15
The Role of Consumer Involvement in Determining Cognitive Response to Broadcast Advertising Laura M. Buchholz Robert E. Smith This paper investigates the role of involvemeni in deurmmitxg consumer response to radio and TV commercials. Afur reviewing reletani Uterature. a summary model thai focuses on the amount and type of cognitive elaboration and subsequent ejects on consumer recognition of the brand and message points is presented. Hypotheses are developed that predict interaction ejfects between the type of brondcasi media and the level of consumer tm-oltement in the commercial. A study is conducted where mode of presentation (radio versus television) and let«l of consumer involvement (low versus high) are experimentally manipulated. Analysis of variance of the data provide general support for the hypotheses. Other results and the impli- cations for advertising research and practice are discussed. Laura Buchholz is a graduate student at Indiana University. Robert E. Smith (Ph.D., University of Wisconsin-Madison) is associate professor of marketing, Indiana University, The avithora chanlc Frank DiSilvestro and Dan McQuiston for their contribution to this study. Advertising research has provided considerable evidence regarding consumer re- sponse to persuasive messages. Numerous models that explain consumers' cog- nitive, affective, and conative reactions have been advanced and tested. As theories become more detailed, greater discrimination is possible in understanding the important dimensions of consumer response. However, further research is needed on many fronts, including investigations of differences that exist in message pro- cessing among alternative advertising media. For example, two types of broadcast media—radio and TV—have obvious differences that could seriously affect the way consumers process persuasive messages. Despite calls for research on these issues (Edell and Keller 1989; Greenwald and Leavitt 1984), few mtxiels distinguish between radio and TV message processing. The importance of these broadcast media to advertisers is evident by the fact that they spent $26 billion on TV and $7-7 billion on radio in 1988 (Marketer's Guide To Media 1989). In addition, these media are often used together in "co- ordinated media campaigns" (Edell and Keller 1989). To most effectively com- municate their message, creative teams and media planners need to understand differences in the way consumers process radio and TV commercials. Knowledge of the particular strengths and weaknesses of different broadcast media would represent a practical advance of marketing communications theory. Accordingly, the purpose of this paper is to review issues relevant to processing differences between broadcast media and to advance a sumtnary model of the important dimensions. In addition, a study is reported that tests the key propositions of the model. Background Literature The most basic difference between radio and TV is the number of sensory modes involved. Radio messages consist of only auditory stimuli and audience processing consists of only listening. TV messages project both auditory and visual stimuli, and audience processing consists of listening and viewing, ©Journal of Advertising Volume 20, Nutnber 1. 1991. Pages 4-17

The Role of Consumer Involvement in Determining … Role of Consumer Involvement in Determining Cognitive Response ... marketing, Indiana University, ... because without some lasting

Embed Size (px)

Citation preview

The Role of Consumer Involvementin Determining Cognitive Responseto Broadcast Advertising

Laura M. BuchholzRobert E. Smith

This paper investigates the role of involvemeni in deurmmitxg consumer response to radio andTV commercials. Afur reviewing reletani Uterature. a summary model thai focuses on theamount and type of cognitive elaboration and subsequent ejects on consumer recognition ofthe brand and message points is presented. Hypotheses are developed that predict interactionejfects between the type of brondcasi media and the level of consumer tm-oltement in thecommercial. A study is conducted where mode of presentation (radio versus television) andlet«l of consumer involvement (low versus high) are experimentally manipulated. Analysis ofvariance of the data provide general support for the hypotheses. Other results and the impli-cations for advertising research and practice are discussed.

Laura Buchholz is a graduate student atIndiana University.

Robert E. Smith (Ph.D., University ofWisconsin-Madison) is associate professor ofmarketing, Indiana University,

The avithora chanlc Frank DiSilvestro and DanMcQuiston for their contribution to this study.

Advertising research has provided considerable evidence regarding consumer re-sponse to persuasive messages. Numerous models that explain consumers' cog-nitive, affective, and conative reactions have been advanced and tested. As theoriesbecome more detailed, greater discrimination is possible in understanding theimportant dimensions of consumer response. However, further research is neededon many fronts, including investigations of differences that exist in message pro-cessing among alternative advertising media. For example, two types of broadcastmedia—radio and TV—have obvious differences that could seriously affect theway consumers process persuasive messages. Despite calls for research on theseissues (Edell and Keller 1989; Greenwald and Leavitt 1984), few mtxiels distinguishbetween radio and TV message processing.

The importance of these broadcast media to advertisers is evident by the factthat they spent $26 billion on TV and $7-7 billion on radio in 1988 (Marketer'sGuide To Media 1989). In addition, these media are often used together in "co-ordinated media campaigns" (Edell and Keller 1989). To most effectively com-municate their message, creative teams and media planners need to understanddifferences in the way consumers process radio and TV commercials. Knowledgeof the particular strengths and weaknesses of different broadcast media wouldrepresent a practical advance of marketing communications theory. Accordingly,the purpose of this paper is to review issues relevant to processing differencesbetween broadcast media and to advance a sumtnary model of the importantdimensions. In addition, a study is reported that tests the key propositions of themodel.

Background Literature

The most basic difference between radio and TV is the number of sensory modesinvolved. Radio messages consist of only auditory stimuli and audience processingconsists of only listening. TV messages project both auditory and visual stimuli,and audience processing consists of listening and viewing,

©Journal of AdvertisingVolume 20, Nutnber 1. 1991. Pages 4-17

Processing Auditory Information—Consumer listening. Listening was de-fined by Barker (1971) as: "the selectiveprocess of attending to, bearing, under-standing, and remembering aural sym-bols." This definition focuses uponcognitive processes that register, com-prehend, and retain auditory informa-tion. The memory component wasconsidered necessary by Barker (1971)because without some lasting vestige ofthe input stimuli, no real evidence of theentire listening process exists. It is alsoimportant to note that people vary intheir motivation and ability to performthe necessary steps in listening, makingthis a complex process not just a simpleskill (Barker 1971). In addition, Spear-ritt (1962) notes that it is important todistinguish between hearing, which isthe "mere physiological reception ofsound," and listening, which involvesboth attending to the sounds and at-taching meanings to them.

Research on listening frequently ex-amines comprehension and/or recall ofauditory stimuli, and results indicatethat listening can be a difficult and de-manding task. For example, Nichols andStevens (1957) tested the ability of peo-ple to understand and remember whatthey hear. Results showed that imme-diately after listening to someone, theaverage person forgets about one halfof what was just said. After two monthselapsed time, the average individualcould recall about 25 percent of whatwas said. Another study by Nichols(1957) showed that while 40 percent ofa white collar worker's day was devotedto listening behaviors, this group lis-tened at only a 25 percent rate of ef-ficiency.

