Вы находитесь на странице: 1из 13

Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of

Author(s): Richard E. Petty, John T. Cacioppo, David Schumann
Source: The Journal of Consumer Research, Vol. 10, No. 2 (Sep., 1983), pp. 135-146
Published by: The University of Chicago Press
Stable URL: http://www.jstor.org/stable/2488919
Accessed: 18/04/2009 09:13

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

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 organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to The
Journal of Consumer Research.

Central and Peripheral Routes
to Advertising Effectiveness:
The Moderating Role of Involvement


Undergraduates expressed their attitudes about a product after being exposed to

a magazine ad under conditions of either high or low product involvement. The
ad contained either strong or weak arguments for the product and featured either
prominent sports celebrities or average citizens as endorsers. The manipulation
of argument quality had a greater impact on attitudes under high than low involve-
ment, but the manipulation of product endorser had a greater impact under low
than high involvement. These results are consistent with the view that there are
two relatively distinct routes to persuasion.

Over the past three decades, a large numberof studies particularattitudinalposition. The theoretical approaches
have examinedhow consumers'evaluationsof issues, following this route emphasize factors such as (1) the cog-
candidates, and products are affected by media advertise- nitive justification of attitude discrepant behavior (Cum-
ments. Research on the methods by which consumers' at- mings and Venkatesan1976; Festinger 1957); (2) the com-
titudes are formed and changed has acceleratedat a pace prehension, learning, and retention of issue- or product-
such that Kassarjianand Kassarjianwere led to the conclu- relevant information(Bettman 1979; Hovland, Janis, and
sion that "attitudes clearly have become the central focus Kelly 1953; McGuire 1976); (3) the nature of a person's
of consumerbehaviorresearch" (1979, p. 3). Not only are idiosyncratic cognitive responses to external communica-
there a large numberof empirical studies on consumer at- tions (Cacioppo and Petty 1980a; Greenwald 1968; Petty,
titudeformationand change, but thereare also a largenum- Ostrom,and Brock 1981; Wright 1980); and (4) the manner
ber of differenttheoriesof persuasionvying for the attention in which a person combines and integratesissue- or prod-
of the discipline (see Engel and Blackwell 1982; Kassarjian uct-relevant beliefs into an overall evaluative reaction
1982). (Ajzen and Fishbein 1980; Lutz and Bettman 1977; Trout-
In our recent reviews of the many approachesto attitude man and Shanteau 1976). Attitudechanges inducedvia the
change employed in social and consumer psychology, we central route are postulated to be relatively enduring and
have suggested that-even though the differenttheories of predictiveof behavior(Cialdini, Petty, and Cacioppo 1981;
persuasionpossess differentterminologies, postulates, un- Petty and Cacioppo 1980).
derlying motives, and particular"effects" that they spe- A second group of theoreticalapproachesto persuasion
cialize in explaining-these theories emphasizeone of two emphasizes a more peripheral route to attitude change.
distinctroutesto attitudechange (Petty and Cacioppo 1981, Attitudechanges that occur via the peripheralroute do not
1983). One, called the central route, views attitudechange occur because an individual has personally considered the
as resulting from a person's diligent considerationof in- pros and cons of the issue, but because the attitudeissue
formationthat s/he feels is central to the true merits of a or object is associated with positive or negative cues-or
because the person makes a simple inference about the
merits of the advocated position based on various simple
*RichardE. Petty is Associate Professorof Psychology at the University
of Missouri, Columbia, MO 65211. John T. Cacioppo is Associate Pro-
cues in the persuasion context. For example, ratherthan
fessor of Psychology at the University of Iowa, Iowa City, IA 52242. diligently considering the issue-relevantarguments,a per-
David Schumannis a graduatestudentin psychology at the Universityof son may accept an advocacy simply because it was pre-
Missouri. The authors are grateful to Rob Greene, Nancy Stabler, Trez sented during a pleasant lunch or because the source is an
Bayer, KarenKing, BrianKinkade,Edith Meredith,Tim Nash, and Todd expert. Similarly, a person may reject an advocacy simply
Nixon for their considerablehelp in conducting the experimentreported
here, and to the Universityof MissouriResearchCouncilfor grantsupport. because the position presentedappearsto be too extreme.
These cues (e.g., good food, expert sources, extreme po-
? JOURNALOF CONSUMERRESEARCH* Vol. 10 0 September 1983

sitions) and inferences (e.g., "If an expert says it, it must tude change must consider that in some situations people
be true") may shape attitudesor allow a person to decide are avid seekers and manipulatorsof information, and in
what attitudinalposition to adopt without the need for en- others they are best described as "cognitive misers" who
gaging in any extensive thought about issue- or product- eschew any difficult intellectual activity (Burnkrant1976;
relevant arguments. The theoretical approachesfollowing McGuire 1969). An importantquestion for consumer re-
the peripheralroute emphasize factors such as (1) whether searchersthen is: when will consumers actively seek and
a simple attitudinalinference can be made based on ob- process product-relevantinformation, and when will they
serving one's own behavior (Bem 1972; Scott 1978); (2) be more cursory in their analysis of ads? Recent research
whether the advocacy falls within one's latitudeof accep- in consumer behavior and social psychology has focused
tance or rejection (Newman and Dolich 1979; Sherif, on the concept of "involvement" as an importantmoder-
Sherif, and Nebergall 1965); (3) whether some transient atorof the amountand type of informationprocessing elic-
situational utility is associated with adopting a particular ited by a persuasive communication (see Burnkrantand
attitude (Schlenker 1978, 1980); and (4) whether an ad- Sawyer 1983; Petty and Cacioppo 1981, 1983). One major
vocated position or product is classically conditioned to goal of the experimentreportedin this paperwas to test the
basic but issue-irrelevantcues, such as food and pain (Janis, hypothesis that under "high involvement," attitudesin re-
Kaye, and Kirschner 1965; Stemthal and Craig 1974), or sponse to an advertisementwould be affectedvia the central
is associatedwith secondarycues, such as pleasantpictures route, but that under "low involvement," attitudeswould
and attractiveendorsers(Kelman 1961; Mitchell and Olson be affected via the peripheralroute.
1981; Mowen 1980). Attitude changes induced under the
peripheralroute are postulated to be relatively temporary INVOLVEMENT AND
and unpredictiveof behavior.'
Unfortunately,none of the unique theories of persuasion ATTITUDE CHANGE
has yet provided a comprehensiveview of attitudechange. Methods of Studying Involvement
For example, cognitive response theory-an approachthat
falls underthe centralroute-assumes that people are usu- Although there are many specific definitionsof involve-
ally interested in thinking about and elaboratingincoming ment within both social and consumerpsychology, there is
information,or in self-generatingissue- or product-relevant considerable agreement that high involvement messages
thoughts (Brock and Shavitt 1983). Yet, as Miller and his have greaterpersonalrelevance and consequences or elicit
colleagues have noted, "It may be irrationalto scrutinize more personalconnections than low involvementmessages
the plethoraof counterattitudinalmessages received daily. (Engel and Blackwell 1982; Krugman1965; Petty and Caci-
To the extent that one possesses only a limited amountof oppo 1979; Sherif and Hovland 1961).2 Various strategies
information processing time and capacity, such scrutiny have been employed in studyinginvolvement.For example,
would disengage the thought process from the exigencies both social (Hovland et al. 1957) and consumer (Newman
of daily life." (Miller, Maruyama, Beaber, and Valone and Dolich 1979) researchers have investigated existing
1976, p. 623). Haines (1974), in fact, has proposeda prin- groupsthat differed in the extent to which an issue or prod-
ciple of information-processing parsimony according to uct was personally important, or have employed designs
which consumersseek to process as little data as necessary allowing subjects to assign themselves to high and low in-
in order to make decisions. volvement groups. These correlational methods may be
The accumulatedresearchon persuasionclearly indicates high in external validity, but they confound involvement
that neither the central nor the peripheralapproachalone with all otherexisting differencesbetween the high and low
can account for the diversity of attitude-changeresults ob- involvement groups (attitude extremity, amount of prior
served. Thus, a general frameworkfor understandingatti- information,and so on), and thus compromiseinternalva-
lidity (Kiesler, Collins, and Miller 1969). Other social
'Our categorizationof the traditionaltheoreticalapproachesunder one (Rhine and Severance 1970) and consumer(Lastovickaand
or the other route to persuasion is meant to be suggestive rather than Gardner 1979) researchers have defined involvement in
absolute. For example, the theoretical process of self-perception (Bem terms of the specific issue or productunder consideration.
1972) mightgenerallylead to attitudechangebecause of a simple inference This procedure,of course, confoundsinvolvementwith as-
(peripheralroute), but might also be capable of initiatingextended issue-
relevant thinking in other circumstances(e.g., when personal relevance pects of the issue or product that may be irrelevantto its
is high; see Liebhart1979). Additionally, we note that the view that there personal importance.Finally, some researchershave stud-
are different "kinds" of persuasioncan be tracedback to Aristotle'sRhet- ied involvement by varying the medium of message pre-
oric, and that the distinction we have made between the central and pe-
ripheralroutes to attitude change has much in common with Kelman's
(1961) earlierview of "internalization"vs. "identification"and with the 2in the present paper, we use the term involvement to refer to "issue"
recent psychological distinctions between "deep" vs. "shallow" pro- or "product" involvement rather than "response" involvement. In the
cessing (Craik and Lockhart 1972), "controlled" vs. "automatic" pro- former, the attitude issue or the product itself has some direct personal
cessing (Schneiderand Shiffrin 1977), "systematic" vs. "heuristic" pro- relevance or consequence, and people are concerned with forming a rea-
cessing (Chaiken 1980), "thoughtful" vs. "scripted" or "mindless" soned opinion (Petty and Cacioppo 1979). In the latter, the attitudere-
processing (Abelson 1976; Langer et al. 1978), and others. For more sponse is important,and people are more concerned with expressing an
details on similarities and differences among the approaches, see Petty attitudethat will produce immediate situationalrewards(such as gaining
and Cacioppo (forthcominga). favor with others) than with forming a veridicalopinion (Zimbardo1960).

