tobacco.

    Study 2 generated nformationabout referentothers by asking respondents o discusssix, specific persons:mother, father, a specific sibling, a specific girlfriend,a specificboyfriend, and a teacher. To measure modeling behavior, respondentswere asked thefollowing questionsfor each referentother:DOES (yourreferentother)DRINK ALCO-HOLICBEVERAGES(andif so) DURINGTHE PAST YEAR HOWOFTENDID (yourreferentother) DRINK ON AVERAGE? and DOES YOUR (referentother) SMOKECIGARETTES andif so) DURING THE PAST YEAR HOWOFTENDID (yourrefer-ent other)SMOKE ON AVERAGE?Questionsdesignedto elicit attributednormsread:DOES (yourreferentother)HOLDAN OPINIONABOUT HOWFREQUENTLYYOUSHOULD DRINK (and if so) WHAT IS THAT OPINION?,and DOES (your referentother)HOLDA SPECIFICOPINIONABOUT HOW MUCHYOU SHOULD SMOKE(andif so) WHAT IS THAT OPINION?Answers to thesefourquestionswerealso codedas six-category,Likert-typescales with low scores indicatinglow use or disapprovalofthe use of alcoholor tobacco. Scores for "parents"were formedby addingresponsesformotherand father;scores for "peers" involved the summingof responsesfor girlfriendandboyfriend.

    DataAnalysisFindingsfrom bothstudiesare to be presentedby meansof pathdiagrams hatarebasedon two-stage regressionanalysesof product-moment orrelationsamongvariables.Thesetechniquespresumeinterval measurementand linearrelationshipsand are used here forconveniencein presenting indings. As indicatedabove, the basicdata for the two studieswere generatedby scales meeting ordinalassumptionsonly. Because assumptionshavebeen made that may have been inappropriateor our data, care was taken to check allreported indings with nonparametricechniques.Regressionanalysisalso requires hat one specify which variablesare to be considered

    exogenous ("causes") and which endogenous ("effects"). In our analyses we haveassumedthatparentalandpeernorms and behaviorswere generatorsof respondents'ownconceptions,and thatboth the formerand latterhadpotential mpactson behavior.Again,we return o these assumptionsbelow.

  • 8/8/2019 smoking&drinking

    12/28

    SocialInfluence,Self-ReferentdentityLabels, nd Behavior 169

    Table 1CorrelationsAmong Respondents'Own Expectations n Studies One and TworLabel.Preference rLabel-Norm rPreference.Norm

    Datafor StudyOneUnited States (Drinking) .56 .27 .47Australia(Drinking) .46 .30 .12Norway (Drinking) .57 .33 .43France(Drinking) .65 .47 .62

    Data or StudyTwoSecondaryStudentDrinking .68 .43 .54UndergraduateDrinking .57 .40 .47High School Smoking .88 .37 .51UndergraduateSmoking .96 .36 .34

    RESULTSPerhaps he first questionone would like answeredconcerningself-referent abels is, towhatextent arethese conceptionsdistinct from the more-oftenstudiedconceptualforms,preferences,and norms?Findingsbearingon thisquestionare availablefrom bothstudiesandaredisplayed n Table I whichprovidesthecorrelationswe foundamong conceptionsforeach of theeight sets of data.As can be seen, correlationsamongall conceptionsweregenerallypositive and of moderate size. Nevertheless, in all eight cases examined, thecorrelationbetween self-referent abels and preferenceswas greaterthan the correlationbetween labels andnorms or betweenpreferencesand norms(p < .01 by sign test). Thissuggeststhatself-referentabels andpreferenceshave greater endencytowardssimilarityfor these issues. In fact, correlations between labels and preferencesfor respondents'smokingwere more thanmoderate n size, being .88 for secondarystudentsand .96 forundergraduates.This means that for smoking (only) either variable accounted for themajorityof the variance of the other. These relationshipsare sufficiently strong as tosuggestthatself-referentabels andpreferences orsmokingarenotdistinguishedby someyoung persons. We returnto this issue in our discussion of results.

    HypothesisOneOurfirsthypothesisdeals with theetiology of self-referent abels. It statesthatbehaviormodeling by parentsand peers have significanteffects on young persons' self-referentlabelingfor drinkingand smokingthatare independentof one anotherand of normativeadvocacy by these two referentgroups. In contrast,normativeadvocacy by parentsandpeers have weak or nill impacton young persons' self-referent abelingfor these issues.