Bostrom and Waldhart (1980) usedthree different measures to test listen-ing abilities: (1) short-term listening, (2)short-term listening with rehearsal, and(3) lecture comprehension. Thesemeasures were then compared withscores measuring general mental abil-ity. Results showed that short-term lis-tening was a good predictor of oralperformance; short-term listening withrehearsal had no apparent relationshipto any other measures; and lecture-

comprehension listening proved to bea good predictor of written test scores.Based on these results, Bostrom andWaldhart (1980) believe that currentconceptualizations of listening behav-ior need to be revised, and better meas-urements of listening abilities need tobe developed, In addition, closer atten-tion needs to be paid to the listener'sshort-term-memory processing in or-der to resolve the many conflictingclaims about listening behavior.

, • • two types ofbroadcast media—radio and TV—haveobvious differencesthat could seriouslyaffect the wayconsumers processpersuasive messages.

In a similar study, Bostrom andBryant (1980) tested a hypothesis com-paring short-term listening and short-term-memory-processing structures.The results showed that short-term lis-tening is distinguishable from the usualdefinitions of listening (an "ability" tohear; to attend to) and is different fromthe kinds of storage processes usuallydescribed as short-term memory Ex-actly how the process works in rela-tionship with other memory functionswas not discussed and awaits furtherinvestigation.

Other studies have compared the ef-fectiveness of listening to that of vis-ually presented information. Forexample, Pauk (1984) found that "re-membering what you have heard isusually more difficult than remember-ing what you have read." This may bedue to the fact that one can slow downthe rate of reading if necessary, but cannot slow down the rate of listening.Another explanation is that seeing thevisual image of printed words processesinformation at a deeper level than

merely hearing words. In the latter case,words often are not given enough timeto allow for the construction of a men-tal image.

Similarly, Siegel and Allik (1973)compared visual and auditory process-ing effects on children's and coUege stu-dents' short-term memory Modality ofstimuli (visual or auditory) and modalityof recall cue (visual or auditory) wereboth manipulated. Results showed thatthe recall of visually presented stimuliwas superior to the recall of auditorystimuli, while the effect of the modalityof the recall cue was negligible. Siegeland Allik (1973) concluded that visualstimuli fecilitate performance because apicture can be stored simultaneously inboth a visual and an auditory-verbalprocessing system.

As the review indicates, several as-pects of listening research are disturb-ing from the advertiser's point of view.First, this topic has received limited at-tention from researchers. Existingmodels are still in the developmentstage and often do not consider theimpact of important mediating factorssuch as the individual's motivation andopportunity to process auditory stim-uli. Moreover, very little research onconsumer listening has appeared inmajor marketing and advertising jour-nals. This is an important gap, sincelistening to a person may involve dif-ferent processing mechanisms than lis-tening to a radio (since visual contactwith the source is not an issue in thelatter case). Another problem is thatmuch of the research designed to meas-ure a person's listening ability seems tobe a better indicator of their generalmental ability or skills; valid and reli-able indicators of listening have yet tobe developed. Certainly, more researchneeds to be conducted so that adver-tisers can better understand and man-age the auditory component of broad-cast messages.

Processing Visual Stimuli—Con'sumer Viewing. With T y advertisershave the opportunity to combine vis-ual and verbal stimuli when construct-ing the persuasive message. Although

significant research has been conduct-ed on consumer processing of visualand verbal material, most of that re-search has used print rather thanbroadcast stimuli. For example, Child-ers and Houston (1984) studied the ef-fects of pictures-only versus picturesand words in print advertisements.They found that pictorial material con-veying brand/product class-associa-tions is recalled better than corre-sponding verbal-only material wheneach is processed at a sensory level (ashallow, non-stringent form of messageprocessing). They also found thatgreater sensory discrimination of pic-tures improves recall over correspond-ing verbal-only material when each isprocessed at a sensory level.

However, the picture-sup erioriry ef-fects were eliminated when material isprocessed at a semantic level (an in-tense form of elaboration pertaining tothe meaning of words or pictures). In-deed, verbal and verbal-plus-visualstimuli were equivalent when pro-cessed at the semantic level.

In a continuation of this research,Houston, Childers, and Heckler (1987)investigated consumers' memory formessages in which the semantic con-tent of the pictorial material is (1) con-sistent with the verbal copy (verbal copyand picture present the same productattributes) or (2) discrepant from theverbal copy (verbal copy states one at-tribute while picture shows a differentattribute). In three separate experi-ments they also manipulated anotherindependent variable: interactive pic-tures (pictures that pictorially representthe brand name and product class) ver-sus non-interactive pictures (pictures thatbear no relationship to the brand nameand may or may not represent theproduct class). Results indicate that in-teractive ads were superior to non-in-teractive ads. Also, using an interactivepicture to convey a different productattribute than that presented in theverbal copy used in the ad (discrepantinformation) increases the audience'srecall of the material.

Maclnnis and Price (1987) reviewedresearch on imagery processing (a pro-

cessing mode in which the informationreceived is represented in a gestalt or to-tal-picture form in the active memory),and contrasted it with discursive pro-cessing (a more abstract, symbolic, lan-guage-like form of processing). They alsosuggested some important ways in whichimagery impacts consumers' learning,choices, and satisfection. The authorsfound that imagery is a process, not astructure, and that imagery affects manycognitive, physiological, and behavioralphenomena. In addition, they found thatimagery enhances incidental learningand helps individuals to anticipate thefuture. Maclnnis and Price (1987) con-clude that imagery can help consumerswith product evaluations, purchase in-tentions, remembering consumption,and intentions to repurchase.

Although significantresearch has beenconducted onconsumer processingof visual and verbalmaterial, most of thatresearch has usedprint rather thanbroadcast stimuli.

Thus, research in print advertisingindicates that visual stimuli (and/orcreating a visual image from stimuli) en-hance message processing and recallunder some conditions. While thisstream of research is more advancedthan the listening research, it also hassome limitations. Since most of this re-search has been conducted with printstimuli, it is reasonable to questionwhether the same effects will occurwith broadcast ads where the consum-er has little control over the transmis-sion rate. Finally, it is time to begindeveloping advertising-response theo-ries that specifically account for pro-cessing differences between radio andTV ads.

Consumer Response toBroadcast Advertising

Background. According to the elab-oration likelihood model (Petty andCacioppo 1983), consumers can followtwo routes to persuasion. "Centralroute" processing occurs when in-volved consumers seek product-relatedinformation to diligently consider.Here, cognitive response to the adver-tising message has been shown to me-diate subsequent brand attitudes (Pettyand Cacioppo 1983; Petty, Cacioppo,and Schumann 1983). "Peripheralroute" processing occurs when unin-volved consumers lack sufficient mo-tivation to pay close attention tomessage points. Here, brand and ad at-titudes have been shown to be me-diated by incidental factors (e.g.,attractiveness of source) or by affectivereactions (e.g., mood states). Othermodels of advertising response (Green-wald and Leavitt 1984; Maclnnis andJaworski 1989) suggest that superficialcognitive effects can be expected evenat lower levels of involvement. Giventhe importance of cognitive effects inunderstanding consumer processing,the model developed below focuses onthe cognitive dimensions of consumerresponse to broadcast advertising.