sentation.Interestingly,some investigatorshave arguedthat likely to fall within the unacceptablerange of a person's

television is a more involving mediumthanprint(Worchel, implicit attitudecontinuum.Krugman(1965) has proposed
Andreoli, and Eason 1975), whereas others have argued an alternativeview that has achieved considerablerecog-
just the opposite (Krugman1967). nition among consumer researchers. According to this
A preferredprocedurefor studying involvement would view, increasing involvement does not increase resistance
be to hold recipient, message, and medium characteristics to persuasion, but instead shifts the sequence of commu-
constant and randomlyassign participantsto high and low nication impact. Krugmanargues that under high involve-
involvement groups. Apsler and Sears (1968) employed an ment, a communicationis likely to affect cognitions, then
ingenious method to manipulateinvolvement:some partic- attitudes, and then behaviors, whereas under low involve-
ipants were led to believe that a persuasive proposal had ment, a communicationis more likely to affect cognitions,
personal implications for them (an advocated change in then behaviors, then attitudes(see also Ray et al. 1973).
university regulations would take effect while the student As noted earlier, a focal goal of this study is to assess
participantswere still in school), while others were led to the viability of a third view of the effects of involvement
believe that it did not (i.e., the change would not take effect on consumer response to advertisements.This view stems
until after the students had graduated).A variationof this from our ElaborationLikelihood Model (ELM) of attitude
procedurewas developed by Wright (1973, 1974) to ma- change (Petty and Cacioppo 1981). The basic tenet of the
nipulate involvement in an advertising study. Participants ELM is that different methods of inducing persuasionmay
in the high involvement group were told that they would work best dependingon whetherthe elaborationlikelihood
subsequentlybe asked to evaluate the productin an adver- of the communication situation (i.e., the probability of
tisement they were about to see, and were given some ad- message- or issue-relevant thought occurring) is high or
ditional background information. Participantsin the low low. When the elaborationlikelihood is high, the central
involvement group did not expect to evaluate the product route to persuasion should be particularly effective, but
and were given no background information. The back- when the elaborationlikelihood is low, the peripheralroute
ground informationprovided to the high involvement sub- should be better. The ELM contends that as an issue or
jects explained the relevance of their productdecisions to product increases in personal relevance or consequences,
"their families, their own time and effort, and their per- it becomes more importantand adaptiveto forming a rea-
sonal finances" (Wright 1973, p. 56). However, it is some- soned and veridical opinion. Thus, people are more moti-
what unclear to what extent this backgroundinformation vated to devote the cognitive effort requiredto evaluate the
made certain product-relevant arguments salient or sug- truemeritsof an issue or productwhen involvementis high
gested appropriatedimensions of product evaluation for ratherthan low. If increased involvement increases one's
high but not low involvement subjects. propensity to think about the true merits of an issue or
In the present experiment, participantsin both the high product,then manipulationsthat requireextensive issue- or
and low involvement groups were told that they would be product-relevant thought in order to be effective should
evaluating advertisementsfor products, but subjects in the have a greaterimpact under high ratherthan low involve-
high involvement group were led to believe that the exper- ment conditions. On the other hand, manipulationsthat al-
imental advertisedproductwould soon be availablein their low a person to evaluate an issue or product without en-
local area, andthatafterviewing a varietyof advertisements gaging in extensive issue- or product-relevant thinking
they would be allowed to choose one brand from the ex- should have a greater impact under low ratherthan high
perimentalproduct category to take home as a gift. Low involvement.
involvement participantswere led to believe that the ex- Research in social psychology has supportedthe view
perimental advertised product would not be available in thatdifferentvariablesaffect persuasionunderhigh and low
theirlocal area in the nearfuture, and that afterviewing the involvement conditions. For example, the quality of the
ads they would be allowed to take home one brand from argumentscontainedin a message has had a greaterimpact
a categoryof productsotherthanthe experimentalcategory. on persuasionunder conditions of high ratherthan low in-
volvement (Petty and Cacioppo 1979; Petty, Cacioppo, and
Heesacker 1981). On the other hand, peripheralcues such
Theories of Involvement as the expertiseor attractivenessof a message source (Chai-
In addition to the methodological differences that have ken 1980; Petty, Cacioppo, and Goldman 1981; Rhine and
plaguedthe involvementconcept, anotherarea of disagree- Severance 1970) have had a greater impact on persuasion
ment concerns the effects on persuasion that involvement under conditions of low ratherthan high involvement. In
is expected to have. Perhapsthe dominantnotion in social sum, under high involvement conditions people appearto
psychology stems from the Sherifs' social judgmenttheory exert the cognitive effort required to evaluate the issue-
(Sherif et al. 1965). Their notion is thaton any given issue, relevantargumentspresented,and theirattitudesare a func-
highly involved persons exhibit more negative evaluations tion of this information-processingactivity (centralroute).
of a communicationbecause high involvementis associated Under low involvement conditions, attitudesappearto be
with an extended "latitude of rejection." Thus, incoming affected by simple acceptance and rejection cues in the
messages on involving topics are thought to have an en- persuasioncontext and are less affectedby argumentquality
hanced probabilityof being rejectedbecause they are more (peripheralroute). Although the accumulatedresearch in