    Behavior modeling and normativeadvocacy by parentsand peers may or may not becorrelatedwithone another,so theproperway to establish the independent ffects of eachof these fourvariables s to control for the other three. In Study 1 this was accomplishedby conductingseparateregression analyses, for each country, in which the behaviors

  • 8/8/2019 smoking&drinking

    13/28

    170 THESOCIOLOGICALUARTERLYol. 26/No. 2/1985

    reportedand norms attributed o parentsand peers were examined simultaneouslyaspredictorsof self-referent abels, preferences,andrespondents'own norms. The results ofthese analyses arepresentedas partof a single pathdiagram n Figure 1. We have usedbold lines in this figure to indicatepathsthatattainedstatisticalsignificancein threeormorecountries.Regressioncoefficients thatwere not statisticallysignificant n any coun-try and theirpathshave been omittedfrom the figure.As can be seen, moderate but significantpaths were obtained linking reportedpeerbehaviorswith self-referent abels for drinkingin all four countries. In contrast,pathslinkingnormsattributedo peers with self-referent abels were either weak or nonsignifi-

    BehaviorsSA30**e p o r t e d A u . 3 0 * * ,Peerso-3o***

    4 0 Self-ReferentVls 4 Labels orSor*)

    ** DrinkingB e h a v i o r s A u

    Reportedsfor

    Fra2P a r e n t s . 2 4 *

    PreferencesforNorms USA.23** Drinking

    USA.4Own NormsA t t r i b u t e d

    Norms ra,5*"I.i

    t r b t d t

    Figure1. Path Coefficients StandardizedRegressionCoefficients)PredictingRespondents'OwnExpectationsor Drinking n Study1.Notes:* = .01

  • 8/8/2019 smoking&drinking

    14/28

    SocialInfluence,Self-ReferentdentityLabels, nd Behavior 171

    cant in each country. Moreover, in all four countries the regressioncoefficients obtainedfor reported peer behaviors were significantly greaterthan those obtained for normsattributed o peers.' Thus, our first hypothesiswas supported orpeers in all four coun-triesin Study 1. It was not, however, supported orparents in any of the fourcountries,sincereportedparentalbehaviorsand normsattributedo parentswere bothfound to haveno significanteffects on self-referent abels.Data from Study2 were subjectedto similarregressionanalyses, and results fromthelatteraredisplayedas partsof pathdiagrams n Figure2 (which concernsdrinking)andFigure3 (which concerns smoking). Within these two figureswe have used bold lines toindicatepathsthatattainedstatisticalsignificancefor bothhigh-schoolandundergraduate

    BehaviorsReportedor

    P e e r s U G. 4 8 * * *

    , Self-Referentr Labels orDrinkingBehaviorsR e p o r t e d

    P a r e n t s

    P r e f e r e n c e sforUG .25* DrinkingNormsAttributedoParents

    Ow n N o r m sU G . 5 2 * *

    NH . 5 1 * * * D r i n k i n gN o r m sA t t r i b u t e d

    P e e r s

    Figure2. Path Coefficients StandardizedRegressionCoefficients)PredictingRespondents'OwnExpectations or Drinking n Study2.Notes: * = .01

  • 8/8/2019 smoking&drinking

    15/28

    172 THESOCIOLOGICALUARTERLYol. 26/No. 2/1985

    B e h a v i o r sR e p o r t e d

    P e e r s UG7 3 * * .

    . Self-ReferentLabels orSmoking

    B e h a v i o r sR e p o r t e d

    P a r e n t s

    PreferencesforSmokingNorms

    A t t r i b u t e dP a r e n t s

    OwnNormsUG .48*** forNorms HS .20 (.12) SmokingAttributedoPeers

    Figure3. Path Coefficients StandardizedRegressionCoefficients)PredictingRespondents'OwnExpectations or Smoking n Study2.Notes: * = .01

  • 8/8/2019 smoking&drinking

    16/28

    SocialInfluence,Self-ReferentdentityLabels, ndBehavior 173

    preferencesfor drinking. Thus, the etiology of self-referentlabels and preferences issimilar, in part. We returnto this effect, also, in our discussion below. (Figure 1 alsoreportsotherfindingsconcerningthe etiologies of preferencesandnorms,some of whichreflect sharp, national differences. Detailed discussion of these lattermay be found inBank et al., forthcoming.)

    HypothesisTwoOursecond hypothesisconcerns the behavioraleffects of self-referent abels. It statesthat self-referentidentity labels for drinkingand smoking have significant effects onyoung persons' behaviorsthat are independentof the effects of otherconceptionsaboutthese issues. Thishypothesiswas tested in Study 1by conductingregressionanalyses, foreach country,in whichrespondents'own self-referentabels, preferences,norms,andthebehaviorsandnormsthey attributedo parentsandpeers were examinedsimultaneouslyfortheirindependent mpacton reporteddrinking requency.The resultsof these analysesarepresented n Figure4 as a single path diagram which, in effect, completesFigure 1).As before, boldlines are used to indicatepathsthat attained tatisticalsignificanceinthreeor more countries,and regressioncoefficients and theirpathshave been omittedif theydid not attainstatisticalsignificance.As canbe seen, moderatebutsignificantpathswere obtained inkingrespondents'self-referent abels with reporteddrinking n all four countries.These pathswere roughlyasstrongas those forrespondents'preferencesand were significantlystronger han those for

