17
The influence of evidence type and product involvement on message-framing effects in advertising Frank E. Dardis 1 * and Fuyuan Shen 2 1 College of Communications, Penn State University, 107 Carnegie Building, University Park, Pennsylvania 16802 2 College of Communications, Penn State University, 22 Carnegie Building, University Park, Pennsylvania 16802 Extrapolating from prior research that describes the persuasive effects of gain- versus loss-framed messages via the heuristic-systematic model (HSM), the current study incorp- orated two advertising-related factors – evidence type (informational vs. exemplar) and product involvement – and examined their influence on message-framing effects in advertisements for commonplace consumer products. A significant interaction in Exper- iment 1 indicated that loss-framed messages were persuasive in a higher-involvement context only when coupled with informational evidence, which enhanced systematic processing among participants and thereby elicited the framing effect. No interaction effects occurred in the lower-involvement context of Experiment 2, in which the hypoth- esized thought-processing patterns did not evince. Consistent with recent theoretical advancements, these results indicate that message-framing effects can be attenuated when both systematic and heuristic processing occur simultaneously. Practical implica- tions are discussed. Copyright # 2008 John Wiley & Sons, Ltd. Introduction Marketing managers and advertising strategists face myriad ways in which to persuade consumers to buy their products. A quite common method of promoting everyday consumer goods is to create advertising messages that emphasize either the benefits of using the product or the risks and/or problems associated with not using the product. For example, many cereal ads inform consumers about how many vitamins and minerals are contained within the food, and how the cereal can be part of a healthy breakfast. Contrast this strategy with the appeal made by a popular trash bag manufac- turer showing garbage and debris spilled all over the kitchen floor, with the exasperated model holding a ripped bag (obviously not the target brand’s) in her hand. In fact, the popular Journal of Consumer Behaviour J. Consumer Behav. 7: 222–238 (2008) Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/cb.247 *Correspondence to: Frank E. Dardis, College of Com- munications, Penn State University, 107 Carnegie Build- ing, University Park, PA 16802. E-mail: [email protected] Copyright # 2008 John Wiley & Sons, Ltd. Journal of Consumer Behaviour, May–June 2008 DOI: 10.1002/cb

The influence of evidence type and product involvement on message-framing effects in advertising

Embed Size (px)

Citation preview

Page 1: The influence of evidence type and product involvement on message-framing effects in advertising

Journal of Consumer BehaviourJ. Consumer Behav. 7: 222–238 (2008)Published online in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/cb.247

The influence of evidence type andproduct involvement onmessage-framing effects inadvertisingFrank E. Dardis1* and Fuyuan Shen2

1College of Communications, Penn State University, 107 Carnegie Building, University Park,

Pennsylvania 168022College of Communications, Penn State University, 22 Carnegie Building, University Park,

Pennsylvania 16802

� E

*Cormuning,E-ma

Cop

xtrapolating from prior research that describes the persuasive effects of gain- versus

loss-framed messages via the heuristic-systematic model (HSM), the current study incorp-

orated two advertising-related factors – evidence type (informational vs. exemplar) and

product involvement – and examined their influence on message-framing effects in

advertisements for commonplace consumer products. A significant interaction in Exper-

iment 1 indicated that loss-framed messages were persuasive in a higher-involvement

context only when coupled with informational evidence, which enhanced systematic

processing among participants and thereby elicited the framing effect. No interaction

effects occurred in the lower-involvement context of Experiment 2, in which the hypoth-

esized thought-processing patterns did not evince. Consistent with recent theoretical

advancements, these results indicate that message-framing effects can be attenuated

when both systematic and heuristic processing occur simultaneously. Practical implica-

tions are discussed.

Copyright # 2008 John Wiley & Sons, Ltd.

Introduction

Marketing managers and advertising strategistsface myriad ways in which to persuadeconsumers to buy their products. A quitecommon method of promoting everydayconsumer goods is to create advertisingmessages that emphasize either the benefits

respondence to: Frank E. Dardis, College of Com-ications, Penn State University, 107 Carnegie Build-University Park, PA 16802.il: [email protected]

yright # 2008 John Wiley & Sons, Ltd. J

of using the product or the risks and/orproblems associated with not using theproduct. For example, many cereal ads informconsumers about how many vitamins andminerals are contained within the food, andhow the cereal can be part of a healthybreakfast. Contrast this strategy with theappeal made by a popular trash bag manufac-turer showing garbage and debris spilled allover the kitchen floor, with the exasperatedmodel holding a ripped bag (obviously not thetarget brand’s) in her hand. In fact, the popular

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 2: The influence of evidence type and product involvement on message-framing effects in advertising

Influencing message-framing effects in advertising 223

‘‘Got Milk’’ campaign has utilized both types ofappeals: many ads have shown physically fitcelebrities espousing the benefits of drinkingmilk, while other ads have comically demon-strated the dangers associated with not drink-ing it – a popular version of which included anolder manwho said he never drankmilk havinghis arms disconnected at the shoulders as hetried to lift a wheelbarrow, thereby promptingon-looking children into anxiously gulpingdown their milk.Another level of strategic consideration that

advertisers face involves the type of evidencethey implement in support of their claims.Many ads use concrete, statistical informationto back up the claims made within them. Forexample, a current cereal ad explains thatusers of the brand have ‘‘lowered theircholesterol levels up to 4% within the firstsix months.’’ Other ads feature exemplars – atypical consumer with whom the audience caneasily identify – to provide claim support. Anexample of this would be a current insurancead that invites viewers to consider ‘‘Ned,’’ a41-year-old, married father of two who hasspecific life-stage and insurance needs. Notethat, although such ads can directly identifyand describe the exemplar, oftentimes theexemplar can be expressed more subtly as an‘‘unnamed’’ consumer in an easily identifiablesituation, much like the distressed bag-holdermentioned above.Although the advertising appeals described

above are implemented quite frequently andperhaps without much knowledge about theireffectiveness on the practical level, theoreticallogic would imply that the varied messagestrategies obviously could have differentialpersuasive impact on consumers’ attitudestoward the brands being advertised. Forexample, perhaps the persuasiveness of eachstrategy type would vary depending on theconsumer’s involvement level with the pro-duct. A major stream of message-framingresearch has demonstrated that, when infor-mation is processed or elaborated upon byindividuals, loss-framed messages (those thatemphasize the potential harms of not perform-ing an action) tend to be more persuasive in

Copyright # 2008 John Wiley & Sons, Ltd. J

eliciting desired behaviors than gain-framedmessages (those emphasizing the benefits ofperforming the action).Though interesting to marketers from a

social-psychology perspective, much of thisresearch focuses on health-related concernsand provides little insight regarding individ-uals’ attitudes and behavior in traditionalconsumer-oriented contexts, in which thepurchase of commonplace products – andnot one’s health or life – hang in the balance.Therefore, the goal of the current study is toprovide linkage among common advertisingpractice, message-framing research, and con-sumer-behavior theory in the realm of everydayconsumer products. Specifically, the currentstudy relates two advertising-based factors –evidence type and consumers’ product invol-vement – to extant message-framing research,then tests these factors under the aegis of theRossiter–Percy Grid (Rossiter et al., 1991) todetermine if the effects of gain- versusloss-framed messages demonstrated in priorresearch can evince within the context ofadvertising for ordinary consumer products. Ifeffects can be replicated in the more mundane,product-based setting, not only would thisprovide theoretical linkage between the realmsof message framing and consumer behavior,but findings also could offer practical advice tomarketers who now could implement specificcommunicative strategies based on insight thatwas quite established in one domain, butwhich was not applied to or incorporated intoother contexts.