social psychology is quite consistent with the ELM, it is The present study was a conceptual replication of pre-
not yet clear whether or not the ELM predictions would vious work (Petty and Cacioppo 1980), except that we em-
hold when involvement concerns a product(such as tooth- ployed a peripheralcue that could not be construed as a
paste) rather than an issue (such as capital punishment), product-relevantargument. In the current study, partici-
and when the persuasivemessage is an advertisementrather pants were randomlyassigned to high and low involvement
than a speech or editorial. conditions and viewed one of four different ads for a fic-
titious new product, "Edge disposablerazors." The ad was
Centraland PeripheralRoutes to presentedin magazine format and was embeddedin an ad-
AdvertisingEffectiveness vertising booklet along with 11 other ads. Two featuresof
the Edge ad were manipulated:the qualityof the arguments
One importantimplication of the ELM for advertising in supportof Edge (strongor weak), and the celebritystatus
messages is that different kinds of appeals may be most of the featuredendorsersof Edge (celebrity or averagecit-
effective for different audiences. For example, a person izen). It is importantto note that preliminarytesting re-
who is about to purchasea new refrigerator(high involve- vealed that for most people, the celebrity status of the en-
ment) may scrutinizethe product-relevantinformationpre- dorserswas irrelevantto an evaluationof the true meritsof
sented in an advertisement.If this informationis perceived a disposable razor, but that because the celebrity endorsers
to be cogent and persuasive, favorableattitudeswill result, were liked more than the average citizens, they could still
but if this informationis weak and specious, unfavorable serve as a positive peripheralcue.
attitudes will result (central route). On the other hand, a We had two major hypotheses. First, we expected the
personwho is not consideringpurchasinga new refrigerator quality of the arguments presented in the ad to have a
at the moment (low involvement)will not expend the effort greaterimpact on product attitudesunder high ratherthan
requiredto think about the product-relevantargumentsin low involvement conditions. Second, we expected the ce-
the ad, but may instead focus on the attractiveness,credi- lebrity status of the product endorsers to have a greater
bility, or prestige of the product's endorser (peripheral impact on productattitudesunder low ratherthan high in-
route). Some evidence in consumer psychology is consis- volvement conditions. If these hypotheses were supported,
tent with this reasoning. For example, Wright(1973, 1974) it would provide the first evidence that the Elaboration
exposed people to an advertisementfor a soybean product Likelihood Model can contributeto understandingthe ef-
underhigh and low involvementconditions (see earlierde- fects of involvement on attitudinalresponses to advertise-
scription) and measured the number of source comments ments.
(derogations) and message comments (counterarguments)
generatedafterexposure. AlthoughWright( 1974) predicted
that involvement would increase both kinds of comments, METHOD
he found that more message comments were made under
high rather than low involvement, but that more source Subjects and Design
comments were made under low involvement conditions. A total of 160 male and female undergraduatesat the
This finding, of course, is consistent with the ELM. University of Missouri-Columbia participatedin the ex-
In an initial attemptto providea specific test of the utility periment to earn credit in an introductory psychology
of the ELM for understandingthe effectiveness of adver- course; 20 subjects were randomlyassigned to each of the
tising messages (Petty and Cacioppo 1980), we conducted cells in a 2 (involvement: high or low) x 2 (argument
a study in which three variables were manipulated:(1) the quality:strongor weak) x 2 (cue: celebrityor noncelebrity
personalrelevanceof a shampooad (high involvementsub- status) factorial design. Subjects participatedin groups of
jects were led to believe thatthe productwould be available three to 15 in a very large classroom. The subjects were
in their local area, whereas low involvement subjects were isolated from each other so that they could complete the
not); (2) the quality of the argumentscontained in the ad; experimentindependently,and subjects in a single session
and (3) the physical attractivenessof the endorsersof the participatedin different experimentalconditions. In fact,
shampoo. Consistentwith the ELM predictions,the quality if enough subjects were present it was possible to conduct
of the arguments contained in the advertisement had a all eight experimentalconditions simultaneously.This pro-
greater impact on attitudes when the product was of high cedureavoided confoundingsession with experimentalcon-
ratherthan low relevance. Contraryto expectations, how- dition.
ever, the attractivenessof the endorsers was equally im-
portant under both the high and low involvement condi- Procedure
tions. In retrospect, in addition to serving as a peripheral
cue underlow involvement, the physical appearanceof the Two booklets were preparedfor the study. The firstcon-
product endorsers (especially their hair) may have served tained the advertisingstimuli and the second containedthe
as persuasive visual testimony for the product's effective- dependentmeasures.The firstpage of the advertisingbook-
ness. Thus, under high involvement conditions, the phys- let explained that the study concerned the evaluation of
ical attractivenessof the endorsers may have served as a magazine and newspaper ads and that the psychology de-
cogent product-relevantargument. partmentwas cooperatingwith the journalismschool in this

endeavor. The first page also containedpartof the involve- * New advanced honing method creates unsurpassedsharp-
ment manipulation(see below). It was explained that each ness
ad in the booklet was precededby an introductorystatement * Special chemically formulated coating eliminates nicks
thattold a little aboutthe advertisementthatfollowed (e.g., and cuts and prevents rusting
"The company of Paris, Francehas just opened an * Handle is taperedand ribbed to prevent slipping
Americanoffice in New York City. This elite men's cloth- * In direct comparisontests, the Edge blade gave twice as
ing company originally sold clothing only in Europe, but many close shaves as its nearest competitor
is now in the process of attemptingto enter the American * Unique angle placementof the blade providesthe smooth-
market. The ad on the next page is one that they will be est shave possible
testing soon in Tampa, Florida before running the ads in
other major cities that will eventually carry their prod- In the weak arguments version of the ad, the razor was
ucts"). The instructionstold subjects to continue through characterized as "designed for beauty," and the following
the booklet at their own pace and to raise their hands when five statements were made about the product:
finished. The ad booklet contained 10 real magazine ads * Floats in water with a minimumof rust
for both relativelyfamiliar(e.g., Aquafreshtoothpaste)and * Comes in various sizes, shapes, and colors
unfamiliar(e.g., Riopan antacid)products, and two bogus
o Designed with the bathroomin mind
ads. The sixth ad in each booklet was the crucial fictitious
ad for Edge razors (the nature of the other bogus ad was * In direct comparisontests, the Edge blade gave no more
variedbut is irrelevantto the presentstudy). When subjects nicks or cuts than its competition
had completed perusingtheir ad booklets, they were given * Can only be used once but will be memorable
a questionnairebooklet to complete. Upon completion of
Peripheral cue. In the "famous endorser" conditions,
the questionnaire,the subjects were thoroughlydebriefed,
the headline accompanying the advertisement read "Profes-
thankedfor their participation,and dismissed.
sional Athletes Agree: Until you try new Edge disposable
razors you'll never know what a really close shave is." In
IndependentVariables addition, the ad featured the pictures of two well-known,
well-liked golf (male) and tennis (female) celebrities. In the
Involvement. Involvementwas embeddedin two places "'nonfamous endorser" conditions, the headline read
in the ad booklet. First, the cover page offered subjects a "Bakersfield, California Agrees: ," and the ad fea-
free gift for participationin the experiment. Subjects were tured pictures of average looking people who were unfa-
either informed that they would be allowed to choose a miliar to the subjects. The average citizens in the ad were
particularbrandof disposablerazor(high involvementwith middle-aged and characterized as coming from California
the fictitious Edge ad) or that they would be allowed to to minimize perceptions of similarity to the subjects (Mis-
choose a brandof toothpaste(low involvementwith Edge). souri college students). Figure A depicts two of the four
A toothpastead did appearin the ad booklet, but it was the Edge ads used in the present study.
same ad for all subjects. To bolster the involvement ma-
nipulation, the page that introducedthe Edge ad also dif-
fered in the high and low involvement conditions. High
involvement subjects were told that the advertisementand On the first page of the dependent variable booklet, sub-
productwould soon be test-marketedin medium-sizedcities jects were asked to try to list all of the product categories
throughoutthe Midwest, including their own city (Colum- for which they saw advertisements, and to try to recall the
bia, Missouri);low involvement subjects were told that the brand name of the product in that category. On the next
advertisementand product were being test-marketedonly page, subjects were given descriptions of the 12 product
on the East Coast. Thus high involvement subjects were categories and were asked to select the correct brand name
not only led to believe that they would soon have to make from among seven choices provided. Although we had no
a decision about the product class, they were also led to specific hypotheses about brand recall and recognition,
believe that the productwould be available in their area in these measures were included because of their practical im-
the near future. Low involvement subjects, on the other portance and for purposes of comparison with the attitude
hand, did not expect to make a decision about razors (but data.
did expect to make one about toothpaste), and were led to Next, subjects responded to some questions about one of
believe thatEdge razorswould not be availablefor purchase the legitimate ads in the booklet; this was followed by the
in their area in the forseeable future. crucial questions about Edge razors. The questions about
Edge were placed relatively early in the booklet to avoid
Argumentquality. A variety of argumentsfor dispos- subject fatigue and boredom and to maximize the effec-
able razors were pretestedfor potency on a sample of un- tiveness of the manipulations. Subjects were first asked to
dergraduates. In the strong arguments ad, the razor was rate, on a four-point scale, how likely it would be that they
characterizedas "scientifically designed," and the follow- would purchase Edge disposable razors "the next time you
ing five statementswere made about the product: needed a product of this nature. " The descriptions for each