    respondents'normsin all four countries. Thus, self-referent abelingwas found to be anindependent eneratorof behavior n each samplestudied,andour secondhypothesiswasconfirmedin Study 1.Similaranalyseswere, again, conductedfor the high-schoolandundergraduate ata inStudy 2, and results for the latteranalyses are reportedin Figures 5 and 6 (which, ineffect, complete the pathdiagramsof Figures2 and 3). As can be seen, significant pathswere again obtained for the independenteffects of self-referent labels upon drinkingfrequency(Figure5) and smokingamount(Figure6). Moreover,pathsfor the effects ofself-referentlabels were comparable to those for respondents'preferences and weresignificantly greaterthan those of respondents'norms in each case. Thus, our secondhypothesiswas also supported or both samples and topics within Study 2.An interestingage effect is also observablewithinFigures5 and 6. As can be seen, theeffect of self-referent abels on reportedbehavioris greater or high-schoolstudentsthanfor undergraduatesor both smokinganddrinking,whereas the effects of preferencesonreportedbehavior is greater in both cases for undergraduates.These differences aregreaterfor smoking than for drinking,however, and only those for smoking achievedstatisticalsignificance. We also returnto this age effect in our discussion.

    DISCUSSIONOurresults suggest that self-referentidentity labels are, indeed, importantconceptionsamongyoung persons, at least for the topics of drinkingand smoking. Such labels werefoundto be relatively independentof respondents'preferencesandnorms, to have some-what differentetiologies than those of the other two conceptions studied, and to haveindependent ndsignificanteffects on reportedbehavior. These results were found to hold

  • 8/8/2019 smoking&drinking

    17/28

    BehaviorsReportedfor

    S e l f - R e f e r e n t

    LabelsAus.32***for

    Preferences Nor .18 (.08)rinking Fra .50**Behaviors

    ReAttributedo

    Parents

    A u . 3 2 O r *

    N o r m s D r i n k i n gA t t r i b u t e d

    P e e r s

    Figure4. Path Coefficients StandardizedRegressionCoefficients)PredictingReportedDrinkingFrequencies n Study 1.Notes: * = .01< p < .05** = .001

  • 8/8/2019 smoking&drinking

    18/28

    B e h a v i o r sReportedor

    P e e r s

    Self-ReferentLabels orDrinking

    Behaviors JOReportedorParentsUG 61***Preferences .4***Reportefor

    DrinkingNormsAttributedoParents

    O wn N o r m s

    N o r m sD r i n k i n g

    A t t r i b u t e dP e e r s

    Figure5. Path Coefficients StandardizedRegressionCoefficients)PredictingReportedDrinkingFrequency n Study2.Notes: * = .01< p< .05** = .001

  • 8/8/2019 smoking&drinking

    19/28

    BehaviorsReportedor

    P e e r s

    S e l f - R e f e r e n tL a b e l s toSmoking Is -JO**B e h a v i o r s i6H

    R e p o r t e d ParentsUG .80***Preferences HS .51***for

    NormsSmokingormsAttributedoParents

    OwnNormsforNormsSmokingonusAttributedoPeers

    Figure 6. Path Coefficients (Standardized Regression Coefficients) PredictingReported Smoking Amount in Study 2.Notes:* = .01

  • 8/8/2019 smoking&drinking

    20/28

    SocialInfluence,Self-ReferentdentityLabels, ndBehavior 177

    amongsamplesof respondentsn four, Westerncountries,for two topicsin behavior,andfor two age groups among American respondents.The strengthand breadthof theseresultscertainly mply support or those theoristswho have arguedfor the importanceofself-referentabelingandsuggest thattheoreticalmodels whichaspireto predictbehaviorfromvariousconceptual componentsshouldgenerallyincludesome type of self-labelingcomponent.At the same time, our hypothesesdid not anticipatethe relative differencesbetweenparentaland peer impact on self-referent abeling, nor did we anticipatethe interestingeffects of age andtopic revealed in resultsfromStudy2. These unexpectedfindingsbeardiscussion.

    Parentsand PeersOurhypothesesfailed to anticipate he relativelyweak influencesof parentsthatthesedatareveal. Considerable ontroversyhas appearedconcerningthe impactof parentsandpeers on youthbehavior, and we have arguedearlierthatin the United States (at least),peers probablyhave more impact throughmodeling whereas parentsare more likely toinfluencethrough he expressionof norms(Biddleet al., 1980a). Ourpreviousargumentwas basedin the likely impactsof peerbehaviorson adolescents'preferencesandparentalnormativeadvocacyon adolescents' own norms. Inthe lightof thedatawe havepresentedin this paper, it appears ikely that peer behaviorswill also affect young persons' self-referent abelingfor such issues as drinkingandsmoking.If so, andif self-referent abelshavethe additional mpacton behaviorthat our datasuggest, thenpeerinfluencefor thesetopics is even strongerthanwe (and some other investigators)have suggested.The reasonswhy young persons should identify with peers ratherthanparentsseemrelatively simple. Western youths today are held out of employment and are largelysegregated romadultsociety in academies, communes, andurbanghettoes whereyoungmen and women pursueactivities that are often antithetical o those they will exhibit at alaterage. Takentogether,these activities form a youthculturethatcapturesthe sympa-thies of many young people. Thus, whereasonce a youth had only to decide whethertocontinueto identifywith his or herparentswhen formingan adultsense of self, today'syouth must first withdrawfrom parental nfluence into the youth cultureand then (onepresumes)withdraw romthe latter n orderto enteradulthood.Drinkingandsmokingareoften a partof the youthculture, so it would be surprising f peers had less impactthanparentson youths' self-conceptsfor these topics. This does not meanthatparental nflu-ence will always be weak. Parentsshould have stronger mpact for children,for youthswhen the topic lies outside of the youth culture, and-paradoxically-when the youngpersonhas left the youth culture and entered adulthood.

