Elaboration Likelihood Model(1)

Embed Size (px)

Citation preview

  • 7/29/2019 Elaboration Likelihood Model(1)

    1/18

    ElaborationLikelihood Model

    he elaboration likclihood rnodel (ELM) of persuasion is an approachI d.u"lopcd by Richard Pctry lohn Cacioppo, and their associates (the-ort .o-p.hensive single rreatment of the ELM is provided by Petty &Cacioppo, 198a; for a briefr recent presentation, see Petty & Wegener,f 99l j. The ELM suggests that i'rporta't variations in the nature of per-suasion are a function of the likelihood that receivers rvill engage in elabo-

  • 7/29/2019 Elaboration Likelihood Model(1)

    2/18

    I38 :N PERSUASION(HSM; see Chaiken, 1987; S' Chen & !fr1\en' 1999)' Althor"rgh the tu'omodels differ ir-r r"rrrt 'niiJ*;;;"y:' the ELM and HSM share the broadidea that persuasion tu"t "tnittedthrough two generai avenLles (var'vingi;,lre #""nt olcareful thinking involved)'N vRRlRrloNs lN THE DEGREE oF ELABoRATIoN:* #uiioi visu prnlpHrnel RourEs ro PERsuAsloNThe Nature of Elaboration

    TheELMisbasedontheideathattrndcrdifferentconditiorrs,receir'ersu'ill vary in the degrce ililil;;u" titttty to engage in elaboration ofinformation relevant to the persuasive ""tt' Ry,"tt*":11T-":S^i::};r;;"r"ru1, in issue_reier,ant thinkine. Thus somctimes rcc(rlversrvill engage in .*t.'-"'* i"t"-'"ltt'ont thinkinlg: They u'ill atten{ closel'v toa presente d me ssage' t;;t;;ll)';tttttinize tl-re "I'go-ttl" it contains' reflecton other issu.-,.ltuu'-'t tonsierations (e'g'' other argLrm:nts i::11.:O tt"-memory or argllnle nt' tf-'ty atui";' *t'.:." But sometimes re ceivers rvillnot r.rndertak. ro -t"n^i-'e-relevant thinking; no one tT:li:-tl t-*cffbrt for el'erv persuasive topic or messagc' and hence sometimes re celversrvill clisplav relatitell' little claborattott"A numbe r or n-r.u,', "=". ..r, deve loped for assessing 'ariatio's i. thedegree of eiaboratio"li-'"'ottuts in a given circumstance (fbr discussion'see Petry s. cu.iopit';''"a;; z +z) Perhaps tire most straight-fbru,ard of these ir rJ'rn".,gf-,i-fir',i"g technique: Immediately follorvingthe receipt of a persuasive [ressage' receivers are simply asked to list thethoughts tl-t"t o..ttt"Jio'tl"- cliri'g the communicatio' (br a more de -Hatf,ins' & Pet6' 198r' pp' 38-47;for a

  • 7/29/2019 Elaboration Likelihood Model(1)

    3/18

    Elaboraton Lkelhood Model NT I 39think that in circumstances in r,vhich little or no elaboration occurs) little orno persuasion'uvill occur (after all, the receiver has not really engaged the-.irug" a'd has not undertake' much issue-relevant thinking). But theELM iuggests that persuasion can take place at any point along the elabo-ration cJitinuum-although the nature of the persuasiotr processe s lviil bedifrent as the degree of elaboration varies. To bring out the differer-rce s inthese persuasion processes, the ELM oflers a broad distinction befiveenftrro ,ot.. to persuasicln: a central and a peripheral route'Central and Peripheral Routes to Persuasion

    The central roLrte to persuasion represents the persuasion proces.ses itl-volr,ed u,hen elaboration is relatively high. Wren pcrsuasion is achievedthrough the central route ) it commonly comes about through extensiveissue-ielevant thinking: careful examination of the informatiou containedi' the message, close sirutiny of the message's arguments) consideration ofothcr issue -relevant material (e.g., arguments recalle d from memory' arglr-ments devised by the receiver), and so on. In short, persuasion through thecentral route is achieved through the receiver's thoughtftil examination ofissue-relcvant cotrsidcratious.The peripheral route represents the persuasion processes involved lvhenelaboraiion is relati'ely loi* Whe. persuasion is achieved through periph-eral rolltes) it commonly comes about because the receiver employs somesirnple decision rule (some heuristic principle ) to evaluate the ad'ocated