Until you try now Until you try new
EDGE d bGE dsposable
razors you'U never razors you'l never
know what a know what a
"really dose shave" "really dose shave"
is. s

* Scientifically Designed * Desgned for Beauty

* New advanced honing method creates unsurpassed sharpness. 0 Floats in water with a minimum of rust.
* Special chemically formulated coating eliminates nicks and cuts and prevents 0 Comes in various sizes, shapes, and colors.
rusting. * Designed with the bathroom in mind.
* Handle is tapered and ribbed to prevent slipping. 0 In direct comparison tests the EDGE blade gave no more nicks or cuts
* In direct comparison tests the EDGE blade gave twice as many close shaves than its competition.
as its nearest competitor. 0 Can only be used once but will be memorable.
* Unique angle placement of the blade provides the smoothest shave possible.

NOTE: Left panel shows celebrity endorser ad for Edge razors employing the strong arguments. Right panel shows average citizen endorser ad for Edge razors employing the weak
arguments. Pictures of celebrities and citizens have been blacked out to preserve propriety and anonymity.

scale value were: 1 = "I definitely would not buy it," 2 to check on the experimentalmanipulations,and subjects
= "I might or might not buy it," 3 = "I would probably were asked to try to list as many of the attributesmentioned
buy it," and 4 = "I would definitely buy it." Following in the ad about Edge razorsas they could recall. Following
this measureof purchaseintentions, subjects were asked to the questionsaboutEdge were severalquestionsaboutsome
rate their overall impression of the producton three nine- of the other productsand ads in the booklet. As a check on
point semantic differentialscales anchoredat -4 and + 4 the involvement manipulation,the very last questionin the
(bad-good, unsatisfactory-satisfactory, and unfavor- booklet asked subjects to recall the free gift they had been
able-favorable). Since the intercorrelationsamong these told to expect.
measures were very high (average r = 0.86), responses
were averaged to assess a general positive or negative at-
titude toward the product. RESULTS
Following some additionalquestionsthat were consistent ManipulationChecks
with the cover story, subjects were instructed to list the
thoughts that crossed their minds as they examined the ad In response to the last question in the dependentvariable
for Edge disposable razors. These thoughts were subse- booklet asking subjects what gift they had been told to
quently scored on several dimensions by trained judges. expect, 92.5 percentof the subjectsin the high involvement
Since subjects listed very few thoughts about the product conditionscorrectlyrecalledthatthey were to select a brand
(M = 1.18) and since the manipulationsfailed to affect of disposable razor. In the low involvement conditions,
this measure, it will not be discussed further. This "cog- none of the subjects indicated a razor and 78 percent cor-
nitive response" measure would probablyhave been more rectly recalledthatthey were to select a brandof toothpaste.
sensitive if it had been administeredimmediatelyafter ex- Thus, subjectspresumablyrealized what productthey were
posure to the Edge ad ratherthan after exposure to all 12 soon to make a decision about as they examined the ad
ads, but in the presentstudy this would have compromised booklet.
the cover story (for an extendeddiscussionof the reliability, To assess the effectiveness of the endorsermanipulation,
validity, and sensitivity of the thought-listingmeasure in two questionswere asked. First, subjectswere asked if they
persuasionresearch, see Cacioppo and Petty 1981). recognized the people in the ad for the disposable razor.
After listing their thoughts, several questionswere asked When the famous athletes were employed, 94 percent in-

dicated "yes," whereas when the average citizens were TABLE

employed, 96 percentindicated"no." In addition,subjects MEANSANDSTANDARDDEVIATIONSFOR EACH
were asked to rate the extent to which they liked the people INDEX
depicted in the ad on an 11-point scale, where 1 indicated
"liked very little" and 11 indicated "liked very much." Lowinvolvement Highinvolvement
An analysis of this measure revealed that the famous en-
dorserswere liked more (M = 6.06) than the average cit- Weak Strong Weak Strong
arguments arguments arguments arguments
izens (M = 3.64; F (1, 143) = 40.81, p < 0.0001); on
average, women reportedliking the endorsersmore (M = Citizen -.12 .98 -1.10 1.98
5.32) than did men (M = 4.44; F (1, 143) = 5.25, p < endorser (1.81) (1.52) (1.66) (1.25)
0.03). Celebrity 1.21 1.85 -1.36 1.80
As a check on the argument-persuasivenessmanipula- endorser (2.28) (1.59) (1.65) (1.07)
tion, two questions were asked. The first requiredrespon-
dents to "rate the reasons as describedin the advertisement NOTE: Attitude scores represent the average rating of the product on three nine-point
for using EDGE" on an 1I-point scale anchoredby "un- semantic differentialscales anchored at -4 and +4 (bad-good, unsatisfactory-satisfactory,
persuasive" and "persuasive"; the second question asked and unfavorable-favorable). Standard deviations are in parentheses.