Literature review

Message framing and involvement

Message framing research fundamentally stemsfromKahneman and Tversky’s (1979) advance-ment of prospect theory, which demonstratedthat the framing of equivalent information interms of either potential gains or lossesinfluenced people’s decision-making processand choices (Tversky and Kahneman, 1981).Although originally conceptualized to implymessage differentiation in terms of gain (loss)

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 3: The influence of evidence type and product involvement on message-framing effects in advertising

224 Frank E. Dardis and Fuyuan Shen

to increase (decrease) perceived risk, messageframing – at least in much behavior-relatedpersuasion literature – typically has beenoperationalized as communication that empha-sizes either the potential positive benefitsassociated with doing something (gain fram-ing) or the potential negative consequences ofnot performing the action (loss framing). Levinet al. (1998) referred to this modified type offraming as ‘‘goal’’ framing, because theultimate question does not involve influencinga choice between two different behaviors, butrather involves determining which of twocommunicative strategies is best in producingthe same desired behavior (i.e., the ‘‘goal’’).Note that ‘‘gain-versus-loss’’ framing there-

fore is dissimilar to the ‘‘positive-versus-negative’’ framing operationalized inmuch research regarding traditional consumerproducts (e.g., Levin, 1987; Levin and Gaeth,1988; Donovan and Jalleh, 1999), in which aparticular product attribute is framed equally,but in terms of positive or negative language(e.g., meat being ‘‘75% lean’’ vs. ‘‘25% fat’’).Specifically, while the simple framing ofproduct attributes in a positive versus negativemanner does not encompass any behavioralcomponent, gain-versus-loss framing inher-ently implies some type of consequenceresulting from some behavioral (in)action.Therefore, the gain/loss mode under examin-ation in the current study is theoretically andconceptually independent from the framingapproach adopted in prior product-relatedresearch.Because it entails a behavioral consequence,

most research on gain/loss framing hasoccurred within health-promotion contexts,including topics like breast cancer (Meyero-witz and Chaiken, 1987; Banks et al., 1995;Schneider et al., 2001a), smoking (Wilsonet al., 1990; Schneider et al., 2001b), and skincancer (Rothman et al., 1993; Detweiler et al.,1999). In fact, most major gain/loss-framingstudies in the marketing literature also havefocused on health-related concerns or pro-ducts (e.g., Maheswaran and Meyers-Levy,1990; Block and Keller, 1995; Cox and Cox,2001; Meyers-Levy and Maheswaran, 2004).

Copyright # 2008 John Wiley & Sons, Ltd. J

Generally, these studies have shown lossframing to be more influential than gainframing in high-risk or high-uncertainty situ-ations, due to greater perceived messageimportance (i.e., higher involvement) andgreater cognitive processing among receivers.The research also has indicated that, wheninformation was less cognitively demanding,either the framing effects disappeared (e.g.,Rothman et al., 1993; Block and Keller, 1995)or gain framing was found to be more effective(e.g., Maheswaran and Meyers-Levy, 1990;Rothman and Salovey, 1997; Rothman et al.,1999). Scholars suggest these findings can beexplained by the heuristic-systematic model(HSM) (Eagly and Chaiken, 1993), which issimilar to the Elaboration Likelihood Model(ELM) (Petty and Cacioppo, 1986) in thatpersuasion is assumed to occur via twodifferent modes of information processing.

When information is deemed important orpersonally relevant, individuals partake insystematic processing via scrutinizing andcognitively elaborating upon the content of amessage (e.g., Petty and Cacioppo, 1983;Maheswaran and Meyers-Levy, 1990). Loss-framed messages are theorized to be morepersuasive under such conditions becausenegative information usually is perceived asmore informative, diagnostic, and consequen-tial than positive information (e.g., Fiske,1980), similar to the so-called ‘‘negativityeffect’’ (e.g., Ahluwalia et al., 2000; Ahluwalia,2002). Conversely, when an individual’s invol-vement is low, less cognitive elaborationoccurs and persuasion is based to a greaterextent on heuristic, peripheral cues. In thiscase, scholars suggest that gain-framedmessages have the potential to be morepersuasive due to their general positivity andhedonic properties (Meyers-Levy and Mahes-waran, 2004; Shiv et al., 2004) as opposed tothe cogency of the information containedtherein.

However, HSM further holds that, inaddition to being able to persuade individualsindependently, the two modes of processingcan occur simultaneously (e.g., Eagly andChaiken, 1993; Chaiken and Maheswaran,

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 4: The influence of evidence type and product involvement on message-framing effects in advertising

Influencing message-framing effects in advertising 225

1994). Applying this notion to message fram-ing, one study (Meyers-Levy and Maheswaran,2004) demonstrated that the persuasive effectsof loss-framed messages in a high-relevancecontext – previously assumed to be quitecustomary – could be eliminated whensimultaneous processing occurred. Specifi-cally, the authors noted an interaction inwhich loss-framedmessages about a product tohelp control cholesterol were more persuasivethan gain-framed messages only when systema-tic processing was at its apex, which occurredwhen: (1) the information was deemed high inpersonal relevance and (2) the messagepresented high perceived-risk implications. Ifthe message was low in either personalrelevance or perceived risk, loss-framedmessages lost their persuasive capacity. Thus,not only did the subject matter have to bepersonally relevant, but the information con-tained in the message itself also had to elicitaugmented systematic processing among indi-viduals for loss-framed messages to demon-strate any effects.