  • 7/29/2019 Elaboration Likelihood Model(1)

    4/18

    I40 N5 PERSUASIONThis distinction between the tlvo routes to persuasion should not bepermitted to obscure the underlying elaboration continuum. The centraland peripheral routes to persuasion are not trvo exhaustive and mutuallyexclusiv caregories or kinds of persuasion (cf. Stiff, 198); they sirnply

    represent prototypical extremes on the high-to-lolv elaboration contin-uum (for an illuminating discussion, see Pettl', Cacioppo, Kasmer, &F{augtvedt, 1987; Petty, I(asmer, l{augtvedt, & Cacioppo,I9BT; Stiff &Bostr, Ig87).The ELM recognizes, for example, that at moderate levelsof elaboration, persuasion involves a mixture of central route and periph-eral roLrte proce;ses, with correspondingly complex patterns of effects (see ,e.g., Petq, & Cacioppo, 198a, pp.206-207; Petty & Wegener, 1999,pp. ++ +s1. Thus in considering the differing character of persuasionainieu.a through cenrral and peripherai routes) it is important to bear inmind that these routes are offered as convenient idealized cases represent-ing diffrent points on the elaboration continuum.A useful illustration of the distinction betrveen central and peripheralroutes to persuasion is provided by Petq', Cacioppo, and Goldman's( l9s 1 ) study of the effects of argument strength and communicator exper-tise on persuasive effectiveness. In this investigation, the personal relevanceof the message topic fbr receivers u'as varied, such that for some receiversthe topic \\,as quite relevant (and so prcsumably disposed receivers to en-gage in high elboration), whereas for orher receivers the topic rvas muchLri r.l.u"t-tt (and hence these receivers would presumabl.v be less likely toengage in elaboration). The design also varied the qr"rality of the message 's"rgo..rt, (strong vs. weak arguments) and the expertise of the communi-cator (high vs. lorv).

  • 7/29/2019 Elaboration Likelihood Model(1)

    5/18

    Elaboraton Llelhood Model N l4lstlrdy; moreover) as r'vill be seen, the consequences of persuasion. are notidentical for the t\vo routes. It thus becomes important to consider whatfactors influence the dcgree of elaboration that receivers are likely to un-..t"L., becar.rse with vriations in elaboration, differe't sorts of persua-sion processes are engaged.N rncrons AFFEcTING THE DEGREE oF ELABoRATIoNTrvo broad classes of factors influence the degree of elaboration that areceiver li,iil likely undertake in any given circumstance. one concerns thereceiver,s motivation for engaging in elaboration, the other the receiver'sability to engage in such elaboration. For extensil'e elaboration to occur,Uotn aUitlty in motivation must be present. High elaboration r'vill not oc-cur if the receiver is motivated to undertake issue-relevant thinking but isunable to do so, nor rvill it occur if the receiver is able to engage in elabora-tion but is unmotivated to do so'Factors Affecting Elaboration Motivation

    A variety of fctors have received research attention as influences onreceil,ers, motivation to engage in issue-relevant thinking, inciuding thereceiver's mood (e.g., BleJs, Bohner, Schrvarz, & Strack' 1990; Bless'Mackie, & Schwarz, 1992; Bohner' Crow, Erb, & Schrvarz' J'992; lVorth& Mackie, ).987; but also see Bless & Schrvarz, I999);the receiver's attitu-dinal ambivaience (that is, the degree to which the attitude is based on amixture ofpositive and ncgative eliments; e'g', Maio, Bell, & Esses' 199;Maio, Esses, & Be 11, 2000! and the presence of multiple sources with mul-Harkins & Pitty, 1987;D' J' Moore & Reardon'

  • 7/29/2019 Elaboration Likelihood Model(1)