them to rate the reasons on an 11-pointscale anchoredby

""weakreasons" and "strong reasons." On the first mea- Each of these main effects must be qualified and inter-
sure, subjects exposed to the strong argumentsrated them pretedin light of two importanttwo-way interactions.First,
as significantlymore persuasive (M = 5.46) than did sub- an Involvement x Endorserinteraction(F (1, 148) = 5.94,
jects exposed to the weak arguments(M = 4.03; F (1, 139) p < 0.02) revealed that the natureof the productendorser
= 12.97, p < 0.0004). Additionally, a main effect for had a significantimpacton productattitudesonly underlow
gender was found such that women ratedthe argumentsas involvement(F (1, 148) = 5.96, p < 0.02), but not under
more persuasive (M = 5.26) than did men (M = 4.28; F high involvement (F < 1; see top panel of Figure B). On
(1, 139) = 5.25, p < 0.02). Finally, an Arguments x the other hand, an Involvement x Argumentsinteraction
Genderinteractionemerged(F (1, 139) = 5.43, p < 0.02), (F (1, 148) = 18.47, p < 0.0001) revealed that although
indicating that the tendency for females to find the argu- argumentquality had an impact on productattitudesunder
mentsmorepersuasivethanmales was greaterfor the strong both low involvement (F (1, 148) = 5.40, p < 0.02) and
than for the weak arguments.On the second manipulation high involvement (F (1, 148) = 71.36, p < 0.0001), the
check measure, subjects rated the strong arguments as impact of argumentquality on attitudes was significantly
"stronger" (M = 5.58) than the weak ones (M = 4.13; greaterunderhigh ratherthan low involvement(see bottom
F (1, 138) = 14.31, p < 0.002). Again, an Arguments panel of Figure B). Neitherthe Endorser x Argumentsnor
x Genderinteractionoccurred, indicatingthat females es- the three-way interaction approached significance (F =
pecially tended to rate the strong argumentsmore highly 0.14 and 0.54, respectively).
than did males. In short, all of the variables were manip- Two significanteffects emergedfrom the questionasking
ulated successfully. The tendency for females to be more subjectsto rate their likelihood of purchasingEdge dispos-
positive in theirratingsof both endorsersand the arguments able razors the next time they needed a product of this
in the ads is generally consistent with previous psycholog- nature.Subjects said that they would be more likely to buy
ical research portraying women as more concerned with the productwhen the argumentspresentedwere strong (M
= 2.23) rather than weak (M = 1.68; F (1, 152) = 25.37,
social harmonythan men (Eagly 1978). Importantly,these
sex differencesdid not lead to any significantgendereffects p < 0.0001). Additionally, an Involvement x Arguments
on the crucial measuresof attitudeand purchaseintention. interactionemerged (F (1, 152) = 4.25, p < 0.04). This
interactionparalleledthat obtainedon the attitudemeasure
and indicatedthat argumentquality was a more important
Attitudesand PurchaseIntentions determinantof purchase intentions under high ratherthan
The Table presentsthe means and standarddeviationsfor low involvement.
each cell on the attitude index. A number of interesting The correlationbetween attitudesandpurchaseintentions
main effects emerged. First, involved subjects were some- for low involvement subjects was 0.36; and for high in-
what more skeptical of the product(M = 0.31) than were volvement subjects it was 0.59. Althoughboth correlations
less involved subjects (M = 0.99; F (1, 148) - 6.64, p are significantly different from zero (ps < 0.001), it is
< 0.01). Second, subjects liked the product significantly interestingto note that the low involvement correlationis
more when the ad containedcogent arguments(M = 1.65) considerablysmaller than the high involvementcorrelation
than when the arguments were specious (M = - 0.35; F (p < 0.07). The fact thatthe argumentqualitymanipulation
(1, 148) = 57.81, p < 0.0001). Third, subjectstended to affected behavioral intentions while the endorser manipu-
like the productmore when it was endorsedby the famous lation did not (althoughit did affect attitudes)-and the fact
athletes (M = 0.86) than by the average citizens of Bak- that attitudeswere betterpredictorsof behavioralintentions
ersfield, California (M = 0.41; F (1, 148) = 2.91, p < under high rather than low involvement provide some
0.09). support for the ELM view that attitudes formed via the

FIGURE B p < 0.03), but had no effect on product category recall

PRODUCT ATTITUDES underhigh involvement (80 versus 82 percent).
Involvementaffected free recall of the brandname of the
1.8 product, increasing it from 42 percent in the low involve-
ment conditions to 60 percentin the high involvementcon-
1.2 Famous endorsers ditions (Z = 2.28, p < 0.01). There was also an effect for
gender on this measure, with males showing greaterbrand
.6 name recall (61 percent) than females (39 percent; Z =
F -
2.78, p < 0.007). The endorser manipulationhad a mar-
Non-famous endorsers ~ ginally significant effect on brand name recall, with the
famous endorsers tending to enhance recall over average
citizens from 43 to 58 percent (Z = 1.89, p < 0.06).
On the measureof brandname recognition,an interaction

patternemerged. Underlow involvement,the use of famous
a) -1.2
endorsersreduced brand name recognition from 85 to 70
percent, but under high involvement, the use of famous
cn endorsersimprovedbrandname recognitionfrom 77 to 87
t5 1.8 - percent (Z = 1.96, p < 0.05).3
To summarize the recall and recognition data thus far,
1.2 Strong arguments it appearsthat increasinginvolvement with the producten-
hanced recall not only of the productcategory, but also of
.6 the brandname of the specific productadvertised.The ef-
fects of the endorsermanipulationwere more complex and
0 Weak arguments depended on the level of involvement. In general, under
low involvement a positive endorser led to increased recall
of the product category but reduced brand name recogni-
tion. Thus, people may be more likely to notice the products
in low involvement ads when they feature prominentper-
- 1.2
sonalities, but because of the enhanced attentionaccorded
Low High the people in the ads and the general lack of interest in
Involvement Involvement assessing the merits of the product (due to low involve-
NOTE: Top panel shows interactive effect of involvement and endorser status on attitudes
ment), reductions in brand recognition may occur. This
toward Edge razors. Bottom panel shows interactive effect of involvement and argument finding is similar to the results of studies on the use of
quality on attitudes toward Edge razors. sexually orientedmaterialin ads for low involvementprod-
ucts the sexual materialenhances recognition of the ad,
central route will be more predictive of behavior than at- but not the brand name of the product (e.g., Chestnut,
titudes formed via the peripheralroute. LaChance, and Lubitz 1977; Steadman 1969). Under high
involvement, however, the use of prominentpersonalities
Recall and Recognition Measures enhanced brand name recognition.Whenpeople are more
interestedin the productcategory, they may be more mo-
Subjects were asked to list all of the productsfor which tivated to assess what brandthe liked personalitiesare en-
they saw ads and all of the brandnames they encountered. dorsing. The manipulationof argumentquality had no ef-
Following this, all subjects were told that they had seen an fect on recall of the product category, brand name recall,
advertisementfor a disposable razor and were asked to se- or brandname recognition.
lect the correct brandname from a list of seven (Gillette, A final recall measureassessed how many of the specific
Wilkinson, Schick, Edge, Bic, Schaffer, and Remington).
The proportionof subjects showing correct recall or rec-
ognition was calculated for each cell. These proportions 3Some authors have suggested that it may be appropriateto analyze
were then subjected to an arcsin transformation (Winer dichotomous data using analysis of variance without biasing the results
1971) and analyzedby the procedurerecommendedby Lan- greatly (e.g., Winer 1971). We subjectedour recall and recognitiondata
ger and Abelson (1972). (scored 0 or 1) to ANOVA, and the following significant effects were
obtained. On the measureof recall of the productcategory, a main effect
The involvement manipulationhad a significant impact of involvement (F (1, 152) = 6.42, p < 0.01) and an involvement x
on free recall of the productcategory, with more high in- endorser interaction(F (1, 152) = 3.28, p < 0.07) were obtained. On
volvement subjects (81 percent) recalling the productcat- the measure of brand name recall, main effects for involvement (F (1,
egory than low involvement subjects(64 percent;Z = 2.4, 145) = 6.34, p < 0.01), gender (F (1, 145) = 7.20, p < 0.008), and
endorser(F (1, 145) = 3.49, p < 0.06) were obtained. On the measure
p < 0.02). Additionally, exposure to the famous endorser of brand name recognition, an involvement x endorser interactionwas
increasedrecall of the productcategory underlow involve- obtained(F (1, 152) = 4.04, p < 0.05). This patternof significanteffects
ment conditions (from 52 percentto 75 percent;Z = 2.14, is identical to the significantpatternof effects reportedin the text.