    Topicand MaturityOur results also indicate differences in the ways in which young persons use self-referent abels dependingon their age and whether the topic is alcohol use or cigarette

    smoking.As we know from Table 1, respondentsdifferentiatedmore between their self-referent abels and theirpreferenceswhen the topic was drinking hanwhen they thoughtaboutsmoking. This suggests that the rhetoricwith which smoking is discussed is morelikely to be phrased n preferential erms alone whereas thatof drinking s morelikely tocontain normative or other components. Such a difference may reflect the fact that

  • 8/8/2019 smoking&drinking

    21/28

    178 THESOCIOLOGICALUARTERLYol. 26/No. 2/1985

    drinking s confined to recreationalcontexts whereassmokingis practiced n many con-texts andthroughout he day. Or, alternatively, t may be thatyoung personsare simplymore awareof the immediateconsequencesof alcoholabuse whereasthoseof tobacco useare thoughtto be delayed until mid-life or later.In addition, Figures 5 and 6 reveal that self-referentlabels had more impact amonghigh-school studentsthan among undergraduates,but this difference reached statisticalsignificanceonly for the topicof smoking.This suggeststhatself-referentabelsare moreinvolved in the initiation of smoking, but the continuationof smoking may be more amatter or preferences.We suspectthat this findingreflects the differingsocial organiza-tions of smokinganddrinkingamongyoung personsandpossibly the differingpotentialsof tobacco and alcohol for generatingaddiction.On the one hand, considerablepublicattentionhasrecentlybeen focused on smoking, andyoung personswho choose to smoketodaymust do so in defiance of widespreadbeliefs thatsmokingis addictive andbad forone's health. This means that smokingis an act of rebellion,and many adolescentswhochoose to smoke(or not to smoke)do so to win solidaritywithpeers. The youngadult, incontrast, learns that adult smoking is not associated with sociality and may have hadpersonalexperiences with the addictive propertyof tobacco. On the other hand, littlepublic attentionis presently focused on drinking in the United States, most personsbelieve thatmoderatedrinkingposes few long-termproblems,morethan90 percentof alladultsdrink,and most alcohol use is social. This meansthatdrinkingamongadolescentsis less a matter of rebellionthan of conformityto adult mores, and alcohol use is lesslikely to be perceived or experienced as addictiveby young adults.

    ImplicationsOurfindingshave both theoreticalandpractical mplications.Among the former,theysuggestthatmany contemporaryheoriesin social psychologyareweakenedby lackof anidentity component. One finds little evidence of interestin identity processes in attitudetheory, for example, in attribution esearch, in most investigationsof social influenceorstudiesof role conflict. Withinthis paperwe have suggestedvariouspropositionsaboutthe etiology and effects of self-referent labels. Whereasour findings providea test foronly some of those propositions,they imply thatself-referent abels areamongthe more

    significantdeterminantsof behavior and deserve an expandedresearcheffort.How much power does one gain by addingself-referent abels to an existing model?This questionis easy to answer for the datapresented n this paper. In previouspublica-tions we have advocated a model for predictingthe behavior of young persons thatcombines normsattributed ndbehaviorsreported orparentsandpeerswithrespondents'own preferencesand norms but ignoredrespondents'self-concepts.Table2 providesdataindicating he variance hat this simplermodelexplainsin reportedbehavior,the variancegainedby addingself-referent abels, and the varianceexplainedby the new model. Ascanbe seen, althoughthe simplermodel is good, the new model is significantlybetter neach case considered.Enthusiasm for the efficiency of our new model should be tempered with caution,however. As readerswill have noted, our data were collected fromresponding ndividualsin single waves of datacollection. Reportsgiven by ourrespondentsof the norms advo-cated and behaviors modeled by parentsand peers should not be confused with data onthose topics that might have been obtained from such others, nor should resoondents'

  • 8/8/2019 smoking&drinking

    22/28

    SocialInfluence,Self-ReferentdentityLabels, ndBehavior 179

    Table 2Variances in ReportedBehavior Explainedby ModelWith and WithoutSelf-Referent Labels

    Variances Variances VariancesExplained by Gained by Explained byModel Without Adding Model IncludingS-R Label S-R Label S-R Label

    Data for StudyOneUnited States (Drinking) .60 .08*** .68Australia(Drinking) .58 .14"** .72Norway (Drinking) .62 .11*** .73France(Drinking) .52 .07*** .59Data for StudyTwo

    SecondaryStudentDrinking .66 .04* .70UndergraduateDrinking .68 .04"* .72High School Smoking .82 .07*** .89UndergraduateSmoking .98 .003** .98