Cognitive Response to BroadcastAdvertising. Only one previous studycould be found that compared cogni-tive response to radio and TV ads. Edelland Keller (1989) present a concep-tualization and empirical test of howconsumers process ads in these twomedia. In their model. "TV and radioadvertising differ in the content, num-ber, and relationship of the sensorymodes the viewer encounters" (Edelland Keller 1989, p. 150).

At first glance, it seems that TV isdestined to be a more effective advertis-ing medium than radio. This conclusionis based on several facts. First, TV hastwo sensory-input modes, compared toonly one for radio, allowing TV to dis-seminate more information in the sameamount of time. Second, TV makes useof the powerful visual mode and, as

noted above, researchers often find apicture-superiority effect over verbal-only stimuli. Third, TV can create ver-bal/visual synergy by coordinating bothstimuli (i.e., "cotisistent" ads). Finally, TVviewers seem less likely to be engaged ina distracting secondary task like driving,talking to a friend, studying, or readingthe paper

However, Edell and Keller (1989)question whether TV will always be su-perior to radio as a persuasive medium.They note that while TV does providemore information to the viewer, this in-formation may require more effort toprocess. As a result, "it is unclearwhether consumers will extract more orless information and meaning from TVads than from their radio counterparts"(Edell and Keller 1989, p. 150). Second,Edell and Keller note that while somestudies show that multiple sensorymodes facilitate learning others havefound an interference effect. Third, re-search has shown that it takes time (upto 1.5 seconds) to switch from process-ing visual to verbal inputs, which maylessen the impact of TV vis-a-vis radio(Edell and Keller 1989). Finally, Edell andKeller note that concurrent distractors(e.g., discrepant verbal and visual com-ponents) can inhibit cognitive elabora-tion and critical thinking. These con-ceptual considerations are bolstered byfindings that auditory presentation ofverbal information can indeed result insuperior retention (Murdock 1967,1968,1969).

What factors might mediate the im-pact of radio commercials versus TVcommercials and allow advertisers tosort out the conflicting theories andresults noted above? Several factorsseem to play a role including: the typeof ad (consistent or discrepant; Hous-ton et al. 1987), whether the consumeris processing the commercial for thefirst time or has seen it before (Edelland Keller 1989), the existence of a sec-ondary (distracting) task (Maclnnis andJaworski 1989), stimulus complexity(Leigh and Menon 1987), and the levelof consumer involvement (Greenwaldand Leavitt 1984; Krugman 1965). Thegoal of this paper is to examine the role

of consumer involvement in the pro-cessing of broadcast commercials.

Consumer Involvement

Media Differences in Involvenient.Krugman (1965) was one of the first tonote the important differences that ex-ist in consumer processing of print ver-sus broadcast advertising. In developinghis "low-involvement-learning hypoth-esis," Krugman (1965) observed thatprint media require the active partici-pation of the audience since readingprinted words is a relatively demandingcognitive task. Indeed, the "informa-tion-processing-parsimony hypothesis"(Holbrook 1978) suggests that con-sumers attempt to minimize demand-ing cognitive endeavors and would beunUkely to read information of littleinterest to them. Greenwald and Leav-itt (1984) also note the limited abilityof print media to get a meaningful re-sponse from uninvolved consumers:"with rapid page turning and only par-tial scanning of page contents . . . , crit-ical cues that could attract higherinvolvement may simply be missed"(Greenwald and Leavitt 1984, p. 590).Thus, print media have limited oppor-tunity to influence uninvolved or pas-sive audience members who aredisinclined to read the message.

Conversely, the nature of broadcastmedia makes them much better suitedfor influencing passive consumers. Theverbal information is spoken and can,therefore, impact consumers who arenot actively seeking exposure to the admessage. Similarly, TV graphics canconvey information in an interestingand effective manner that requires littlecognitive effort from the viewer. Thus,while consumers rarely read print adsthat do not interest them, they are oftenexposed to (i.e., see and hear) broadcastcommercials for products with whichthey are not involved. Indeed, this sit-uation was recognized as the basis forKrugman's (1965) low-involvement-learning model which he specificallylimited to broadcast media.

As a result, the consumer's level ofinvolvement with the ad, brand, or

product category can be expected toplay a major role in determining theamount and type of cognitive response.Of significant importance is the fact thatboth low- and high-involvement con-sumers can be impacted by broadcastcommercials, though perhaps, in dif-ferent ways. This proposition is con-sistent with Wright's (1973) model thatthe consumer's motivation to cogni-tively process message points mediatessubsequent cognitive and affective re-actions.

Defining Consumer Involvement. In-volvement is a construct that has beendefined and operationalired in many dif-ferent ways. While no consensus ex-ists, many researchers define involve-ment as the extent to which a stimulusor task is relevant to the consumer'sexisting needs and values (Maclnnis andJaworski 1989; Petty and Cacioppo1983; Petty, Cacioppo, and Schumann1983; Wright 1973). Laczniak, Muehl-ing, and Grossbart (1989) reviewed pastconceptualizations in advertising stud-ies and established two basic compo-nents of involvement in advertising. Asconsumers become more involved theyshould (1) pay more attention to the admessage, and (2) focus more on brandprocessing as opposed to nonbrandprocessing. Since this conceptualiza-tion was specifically developed for ad-vertising research, and captures theessential components of involvement,it will be used as the conceptual basisfor operationalizing involvement in thisstudy.

Effects of Consumer Involvement. Inmany advertising response models, thedegree of consumer involvement is ex-pected to influence both the amountand the quality of the consumer's cog-nitive response (Wright 1973, 1974,1975; Greenwald and Leavitt 1984;Maclnnis and jaworski 1989; Edell andKeller 1989). In terms of the amountof cognitive response, it is clear thatinvolved consumers engage in more ef-fortful information-search and acqui-sition strategies. When activelysearching for product-related infor-mation, interest in brand advertising is

at a maximum, as is the desire to thinkabout the ad, its claims, and the brand.

In addition to increased cognitive ef-fort, most advertising models predict aqualitative difference in message pro-cessing as involvement increases. Spe-cifically, involved consumers are morelikely to process brand information ata "deep level." As consumers begin toengage in deeper processing, they startto actively evaluate the message. Thesecognitive evaluations can be directedtoward the importance, persuasive-ness, or relevance of the ad's content,and are often measured by countingthe number of support arguments orcounterarguments (Maclnnis and Ja-worski 1989). For example, in a natur-alistic viewing environment, just gettingthe consumer to evaluate the contentsof the ad is a major accomplishmentthat goes beyond superficial attentionor association.