Evidence type and cognitive processing

From a theory-driven standpoint, informationalcontent itself also has the potential to influencethe amount of elaboration that individualsexpend in processing a message. According toChaiken and Maheswaran (1994), when sys-tematic and heuristic processing co-occur, thelatter can mitigate or influence the former‘‘when message content . . . is ambiguous andthus amenable to differential interpretation orevaluation’’ (p. 469). On the other hand, theauthors also found that the presentation ofcogent, unambiguous information led directlyto systematic processing among subjects (incertain conditions), in which case the influ-ence of heuristic cues was duly eliminated andattitudes were formed via information-basedscrutiny. Thus, if the informational content ofan advertisement can affect message proces-sing, this implies that individuals may partakein simultaneous systematic and heuristicprocessing when perceived involvement or

Copyright # 2008 John Wiley & Sons, Ltd. J

relevance is high but the informational contentdoes not elicit the required level of systematicprocessing. In this case, higher perceivedinvolvement is a necessary – but not sufficient– condition for systematic processing to occur;the type of information being offered also mustaffect consumers’ thought processing appro-priately.Hence, the way in which information is

presented in a message is an important choicethat can greatly affect cognitive activity amongreceivers and, subsequently, the message’spersuasiveness. Wood et al. (1985) demon-strated that strong messages (manipulated viastatistical evidence and credible sources) weremore persuasive than weaker claims (e.g., anopinion in a testimonial) when subjectsprocessed and evaluated the validity of theinformational content. In contrast, subjectswho did not actively scrutinize the messagewere not persuaded by the informationcontained in the message but rather byperipheral, heuristic message-based cues(e.g., length, structural features). Hafer et al.(1996) also held that if subjects were motiv-ated to engage in scrutiny, clear arguments ledto greater persuasion via systematic processingthan did arguments in which the claims werenot as easily understood. The authors alsodemonstrated that persuasion was based onnon-informational cues when argument claimswere not clearly supported in the message.From a practical standpoint, many adver-

tisements employ concrete, factual data tosupport their claims, as described in theprevious section. However, researchers andpractitioners obviously advocate the imple-mentation of other message tactics. TheRossiter–Percy Grid (Rossiter et al., 1991)suggests that informational appeals are notrecommended in all situations and that other(transformational) appeals are best used incertain contexts (see Figure 1). Interestingly,the Rossiter–Percy Grid cross-sections advo-cated appeal type by consumers’ involvementwith the product purchase, a factor also underexamination in the current study. Therefore,by investigating these two factors simul-taneously while also incorporating message

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 5: The influence of evidence type and product involvement on message-framing effects in advertising

Figure 1. The Rossiter–Percy Grid with typical product categories in each quadrant. Note. Adapted from ‘‘A BetterAdvertising Planning Grid,’’ by Rossiter JR, Percy L, Donovan RJ, 1991, Journal of Advertising Research, 31(5), 13.*Product categories added by current authors.

226 Frank E. Dardis and Fuyuan Shen

framing, the current study adds a thirddimension – valence of message – to the gridand empirically tests it, which could providemore specific insight to scholars and prac-titioners alike.As opposed to offering concrete information

as evidence, another popular advertisingstrategy is to provide claim support througha consumer exemplar – an everyday personwith whom the audience can easily identify.Although the influence of base-rate infor-mation versus individual cases (exemplars)has been studied a great deal in the fields ofbehavioral decision-making (e.g., Lynch andOfir, 1989; Markoczy and Goldberg, 1998) andnews reports (e.g., Zillman, 2002), there is adearth of research investigating the persua-siveness of informational versus exemplarevidence in common product advertising,especially in conjunction with message-framing. However, some researchers haveexamined the impact of informational versustransformational advertising, two conceptssimilar to the evidence-type conceptualiz-ations in the present study.According to scholars, informational adver-

tising emphasizes the attributes of the product

Copyright # 2008 John Wiley & Sons, Ltd. J

through explicit or testable claims (Laskeyet al., 1989), whereas transformational adver-tising focuses on consumer associations orimagery (e.g., exemplars), among other strat-egies (Pae et al., 2002). There has been somesupport for the use of factual, informationaladvertising. Two studies have found thatinformational messages were more persuasive(Laskey et al., 1992) and produced morefavorable attitudes toward the ad and greaterbuying intentions (Laskey et al., 1994) than didtransformational ads. Additionally, although itdid not involve advertising per se, anotherstudy demonstrated that statistical evidence ina textual passage was more persuasive thananecdotal evidence, even over an extendedtime period (Baesler and Burgoon, 1994).

It is not known if the effects reported in theabove literature were attributable to differ-ences in subjects’ information processingbecause no cognitive measures were attained.However, it would seem that the presentationof statistical evidence in support of productclaims logically requires more cognitive elab-oration than do anecdotal or associativemessages. Such logic is implicit in thedefinition of informational advertising, which

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 6: The influence of evidence type and product involvement on message-framing effects in advertising

Influencing message-framing effects in advertising 227

provides ‘‘consumers with factual. . . data in aclear and logical manner such that they havegreater confidence in their ability to assess themerits of buying the brand’’ (Puto and Wells,1984, p. 638). Therefore, we aim to investigatethe impact that evidence type might exertupon message-framing effects through thedifferential cognitive processing it elicitsamong individuals in both higher- and low-er-involvement contexts, just as perceived riskhas been examined in other message-framingcontexts. As stated, although perceived invol-vement may be higher, a lack of systematicprocessing and an increase in simultaneousprocessing can occur if the informationalcontent of an advertisement does not elicitproper cognitive activity among individuals.We operationalize product-class involve-

ment based on Zaichkowsky’s (1985)Personal Involvement Inventory, which gauges‘‘perceived personal relevance of the objectbased on inherent needs, values, and interests’’(p. 342). The current operationalization(described in the Methodology Section)includes elements extremely similar to thoseused inmost prior message-framing research tooperationalize perceived involvement orpersonal relevance. Therefore, we expectproduct-class involvement to influence messa-ge-framing effects in a manner similar to thepreviously studied factors of perceivedpersonal relevance or felt involvement.

Hypotheses

Based upon the above review, we propose thatinformational evidence will elicit greatersystematic processing among individuals in ahigher-involvement condition than will exem-plar evidence, thereby influencing message-framing effects in a manner theoreticallyconsistent with the perceived-risk scenariosexamined in prior research:

H1: When product involvement is higher,

loss framing will be more effective than

gain framing, but only when informa-

tional evidence is employed.

Copyright # 2008 John Wiley & Sons, Ltd. J

In addition to measuring attitudinal judg-ments, prior research also has reported parti-cipants’ thought measures to affirm that theanticipated types of information processingoccurred in appropriate treatment conditions(Maheswaran and Meyers-Levy, 1990; Meyers-Levy and Maheswaran, 2004). The authorsfound that ‘‘message-related thoughts’’ (opera-tionalized as an indicator of systematic proces-sing) were greater among subjects in a high-relevance/high-risk condition when comparedto subjects in a high-relevance/low-risk con-dition. Accordingly, the high-relevance/high-risk subjects also generated fewer ‘‘simpleevaluative thoughts,’’ an indicator of heuristicprocessing. Using identical operationaliza-tions, we expect individuals in the higher-involvement/informational condition to demon-strate a similar pattern of thought processing:

H2: When product involvement is higher,

individuals who read an informational ad

will exhibit (a) more systematic processing

and (b) less heuristic processing than

individuals who read an exemplar ad.