arguments for Edge razors the subjects could spontaneously the ad was a powerful determinantof productevaluations.4
recall after they had examined the entire ad booklet. Over- These data clearly suggest that it would be inappropriate
all, subjects were able to correctly reproduce only 1.75 of for social and consumer researchersto overemphasizethe
the five arguments presented. This was not affected by any influence of issue-relevant argumentsor product-relevant
of the experimental manipulations. attributesand ignore the role of peripheralcues. Each type
Clearly, the manipulations produced a very different pat- of attitudinalinfluence occurs in some instances, and the
tern of effects on the recall and recognition measures than level of personalinvolvement with an issue or productap-
on the attitude and purchase intention measures. In addi- pears to be one determinantof which type of persuasion
tion, the recall and recognition measures were uncorrelated occurs.
with attitudes or intentions toward Edge razors. This finding Accordingto the ElaborationLikelihoodModel, personal
is consistent with a growing body of research indicating relevanceis thoughtto be only one determinantof the route
that simple recall or recognition of information presented to persuasion. Personal relevance is thought to increase a
about an attitude object is not predictive of attitude for- person's motivationfor engaging in a diligent consideration
mation and change (e.g., Cacioppo and Petty 1979; Green- of the issue- or product-relevantinformationpresented in
wald 1968; Insko, Lind, and LaTour 1976). orderto form a veridicalopinion. Justas differentsituations
The present data also argue against using measures of may induce different motivationsto think, differentpeople
brand name recall or recognition as the sole indicants of may typically employ different styles of informationpro-
advertising effectiveness. For example, in the present cessing, and some people will enjoy thinking more than
study, enhancing involvement led to a significant improve- others (Cacioppo and Petty 1982, forthcoming).However,
ment in brand name recall, but increasing involvement led a diligent considerationof issue- or product-relevantinfor-
to a decrement in attitude toward the brand when the ar- mation requiresnot only the motivation to think, but also
guments presented were weak. the ability to process the information.Thus situationalvari-
ables (e.g., distraction;Petty, Wells, and Brock 1976) and
DISCUSSION individualdifference variables (e.g., priorknowledge; Ca-
cioppo and Petty 1980b) may also be importantmoderators
As we noted earlier, previous research on attitude for- of the route to persuasion. In the present study, subjects'
mation and change has tended to characterize the persuasion ability to think about the product was held at a high level
process as resulting either from a thoughtful (though not across experimentalconditions-that is, the messages were
necessarily rational) consideration of issue-relevant argu- easy to understand,the presentationwas self-paced, and so
ments and product-relevant attributes (central route), or on. Thus the primarydeterminantof the routeto persuasion
from associating the attitude object with various positive was motivationalin nature.5
and negative cues and operating with simple decision rules
(peripheral route). Over the past decade, investigators in
both social psychology and consumer behavior have tended 4Althoughnot tested in the present study, the ELM predictsthat under
moderateinvolvement conditions, source informationserves neither as a
to emphasize the former process over the latter. Consider simple cue (as under low involvement) nor is it ignored (as under high
the recent comments of Fishbein and Ajzen (1981, p. 359): involvement). Instead, source informationhelps a person determinehow
much thinkingto do about the message (Petty and Cacioppo 1981, forth-
The generalneglect of the informationcontainedin a message coming a).
. . .is probablythe most serious problemin communication 'An anonymousreviewerof this articletook issue with our motivational
and persuasionresearch. We are convinced that the persua- interpretationof the effects of involvementand suggested thatperhapsour
siveness of a communicationcan be increased much more effects resulted because our experimental task overtaxed our subjects'
easily and dramaticallyby paying careful attentionto its con- cognitive abilities. This suggestion assumes that subjectslackedthe ability
tent . . . than by manipulationof credibility, attractiveness to evaluate both the source and the message, and thereforehad to choose
. . .or any of the other myriadfactors that have caught the one over the other. We find this explanation implausiblefor several rea-
fancy of investigatorsin the area of communicationand per- sons. First, since the subjects paced themselves throughthe ad booklet,
they could spend as much time as they wished evaluating each ad: the
"overtaxed" explanationmay thus be more plausiblefor researchin which
The present study suggests that, although the informa- the message is externallypaced. Second, our experimentincluded several
checks on whether or not subjects attended to the source and message
tional content of an advertisement may be the most irnpor- information.For example, all subjects were asked if they recognized and
tant determinant of product attitudes under some circum- liked the endorsers appearing in the ad. If the reviewer's suggestion is
stances, in other circumstances such noncontent manipulations correct, we would expect subjects in the high involvement group (who
as the celebrity status (likeability) or credibility of the prod- diligently processed the message content) to be less likely to reportrec-
ognizing the endorsersin the ad, and hence to show less liking for the ad
uct endorsers may be even more important. Specifically, endorsers. However, the involvement manipulationfailed to affect either
we have shown that when an advertisement concerned a the recognitionor the liking measure. In fact, subjectsin the high involve-
product of low involvement, the celebrity status of the prod- ment group reportedslightly (thoughnot significantly)greaterrecognition
uct endorsers was a very potent determinant of attitudes and liking of the famous endorsersthan did the low involvementsubjects.
about the product. When the advertisement concerned a Thus high involvement subjects were not overtaxed.They recognizedand
liked the famous endorsers to the same extent as did low involvement
product of high involvement, however, the celebrity status subjects. It is just that the productattitudesof the high involvement sub-
of the product endorsers had no effect on attitudes, but the jects were not affected by this liking, while the product attitudesof the
cogency of the information about the product contained in low involvement subjects were.

It is importantto note that although our "peripheral" titudinal effects under low involvement. In the present
manipulationwas a source variablepresentedvisually and study, both attitudinal and behavioral (intention) effects
our "central" manipulationwas a message variable pre- were observed underhigh involvement, which is consistent
sented verbally, neither the source/messagenor the visual/ with both models. Under low involvement, however, ef-
verbal dichotomy is isomorphicwith the central/peripheral fects were obtained on the measure of attitudebut not on
one. Thus a source variablemay induce persuasionvia the the measureof behavioralintentions. This findingis incon-
central route, and a message variable may serve as a pe- sistent with Krugman'sformulation,which expects stronger
ripheralcue. For example, in one study described previ- behavioraleffects under low involvement than under high
ously (Petty and Cacioppo 1980), we observedthat a phys- involvement, but it is consistent with the ELM, which pos-
ically attractivemessage endorsermight serve as a cogent tulates a greatercorrespondencebetween attitudesand be-
product-relevantargumentfor a beauty product.In another haviors under high involvement (centralroute) than under
study (Petty and Cacioppo, forthcomingb), we found that low involvement (peripheralroute).
the mere numberof message argumentspresentedmay ac- In sum, the present study has provided supportfor the
tivate a simple decision rule (the more the better)underlow view that different features of an advertisement may be
involvement, but not under high involvement, where ar- more or less effective, dependingupon a person's involve-
gument quality is more importantthan number. Similarly, ment with it. Under conditions of low involvement, pe-
a "central" manipulationmay be presentedvisually-e.g., ripheralcues are more importantthan issue-relevantargu-
depicting a kitten in an advertisementfor facial tissue to mentation,but underhigh involvement, the oppositeis true.
convey the product-relevantattribute"softness" (Mitchell The realizationthat independentvariablesmay have differ-
and Olson 1981)-and a "peripheral" manipulationmay ent effects, dependingon the level of personalrelevanceof
be presentedverbally-e.g., providinga verbaldescription a message, may provide some insight into the conflicting
of a message source as an expert or as likeable (Chaiken patternof results that is said to characterizemuch attitude
1980; Petty, Cacioppo, and Goldman 1981). The critical research.It may well be thatattitudeeffects can be arranged
featureof the central route to persuasionis that an attitude on a continuum, depending on the elaborationlikelihood
change is based on a diligent considerationof information of the particular persuasion situation. This continuum
that a person feels is central to the true merits of an issue would be anchoredat one end by the peripheralroute and
or product. This information may be conveyed visually, at the other end by the central route to persuasion. Fur-
verbally, or in source or message characteristics. In the thermore, these two routes may be characterizedby quite
peripheralroute, attitudeschange because of the presence different antecedents and consequents. If so, future work
of simple positive or negative cues, or because of the in- could be aimed at uncoveringthe various moderatorsof the
vocation of simple decision rules which obviate the need route to persuasionand at trackingthe various consequents
for thinking about issue-relevant arguments. Stimuli that of the two differentroutes.
serve as peripheralcues or that invoke simple decision rules
may be presented visually or verbally, or may be part of [Received March 1982. Revised April 1983.]
source or message characteristics.
In the presentstudy, the overall patternof results on the
attitudeand purchaseintentionmeasuresis more consistent
with the ElaborationLikelihood Model formulationthan Abelson, Robert P. (1976), "Script Processing in Attitude For-
with the Sherif, Sherif, and Nebergall (1965) social judg- mation and Decision Making," in Cognition and Social Be-
ment model or with Krugman's (1965, 1967) sequence havior, eds. John S. Carrolland John W. Payne, Hillsdale,
model of involvement. Although increasing involvement NJ: Lawrence Erlbaum.
Ajzen, Icek and MartinFishbein (1980), UnderstandingAttitudes
did produce a main effect on the attitude measure (more
and Predicting Social Behavior, Englewood Cliffs, NJ: Pren-
resistance to the product under high rather than low in- tice-Hall.
volvement), as anticipatedby social judgment theory, the Apsler, Robert and David 0. Sears (1968), "Warning, Personal
more complicated interactions of endorser and argument Involvement, and AttitudeChange," Journal of Personality
quality with involvement cannot be accounted for by the and Social Psychology, 9 (June), 162-166.
theory. Thus the social judgmenttheoryview thatit is more Bem, Daryl J. (1972), "Self-PerceptionTheory," in Advancesin
difficult to change attitudesunder high involvement is, at Experimental Social Psychology, Vol. 6, ed. Leonard Ber-
best, only partiallycorrect, and is unableto accountfor the kowitz, New York: Academic Press, 2-57.
complete patternof attitudedata. The attitudeand behav- Bettman, James R. (1979), "Memory Factors in Consumer
ioral data are genierallyinconsistent with Krugman's se- Choice: A Review," Journal of Marketing, 43 (Spring),
quence formulation. Krugman suggested that under high
Brock, Timothy C. and Sharon Shavitt (1983). "Cognitive Re-
involvement, attitude change preceded behavior change, sponse Analysis in Advertising," in Advertising and Con-
but that underlow involvement, behaviorchange preceded sumer Psychology, eds. Larry Percy and Arch Woodside,
attitudechange. This reasoning would suggest that on im- Lexington, MA: Lexington Books, 91-116.
mediate measures, attitudinaleffects should be easier to Burnkrant,Robert E. (1976), "A MotivationalModel of Infor-
detect than behavioral effects under high involvement, mation Processing Intensity," Journal of Consumer Re-
while behavioraleffects should be easier to detect than at- search, 3 (June), 21-30.