    Notes: * = .01

  • 8/8/2019 smoking&drinking

    23/28

    180 THESOCIOLOGICALUARTERLYol. 26/No. 2/1985

    with techniques for measuringself-referentlabels may wonder whether they are notsimply anotherway of measuringreportedbehavior. In fact, measuresof self-referentlabelsare not always closely associatedwith reportedbehavior.Indeed,as Figures4, 5,and 6 suggest, we found preferencesmore highly correlatedwith reportedbehaviors inseveral instances. The reason for this is that self-labelingcan representother sources ofinfluence as well as the experience of behaving, and a person may choose to followpreferences, norms, or other sources of influence when taking action rather than toconfirmhis or her self-referent abels.Ourfindingsalso imply practicalconclusions. Most of these reflect the fact that self-referent abeling has often been ignoredby those who discuss reasons for drinkingandsmokingby young persons. Programsdesignedto affect rates of substanceabuseamongyouthareusuallyfocused on changingthosepersons'preferences "attitudes") or normsin the assumption hatsuch changes will also inducecorresponding hifts in behavior.Ifself-referent abels are also potent generatorsof conduct, then programsfor modifyingbehaviormightprofitablyfocus on the latter. Variousstudies (e.g., Miller et al., 1975)have shown that behaviors of school pupils may be shifted more easily by strategiesdesignedto change self-conceptsthanthroughother influencestrategies,andmodificationof identities is clearly one of the primarygoals of counseling. More thoughtshould begiven to effective ways of modifying the identitiesof youths with regardto drinking,smoking, and the use of otherquestionablesubstances.

    The Paucityof ResearchGiventhe strengthof ourfindingsandthe fact thatvarioustheoristshaveargued ortheimportanceof self-referent abels, one wonderswhy so little researchhas yet appearedconcerning hem. Ourguess is that this relativelack reflectsboth historicalaccidentsandintrinsicprocesses. Among the former, many advocates for the importanceof identityprocesseshave been Symbolic Interactionistsor other authorswho have been criticalofthequantitative esearchmethodsneeded to disentangle heimpactof norms,preferences,and self-referentlabels. In addition, gross confusion seems to have arisen concerningrelationshipsbetween identity, the self-concept, and self-esteem, and most research onThe Self to date has used instrumentsdesigned to measure esteem only (Wylie, 1979).Suchreasonsareaccidental,butwe suspectthat intrinsicreasons arealso partlyrespon-sible for avoidanceof self-referent abels by researchers.These latterare associated withthe "obviousness" of the belief-behaviorrelationship.What could be more straightfor-wardthanthe assumptionthat beliefs aboutone's own behavior are closely tied to thelatter?Onepresumablyhasall theevidence one needsto base beliefs aboutoneself on pastbehaviors, and unless the context changes, those beliefs should be a good predictorofone's behavior in the future.Indeed, some social psychologistsassume that self-referentbeliefs are suchpotentpredictorsof behavior hatthey subsumeall otherconceptualforms(Fishbeinand Ajzen, 1975). Moreover, self-referentbeliefs are not assumed by somecognitivepsychologists to motivateconduct. Preferencesmotivate, so do norms, and sodo beliefs about the likely conduct of others, but self-referentbeliefs are assumed to bemere statementsof subjectiveprobabilityand to have few motivating properties.Ourfindingssuggestthatthese assumptionsarein error.Norms, preferences,and self-referent abels appear o be generatedby somewhat differentexperiences. Moreover, wehave found that self-referent labels have associationswith behaviorthat are strongand

  • 8/8/2019 smoking&drinking

    24/28

    SocialInfluence,Self-ReferentdentityLabels, ndBehavior 181

    independentof those we found for preferences, norms, and otherconceptions. In short,self-referentabelsshouldalsobe treatedas conceptionsthatcananddo motivateconduct.Identityprocessesdeservea lot more researchamongsocial psychologistsconcernedwithpredictingbehavior.

    ACKNOWLEDGMENTSThis research was supported, in part, throughgrants (in the United States) from theNationalInstituteof Alcohol Abuseand Alcoholism(AA 00642) andtheNationalScienceFoundation INT76-15639). Additional resourceswere providedby the Psychology De-partment,Universityof Newcastle and the EducationResearchUnit, AustralianNationalUniversity (Australia);the National Institute for Alcohol Research (Norway); and theCenter For Research in Social Behavior, University of Missouri-Columbia U.S.A.).Thanks are due to Diane Chappellfor drawingthe figuresfor this article andto TerryBrown and PatriciaShanksfor typing its several versions.An earlier version of this article was presentedat the Annual Meeting of the SocialPsychology Section of the AustralianPsychologicalSociety, Sydney, May, 1983.

    NOTE1. Tests for the significanceof differencesbetweenregressioncoefficients seem not to be familiarto manysocial scientists today. Marlin and Biddle (1983) provide a discussionof these tests andtheiruses.