    6/18

    ID N\ PERSUASIONearlier studies ofthe effect oftopic relevance variations on persuasive pro-cesses, researchers commonly employed two message topics, one presLlm-ably qirite relevant for the population from rvhich receivers were draw. ando., ,-toa so relevant. This obviously creates difficulties in interpretingexperimenral results because any observed differences between high- andlow-relevance conditions might be due not to the relevance differences butto some ctor connected to the topic differences (e.g., the necessarily dif-ferent arguments used in the messages on the two topics)'The piocedure follorved by ELM researchers is exemplified in a study bypetty and cacioppo (lg79b). Receivers in this investigation were collegeundergraduar.rf. persuasive messages adt'ocated the adoption of seuior.o-pr-.h.nrive examinations as a graduation requirement-either at the,ec.iv.rs' college (the high-relevance condition) or at a different, distantcollege (the lorv-relevance condition). With this form of manipulation, re-ceiveis in parallel high- and low-relevance conditions could hear messagesidentical in every respect (e.g., with the same arguments and evidence) savefor the name of- the college involved, thus simpli{,ing interpretation ofexperimental findings.^d rrot. abogt terminology: In ELM research reports, these variations inpersonal relevance have often been labeled as variations in the receiver'sievel of ,,involvement" r,vith the message topic (and so, for instance, in thehigh-relevance condition, receivers r,vould be said to be "highly involved"-iih ,h. topic). But in persuasion research, the term "involvement" hasalso been used to cover othervariations in the sort of relationship that mes-sage recipients have to the topic of advocacl', including the person's judg-ofih. importance of the issue, the degree to u'hich the person is

  • 7/29/2019 Elaboration Likelihood Model(1)

    7/18

    Elaboraton Lkelhood Model N 143whereas others are not. Need-for-cognition scales have bcen deveioped toassess this individual difference (e.g., Cacioppo & Petq', 1982). Personshigh in need for cognition tend to agree lvith statements such as "I reall,venjoy x task that involves coming up with neu'solutions to problems" and"I like to have the responsibility of handiing a situation that requires a lot ofthinking," r,vhe re as individuals lorv in ne ed br cognition are more likely toagree rvith statements such as "I like tasks that require iittle thought onceI've iearned them" and "I think onlv as hard as I have to." (For a generalrevierv of research on nee d br cognition, see Cacioppo, Petty', F'einste in, &Jarvis, I996.)As one might suppose, a good deal of research suggests that need forcognition inf'luences elaboration likelihood. Persons high in need for cog-nition are likely to report a larger number of issue-relevant thoughts (bl-lor,ving message exposure) than are persons lorv in need for cognition (e.9.,S. M. Smith, F{augtvedt, & Petq', 1994 br a revierv, see Cacioppo, Petty,etal., 1996, pp. 230-23I). Relatedlr', those high in need for cognitionare more influenced by the quality of the message's arguments than arethose lorv in need for cognition (e.g., Axsom, Yates, & Chaiken, 1987;Cacioppo, Petty', I(ao, & Rodrigue z, 1986; Cacioppo, Pettl', & Morris,1983; Haugtvedt, Petty, Cacioppo, & Steidle)', 1988; for a revieu', seeCacioppo, Pettl', et al., 199, pp.229-230).5 Such findings, of course , areconsistent rvith the supposition tl-rat persons high in need for cognitionhave generaily greater motivation fbr cngaging in issue-relevant thinkingthan do persons lorv in need for cognition.6Factors Affectng Elaboration Ability

  • 7/29/2019 Elaboration Likelihood Model(1)

    8/18

    144 N PERSUA5IONto time on a screen in front ofthem (for a general discussion of such manip-ulations, see Petty & Brock, 1981).The theoretical importance of distraction effects to the ELM should beplain. Under conditions thatrvould otherlvise produce relatively high elab-tr"don, distraction should interfere r,vith such issue-relevant thinking.Such interference should enhance persuasion in some circr-rmstances andreduce it in others. Specifically, if a receiver rvould ordinarily be inclined toengage in favorable elaboration (that is, to predominantly have thoughtsfu.|"g the advocated position), then distraction, by interfering with suchelaboraiion, rvould prsumably reduce persuasive effectiveness. But if areceiver rvould ordinarily be inclined to predominantly have thoughts un-favorable to the position advocated, then distraction should presumablyenhance the succss of the message (by interfering with the having of thoseunfavorable thoughts).7Quite a bit of research concerns distraction's effects on persuasion,althugh regrettably little of it is completely suitable for assessing the pre-dictions of the ELM (for some general discussions of this literature, seeBaron, Baron, & Miller, J.973; Buller, 198; Buller & Hali, 1998; Petty &Brock, 19 8 L ) . But rvhat relevant evidence e xists does see m largely compat-ible with the ELM. For example, studies reporting that distraction en-hances persuasive elfects have commonly relied on circumstances in rvhichelaboraiion likelihood was high and predominantly unfavorable thoughtsli'ould be expected (see Petty & Brock, I981, p. 5). More direct tests ofthe ELM's predictions have also been generally supportive (for a review,see Petty & cacioppo, 198a, pp. 1-8). For instance) Pett)" wells, andnrock (i9Zo, Expiment l) found that increasing distraction increased