The above hypotheses were investigatedin Experiment 1, which used a higher-involvement scenario. However, we alsowanted to examine any potential relationshipsin a lower-product-involvement condition.Research has indicated that gain-framedmessages may exert greater influence in alow-involvement context (e.g., Maheswaranand Meyers-Levy, 1990; Martin and Marshall,1999; Cox and Cox, 2001) because, whenindividuals do not systematically processinformation, the innately positive and hedonicproperties contained within gain-framedmessages can be more appealing and, thusly,more persuasive. However, other studies havedemonstrated that gain-framedmessages mightnot have any relative influence over loss-framed messages in low-involvement contextsbecause individuals may not always feel theneed to attend to a message (Gleicher andPetty, 1992) and, when this occurs, thepersuasiveness of any particular message-based strategies or factors simply can disap-

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 7: The influence of evidence type and product involvement on message-framing effects in advertising

228 Frank E. Dardis and Fuyuan Shen

pear due to the lack of perceived importanceof the information (Rothman et al., 1993; Blockand Keller, 1995). In acknowledging that nullmessage-framing effects are not ‘‘amenable to aunique explanation’’ (p. 160), Meyers-Levy andMaheswaran (2004) showed that any relativesuperiority of gain-framed messages in a low-involvement context is predicated upon themessage processing undertaken by individuals.Specifically, the authors demonstrated thatgain-framed messages were superior to loss-framed messages in a low-relevance contextonly when individuals partook in enhancedheuristic (and lower systematic) processing.Therefore, we offered the following hypoth-eses for the lower-involvement context ofExperiment 2:

H3: When product involvement is lower,

gain framing will be more effective than

loss framing, particularly when exemplar

evidence is employed.

H4: When product involvement is lower,

individuals who read an exemplar ad will

exhibit (a) more heuristic processing and

(b) less systematic processing than indi-

viduals who read an informational ad.

Methodology

Before conducting the two experiments,39 individuals who did not participate in themain experiments were recruited to completea questionnaire designed to measure theirinvolvement with several product categories.Respondents indicated their level of agreement(1¼ disagree, 7¼ agree) with the following sixstatements (see Zaichkowsky, 1994): (1) I havea lot of experience with this product, (2)Compared to the average person, I would saythat I was highly knowledgeable about thisproduct category, (3) I would describe myselfas being very familiar with this productcategory, (4) Compared to other products,this product is important to me, (5) I am notinterested in this product, and (6)When I buy abrand from this product category, I choose

Copyright # 2008 John Wiley & Sons, Ltd. J

very carefully (a¼ 0.89). A paired-samplest-test indicated that subjects’ involvementwith paper shredders was lower than that ofmouthwash,M: 2.68 versus 4.80, t(38)¼ 6.22,p< 0.01. Accordingly, mouthwash wasselected as the higher-involvement productfor Experiment 1 and a paper shredder waschosen as the lower-involvement product forExperiment 2.

It is important to note that, although itobviously was rated as more involving thanpaper shredders, mouthwash may not seemlike a typical ‘‘high’’ involvement product.However, the authors purposely selectedmundane consumer goods because the goalof the current study was to examine the effectsof advertising on commonplace purchases thatwere realistically pertinent to the participantsunder study. Further, if differential effectscould be found in the current mundane-product contexts, it seems likely that theeffects still would hold – and perhaps evenincrease – in cases of higher involvementpurchases that occur less frequently.

Experiment 1

Method

Participants

For extra credit in a university course, 196undergraduate students participated in Exper-iment 1, which used a 2 (informational orexemplar evidence)� 2 (gain or loss framing)between-subjects design. Participants wereassigned randomly to one of four experimentalconditions in which the stimulus advertise-ment was created via manipulating the twomain factors within its copy: gain/informa-tional, gain/exemplar, loss/informational, andloss/exemplar.

Procedure

After being assigned to a specific condition,each participant was presented with anexperimental booklet explaining that s/hewas agreeing to participate in a print advertis-

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 8: The influence of evidence type and product involvement on message-framing effects in advertising

Influencing message-framing effects in advertising 229

ing study. Participants were instructed thatthey would then be exposed to several printadvertisements that were still in development,and therefore would be in ‘‘mock-up’’ ordummy layout, containing no pictures orvisual elements but resembling full-pageform. The three advertisements used black-and-white, text-only messages. With the ex-ception of the message manipulations, thelayout and content of the ads were heldconstant across conditions.Participants were instructed to read the ads

as they normally would at their own speed.Each ad introduced a fictitious new brandname in its particular product category.Presentation order of the ads was the samein all conditions: the manipulated ad for Rinze

mouthwash was inserted between two fillerads. After reading the ads, participants weretold that they would be asked questions aboutonly one of the ads/products to which theywere just exposed, but were unaware thateveryone was assigned to the mouthwash ad.Participants then responded to several scaleditems comprised of dependent measures andmanipulation checks, and completed a thought-listing task. Upon completion of the post-stimulus questionnaire, they were debriefedand thanked for their participation.

Independent variables

Manipulations of independent variables withinthe messages were modeled after the Cox andCox (2001) study. Ads explained either thebenefits of using the mouthwash (gain fram-ing) or the unfavorable consequences of notusing it (loss framing), and evidence type wasmanipulated via using either statistical infor-mation (informational) or an example of aproduct user (exemplar) to support productclaims.

Dependent measures

To determine the persuasive effects of the ad,three dependent variables derived from scalemeasures were attained: purchase intention,attitude toward the brand, and attitude toward

Copyright # 2008 John Wiley & Sons, Ltd. J

the ad. For all dependent variables, measureswere collected via 7-point, semantic differen-tial scales. Purchase intention (see Zhang andBuda, 1999) was measured via three itemsanchored by unlikely/likely, improbable/prob-able, and impossible/possible (a¼ 0.92). Atti-tude toward the brand and attitude toward thead were measured via two scales adapted fromShiv et al. (1997). Attitude toward the brandwas measured via five items in which partici-pants rated their overall evaluation of Rinze

mouthwash: bad/good, low quality/high qual-ity, not likeable/likeable, unpleasant/pleasant,and unappealing/appealing (a¼ 0.90). Atti-tude toward the ad was measured via threeitems with anchors of bad/good, unappealing/appealing, and not likeable/likeable (a¼ 0.94).Systematic/heuristic processing was mea-

sured by an open-ended question in whichparticipants were asked to list up to fourthoughts that ‘‘crossed their mind’’ whenreading the mouthwash ad. Using a thought-coding scheme identical to Meyers-Levy andMaheswaran (2004) and Maheswaran andMeyers-Levy (1990), two independent judgesblind to the experimental design codedresponses into the following three categories:message-related thoughts that contained expli-cit reference to information contained in the ad(e.g., ‘‘Users 50% less likely to have gumdisease’’), simple evaluative thoughts thatrepresented global liking (e.g., ‘‘Rinze is agood product’’), and ‘‘other’’ thoughts notrelevant to either the ad or the informationcontained therein (e.g., ‘‘My dad is a dentist’’).Intercoder reliability was 0.88 (Holsti, 1969).