    REFERENCESAlexander,C. Norman,Jr., and Mary Glenn Wiley1981 "Situatedactivity and identity formation." Pp. 269-89 in M. RosenbergandR. H. Turner(eds.),Social Psychology: Sociological Perspectives.New York:Basic Books.Andrews,KennethandDenise B. Kandel1979 "Attitude and behavior: a specification of the contingentconsistency hypothesis." American So-

    ciological Review 44:298-310.Aronson, Elliot1969 "The theory of cognitive dissonance: a currentperspective." Pp. 1-34 in L. Berkowitz (ed.),Advances in ExperimentalSocial Psychology, Vol. 4. New York: AcademicPress.Backman, Carland Paul Secord1968 "The self androle selection." Pp. 289-96 in C. GordonandK. J. Gergen(eds.), The Self in SocialInteraction.New York:Wiley.Backman,Carl, Paul Secord, and JerryPierce1963 "Resistance to change in the self-concept as a function of consensus among significantothers."

    Sociometry 26:102-11.Bandura,Albert1977 Social LearningTheory. EnglewoodCliffs, NJ: Prentice-Hall.1978 "The self system in reciprocaldeterminism." AmericanPsychologist 33:344-58.Bank, BarbaraJ., Bruce J. Biddle, Don S. Anderson,RagnarHauge, DaphneM. Keats, John A. Keats,MarjorieM. Marlin, and Simone ValantinForth- "Comparative research on the social determinantsof adolescent drinking." Social Psychology

    com- Quarterly.ingBem, Daryl J.1972 "Self-perception heory." Pp. 1-62 in L. Berkowitz(ed.), Advancesin ExperimentalSocial Psychol-

    ogy, Vol. 6. New York: Academic Press.

  • 8/8/2019 smoking&drinking

    25/28

    182 THESOCIOLOGICALUARTERLYol. 26/No. 2/1985

    Bem, DarylJ. and AndreaAllen1974 "On predictingsome of the people some of the time:the searchfor cross-situational onsistenciesinbehavior." PsychologicalReview 81:506-19.Biddle, Bruce J.1979 Role Theory: Expectations, Identities, andBehaviors. New York: Academic Press.Biddle, Bruce J., BarbaraJ. Bank, Don S. Anderson,JohnA. Keats, and DaphneM. Keats1981 "The structureof idleness: in-school and dropoutadolescentactivities in America and Australia."Sociology of Education54:106-19.Biddle, Bruce J., BarbaraJ. Bank, and MarjorieM. Marlin1980a "Parentaland peer influenceon adolescents." Social Forces 58:1057-79.1980b "Social determinantsof adolescentdrinking:what they think, what they do, and what I thinkanddo." Journalof Studies on Alcohol 41:215-41.

    Blane, HowardT. and Linda E. Hewitt1977 Alcohol and Youth: An Analysis of the Literature:1960-1975. Springfield,VA: NationalTechnicalInformationService.Burke, PeterJ. and Donald C. Reitzes1981 "The link between identityandrole performance."Social Psychology Quarterly44:83-92.Burke, PeterJ. andJudye Tully1977 "The measurementof role identity." Social Forces 55:881-97.Carmines,EdwardG.1978 "Psychological origins of adolescent political attitudes:self-esteem, political salience and politicalinvolvement." American Politics Quarterly6:167-86.Chassin, LaurieForth- "Adolescent substance use and abuse." In P. Karoly and J. Steffen (eds.), Advances in Childcom- Behavioral Analysis and Therapy: Vol. 3, Adolescent Behavior Disorders:CurrentPerspectives.

    ing Lexington, MA: D. C. Heath.deCharms,Richard1968 PersonalCausation. New York:Academic Press.DeJong, William1979 "An examination of self-perceptionmediation of the Foot-In-The-Dooreffect." Journalof Person-ality and Social Psychology 37:2221-39.1981 "Consensus informationand the Foot-In-The-Door ffect." PersonalityandSocial Psychology Bul-letin 7:423-30.Dipboye, Robert1977 "Alternativeapproaches o deindividuation." Psychological Bulletin 84:1057-75.Erikson,Erik H.1968 Identity,Youth and Crises. New York: Norton.Erikson,Kai T.1962 "Notes on the sociology of deviance." Social Problems 9:307-14.Fazio, Russell H. andJoel Cooper1983 "Arousal in the dissonanceprocess." Pp. 122-52 in J. T. Cacioppoand R. E. Petty (eds.), Social

    Psychophysiology. New York:Guilford Press.Festinger,Leon1957 A Theoryof CognitiveDissonance. Evanston, IL: Row, Peterson.Fishbein,Martinand Isaac Ajzen1975 Belief, Attitude, Intention, and Behavior: An Introductiono Theory and Research. Reading, MA:

    Addison-Wesley.Flay, BrianR., Josie R. d'Avernas, J. Allan Best, MaryW. Kersell, and KatherineB. Ryan1982 "Cigarettesmoking: why young people do it and ways of preventingit." In P. Firestone and P.McGrath eds.), PediatricBehavioral Medicine. New York: Springer-Verlag.Foote, Nelson1951 "Identification as the basis for a theory of motivation." American Sociological Review26:14-21.Freedman,JonathanL. and Scott C. Fraser

    1966 "Compliancewithoutpressure: he Foot-In-The-Door echnique." Journalof Personalityand SocialPsychology 4:195-202.