  • 7/29/2019 Elaboration Likelihood Model(1)

    9/18

    Elaboraton Lkelhood Model N= 145Kallgren, 1988; Wood, I(ailgren, & Preisler, 1985; fbr related work, seeCacioppo, Petq', & Sidera, 1982).8 As one might expect) this suggeststhat rvhen receivers lvith extensive prior knowledge encounter a counter-attitudinal message, such receivers are better able to generate counterargu-ments (i.e., arguments opposing the message's advocated position) andhence in general are less likely to be persuaded (in comparison u'ith re-ceivers with less extensive topic knorvledge). But receivers with extensiveprior knorvledge are also more affected by variations in message argumentstrength; hence increasing the strength of a counterattitudinal mes-sage's arguments will presumably enhance persuasion for receivers withextensive knorvledge but rvill have little effect on receivers rvrth less exten-sive knorvledge.e5ummary

    As should be apparent, a variety offactors can influence the likelihood ofelaboration in a given circumstance by affecting the motivation or the abil-ity to engage in issue-relevant thinking. With variations in elaboration like-lihood, of course, different sorts of persuasion processes are engaged: Aselaboration increases, peripheral cues have diminished effects on persua-sive outcomes, and central route processes play correspondingly greaterroles. But the factors influencing persuasive effects are different, depend-ing on whether central or peripheral routes to persuasion are followed'Thus the next two se ctions consider rvhat factors influence persuasive out-comes rvhen elaboration likelihood is relatively high and when it is rela-tively lorv.

  • 7/29/2019 Elaboration Likelihood Model(1)

    10/18

    146 N PERSUASIONchange in the desired direction, but if the receiver has predorninantly un-favorable thoughts, then the message u'ill presumably be relatively unsuc-cessful. Thus the question becomes this: Given relatively high elaboration,what influences the predomir-rant valence (the overall evaluative directior-r)of elaborationllnfluences on Elaboration Valence

    Of the many influences on the evaluative direction of receivers' issue-relevant thinking, two factors merit attention here: 1'hether the message'sadvocated position is proattitudinal or colrnterattitudinal and the strength(quality) of tl-re message's arguments.

    Proattitwtinal Wrsys Counterattitur.inal Messnges. The receiver's ini-tial attitude and the message 's advocated position, considered jointll', rvillsurely influence the valence of elaboration. When the advocated position isone torvard rvhich the receiver is already fvorably inciined-that is, whenthe message advocates a proattitudinal position-the receiver rvill pre-surnably ordinarily be inclined to have favorable thoughts about the posi-tion advocated. By contrast) u'hen the message advocates a counteratti-tudinal position, receivers u'ill ordinarily be inclined to have unfavorablethoughts about the vierv being advocated. That is, everything else beingequal, one expects proattitudinal messages to evoke predominantly r'or-able thoughts and counterattitudinal messages to evoke predominantlVunfr'orable thoughts. the rvhole storlr-e11t..lvise nobody rvould

  • 7/29/2019 Elaboration Likelihood Model(1)

    11/18

    Elaboraton Lhelhood Model NX 141That is, Lrnder conditions ofhigh elaboration, the strength (the quality)of the mcssage's arguments should influence the evaluative directionof elaboration (and hence should influence persuasive success). Manyinvestigations have reported results indicating just such effects (e'g.,Fleesacker, Petty, & Cacioppo, 1983; Petty & Cacioppo,I979b,1984;

    Petq,, Cacioppo, & Goldman, 1981; Petty, Cacioppo, & Schumar-rn, 1983).Unhappily, this research et'idence is not as illuminating as one mightsuppose be cause of the u'ay in which arglrment strength is operationally de -fined. To obtain experimental messages containing strong or weak argu-ments, ELM researchers commonly pretest l'arious messages: A strong-argument message is defined as "one containing argLrments such that whensubjects are irxstru.ctel to think about the message) the thoughts that theygellerate are predominantly favorable," and a rveak-argument message isde fined as one in which the arguments "are such that rvhen sr-rbjects are in-srrucred to think about them, the thoughts that they generate are pre domi-nantly unfvorable." That is, ELM research has "postponed the questionofwhat specific qualities make arguments persuasive by defining argumentquality in an empirical manner" (Pett,v & Cacioppo, 198a, p.32).The consequence of this research practice is that it is not possible to sayjust what made these strong-argument messages effective under condi-tions of high elaboration. (It's not that the rnessages contained strongarguments-if the messages had not been effective under conditions ofclose scrutinlr, they lvould not have been labeled strong-argument mes-sages in the first place' ) F{ence it is not )'et possible to provide much direc-tion to persuaders about just how to compose effective messages underconditions of high elaboration likelihood. One can say, "Use strong argu-