Manipulation checks

Prior to being exposed to any stimuli,participants were asked to indicate their levelof involvement with paper shredders andmouthwash, among other products, usingthe same questionnaire as the pretest group.A paired-samples t-test indicated that partici-pants were more involved with mouthwashthan with paper shredders, Ms¼ 4.44 versus1.78, t(192)¼ 21.24, p< 0.001. During thepost-stimulus questionnaire, participants also

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 9: The influence of evidence type and product involvement on message-framing effects in advertising

Table 1. Means and standard deviations on scaled dependent measures: Experiment 1(higher-involvement product)

n Purchase int. A-brand A-ad

Gain framingInformational 49 3.67 (1.67) 4.58 (1.10) 4.40 (1.28)Exemplar 49 3.63 (1.50) 4.19 (1.25) 4.10 (1.55)

Loss framingInformational 49 4.51 (1.65) 4.73 (1.21) 4.57 (1.25)Exemplar 49 3.55 (1.58) 4.02 (1.24) 3.12 (1.42)

Note: Standard deviations are in parentheses. Scores are presented on a 1–7 scale to enhancecomparability.

230 Frank E. Dardis and Fuyuan Shen

indicated their level of agreement with twoitems gauging whether the stimulus ad con-tained primarily statistical information or useda consumer example. Results revealed thatparticipants in each condition correctly ident-ified the type of evidence presented in thestimulus ad: F(1,194)¼ 138.87, p< 0.001, onthe first item; F(1,194)¼ 153.97, p< 0.001, onthe second item. Similar analysis indicated thatmessage framing also was manipulated suc-cessfully. When asked on seven-point scale ifthe ad emphasized the advantages of using theproduct (7) or the disadvantages of not usingthe product (1), the mean of participants in thegain-framed condition was significantly higherthan the mean of those in the loss-framedcondition, t(194)¼ 4.24, p< 0.001.

Results

Descriptive statistics are presented in Table 1.To examine H1, the three dependent variables

Table 2. Results of MANOVA and ANOVA on scaled depeproduct)

Multivariate

Effect Wilks’s l Effect size df F

F 0.915 0.085 3/190 5.86E 0.904 0.096 3/190 6.70F� E 0.944 0.056 3/190 3.75

Note: F, framing (gain/loss); E, evidence type (information�p< 0.05; ��

p< 0.01; ���p< 0.001.

Copyright # 2008 John Wiley & Sons, Ltd. J

were submitted to MANOVA with messageframing (gain/loss) and evidence type (infor-mational/exemplar) as the independent vari-ables. As demonstrated in Table 2, resultsindicated significant main effects for messageframing (p< 0.001) and evidence type(p< 0.001), as well as a significant interactioneffect (p< 0.05).

Subsequent univariate analyses indicated amain effect of evidence type on all threedependent variables, for which means ofparticipants who received informational evi-dence were higher than means of thosereceiving exemplar evidence. A main effectalso was revealed for message framing onattitude toward the ad, in which gain-framedmessages were rated higher than loss-framedmessages. However, main effects were over-ridden by significant interaction effects on twoof the three dependent measures: purchaseintention (p< 0.05) and attitude toward the ad(p< 0.01). Therefore, these results providedlimited support for H1.

ndent measures: Experiment 1 (higher-involvement

Univariate F (df¼ 1/192)

p Purchase int. A-brand A-ad

0.001 2.72 0.01 4.13�

0.000 4.78� 10.34�� 19.46���

0.012 4.03� 0.86 8.58��

al/exemplar).

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 10: The influence of evidence type and product involvement on message-framing effects in advertising

Influencing message-framing effects in advertising 231

As hypothesized, the interactions demon-strated that loss-framed ads were persuasiveonly when the evidence was informationalrather than exemplar (see Figure 2). Beyondinteraction effects, post-hoc comparisons ofmeans indicated that the loss/informationalmessage significantly outperformed the loss/exemplar message on all dependent measures(for all t values, p< 0.01). Thus, we can assertthat the loss-framed ad was more persuasiveonly when it was coupled with informational,rather than exemplar, evidence. Conversely,similar post-hoc comparisons indicated thatevidence type had no effect on the persua-siveness of the gain-framed messages (p>

Figure 2. Interaction of evidence type and message frami

Copyright # 2008 John Wiley & Sons, Ltd. J

0.05). These comparisons add support to H1 byaffirming that evidence type is an importantfactorwhen presenting loss-framedmessages ina higher-involvement context, but it does notinfluence the effects of gain-framed messages.In support of the explanation that these

findings were due to differential cognitiveprocessing undertaken by participants, H2aand H2b were supported regarding thoughtmeasures. Participants exposed to the informa-tional ad generated significantly more messa-ge-related thoughts (M¼ 1.22) than did thoseexposed to the exemplar ad (M¼ 0.57),F(1,192)¼ 18.90, p< 0.001, and also gener-ated significantly fewer simple evaluative

ng in Experiment 1.

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 11: The influence of evidence type and product involvement on message-framing effects in advertising

232 Frank E. Dardis and Fuyuan Shen

thoughts (Ms¼ 0.99 vs. 1.51), F(1,192)¼9.80, p< 0.01. This confirms that participantswho received an ad with informationalevidence underwent more systematic proces-sing (and also less heuristic processing) thanparticipants who received an ad containingexemplar evidence.

Discussion

In support of prior research (Meyers-Levy andMaheswaran, 2004), our findings confirmedthat message-framing effects can be influenc-ed by factors that either enhance or mitigateheuristic and systematic processing. Resultsindicated that evidence type differentlyaffected the persuasiveness of loss-framedmessages in a higher-involvement context viaeliciting differential cognitive processing.Specifically, loss-framed messages using infor-mational evidence were more persuasivethan loss-framed messages using exemplarevidence. This finding agrees with the theo-rized notions that information-based argu-ments are more persuasive when personsare motivated to process and elaborate onthe content of a message, in which casetheir attitudes are affected more by thesystematic rather than heuristic route (Haferet al., 1996).On the other hand, claims thatwere based on

exemplar evidence rather than concrete evi-dence did not directly present motivatedreaders with information that could be system-atically elaborated upon, in which case heur-istic processing biased systematic processing(Chaiken and Maheswaran, 1994) and sub-sequently attenuated potential framing effects.As Maheswaran and Meyers-Levy (1990) noted,‘‘the content of the negative informationshould be both highly informative [italicsadded] and of high perceived consequence’’ (p.362) in order for any subsequent framingeffects to occur. Thus, even in a highly relevantor higher-involvement context, a loss-framedmessage containing information that does notelicit adequate systematic-route processing canbe devoid of persuasiveness.

Copyright # 2008 John Wiley & Sons, Ltd. J

Experiment 2

Method

Procedure

A total of 198 undergraduate students partici-pated in this experiment for extra credit. The2� 2 message framing by evidence type designproduced four treatment conditions identicalto Experiment 1 into which participantswere assigned randomly. The filler ads werethe same as in Experiment 1 and all ads werepresented in the same order as in the previousexperiment, the only difference being that thefictitious Docu-Shred office paper shredderwas substituted for the fictitious mouthwashbrand.