  • 8/8/2019 smoking&drinking

    26/28

    SocialInfluence,Self-ReferentdentityLabels, ndBehavior 183

    Gabrenya,William K.1979 The Effects of Altercasting:Self-concept, Mood, Liking, andBehavioralExpectancies.Ph.D. disser-tation, Departmentof Psychology, Universityof Missouri-Columbia.Gecas, Victor1982 "The self-concept." Annual Review of Sociology 8:1-33.

    Gergen, Kenneth J.1965 "Interactiongoals andpersonalisticfeedbackas factorsaffectingthe presentation f self." JournalofPersonalityand Social Psychology 1:413-24.Gove, Walter R.1975 "The labellingperspective:anoverview." Pp. 3-20 in W. R. Gove (ed.), TheLabellingof Deviance:Evaluatinga Perspective.New York: Halstead Press.Greenwald,AnthonyG.1980 "The totalitarian ego: fabrication and revision of personal history." American Psychologist35:603-18.

    Hauge, Ragnar1977 "Ungdoms holdning til bruk og misbruk av alkohol." ("Young people's attitudestoward use andabuse of alcohol.") Tidsskrift om Edruskapssplrsmil29:32-3.1978 "Drikkepressblantungdom:for ventignerom avhold og bruk." ("Drinking pressureamong youth:expectations or abstinenceandindulgence.")Tidsskriftom Edruskapssporsml30(2):33-5, 30(3):4-5, 20-1, 30(4):10-1.Heider, Fritz1958 The Psychology of InterpersonalRelations. New York:Wiley.Heiss, Jerold1968 Family Roles and Interaction.Chicago: RandMcNally.Jensen, Roger and Shirley Moore1977 "The effect of attributestatements on cooperativenessand competitivenessin school-age boys."Child Development48:305-7.Jones, StephenC.1973 "Self and interpersonalevaluations: esteem theories versus consistency theories." PsychologicalBulletin 79:185-99.Kahle, Lynn and John Berman1979 "Attitudescause behaviors: a cross-lagged panel analysis." Journalof Personalityand Social Psy-chology 37:315-21.Kandel, Denise B.1978 LongitudinalResearch on Drug Use: EmpiricalFindings and Methodological Issues. New York:Wiley.1980 "Drug anddrinkingbehavioramong youth." Annual Review of Sociology 6:235-85.Kaplan, Howard1976 "Self-attitudesanddeviant response." Social Forces 54:788-801.1980 "Self-esteemandself-derogationheoryof drugabuse." Pp. 128-31 in D. J. Letterei,M. Sayers,andH. W. Person (eds.), Theories on Drug Abuse: Selected ContemporaryPerspectives. Washington,DC: NIDA Monograph#301.Keats,JohnA., Bruce J. Biddle, DaphneM. Keats, Barbara . Bank, RagnarHauge, WanRafaei, and SimoneValantin

    1983 "Parents, friends, siblings, and adults:unfolding referent other importancedata for adolescents."International ournalof Psychology 18:239-62.Kelley, Harold H.1955 "Salience of membershipand resistance to change of group-inducedattitudes." HumanRelations8:275-89.Kelman, Herbert1958 "Compliance, identification, and internalization: hree processes of opinion change." Journal ofConflictResolution2:51-60.Korman,AbrahamK.

    1966 "Self-esteem variable in vocational choice." Journalof AppliedPsychology 50:479-86.1967 "Self-esteem as a moderator of the relationshipbetween self-perceived abilities and vocationalchoice." Journalof Applied Psychology 51:65-7.

  • 8/8/2019 smoking&drinking

    27/28

    184 THE SOCIOLOGICALQUARTERLYVol. 26/No. 2/1985

    Kraut,Robert E.1973 "Effects of social labelling in giving to charity." Journal of ExperimentalSocial Psychology9:551-62.Kuhn, ManfordH. and Thomas S. McPartland1954 "An empiricalinvestigationof self-attitudes." AmericanSociological Review 19:68-76.Latand,Bibb1966 "Studies in social comparison." Journalof ExperimentalSocial Psychology, Supplement1.Lefcourt,HerbertM.1976 Locus of Control: CurrentTrendsin Theoryand Research. Hillsdale, NJ: LawrenceErlbaum.Leonard, Russell, W. Bruce Walsh, and Samuel H. Osipow1973 "Self-esteem, self-consistency, and second occupationalchoice." Journalof Counseling Psychology20:91-3.Liccione, WilliamJ.1977 The Nature of Adolescent Drinking. Ph.D. dissertation,Departmentof Sociology, University ofMissouri-Columbia.McArthur,Leslie A., CharlesKiesler, and BarryP. Cook1969 "Acting on an attitudeas a function of self-perceptandinequity." Journalof PersonalityandSocial

    Psychology 12:295-302.McCall, George and J. L. Simmons1966 Identities and Interactions.New York:Free Press.Marlin, MarjorieM. and Bruce J. Biddle1983 "LISREL,credulityanddifferences betweenregressioncoefficients." Unpublishedmanuscript.Cen-ter for Researchin Social Behavior, Universityof Missouri-Columbia.Merton,Robert K.1949 Social Theoryand Social Structure:Towarda Codificationof TheoryandResearch.Glencoe, IL:TheFree Press.Miller, Richard,PhillipBrickman,and Diana Bolen1975 "Attributionversuspersuasionas means for modifying behavior." Journalof PersonalityandSocial