Beyond evaluative responses to adstimuli are even deeper and more pow-erful elaborative processes. In theMaclnnis and Jaworski (1989) model thedeepest form of consumer processingis represented by "constructive pro-cessing." Here, the consumer goes be-yond the ad's content and connects itin some meaningful way to his or herown life. Examples of these construc-tive elaborations include thinking upnovel uses for the product and/orimagining the product in use.

Past models of involvement have alsogiven "deeper" processing status to"connective elaborations" (as opposedto evaluative processes). Krugman(1965) suggested that at the highest lev-el of involvement consumers produce"personal connections," or "bridgingexperiences," whereby they relate thead content to meaningful aspects oftheir own life. A well-known model ofconsumer involvement by Greenwaldand Leavitt (1984) also makes this dis-tinction: "In Krugman's conception ofhigh involvement, the audience expe-riences 'personal references' or 'con-nections' to the advertising message.This corresponds well to our highestlevel of involvement, elaboration"(Greenwald and Leavitt 1984, p. 590).

Accordingly, the term elaboration asused in this paper will refer to the spe-cific constructive process of producingpersonal connections and not to eval-uative processes like support arguingand counterarguing.

Summary Model ofCognitive Responseto Broadcast Advertising

Based on the review above, a summarymodel of consumer response to broad-cast advertising can be proposed. Es-sentially, when consumers are exposedto a TV or radio commercial, process-ing begins at relatively superficial levelslike orienting reactions, selective atten-tion, feature discrimination, or cate-gorization (Greenwald and Leavitt 1984;Maclnnis and Jaworski 1989). Whileprint media may not get far (or any-where) in terms of processing depth forlow-involvement consumers (unwillingto read), broadcast media can be missed"only by walking out of the room orby turning down the sound" (Green-wald and Leavitt 1984, p. 590). Thisconclusion may be somewhat overstat-ed, since distracting secondary tasks ornonattention to the stimulus (e.g., clut-ter, day dreaming) could create condi-tions similar to print (i.e., non-registration of the information) forsome viewers.

Notwithstanding this possibility,broadcast media are expected to havesome meaningful cognitive impact ona greater number of non-interestedconsumers. In this respect, low-in-volvement consumers could engage insuperficial processing consistent withPetty and Cacioppo's (1983) "periph-eral route" to persuasion. Here, low-involvement audience members focuson executional aspects of the ad likecelebrity endorsers and source attrac-tiveness rather than processing brand-related data. More recent models(Maclnnis and Jaworski 1989) describesix levels of processing, and suggest thatsome superficial brand processing canoccur even at low levels of involve-ment. If consumers are moderately in-volved with the advertising for a brand

they will move beyond superficial anal-ysis and perform more sophisticatedprocessing including comprehension ofad points and evaluative/comparativeprocessing (Maclnnis and Jaworski1989). At high levels of involvement,consumers will elaborate message claimsby forming personal connections,bridging experiences, or imagining theproduct in use (Greenwald and Leavitt1984; Krugman 1965; Maclnnis and Ja-worski 1989).

As consumers move to deeper levelsof processing, recognition, under-standing, and memory of the inputstimuli are enhanced (Greenwald andLeavitt 1984; Krugman 1965). Thismeans that as consumers become moreinvolved they can better understandand remember brand-related informa-tion. In addition, Maclnnis and Jawor-ski (1989) state that connectiveelaborations produce "self-generatedpersuasion" resulting in strong beliefsabout the brand, These conclusions in-tegrate well with Craik and Lockart's(1972) original exposition of the levels-of-processing framework where se-mantic processing is expected to im-prove the memory of input stimuli. Thisis an important point because adver-tisers are very concerned with con-sumer's recall and recognition of adsand frequently measure them with day-after recall and recognition measures.Thus, understanding how involvementand broadcast media interact to effectconsumer's recognition of ad pointswould be a practical addition to adver-tising response models.

Hypotheses

Cognitive Elaboration of Advertis-ing. When consumer involvement islow, there is little motivation to cog-nitively elaborate message claims. Ac-cordingly, low-involvement consumerscould be expected to rely on more su-perficial processing techniques (Green-wald and Leavitt 1984) and/or to beinfluenced by peripheral ad factors(Petty and Cacioppo 1983). In thesecircumstances, little evidence of elab-orative processing is expected in

consumers' cognitive responses. Spe-cifically, when only rudimentary pro-cessing occurs, there should be fewpersonal connections in response toeither radio or TV commercials.

However, at high levels of consumerinvolvement, greater motivation existsto cognitively elaborate message claims.In this case, radio seems to offer moreopportunity to generate meaningfulelaborations than TV Without strongvisual cues to guide them, involvedconsumers are likely to generate theirown visualizations/connections/elabo-rations in response to radio ads. Thesehighly personal, self-based elaborationsare generally expected to provide a verydeep level of processing and to producea powerful cognitive impact. For ex-ample, Edell and Keller (1989, p. 151)note that " . . . self-generated visualiz-ations from the radio ad may be morepersonally relevant than the TV ad'svideo." Accordingly, when highly in-volved consumers are exposed to radiocommercials, they should freely gen-erate powerful self-based cognitiveelaborations of message points.

T y however, seems to place two lim-itations on the amount and type of cog-nitive elaboration performed. First, TVmay offer less opportunity for cognitiveelaboration because by presenting moreinformation (two sensory input modes),it places greater demands on the lim-ited-capacity short-term memory Inthis case, processing TV commercialsrequires more of the short-term mem-ory's capacity, thereby reducing pro-cessing resources available forelaborations. Second, because of thevisual cues provided, TV seems to en-courage ad-based elaborations as op-posed to the more potent self-basedelaborations.

When considered together, the re-lationships reviewed above suggest aninteraction effect between the type ofbroadcast media and the level of theconsumer's involvement in the ad.

H,; Under conditions of low in-volvement there should be nodifference in the number ofpersotial connections betweenradio and TV Under condi-

tions of high involvement, con-sumers exposed to radiocommercials should produce asignificantly greater number ofpersonal connections thanthose exposed to TV commer-cials.

If these effects occur, important im-pHcations exist for advertising effec-tiveness. As noted above, visual stimulienhance the consumer's ability to pro-cess superficial aspects of ad informa-tion. It seems reasonable to extend thisproposition to broadcast commercialsas well. Thus, when consumers lacksufficient motivation to elaborate mes-sage points, the cognitive impact ofcommercials will be determined by rel-atively superficial processing, since fewelaborations occur (H,). In these cases,picture-superiority effects are predict-ed, since feature discrimination of mes-sage points should be aided by visualstimuli. Radio, with only one sensorymode, will have a more difficult timeattracting attention and creating mean-ingful feature recognition among low-involvement consumers. This reason-ing is consistent with previous findingsof picture-superiority effects at super-ficial processing levels in print media(Houston, Childers, and Heckler 1987).