and Alan G. Sawyer (1983), "Effects of Involvementand logical Foundations of Attitudes, eds. Anthony G. Green-
Message Content on InformationProcessing Intensity," in wald, Timothy C. Brock, and Thomas Ostrom, New York:
Information Processing Research in Advertising, ed. Richard Academic Press, 147-170.
Harris, Hillsdale, NJ: LawrenceErlbaum. Haines, George H. (1974), "Process Models of Consumer De-
Cacioppo, John T. and RichardE. Petty (1979), "Effects of Mes- cision Making," in Buver/Consumer Information Process-
sage Repetitionand Position on Cognitive Responses, Recall, ing, eds. G. David Hughes and Michael L. Ray, ChapelHill,
and Persuasion," Journal of Personality and Social Psy- NC: University of North CarolinaPress, 89-107.
chology, 37 (January),97-109. Hovland, Carl I., 0. J. Harvey, and Muzifer Sherif (1957),
and RichardE. Petty (1980a), "Persuasivenessof Com- "Assimilation and Contrast Effects in Reactions to Com-
municationsis Affected by ExposureFrequencyand Message munication and Attitude Change," Journal of Abnormal and
Quality: A Theoreticaland EmpiricalAnalysis of Persisting Social Psychology, 55 (September), 244-252.
Attitude Change," in Ciurrent Issues and Research in Ad- Irving Janis, and HaroldKelley (1953), Communication
vertising, eds. James H. Leigh and Claude R. Martin, Ann and Persuasion, New Haven, CT: Yale University Press.
Arbor, MI: Universityof Michigan, 97-122. Insko, Chester A., E. Allen Lind, and Stephen LaTour (1976),
and RichardE. Petty (1980b), "Sex Differences in Influ- "Persuasion, Recall, and Thoughts," Representative Re-
enceability: Toward Specifying the Underlying Processes," search in Social Psychology, 7, 66-78.
Personality and Social Psychology Bulletini, 6 (December), Janis, Irving L., D. Kaye, and P. Kirschner(1965), "Facilitating
651-656. Effects of 'Eating while Reading' on Responsivenessto Per-
and RichardE. Petty (1981), "Social Psychological Pro- suasive Communication," Journal of Personality and Social
cedures for Cognitive Response Assessment: The Thought Psychology, 1 (February),181-186.
Listing Technique," in Cognitive Assessment, eds. Thomas Kassarjian,HaroldH. (1982), "ConsumerPsychology," Annual
V. Merluzzi, Carol R. Glass, and Myles Genest, New York: Review of Psychology, 33, 619-649.
Guilford Press, 309-342. and Waltraud M. Kassarjian (1979), "Attitudes under
and RichardE. Petty (1982), "The Need for Cognition," Low Commitment Conditions," in Attitude Research Plays
Journal of Personalitv and Social Psychology," 42 (Janu- for High Stakes, eds. John C. Maloney and BernardSilver-
ary), 116-131. man, Chicago: American MarketingAssociation, 3-15.
and RichardE. Petty (forthcoming), "The Need for Cog- Kelman, Herbert C. (1961),."Processes of Opinion Change,"
nition:Relationshipsto Social Influenceand Self Influence," Public Opinion Quarterly, 25 (Spring), 57-78.
in Social Perception in Clinical and Counselinig Psychology, Kiesler, CharlesA., BarryE. Collins, and NormanMiller (1969),
eds. Richard P. McGlynn, James E. Maddux, Cal D. Stol- Attitude Change: A Critical Analysis of Theoretical Ap-
tenberg, and John H. Harvey, Lubbock, TX: Texas Tech proaches, New York: John Wiley.
Press. Krugman,HerbertE. (1965), "The Impactof Television Adver-
Chaiken, Shelly (1980), "Heuristic Versus Systematic Informa- tising: Learning without Involvement," Public Opinion
tion Processing and the Use of Source Versus Message Cues Quarterlv, 29 (Fall), 349-356.
in Persuasion," Journal of Personality and Social Psychol- (1967), "The Measurementof AdvertisingInvolvement,"
ogy, 39 (November), 752-766. Public Opinion Quarterlv, 30 (Winter), 583-596.
Chestnut, Robert W., Charles C. LaChance, and Amy Lubitz Langer, Ellen J. and Robert P. Abelson (1972), "The Semantics
(1977), "The Decorative Female Model: Sexual Stimuli and of Asking a Favor: How to Succeed in Getting Help without
the Recognition of Advertisements," Jouirnal of Advertising Really Dying, "Journal of Personality and Social Psvchol-
Research, 6, 11-14. ogy, 24 (October), 26-32.
Cialdini, Robert B., Richard E. Petty. and John T. Cacioppo , Arthur Blank, and Benzion Chanowitz (1978), "The
(1981), "Attitude and Attitude Change," Annual Reviewvof Mindlessness of Ostensibly ThoughtfulAction: The Role of
Psychology, 32, 357-404. 'Placebic' Informationin InterpersonalInteraction,"Journal
Craik, Fergus M. and Robert S. Lockhart (1972), "Levels of of Personality and Social Psychology, 36 (June), 635-642.
Processing: A Frameworkfor Memory Research," Journal Lastovicka,JohnL. and David M. Gardner(1979), "Components
of Verbal Learning and Verbal Behavior, 11 (October), of Involvement," in Attitude Research Plays for High Stakes,
671-684. eds. John C. Maloney and Bernard Silverman, Chicago:
Cummings, William H. and M. Venkatesan(1976), "Cognitive AmericanMarketingAssociation, 53-73.
Dissonance and Consumer Behavior: A Review of the Evi- Liebhart,Ernst H. (1979), "InformationSearch and Attribution:
dence," Journal of Marketing Research, 13 (August), Cognitive Processes Mediatingthe Effect of False Autonomic
303-308. Feedback," European Journal of Social Psychology, 9, (Jan-
Eagly, Alice H. (1978), "Sex Differences in Influenceability," uary-March), 19-37.
Psychological Bulletin, 85 (January), 86-116. Lutz, RichardJ. and James R. Bettman (1977), "Multiattribute
Engel, James F. and Roger D. Blackwell (1982), ConsumerBe- Models in Marketing:A BicentennialReview," in Consumer
havior, Hinsdale, IL: Dryden Press. and Industrial Buying Behavior, eds. Arch Woodside, Jag-
Festinger, Leon (1957), A Theory of Cognitive Dissoniance, Stan- dish Sheth, and Peter D. Bennett, New York:Elsevier North-
ford, CA: StanfordUniversity Press. Holland, 137-150.
Fishbein, Martinand Icek Ajzen (1981), "Acceptance, Yielding, McGuire, William J. (1969), "The Nature of Attitudes and At-
andImpact:Cognitive Processes in Persuasion," in Cognitive titude Change," in The Hanidbook of Social Psychology, Vol.
Responses in Persuasion, eds. Richard E. Petty, Thomas 3, eds. GardnerLindzey and Elliot Aronson, Reading, MA:
Ostrom, and Timothy C. Brock, Hillsdale, NJ: Lawrence Addison-Wesley, 136-314.
Erlbaum,339-359. (1976), "Some InternalPsychologicalFactorsInfluencing
Greenwald,Anthony G. (1968), "Cognitive Learning, Cognitive Consumer Choice," Journial of Consumer Research, 2
Response to Persuasion, and Attitude Change," in Psycho- (March), 302-319.