  • 7/29/2019 Elaboration Likelihood Model(1)

    12/18

    148 NT PERSUASIoNSummary: Central Routes to Persuasion

    Under conditions of high elaboration (e.g., high personal relevance ofthe topic to the receiver), the outcome ofpersuasive efforts depends on thevalence ofre ceivers' elaboration: When a persuasive message leads receiversto have predominantly favorable thoughts about the position being advo-cated, persuasir.e success is correspondingly more likely. And the valence ofreceivers' elaboration rvill depend (at least in part) on the character ofthemessage's arguments. 1o

    N rnruurncEs oN pERsuAstvE EFFEcrsUNDER CONDITIONS OF LOW ELABORATION:PERIPHERAL ROUTES TO PERSUASIONThe Critical Role o Heuristic Principles

    The ELM suggests that under conditions of relatively lor'v elaboration,the outcomes of persuasive efforts will not generally turn on the results ofthe receiver's thoughtful consideration of the message's arguments orother issue-relevant information. Instead, persuasive effects will be muchmore influenced by the receiver's use of simple decision rules or heuristicprinciples.it These heuristic principles (or heuristics, for short) representsimple decision procedures requiring little information processing. Theprinciples are activated by peripheral cues, that is, by extrinsic features ofthe communication situation such as the characteristics of the commutlica-tor (e.g., credibility). For example, in a circumstance in which elaboration

  • 7/29/2019 Elaboration Likelihood Model(1)

    13/18

    Elaboraton Lh.elhood Model N 149The primary evidence for the operation of heuristic principles consists of'..reai.h resulrs conforming to just such patterns of effect (for some discus-sion, see Bless & Schwarz, 7999).Varieties of Heuristic Principles

    Although a number of heuristic principles have been suggested, threeheuristics have received relatively more extensive research attention: thecredibility, iiking, and consensus heuristics.l2creri,bitity Heuristic. one heuristic principle is based on the apparent

    credibility of the communicator and amounts to a belief that "statementsby credible sources can be trusted" (for alternative expressions of re-lated ideas, see Chaiken, 1987,p.4; Cialdini, 1987,p' f 75)' As discussedin chapter 8, studies have indicated that as the personal relevance ofthe topic to the receiver increases, the effects of communicator credibilirydiminish (..g., H. H. lohnson & Scileppi, 199; Pettl', Cacioppo, &Goldman, tat; nhine & Severance,1970). Similar results have beenobtailed rvhen elaboration likelihood has been varied in other rval's (e'g',S. B. Kiesler & Mathog, I98; Ratneshrvar & Chaiken, 1986)' Thus con-sistent r,vith ELM expectations, the peripheral cue of credibility has becnfound to have greater impact on persuasive outcomes when elaborationlikelihood is retively lorv. Moreover) some research suggests that varia-tions in the salience of credibility cues lead to corresponding variations incredibiliw,s effects (Andreoli & worchel, l97B; Worchel, Andreoli, &Eason, lg75).All told, there looks to be good evidence for the existence ofa credibiliry he uristic in persursion.

  • 7/29/2019 Elaboration Likelihood Model(1)

    14/18

    150 N PERSUASIONEagly, 1983): As elaboration likelihood declines or cLre saliency increases,the-ipact of liking cues on persuasion increases. Take' together, then)these studies pointio the operation of a liking heuristic that can influencepersuasive e1-fects.