However, when consumers are high-ly involved in the ad message the su-perficial processing advantage of TVshould be counterbalanced by the elab-orative advantages of radio (H,). In thesequence of advertising response, TVgains its advantage in the early stagesof processing, while radio gains its ad-vantage at the deeper levels. Whenconsumers go all the way to elaborativeprocessing, TV is still expected to re-alize advantages in the early stages, but,these gains will be offset by radio's ad-vantages at the deeper levels. Thus,measures of ad impact, like recognitionof the brand or ad points, should showno difference in the effectiveness be-tween radio and TV commercials, forhighly involved consumers. Again, in-teraction effects are predicted betweenlevel of involvement and type of mediawith respect to ad and brand recog-nition.

Hj: In conditions of low consumerinvolvement, TV commercialswill have significantly greatercognitive impact (in terms cfrecognition of brand and adpoints) than will radio com-mercials. However, in condi-tions of high consumer in-volvement, there will be no sig-nificant difference in the cog-nitive impact (recognition ofbrand and ad points) of TV ver-sus radio commercials.

Methodology

To test the hypotheses, a 2 X 2 factorialdesign was used. The main independ-ent variables were type of broadcastmedia (radio or television) and level ofconsumer involvement (low or high).

Target Commercials. To test the hy-potheses, a commercial was needed thathad several important characteristics.First, the commercial should containverbal material that was relevant foreither broadcast mode. Second, thecommercial should be new to the sub-jects to avoid previous exposure effects.Third, the commercial should be re-alistic and considered reasonably effec-tive to make the test fair. Fourth, thecommercial should be for a product thatthe subject poo! (college students) usesfrequently and considers buying tomake the experiment realistic, A targetad that met these criteria was selectedfrom 45 award-winning regional com-mercials as reviewed by Advertising Age(1986).

The test commercial selected was a30-second informational computercommercial for a major national brand.As summarized in the Exhibit, thecommercial had both strong verbal copy(which mentioned several importantproduct attributes), and strong visualelements (that were consistent with theverbal copy), making it potentially ef-fective as either a radio or TV spot.The commercial was three years oldand had been played only in west coastmarkets, None of the subjects reportedseeing the commercial previous to theexperiment.

EXHIBITSummary of Experimental Advertisement

of Visual Stimuli: Verbal Stimuii:

L Man standing facing chalkboard. His lefthand is drawing a profile of a man, hisright hand is writing an equation.

2. Zoom to right hand calculating amathematical equation.

3. Zoom to left hand drawing the eye areaof the portrait.

4. Scan chalkboard from left to rightshowing man drawing and solving amathematical problem at the same time.

5- Picture of computer monitor withmulticolored geometric ball bouncingaround the screen.

6. Butterfly flapping its wings now oncomputer screen.

7. Man playing saxophone with horizontallines and dots flashing in the background.

8. Return to chalkboard and man. Pan outto show complete drawing and equation.

"Imagine a brain whose left side is asbrilliant as its right. A brain as artisticas it is logical.That can calculate . . .

and create.

Such a brain exists in the remarkablenew . . .

Apple n GS. Brilliant graphics . . .

brilliant color . . .

brilliant sound.

To help you use both sides of themost personal computer of all—yourmind."

To enhance external validity, the tar-get commercial was preceded by a 60-second beer commercial and followedby a 30-second wine commercial. Thisthree-spot commercial pod was theninserted into a four-minute section ofvideo material taped from Entertain-ment Tonight. The sequence of expo-sure was: three minutes of programcontent (discussing Carol King's mu-sical career), two-minute commercialpod (beer commercial, target commer-cial, wine commercial), one minute ofconcluding comments from the pro-gram (summary and sign off).

Type of Broadcast Media. Manipu-lation of the media variable was verydirect. In the TV treatment, the vi-deotape was played on a large TV mon-itor so subjects could see and hear thestimuli. For the radio treatment, theaudio portion of the tape was trans-ferred directly to an audio cassette toproduce a high quality message. Thiscassette was played on a large portableradio similar to the type used by manystudents.

Level of Involvement. Manipulationof the involvement variable was accom-

plished by giving different experimen-tal instructions to the groups. As noted,the conceptualization of Laczniak,Muehling, and Grossbart (1989) sug-gests that two factors must be manip-ulated to create differences ininvolvement (the amount of attentionpaid to the commercial, and the amountof brand processing performed). Ac-cordingly, high-involvement subjectswere given instructions to maximize at-tention to the ad and the amount ofbrand processing:

Thank you for your time. This is astudy about how consumers use ad-vertising information to form opin-ions about products. You will beshown (hear) a segment from theprogram Entertainment Tonight thatcontains a block of three advertise-ments. The second commercial is fora computer. We would like you tocarefully evaluate the brand of com-puter advertised, so pay close atten-tion to the second commercial. Tomake your evaluation as realistic aspossible we would like you to imag-ine that you are about to purchasea computer and are considering var-ious brands. Therefore, after you

view (listen to) the ad, think abouthow the advertised brand would suityour own computing needs. In oth-er words, evaluate the product ad-vertised as you actually would beforebuying one.

To create a low-involvement condi-tion, subjects were given instructionsto minimize attention to the targetcommercial. Specifically, they were di-rected to evaluate the beer and winecommercials that surrounded the com-puter spot:

Thank you for your time. This is astudy about the advertising of al-coholic beverages on shows watched(listened to) by teenagers. We aretrying to determine if you believe itis appropriate for beer and wines tobe advertised on programs like En-tertainment Tonight. To accomplishthis, you will be shown (hear) a seg-ment that we taped from Entertain-ment Tonight that includes a blockof three advertisements. The first adis for a brand of beer and the thirdad is for a brand of wine. Your taskis to view (listen to) this programsegment and determine if these al-cohol ads are appropriate for teensto watch (hear). Therefore, afterviewing (listening to) the alcoholicbeverage ads, think about whetherthey are suitable for programs withteenage viewers in the audience.

To verify that these instructions pro-duced the intended results, manipu-lation check measures were included inthe study Laczniak, Muehling, andGrossbart (1989) recommend testinginvolvement manipulations by using a5-item scale to measure the amount ofattention allocated to the target ad, anda 4-item scale to measure the degree ofbrand processing. These measures havebeen examined for validity and relia-bility (Laczniak, Muehling, and Gross-bart 1989), and represent a method fordirectly checking involvement manip-ulations. These scales required onlyslight modification for use with broad-cast media and were placed at the endof the questionnaire to minimize sen-sitizing subjects.