    Psychology 31:430-41.Miyamoto, S. Frankand SanfordM. Dornbusch1956 "A testof the symbolic interactionisthypothesisof self-conception."AmericanJournalof Sociology61:399-403.Morse, Stan and KennethJ. Gergen1970 "Social comparison, self-consistency, and the concept of self." Journalof Personalityand Social

    Psychology 16:148-56.Parsons,Talcott1951 The Social System. New York:The Free Press.Pettigrew,Thomas1967 "Social evaluation theory: convergences and applications." NebraskaSymposium on Motivation15:241-311.Phares,E. Jerry1976 Locus of Controlin Personality.Morristown,NJ: GeneralLearningPress.Purkey,William W.1970 Self Conceptand School Achievement. EnglewoodCliffs, NJ: Prentice-Hall.Reckless, Walter,Simon Dinitz, and Ellen Murray1956 "Self-concept as an insulatoragainst deviancy." AmericanSociological Review 21:744-46.Reeder, Leo, George Donohue, and ArturoBiblarz1960 "Conceptions of self and others." American Journalof Sociology 66:153-59.Rittle, Robert H.1981 "Changes in helping behavior:self- versus situationalperceptionsas mediatorsof the Foot-In-The-Door effect." Personalityand Social Psychology Bulletin7:431-37.Rosenberg,Florence and Morris Rosenberg1978 "Self-esteem anddelinquency." Journalof Youth and Adolescence 7:279-94.Rosenberg,Morris1954- "Some determinantsof political apathy." PublicOpinionQuarterly18:349-66.1955

  • 8/8/2019 smoking&drinking

    28/28

    SocialInfluence,Self-ReferentdentityLabels, nd Behavior 185

    1979 Conceiving the Self. New York: Basic Books.1981 "The self-concept:social productandsocial force." Pp. 593-624 in M. RosenbergandR. H. Turner(eds.), Social Psychology: Social Perspectives.New York: Basic Books.Rotter,Julian1966 "Generalized expectancies for internal versus external control of reinforcement." PsychologicalMonographs80.

    Ryle, Gilbert1949 The Conceptof Mind. London:Hutchinson.Saltzer, Eleanor B.1981 "Cognitive moderatorsof the relationshipbetween behavioral ntentionsand behavior." JournalofPersonalityand Social Psychology 41:260-71.Scheff, Thomas J.1966 Being MentallyIll: A Sociological Theory. Chicago: Aldine.Sherwood,John J.1965 "Self-identityand referent others." Sociometry28:66-81.1967 "Increasedself-evaluation as a function of ambiguousevaluations by referent others." Sociometry30:404-9.

    Shrauger,J. Sydney1975 "Responses to evaluations as a function of initial self-perceptions." Psychological Bulletin82:581-96.Shrauger,J. Sydney and Thomas J. Schoeneman1979 "Symbolic interactionist view of self-concept: throughthe looking glass darkly." PsychologicalBulletin 86:549-73.Sniderman,Paul M.1975 Personalityand Democratic Politics. Berkeley: Universityof CaliforniaPress.Steele, ClaudeM.1975 "Name-callingand compliance." Journalof Personalityand Social Psychology 31:361-69.Stone, GregoryP.1962 "Appearanceand the self." Pp. 86-118 in A. Rose (ed.), Human Behavior and Social Processes.Boston: HoughtonMifflin.Strenta,Angelo and WilliamDeJong1981 "The effect of a prosocial label on helping behavior." Social Psychology Quarterly44:142-47.Stryker,Sheldon1968 "Identity salience and role performance:the relevance of symbolic interactiontheory for familyresearch." Journalof Marriageandthe Family 30:558-64.1980 Symbolic Interactionism:A Social StructuralVersion. Palo Alto, CA: Benjamin/Cummings.Stryker,Sheldonand RichardT. Serpe1982 "Commitment,identitysalience, androle behavior: heoryandresearchexample." Pp. 199-218 inW. Ickes and E. S. Knowles (eds.), Personality,Roles, andSocial Behavior. New York:Springer-Verlag.Turner,R. H.1971 "Deviance avowal as neutralizationof commitment." Social Forces 19:308-21.Videbeck, Richard1960 "Self-conceptionsand the reactions of others." Sociometry23:351-59.Weinstein,Eugene and Paul Deutschberger1963 "Some dimensions of altercasting." Sociometry26:454-66.Wells, L. Edward1978 "Theoriesof deviance and the self-concept." Social Psychology Quarterly41:189-204.Wicklund,RobertA. and Peter M. Gollwitzer1982 Symbolic Self-Completion.Hillsdale, NJ: Lawrence Erlbaum.Wylie, Ruth1979 The Self-Concept: Theory and Research on Selected Topics. Lincoln: NebraskaUniversity Press.