    Consensus Hewristic. Athird heuristic principle is based on the reactionsof other people to rhe message and could be expressed as a belief that "ifother peple believe it, then it's probably true" (for 'ariant phrasings of,.rch a he.ttlstic, see Chaiken, 1987,p.4; Cialdini, 1987, p' 174)' \4renthis heuristic is employed, the approving reactions of others should en-hance message effectiveness (and disapproving reactions should_impaireffectivenessl. A number of studies norv indicate the operation of such aconsensus heuristic in persuasion (fbr a more careful revierv, see Arsomet al., J.987). For example, several investigations hal.e found that receiversare less persuaded rvhen they overhear an audience expressing disapproval(versus approval) of the communicator's tnessage (e'g', Hylton,,197I;i.urrdy, yOZZ;stlurtthorne & Mazmanian, I975). (For some related r,r'ork,see Darke et al., 1998; Hocking' Margreiter' & Hylton, 1977; Reingen,1982. For complexities, see Beatty & IGuger, 1978')

    other Hewristics. Yarous other principles have been suggested asheuristics that receivers may employ in reacting to persuasive messages' Forexample, it may be that the number ofarguments in the message (Chaiken,2; Petq' & Cacioppo, 1984) or the sheer length of the

  • 7/29/2019 Elaboration Likelihood Model(1)

    15/18

    Elaborotot't Lkelhood Model =.N I 5 I

    N coruplrxtrtEs AND coNsEouENcEsIN PERSUASION PROCESSESComplexities n Persuasion

    In at least rhree rvays, thc ELM ider-rtifies important complexities in per-suasion processes. Fiist, alcl most obviousl,v, it poiuts to the existence offundam.tally diflre't possible rolrtes to persuasion. Persuasio't is com-plicatecl in th sense thatlt does not co*e about ir-r jlist one ll'a' but mightte accomplished through eitl-rer centrai or pcriphcral routes'Secon, the ELM acknorvledges persuasion's complexitv by its recogni-tion of the broad trade-off betwcen elaboration valence and peripheralcues as influences on persuasiol (a trade-offthat creatcs the possibiliW thata mixture of central and peripheral processes might operate sin-rulta-neoushr). Recall that there is a continLrum of elaboration; the distinctionbetlrer-r central roLlte and peripheral route persuasion is simply a conve-nient rvay of expressi'g th uariation representcd at the extre mes of thiscontinuum. what is crucial is not the (useful but too-simple) distinctiorrbetg,een the tq,o persuasion routes but rather tl-re ulderl,ving elaborationcontinuum, tvhich creates something of a trade-of benveen peripheralcues and elaboration valence as influences on persuasive otttcomes: Aselaboration increases, the impact of peripheral cues declines, and thc irn-pnct of th. r,alence ofithe recei..e r's isiue-relevant thinking increases'13 Fori"ar-,p|., as variations in argument strength make tnore diffcrence in out-.or,'.r, riariations in communicaror expertise play smaller roles _as.periph-eral cues (e.g., Petty, Cacioppo, & Goldrnan, I98I)' The ELM does notclaim that (for instance) ,,"riatiour in argument strength *'ill make no dif-in peripheral cues rvill

  • 7/29/2019 Elaboration Likelihood Model(1)

    16/18

    lfl N PERSUASIONwhichcentrairouteorperipheralrouteprocessesareengaged).Second,it-ifn. ,..u. as a periphral ue (and so inhuence persuasive- outcomes r'vhenperipheral route persuasion is occurring)' Third' it might influence thevalence of elaboration (and so influence persuasive outcomes rvhen centralroute persuasion is occurring), by being an a-rgumerlt

    :Py "t1:i-tise

    bias-ing (tat is, encouraging one or anothervalence of) eiaborauon''"The ELM emphasize that a given variable need not play one and onlyone of these roles (e.g., Petty &*Cacioppo, 198a' pp' 204'215; Petty &W.g.n.., 1998a, I'ggb).In diffe'ettt ciicumstances' a variable might affect;"ri;" through diferent mechanisms. For example, consider the vari-'*i" f-.rr"ge lgth (the simple length ofawritten message). This might,..u. u, u ppn.t ..t. th"t "ttiu"tt a length-based heuristic (such as..longermessagesprobablyhavelotsofgoodreasonsfortheadvocatedyi"-;, ,e. wod t l.' l4s1. When message length operates this way'longe. messages will be more persuasive than shorter ones'- o, _.rrui. length might also (or instead) influence elaboratior moti-vation. For eiampl, o., Jhighly technical subject' the length of.the mes-sage might serve as u ,ig" o'"ttether the message was likely to be rvorthclJs. .*Imir1ation. Shorler messages might get little attention (because re-ceivers rvould think that the messge could not possibly contain the neces-.u.y "-orrrra of technical informatn), whereas longer messages would beexaminedmorecarefully'(Forsomeevidenceofsuchaphenomenon)Seeii.v, rqso.) In such a'circ"m't"'-'ce, obviously' a longer message wouldnot necessarily be more persuasive than a shorter one; the persuasiveness ofthelongermessage*'ootdturnontheoutcomeofthecloserscrutinyen-gendered bY the message's length'