10

Dependent Variables. Cognition lut-ing. Subjects were asked to list allthoughts, reactions and/or ideas thatwent through their minds while watch-ing (or listening) to the computer ad.This technique is a common methodfor extracting consumer cognitive re-sponse to advertising, and has beenused in many studies (see Wright 1980for a review). As usual, three minuteswere allowed for subjects to completethis task.

Each of the resulting "primarythoughts" was coded on three specificdimensions. First, was the type of re-sponse: product-related, message-relat-ed, source-re Iated, or unrelated. Secondwas the intent of the thought: positive,negative, neutral, curiosity These twocoding schemes were selected becausethey have provided useful categories inpast cognitive-response studies (Smithand Swinyard 1983, 1988, MacKenzieand Lutz 1989). Third, each thoughtwas examined for the presence of per-sonal connections or elaborations. Ad-vertising elaborations or "bridgingexperiences" occur when consumersrelate the product or ad to meaningfulaspects of their own lives.

Definitions of these variables (shownin the Appendix) were given to twograduate students who served as judges.Each learned the necessary definitionsand then practiced the coding schemeon example statements. Finally, eachjudge was given a photocopy of thesubject's cognitive responses (to avoidrevealing the treatment condition) forfinal coding. Interjudge agreement was86.3 percent for the type of thought,88.0 percent for the intent of thethought, and 92.4 percent for the pres-ence of personial connections. Disa-greements were resolved by a thirdjudge.

Recognition of brand name. After thesubjects completed the cognition-list-ing task, they were given a recognitionmeasure for the brand name of the ad-vertised computer. This measure was amultiple-choice question followed byfour alternatives. Credit was given forcorrect answers while no credit wasgiven for incorrect or missing answers.

Recognition of commercial points.Subjects were then tested on their abil-ity to recognize the product-related in-formation provided in the computercommercial. This recognition test con-sisted of five multiple-choice questionsregarding the content of the commer-cial. Each question was followed by fouralternatives, one of which was correct.Again, credit was given to each correctresponse while no credit was given forincorrect responses or unansweredquestions.

Attitude Toward the Advertisement.This variable was included in the studybecause it has been shown to mediatecommercial effectiveness in some sit-uations (MacKenzie and Lutz 1989).Subjects were asked to rate their atti-tude toward the commercial on threesemantic-differential scales (extremelygood-extremely bad, extremely pleas-ing-extremely irritating, extremely in-teresting-extremely uninteresting).These scales were selected because theyhave been used effectively in past A^studies (MacKenzie and Lutz 1989).

Procedure . Eighty undergraduateseniors at a large midwestern univer-sity were given course credit for com-pleting the study. Subjects wererandomly assigned to one of the fourtreatment conditions. Results of ques-tions designed to verify that studentswere appropriate subjects showed that82.5 percent currently used personalcomputers, and 97.5 percent plannedto use a personal computer in the nextyear. In addition, 10 percent currentlyowned their own computer and 17 per-cent planned to buy one within thenext year. This suggests that collegeseniors are a prime target market forpersonal computers, and thus, were ac-ceptable subjects for the study

Upon arrival at the experimentroom, subjects were read a statementdesigned to reduce demand character-istics. Then, subjects both read and lis-tened to the experimental instructionsnoted above. At this point, the subjectsviewed or listened to the program andcommercials. Viewing/listening tookplace in groups of ten subjects spaced

far apart. After the media exposure,subjects completed the cognition-list-ing task and filled out the remainingquestionnaire items. The entire pro-cedure took approximately 20 minutesper group and the same two graduateassistants administered each of the eightgroups. Total elapsed time for the ex-periment was about three hours. Aftercompletion of the testing process, stu-dents were excused and given no fur-ther information regarding the purposeof the experiment until all subjects hadfinished.

Validity. Three precautionary stepswere taken to reduce demand effects.First, the introduction contained a dis-claimer stating that the researchers werein no way connected with any broad-cast programs and that the tone (po-sitive or negative) of the subjects'responses was not important; onlyhonest opinions mattered. Second, theorder of scale responses (positive tonegative) was alternated in order toavoid "yea saying" response patterns.Third, subjects were not aware that dif-ferent media and involvement treat-ments were applied to other groups.Finally, students were asked if theyguessed the purpose of the study. Nostudent was able to guess the true pur-pose of the study

Reliability. Coefficient alpha was com-puted for the multiple-item scales. Re-sults show that for the 3-item A j scale,a = .92; for the 4-item brand-process-ing scale, a = .90; and for the 5-itemattention-to-the-ad scale, a = .95 .These values indicate the scales hadhigh internal consistency.

Analysis and Results

Manipulation Check. The 5-itemconsumer-attention scale and the 4-itembrand-processing scale were analyzedusing ANO\A. Results, shown in Ta-ble 1, indicate that the main effect forconsumers' attention to the ad was sig-nificant (p < .001) between the low- andhigh-involvement groups. Brand-pro-cessing responses were also significant-ly different between involvement

11

groups (p<.001). In addition, groupmeans showed that low-involvementsubjects were below the midpoint ofthe scales for both attention (3.39) andbrand processing (3.01). Conversely,high-involvement subjects were abovethe midpoint of the scales for attention(5.62) and brand processing (5.02).

In addition to the manipulation-check items, examination of the cog-nitive responses also supports the va-lidity of the involvement treatment. Asshown in Table 1, low-involvementconsumers generated significantly moreunrelated thoughts, while high-in-volvement subjects generated signifi-cantly more total thoughts, productthoughts, and message thoughts. To-gether these findings provide strongsupport for the success of the involve-ment manipulation.

Hypothesis Testing. The number ofelaborations and the adA>i'and-recog-nition measures were subjected to anal-ysis of variance. The results, shown inTable 2. indicate that the predicted in-teraction effect is marginally significant(p < , 10) for the number of elabora-tions, and significant (p < .05) for thead- and brand-recognition measures.These findings are taken as moderatesupport for H, and support for H,.

However, because of the directivenature of the hypotheses, an appro-priate and more sensitive analysis tech-nique is a priori contrasts. Here, thekey comparisons stated in the hypothes-es are tested using the t statistic. Hy-pothesis 1 predicted that there would beno difference in the number of elabora-tions between TV and radio for low-involvement consumers. As shown inTable 3, the cell means are similar andthe a priori contrast is not significant aspredicted in H,. Hypothesis 1 also pre-dicted that radio subjects would pro-duce significantly more elaborations thanTV subjects under conditions of highinvolvement. As shown in Table 3, thecell means are in the expected directionand the a priori contrast is significant att hep< .01 level.