Dependent measures

Scales identical to those used in Experiment 1were used to measure the three persuasion-based dependent variables in Experiment 2:purchase intention (a¼ 0.90), attitude towardthe brand (a¼ 0.92), and attitude towardthe ad (a¼ 0.94). Thought measures wererecorded by the same coders from Experiment1 and were classified into the same threecategories. Intercoder reliability for thoughtmeasures in Experiment 2 was 0.87.

Manipulation checks

Following the same procedures described inExperiment 1, participants in Experiment 2indicated that they were more involved withmouthwash than with paper shredders,Ms¼ 4.39 versus 2.03, t (194)¼ 21.24,p< 0.001. As in Experiment 1, results alsorevealed that participants correctly identifiedthe type of evidence presented in the stimulusad: F(1,197)¼ 161.07, p< 0.001, on the firstitem; F(1,197)¼ 66.34, p< 0.001, on thesecond item. Also, participants correctlyidentified whether the ads emphasized theadvantages of product usage or the disadvan-tages of non-usage, t(197)¼ 5.19, p< 0.001.

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 12: The influence of evidence type and product involvement on message-framing effects in advertising

Table 3. Means and standard deviations on scaled dependent measures: Experiment 2(lower-involvement product)

n Purchase int. A-brand A-ad

Gain framingInformational 49 4.76 (1.39) 4.70 (1.00) 4.33 (1.32)Exemplar 50 4.42 (1.20) 4.38 (1.10) 3.81 (1.33)

Loss framingInformational 50 4.65 (1.45) 4.45 (1.21) 3.75 (1.69)Exemplar 49 3.97 (1.45) 4.04 (1.35) 3.33 (1.49)

Note: Standard deviations are in parentheses. Scores are presented on a 1–7 scale to enhancecomparability.

Influencing message-framing effects in advertising 233

Results

Descriptive statistics are presented in Table 3.As indicated in Table 4, multivariate analysisyielded no main or interaction effects. Norwere there any significant interaction effectson any dependent measures in subsequentunivariate analyses. Therefore, H3 was notsupported.Interestingly, univariate analyses indicated

that a significant main effect emerged forevidence type on all three dependentmeasures, for which means of participantsexposed to informational evidence weresignificantly higher than those of participantsexposed to exemplar evidence. Additionally, asignificant main effect was revealed formessage framing on attitude toward the ad,in which gain-framed messages were ratedhigher by participants than were loss-framedmessages.In support of H4a, thought measures

revealed that participants who received the

Table 4. Results of MANOVA and ANOVA on scaled depeproduct)

Multivariate

Effect Wilks’s l Effect size df F

F 0.968 0.032 3/192 2.13E 0.961 0.039 3/192 2.62F� E 0.992 0.008 3/192 0.49

Note: F, framing (gain/loss); E, evidence type (information�p< 0.05; ��

p< 0.01.

Copyright # 2008 John Wiley & Sons, Ltd. J

exemplar ad (M¼ 1.54) generated significantlymore simple evaluative thoughts than didparticipants exposed to the informational ad(M¼ 1.15), F(1,184)¼ 5.98, p< 0.05. How-ever, H4b was not supported in that messa-ge-related thoughts did not differ significantlybetween the two groups. Thus, the theoreticalpattern of systematic versus heuristic proces-sing anticipated and confirmed in the high-er-involvement context of Experiment 1 (H2)was only partially revealed in the lower-involvement context of Experiment 2 (H4).

Discussion

Results indicated that the combination ofmessage framing and evidence type did notelicit the hypothesized persuasiveness in thelower-involvement condition. Although H3was not supported, findings are not entirelyinconsistent with prior research demonstrat-ing that framing effects simply may disappear

ndent measures: Experiment 2 (lower-involvement

Univariate F (df¼ 1/194)

p Purchase int. A-brand A-ad

0.097 1.97 3.14 6.46�

0.052 6.75�� 4.75� 5.08�

0.690 0.78 0.07 0.07

al/exemplar).

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 13: The influence of evidence type and product involvement on message-framing effects in advertising

234 Frank E. Dardis and Fuyuan Shen

in low-involvement contexts. In describingtwo seminal studies in which gain frames wereshown not to be more persuasive in low-involvement contexts, Meyers-Levy andMahes-waran (2004) posited that the condition mayhave been so low in personal relevance that it‘‘greatly reduced [subjects’] message proces-sing, making [subjects] entirely insensitiveto variations in message framing’’ (p. 161).Similarly, in the current study, message-framing manipulations were rendered incon-sequential.

General discussion

The main contribution of the current study isits demonstration that two advertising-basedfactors could reproduce persuasive results thatpreviously had been based on other factors inmessage-framing literature, thereby providinglinkage and insight – on both theoretical andpractical levels – to more common, consumer-behavior-based contexts. This study proposedthat two factors important in traditionalproduct advertising – evidence type andproduct involvement – could influence messa-ge-framing effects via eliciting differentialcognitive-processing patterns among individ-uals. As hypothesized, we found that evidencetype differently affected systematic thoughtprocessing in a higher-product-involvementcontext and therefore impacted message-framing effects. However, the hypothesizedrelationships between evidence type andcognitive processing were not revealed inthe lower-involvement context, therebyobviating message-framing effects. The currentresults provided support to other findingsindicating that loss-framed messages may notalways be persuasive in a higher-involvementcontext, due to variations in systematic versusheuristic processing among individuals(Meyers-Levy and Maheswaran, 2004).The current study also demonstrated con-

ditions under which loss framing does and,rather importantly, does not persuade individ-uals in a higher-involvement context. Whenparticipants in the higher-involvement context

Copyright # 2008 John Wiley & Sons, Ltd. J

were presented with loss-framed messageclaims that did not elicit adequate systematicprocessing (exemplars), their attitudes werebiased by simultaneous heuristic and systema-tic processing (Chaiken and Maheswaran,1994), which subsequently obviated anyframing effects (Meyers-Levy and Maheswaran,2004). Shiv et al. (2004) suggested that,‘‘When the level of processing motivation ishigh, consumers are likely to engage insystematic processes where they scrutinizethe claims in greater detail, since they caremore about making accurate judgments’’ (p.200). Accordingly, in the higher-involvementcontext, appropriate levels of systematicprocessing manifested only when messageclaims were supported by informational ratherthan exemplar evidence, and only then wascognitive elaboration sufficient enough for theloss-framed message to be more persuasive.

In support of some prior research (Gleicherand Petty, 1992; Block and Keller, 1995), thisstudy also demonstrated that no such inter-action between message framing and evidencetype occurred in the lower-involvement con-text because individuals were not motivated toscrutinize a great deal of information about thebrand. It is possible that perceived personalinvolvement with a product could be so lowthat any information, regardless of how it isframed or presented, would not produceconsequential effects because it would notbe sufficiently elaborated upon.

On a practical level, the current study canassist marketers in determining effective waysin which to relay the benefits of product usageto consumers. Specifically, the current resultsoffer suggestions regarding when and howmarketers should utilize loss-framed messagesin a consumer-goods context. If an advertisedbrand is in a product class that is relevantto a consumer, current findings indicatethat the most effective way to enhancepurchase intention is to employ loss-framedmessages that are coupled with informationalevidence.