Miller, Norman, Geoffrey Maruyama, Rex Julian Beaber, and Ray, Michael L., Alan G. Sawyer, Michael L. Rothschild, Roger
Keith Valone (1976), "Speed of Speech and Persuasion," M. Heeler, EdwardC. Strong and Jerome B. Reed (1973),
Journal of Personality and Social Psychology, 34 (October), "MarketingCommunicationand the Hierarchyof Effects,"
615-624. in New Models for Mass Communication Research, Vol. 2,
Mitchell, Andrew A. and Jerry C. Olson (1981), "Are Product ed. Peter Clarke, Beverly Hills, CA: Sage Publications,
AttributeBeliefs the Only Mediator of Advertising Effects 147-176.
on Brand Attitude?" Journal of Marketing Research, 18 Rhine, Ramon, and LaurenceJ. Severance(1970), "Ego-Involve-
(August), 318-332. ment, Discrepancy, Source Credibility, and Attitude
Mowen, John C. (1980), "On ProductEndorserEffectiveness:A Change," Journal of Personality and Social Psychology, 16
Balance Model Approach," in Current Issues and Research (October), 175-190.
in Advertising, eds. James H. Leigh and Claude R. Martin, Schlenker,BarryR. (1978), "Attitudesas Actions: Social Identity
Ann Arbor, MI: University of Michigan, 41-57. Theory and Consumer Research," Advances in Consumer
Newman, Larry M. and Ira J. Dolich (1979), "An Examination Research, Vol. 5, ed. H. Keith Hunt, Ann Arbor, MI: As-
of Ego-Involvement as a Modifier of Attitude Changes sociation for ConsumerResearch, 352-359.
Caused from Product Testing," Advances in Consumer Re- (1980), Impression Management: The Self-Concept, So-
search, Vol. 6, ed. William L. Wilkie, Ann Arbor, MI: cial Identity, and Inteipersonal Relations, Monterey, CA:
Association for ConsumerResearch, 180-183. Brooks/Cole.
Petty, RichardE. and John T. Cacioppo (1979), "Issue Involve- Schneider, Walter and Richard M. Shiffrin (1977), "Controlled
ment Can Increase or Decrease Persuasion by Enhancing and AutomaticHumanInformationProcessing:I. Detection,
Message-Relevant Cognitive Responses," Journial of Per- Search, and Attention," Psychological Reviewv, 84 (January),
sonality anid Social Psychology, 37 (October), 1915-1926. 1-66.
and John T. Cacioppo (1980), "Effects of Issue Involve- Scott, Carol A. (1978), "Self-PerceptionProcesses in Consumer
ment on Attitudes in an Advertising Context," in Proceed- Behavior: InterpretingOne's Own Experiences," Advanices
ings of the Division 23 Program, eds. Gerald G. Gorn and in Consumer Research, Vol. 5, ed. H. Keith Hunt, Ann
Marvin E. Goldberg, Montreal, Canada:American Psycho- Arbor, MI: Association for ConsumerResearch, 714-720.
logical Association, 75-79. Sherif, Muzifer and Carl I. Hovland (1961), Social Judgment,
and John T. Cacioppo (1981), Attitudes and Persuasion: New Haven: Yale University Press.
Classic and Conitemporarn Approaches, Dubuque, IA: Wil- Sherif, Carolyn W., Muzifer Sherif, and Roger E. Nebergall
liam C. Brown. (1965), Attitude and Attitude Change, Philadelphia: Saun-
and John T. Cacioppo (1983), "Central and Peripheral ders.
Routes to Persuasion:Application to Advertising," in Ad- Steadman, Major (1969), "How Sexy IllustrationsAffect Brand
vertising anid Conisumer Psychology, eds. Larry Percy and Recall," Journal of Advertising Research, 9 (1), 15-19.
Arch Woodside, Lexington, MA: Lexington Books, 3-23. Sternthal, Brian and C. Samuel Craig (1974), "Fear Appeals:
and John T. Cacioppo (forthcominga), AttitudeChange: Revisited and Revised," Journal of ConsumnerResearch, 1
Central and Peripheral Routes to Persuasion, New York: (December), 22-34.
Springer/Verlag. Troutman, C. Michael and James Shanteau (1976), "Do Con-
and John T. Cacioppo (forthcomingb), "The Effects of sumers EvaluateProductsby Adding or Averaging Attribute
Involvementon Responses to ArgumentQuantityand Qual- Information," Journal of Consumer Research, 3 (Decem-
ity: Central and PeripheralRoutes to Persuasion," Journlal ber), 101-106.
of Personality atid Social Psychology. Winer, B. J. (1971), Statistical Principles in Experimental De-
John T. Cacioppo, and Rachel Goldman (1981), "Per- sign, New York: McGraw-Hill.
sonal Involvementas a Determinantof Argument-BasedPer- Worchel, Stephen, Virginia Andreoli, and Joe Eason (1975), "Is
suasion," Journtal of Personality and Social Psychology, 41 the Medium the Message? A Study of the Effects of Media,
(November), 847-855. Communicator, and Message Characteristics on Attitude
, John T. Cacioppo, and MartinHeesacker (1981), "The Change,'' Journal of Applied Social Psychology, 5
Use of RhetoricalQuestions in Persuasion:A Cognitive Re- (April-June), 157-172.
sponse Analysis,"' Journial of Personality anid Social Psy- Wright, Peter L. (1973), "The Cognitive Processes Mediating
chology, 40 (March), 432-440. Acceptance of Advertising," Journlal of Marketing Research,
Thomas Ostrom, and Timothy C. Brock (1981), "His- 10, (February),53-62.
torical Foundationsof the Cognitive Response Approachto (1974), "Analyzing Media Effects on Advertising Re-
Attitudes and Persuasion," in Cogniitive Responises in Per- sponses,'' Public Opiniion Quarterly, 38 (Summer),
sluasion, eds. RichardE. Petty, Thomas Ostrom, and Tim- 192-205.
othy C. Brock, Hillsdale, NJ: LawrenceErlbaum,5-29. (1980), "Message-evokedThoughts:PersuasionResearch
, Gary L. Wells, and Timothy C. Brock (1976), "Dis- Using Thought Verbalizations. Journial of Conisumer Re-
traction Can Enhance or Reduce Yielding to Propaganda: search, 7 (September), 151-175.
Thought DisruptionVersus Effort Justification,"Journial of Zimbardo,Phillip (1960), "Involvementand CommunicationDis-
Personiality and Social PsVchology, 34 (November), crepancyas Determinantsof Opinion Conformity," Journal
874-884. of Abnormal anid Social Psychology, 60 (January), 86-94.