Hypothesis 2 predicted that in theconditions of low involvement, TVcommercials would generate more

TABLE 1ANOVA Summary for Manipulation Checks

Measure

5-itemAttention-to-AdScale

4-itemBrand-ProcessingScale

Total Thoughts

Product Thoughts

Message Thoughts

Unrelated Thoughts

Mean:

Low HighInvoltement Involvement

3.39

3.01

4.75.65

2.50

1.50

5.62

5.02

6.10

2.68

3.30

0.00

Sum ofSquares

99.90

80.50

36.45

82.01

12.80

49.49

DF

1

1

1

1

1

1

F

110.47

65.02

15.68

28.55

4.05

49.49

Sig.

.000

,000

.000

.000

.048

.000

TABLE 2Analysis of Variance Results

Source of variation

For Number of elaborations:

Main Effects;a. Mediab. Involvement

2-'Way Interactions (a X b):

ExplainedResidual

Total

For Brand Recognition:Main Effects:

a. Mediab. Involvement

2-Way Interactions (a X b);

ExplainedResidual

Total

For Ad Recognition:

Main Effects:a. Mediab. Involvement

2-Way Interactions (a X b)i

ExplainedResidual

Total

Sum ofSquares

5.1254.513

.613

2.813

7.93875.45083.388

1.250.450.800

.450

1.7008.500

10.200

46.9259.113

37,813

3.613

50.53860.950

111.488

DF

211

13

7679

2I1

1

37679

211

1

37679

MeanSquare

2.5634.513

.613

2.813

2.646.993

1.056

.625,450.800

.450

.567

.112

.129

23.4639.113

37.813

3.613

16.846.802

1.411

F

2.5854.55.62

2.83

2.66

5.584.027.15

4.02

5.07

29.2611.3647.15

4.505

21.01

Sig.ofF

.082

.036

.435

.096

.054

.005

.048

.009

.048

.003

.000

.001

.000

.037

.000

12

brand and ad recognition than radiocommercials(i.e., picture-superiority ef-fects at superficial processing levels). Assbown in Table 3, tbe cell means arein the expected direction and the apriori contrasts are significant for bothbrand recognition (p < .01) and ad rec-ognition (p < .001). Hypothesis 2 pre-dicted no difference in the brand- andad-recognition scores for radio and TVsubjects who have high involvementwith the commercials. This was be-cause TV's cognitive advantages at su-perficial processing levels would beoffset by radio's cognitive advantages atthe elaborative processing levels for in-volved consumers. As shown in Table3, the cell means are similar and fail toapproach statistical significance as pre-dicted in Hj.

The effects of the media-by-involve-ment interactions on the dependentvariables of elaborations, and brand andad recognition are summarized in Fig-ures 1-3. Together these findings areconsidered to support the hypotheses.

Post Hoc Analysis. In addition to for-mal hypothesis testing, a post-hoc anal-ysis was conducted by examining thecognitive responses for significant (AN-OVA) main effects. Results showed in-teresting patterns of response on twodimensions: (I) the focus of cognitiveresponse and, (2) the favorability ofcognitive response. As shown in Table4, subjects exposed to TV commercialsproduced significantly more message-related thoughts and marginally fewerproduct-related thoughts. This indi-cates that the focus of consumer's cog-nitive responses reflect more of a brand-processing orientation for radio com-mercials and more of a message-pro-cessing orientation for TV commercials.

In terms of the favorability of cog-nitive response, analysis (shown in Ta-ble 4) indicates that, in this study,subjects exposed to TV commercialsreported more positive thoughts,though this effect was only marginallysignificant (p < .10). No difference wasfound in the number of negativethoughts between the two broadcastmedia. While Edell and Keller (1989)did not extract cognitive responses,

TABLE 3A Priori Contrasts Analyses

Measure

ElaborationsFor Low Involvement

ElaborationsFor High Involvement

Brand RecognitionFor Low Involvement

Brand RecognitionFor High Involvement

Ad RecognitionFor Low Involvement

Ad RecognitionFor High Involvement

Mean:

Radio

.55

1.10

.60

.95

1.90

3.70

TV

.45

.25

.90

.95

3.00

3.95

Value

0.100

0.850

- .300

0.000

-1 .100

- .250

t

Valu<

0.315

2.698

- 2.837

0.000

-3 .884

- .883

DF

76

76

76

76

76

76

Sig-

NS

.009

.006

NS

.000

NS

TABLE 4ANOVA Summary for Post Hoc Analysis

Measure

Message Thoughts

Product Thoughts

Positive Thoughts

Negative Thoughts

Attitude Toward the Ad

Mean:

Radio

2.38

2.00

2.40

1.50

1.35

TV

3.43

1.33

3.20

0.00

2.08

Square

22.05

9.11

12.80

49.49

99.90

DF

1

11

1

1

F

6.98

3.17

2.97

.43

12.44

Sig.

.010

.079

.089

NS

.000

they did find that TV subjects had morepositive attitudes toward the commer-cials than radio subjects. This is im-portant because A^j has been shown tomediate brand attitudes in a variety ofsituations (MacKenzie and Lutz 1989;MacKenzie, Lutz, and Belch 1986;Mitchell and Olson 1981). Examiningthis issue in terms of the data collectedhere (Table 4), TV subjects developedsignificantly more favorable attitudestoward the ad than radio subjects, rep-licating the findings of Edell and Keller(1989).

Limitations

Before discussing the impUcations ofthis study it is important to properlydelimit its scope. First, the results re-ported here were obtained after onlyone exposure to the target ad. Thus,

FIGURE 1Media by Involvement Interaction

Effect for Elaborations

1.5 --

M 1,0--

.5 --

0.0Low

InvolvementHigh

Involvement

13

FIGURE 2Media by Involvement Interaction

Effects for Brand Recognition

1.0 --

.75--

ICO

.50--

.25--

0.0Low High

Involvement Involvement

FIGURE 3Media by Involvement InteractionEffects for Ad Point Recognition

8 4.0 t

Iou

S

3.0 •-

2.0 --

1.0 •-

0.0

LowInvolvement

HighInvolvement

the involvement/media interactionsfound here can be predicted with con-fidence on!y during early exposure toan advertising campaign. What hap-pens to consumer elaboration and rec-ognition of brand and copy points aftermultiple exposure is certainly a worthyarea for future research along the linesof Edell and Keller (1989).

A second limitation of this study isthat mediating factors other than con-sumer involvement were held constant.It is possible that variables such as de-gree of message consistency, presenceof distracting secondary tasks, or dif-ferences in message execution could in-teract with involvement to formsystematic response patterns, Again,this is considered a fertile area for fu-ture broadcast advertising research.

Third, this study investigated cog-nitive reactions to broadcast commer-cials. Affective responses such asemotions or mood states may show dif-ferent response patterns. The extent towhich radio commercials can evokeelaborations of an affective nature vis-a-vis TV is another area where futureresearch is needed.

Finally, this study employed a con-venience sample of college seniors. Itis possible that this group has bettercognitive skills than average con-sumers, which may enhance cognitiveprocessing, especially for subjects in thelow-involvement treatment. If true,however, this artifact would make itmore difficult to observe the effects hy-pothesized in tbis study.