In addition, the current research furtherinforms marketers that their desired advertis-ing effects may actually boomerang when

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 14: The influence of evidence type and product involvement on message-framing effects in advertising

Influencing message-framing effects in advertising 235

loss-framed messages are coupled with exem-plar appeals – a strategy that, at first glance,might seem like a logical or reasonable way toadvertise a brand in a higher-involvementcontext. To the contrary, the current findingssuggest that such appeals actually can bedetrimental to a brand rather than beneficial.For example (assuming that insurance isimportant to most married men over 40 withtwo children), creators of the exemplar-based‘‘Ned’’ ad described in the introduction mightbe better served elucidating the negativeconsequences of not having insurance viastatistical or informational claims, such asthe likelihood of having health problems asone gets older, the percentage of families thatare uncovered or that receive poor reimburse-ments from current insurance providers, etc.Thus, the current study provides valuablepractical insight into not only when and howto best utilize loss-framed appeals in advertis-ing messages, but also when and how not touse them. By seemingly using message-framingstrategies and different types of evidencealmost interchangeably to promote traditionalconsumer products, advertising creators maynot fully understand the impact that eachparticular strategic choice has on consumers’reactions to their messages.

Limitations and future research

As with any experimental study, there arecaveats and limitations to these findings;although useful insight is offered, we are notsuggesting that the above results are absolutein all product- or consumer-based contexts.The stimulus materials used in this studyrepresented only two product classes andindividuals were exposed to only one productcategory and one advertisement of interest, sothere may be a chance that the results couldhave been influenced by the particular stimuliin addition to the manipulated involvementconditions. Future research could includemultiple advertisements across multiple pro-duct categories to produce more generalizablefindings.

Copyright # 2008 John Wiley & Sons, Ltd. J

In this regard, an interesting query would bedetermining the degree to which the effectscurrently reported would evince for productsthat elicit even higher-involvement purchasingdecisions. Recall that the current manipulationof a mundane product like mouthwash wasoperationalized as the ‘‘higher’’ involvementproduct because it was rated higher on theinvolvement scale than paper shredders,although the mean score for mouthwash wasnot extremely high in absolute terms. Thus,although the current results did show thatevidence type would interact with messageframing within the context of a product thatwas higher in perceived involvement, futureresearch likely would demonstrate even greatereffects in cases in which products or stimulusmanipulations inherently possess stronger‘‘high-involvement’’ attributes in general.Additionally, there are other, multiple

dimensions on which products may vary. Forexample, perhaps the innovativeness of theproduct (such as an Apple iPhone) wouldimpact the results shown currently. The samelikely is true for other factors that were notstudied here (e.g., the hype that might precedea real product’s introduction into the market-place). Although any one experiment usuallycannot incorporate many types of factors,future studies could examine such relation-ships to determine instances in which thecurrent results could be either replicated orqualified based on specific context-relatedelements.Also, because the current study introduced

participants to fictitious brand names tocontrol for various extraneous factors, findingscould be advanced by conducting similarresearch that uses existing brands, whichpatently possess many more extant variables.Such an effort could provide deeper insightinto the thought processes that affect brandevaluations among consumers. For example,the iconic brand image of a real, well-established brand likely would affect individ-uals’ reactions to its advertising messages andcould qualify the effects of framing or evidencetype employed therein. Lastly, to remainconsistent with the message-framing research

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 15: The influence of evidence type and product involvement on message-framing effects in advertising

236 Frank E. Dardis and Fuyuan Shen

on which the current study was based, wefocused on analyzing solely the informationcontained inmessages, and did not incorporateother advertising elements – such as differentvisuals – that obviously have shown effects inprior research.In sum, the current study demonstrated that

two important advertising-related factors –evidence type and product involvement – caninfluence the effects of message framingwithinadvertisements for commonplace consumerproducts. This is important because many adsfor everyday consumer goods seem to combinemessage-framing elements and types of evi-dence rather arbitrarily within their appeals,apparently without fully understanding howthe different information strategies can affectconsumers’ attitudes. Future research canbuild on our findings by focusing on instancesin which consumers’ involvement with pro-ducts or purchase decisions is even higher thanthat which was studied currently. Also, futureinquiry can add to knowledge in this area byexamining other product-based and context-based factors that were not incorporated intothe current research design.

Biographical notes

Frank E. Dardis (Ph.D., University of SouthCarolina) is Assistant Professor in the Depart-ment of Advertising & Public Relations atPennsylvania State University. He has pub-lished several journal articles relating tostrategic message effects in both consumer-oriented and sociopolitical contexts. His cur-rent research involves the role of informationprocessing in affecting individuals’ attitudesand beliefs.

Fuyuan Shen (Ph.D., University of NorthCarolina) is Associate Professor in the Depart-ment of Advertising & Public Relations at Penn-sylvania State University. He has publishedseveral journal articles relating to framing andthe psychological reactions to strategicallyframed messages. His current research involvesthe use of strategic message framing in shapingconsumers’ or individuals’ opinions.

Copyright # 2008 John Wiley & Sons, Ltd. J

References

Ahluwalia R. 2002. How prevalent is the negativity

effect in consumer environments? Journal of

Consumer Research 29(2): 270–279.

Ahluwalia R, Burnkrant RE, Unnava RH. 2000.

Consumer response to negative publicity: the

moderating role of commitment. Journal of Mar-

keting Research 37: 203–214.

Baesler EJ, Burgoon JK. 1994. The temporal effects

of story and statistical evidence on belief change.

Communication Research 21(5): 582–602.

Banks SM, Salovey P, Greener S, Rothman AJ, Moyer

A, Beauvais J, Epel E. 1995. The effects of

message framing on mammography utilization.

Health Psychology 14: 178–184.

Block LG, Keller PA. 1995. When to accentuate the

negative: the effects of perceived efficacy and

message framing on intentions to perform a

health-related behavior. Journal of Marketing

Research 32(2): 192–203.

Chaiken S, Maheswaran D. 1994. Heuristic proces-

sing can bias systematic processing: effects of

source credibility, argument ambiguity, and task

importance on attitude judgment. Journal of

Perosonality and Social Psychology 66(3):

460–473.

Cox D, Cox AD. 2001. Communicating the con-

sequences of early detection: the role of evidence

and framing. Journal of Marketing 65: 91–103.

Detweiler JB, Bedell BT, Salovey P, Pronin E. 1999.

Message framing and sunscreen use: gain-framed

messages motivate beach-goers. Health Psychol-

ogy 18(2): 189–196.

Donovan RJ, Jalleh G. 1999. Positively versus nega-

tively framed product attributes: the influence of

involvement. Psychology and Marketing 16(7):

613–630.

Eagly AH, Chaiken S. 1993. The Psychology of

Attitudes. Harcourt Brace: Orlando, FL.