  • 7/29/2019 Elaboration Likelihood Model(1)

    17/18

    ElahorotronLhelhood Model N 153The possibility of differcnt persLrasion roles lbr a single variable directlyimplies considcrable cornplexity ir-r persuasion. Consider, for instance:What u'ill be the efct (on persuasivc olrtcomes) of varying the commu-ruicator's attractivcnessi The ELM's analysis irnplies that no simple pre-cliction can be rnade; instead, the effects rvill be expected to var,v de pendingop (among other things)u'hethe r attractiveness operates as au influe nce ontire extent of elaboration) as an influence on the r,alence of elaboration,or as a peripheral cue. So, for iustance, increasing the communicator's;lttractiveness might enhance perslrasion (e.g., if attractivencss oPe rates

    as a peripheral cue that activates a liking-implies-correctness heuristic, ifattractiveness enhances message scrr,rtiny'and the message contains strongarguments, if attractiveness reduces message scrutiny and the messagecontains u'e ak arguments, or if greater attracfivcness ellcourages positiveelaboratior-r b1'sen'ing as an argllment) or inhibit persuasion (e.g., if attrac-tiveness enhances message scrlltirly and the message contains rveak argu-ments or if attractiver-ress reduccs message scrutiny and thc message con-tains strong arguments).

    Obr.ioush', the ke1'question that arises conccms spe ciying e xactl)'$'he na r-ariable is likelv to pla1. one or another role (see, e.g., Pettl', Wegener,Fabrigar, Priester, & Cacioppo ,)-993,p. 35a). Direct researcll evidence onsuch matte rS is sparse , and he nce confldent ger-re ralizations are surell'somedistance in the fr.rture. Even in the absence of sotne rvell-articulated ac-collnt of the circumstances under rvhich a given variable u,ill serve in this orthat persuxsion role, persuaders n'ill be g,ell-advised to be alert to suchcomplexities (1br a general discussion, see Pettv & Wegcner, l998a)'Consequences o Different Routes to Persuasion

  • 7/29/2019 Elaboration Likelihood Model(1)

    18/18

    154 =-N PERSUASIONsuch effects. For example, Pettl', Cacioppo, and Schumann (1983) re-ported that attitudes were more strongly correlated rvith intentions u'henthe attitudes r.l'ere fbrmed under conditions of high (as opposed to lou')personal relevance of the topic; Cacioppo, Pett1., Kao, and Rodriguez(1986)found that persons high in need fbr cognition (and so presumablyhigher in elaboration motivation) displayed greater attitude-intention andattitude-behavior consistency than did persons lor.r'er in need for cogni-tion; Verplanken (I991) reported greater persistence of attitudes andgreater attitude-intention consistency under conditions of high (ratherthan lorv) elaboration likelihood (as indicated b), topic relel'ance and needfor cognition); Macl(enzie and Spreng (1992) experimentallyvaried elab-oration motivation and found stronger attitude-intention relationshipsunder conditions of higher (as opposed to lorver) elaboration motivation.(For some generai reviews and discr-rssions, see Petry & Cacioppo, l9Ba,pp.773-195; Petq., Flaugtvedt, & Smith, 1995; Petty & Wegener, 1999,pp. 1-3.)These effects may seem intuitivelyplausible (in the sense that the greaterissue-relevant thinking affiliate d with central route processe s might well beexpected to yield attitudes that are stronger in these rvays), but it is not yetclear exactly rvhat accounts for them. That is, the mechanism by rvhichthese outcomes arise is not entirely rvell understood (for some discussion,see Petq', l{augn'edt, & Smith, 1995, pp. I'I9-I23). Nevertheless, there isgood reason for persuaders to presulne that persuasion accomplishedthrough high elaboration is likel,v to be more enduring (less likely to de -cav through time , less likely to succumb to counterpersuasion) and to bemore directive of behavior than is persuasion accomplishe d through lolr'elaboration.