Fiske ST. 1980. Attention and weight in person

perception: the impact of negative and extreme

behavior. Journal of Personality and Social

Psychology 38: 889–906.

Gleicher F, Petty RE. 1992. Expectations of reassur-

ance influence the nature of fear-stimulated atti-

tude change. Journal of Experimental Social

Psychology 28(1): 86–100.

Hafer CL, Reynolds KL, Obertynski MA. 1996.

Message comprehensibility and persuasion:

effects of complex language in counterattitudinal

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 16: The influence of evidence type and product involvement on message-framing effects in advertising

Influencing message-framing effects in advertising 237

appeals to laypeople. Social Cognition 14(4):

317–337.

Holsti O. 1969. Content Analysis for the Social

Sciences and Humanities. Addison-Wesley:

Reading, MA.

Kahneman D, Tversky A. 1979. Prospect theory: an

analysis of decision under risk. Econometrica 47:

263–291.

Laskey HA, Day E, Crask MR. 1989. Typology of

main message strategies for television commer-

cials. Journal of Advertising 18(1): 36–41.

Laskey HA, Seaton B, Nicholls JAF. 1992. Strategy

and structure in bank advertising: an empirical

test. The International Journal of Bank Market-

ing 10(3): 3–9.

Laskey HA, Seaton B, Nicholls JAF. 1994. Effects of

strategy and pictures in travel agency advertising.

Journal of Travel Research 32(4): 13–19.

Levin IP. 1987. Associate effects of information

framing. Bulletin of the Psychonomic Society

25: 85–86.

Levin IP, Gaeth GJ. 1988. How consumers are

affected by the framing of attribute information

before and after consuming the product. Journal

of Consumer Research 15: 374–378.

Levin IP, Schneider SL, Gaeth GJ. 1998. All frames

are not created equal: a typology and critical

analysis of framing effects. Organizational

Behavior and Human Decision Processes

76(2): 149–188.

Lynch JGJ, Ofir C. 1989. Effects of cue consistency

and value on base-rate utilization. Journal of

Personality and Social Psychology 56(2):

170–181.

Maheswaran D, Meyers-Levy J. 1990. The influence

of message framing and issue involvement. Jour-

nal of Marketing Research 27(3): 361–367.

Markoczy L, Goldberg J. 1998. Women and taxis

and dangerous judgments: content sensitive use

of base-rate information. Managerial and

Decision Economics 19(78): 481–493.

Martin B, Marshall R. 1999. The interaction of

message framing and felt involvement in the

context of cell phone commercials. European

Journal of Marketing 33(1/2): 206–218.

Meyerowitz BE, Chaiken S. 1987. The effect of

message framing on breast self-examination,

attitudes, intention, and behavior. Journal of

Personality and Social Psychology 52: 500–

510.

Copyright # 2008 John Wiley & Sons, Ltd. J

Meyers-Levy J, Maheswaran D. 2004. Exploring

message framing outcomes when systematic,

heuristic, or both types of processing occur.

Journal of Consumer Psychology 14(1&2):

159–167.

Pae JH, Samiee S, Tai S. 2002. Global advertising

strategy: the moderating role of brand familiarity

and execution style. International Marketing

Review 19(2): 176–189.

Petty RE, Cacioppo JT. 1983. Central and peripheral

routes to persuasion: application to advertising.

In Advertising and Consumer Psychology,

Percy L, Woodside A (eds). Lexington Books:

Lexington, MA; 3–23.

Petty RE, Cacioppo JT. 1986. The elaboration

likelihood model of persuasion. In Advances

in Experimental Social Psychology, Vol. 19, L

Berkowitz (ed.). Academic Press: New York;

123–205.

Puto CP, Wells WD. 1984. Informational and

transformational advertising: the differential

effects of time. In Advances in Consumer

Research, Vol. 11, Kinnear TC (ed.). Associa-

tion for Consumer Research: Provo, UT; 638–

643.

Rossiter JR, Percy L, Donovan RJ. 1991. A better

advertising planning grid. Journal of Advertising

Research 31(5): 11–21.

Rothman AJ, Martino SC, Bedell BT, Detweiler JB,

Salovey P. 1999. The systematic influence of gain-

and loss-framed messages on interest and use of

different types of health behavior. Personality

and Social Psychology Bulletin 25(11):

1355–1369.

Rothman AJ, Salovey P. 1997. Shaping percep-

tions to motivate healthy behavior: the role of

message framing. Psychological Bulletin 121(1):

3–19.

Rothman AJ, Salovey P, Antone C, KeoughK,Martin

CD. 1993. The influence of message framing on

intentions to perform health behaviors. Journal

of Experimental Social Psychology 29: 408–

433.

Schneider TR, Salovey P, Apanovitch AM, Pizarro J,

McCarthy D, Zullo J, Rothman AJ. 2001a. The

effects of message framing and ethnic targeting

on mammography use among low-income

women. Health Psychology 20(4): 256–266.

Schneider TR, Salovey P, Pallonen U, Mundorf N,

Smith NF, Steward WT. 2001b. Visual and

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb

Page 17: The influence of evidence type and product involvement on message-framing effects in advertising

238 Frank E. Dardis and Fuyuan Shen

auditory message framing effects on tobacco

smoking. Journal of Applied Social Psychology

31(4): 667–682.

Shiv B, Edell Britton JA, Payne JW. 2004. Does

elaboration increase or decrease the effective-

ness of negatively versus positively framed

messages? Journal of Consumer Research 31:

199–208.

Shiv B, Edell JA, Payne JW. 1997. Factors affecting

the impact of negatively and positively framed ad

messages. Journal of Consumer Research 24(3):

285–294.

Tversky A, Kahneman D. 1981. The framing of

decisions and the psychology of choice. Science

211: 453–458.

Wilson DK, Wallaston KA, King JE. 1990. Effects

of contract framing, motivation to quit, and

self-efficacy on smoking reduction. Journal of

Applied Social Psychology 20: 531–547.

Copyright # 2008 John Wiley & Sons, Ltd. J

WoodW, Kallgren CA, Preisler RM. 1985. Access to

attitude-relevant information in memory as a

determinant of persuasion: the role of message

attributes. Journal of Experimental Social Psy-

chology 21(1): 73–85.

Zaichkowsky JL. 1985. Measuring the involvement

Construct. Journal of Consumer Research 12:

341–352.

Zaichkowsky JL. 1994. The personal involvement

inventory: reduction, revision, and application to

advertising. Journal of Advertising 23(4): 59–69.

Zhang Y, Buda R. 1999. Moderating effects of need

for cognition on responses to positively versus

negatively framed advertising messages. Journal

of Advertising 28(2): 1–15.

Zillman D. 2002. Exemplification theory of media

influence. In Media Effects: Advances in Theory

and Research, 2nd edn, Bryant J, Zillman D

(eds). Lawrence Erlbaum: Mahwah, NJ; 19–41.

ournal of Consumer Behaviour, May–June 2008

DOI: 10.1002/cb