Implications

Past Research. A review of past re-search on listening and viewing dem-onstrated significant gaps in ourknowledge of how consumers processbroadcast advertising. Listening is a veryimportant dimension of consumer re-sponse to radio and TV commercials,but little empirical or theoretical workhas been reported in the marketing andadvertising journals. While some inter-esting research has developed for verbaland visual processing for print ads. testsusing broadcast media are just begin-ning. Given the importance of broad-cast media it is clear that substantiallymore research is needed. Future broad-cast-advertising research is needed inorder to develop a conceptual model ofprocessing differences. This will allowadvertisers to better understand and re-spond to consumers' information-pro-cessing needs.

Personal Connections. The degree ofcognitive elaboration a consumer un-dertakes with regard to an advertisingmessage has been theorized to signifi-cantly impact other important re-sponse variables like recognition ofbrand and ad points. T^is study foundthat the amount of cognitive elabora-tion, in the form of personal connec-tions or bridging experiences, did vary

systematically depending on type ofmedia and level of involvement. More-over the differences in personal con-nections predicted subsequentdifferences in consumers' recognitionof brand and message points. Thismeans that the amount of personalelaboration is an important dimensionof consumer cognitive response thatshould be studied in much more depth.

Especially under conditions of highinvolvement, the amount and nature ofcognitive elaboration is crucial. Futureresearch is needed on this topic to pro-vide more specific scaling techniquesfor personal connections. For example,it is common to use the cognition-list-ing method to collect consumers' cog-nitive responses, but analyzing thoseresponses for personal connections israre. In this study, the mere number ofpersonal connections was found to beimportant. Future research should in-vestigate other meaningful aspects. Forexample, self-based connections havebeen hypothesized as more personallyrelevant and thus, could have a differ-ent effect than cue-based elaborations.This is a fertile area for future researchand would represent continued devel-opment in the coding of cognitive re-sponses.

Involvement Manipulation Checks.Another implication drawn from thisresearch regards the involvement ma-nipulation and the scales recommendedby Laczniak, Muehling, and Grossbart(1989) to validate involvement manipu-lations. The scales, slightly altered toadapt them from print ads to broadcastcommercials, performed well. Coeffi-cient alphas were high for both the 5-item ad-attention scale and the 4-itembrand-processing scale. These findingsare similar to those reported by La-czniak, Muehling, and Grossbart {1989),and extend tbe external validity of thescales to media other than print. In ad-dition, the results from the cognitive-re-sponse analysis showed that high-involvement subjects generated signifi-cantly more total thoughts, more prod-uct thoughts, more message thoughts,and fewer unrelated thoughts than did

14

the low-involvement subjects. These re-sults using a different scaling method addcovergent validity to the Laczniak,Muehling, and Grossbart (1989) scales.This suggests that researchers seeking tomanipulate audience involvement can ef-fectively validate their treatment by us-ing these scales.

Involvement by Media Effects. Re-sults of this study showed that at lowlevels of consumer involvement, adprocessing is superficial, and few per-sonal connections occur, In these cases,TV, with dual input modes, can be ex-pected to achieve more cognitive im-pact on consumers, resulting in greaterrecognition of brand and messagepoints. Thus. TV with visual and au-ditory stimuli appears well suited to at-tract meaningful attention to thecommercial, even from uninvolvedconsumers, ^)fl ile the processing at thislevel is superficial, it can nonethelessbe effective, especially with repetitionover time. Particularly in product cat-egories where brand differences aresmall and the consequences of a non-optimal choice minimal, superficial rec-ognition of brand and ad points couldplay a significant role in determiningchoice. In these cases the higher costof TV commercials seems reasonableas they provided significantly greaterrecognition of brand and ad points thanthe radio commercials tested here.

However, when consumers are in-volved, the cognitive impact of radioversus TV messages changes signifi-cantly. Here, radio was shown to pro-duce significantly more personalelaborations than the TV commercials.These results are thought to be causedby the greater opportunity to elaborateradio claims made possible by the factthat no visual stimuli are presented to(1) use short-term-memory capacity,and (2) cue elaborations to visual ele-ments that may not be personally rel-evant to the consumer. Whenconsumers are motivated to elaboratemessage claims, recognition of brandand ad points for radio commercials isequivalent to that of TV Here, the costdifferences between radio and TV may

not be warranted, suggesting that radiomight be a very cost-effective mediumfor advertisers promoting products andservices to involved consumers.

Managing Involvement. Research hasdemonstrated that consumer involve-ment is a dynamic concept that candepend on aspects of the situation, theproduct, the advertisement, or the con-sumer's response process. In addition,involvement can be temporary and/orrecurring (e.g., buying a new car everyfive years), or can have its own life cycle(e.g., hobbies in which one loses inter-est). This makes managing involvementa challenging task for advertisers.Nevertheless, some generalizations canbe drawn. First, some products are nat-urally more involving than others.Those with high cost, high perceivedrisk, a lengthy life expectancy, or sig-nificant consequences of a sub-optimalchoice can be expected to receive moreattention and elaboration in the con-sumer's decision-making process, Forproducts that fit these criteria, the useof radio in the media mix should beconsidered, since consumers will oftenbe motivated to engage in elaborativeprocessing.

Products that are less naturally en-gaging (i.e.. low cost, low risk, short lifeexpectancy, little consequence of a sub-optimal choice) are less likely to induceelaborations in the decision-makingprocess. Here, two recommendationscan be made. First, higher degrees ofinvolvement can be attained by creat-ing commercials that are themselves in-volving. Such commercials would beespecially useful for advertisers of low-involvement products that include ra-dio as a significant element in the mediamix. Radio commercials that stimulateconsumers to elaborate the copy pointsand the brand name can be very effec-tive. Second, the results here indicatethat TV is a very effective medium foradvertising products to low-involve-ment consumers.

Post Hoc Analysis. Post hoc analysisof media main effects found that con-

sumers exposed to TV commercialsproduced significantly more messagethoughts, but significantly fewer prod-uct thoughts. This suggests that thedual input modes of TV can be usefulin attracting attention, but that this in-creased focus on the ad can come atthe expense of brand-processing activ-ities. For advertisers traversing the pe-ripheral route, this is not necessarily adisadvantage since ad cues, or affectivemoods created, could be the basis foran effective campaign.

However, when the advertiser seeksto follow the central route to persua-sion, the situation changes. Here, high-involvement consumers seeking to dil-igently consider product-related infor-mation do not need to be cajoled intolistening/viewing. In these cases, heavyuse of executional elements to attractattention is not necessary and may in-terfere with brand processing sincesome attention is diverted to the exe-cutional aspects. Thus, advertisersseeking to follow the central route topersuasion are advised to include radioin the media mix, since it can faciliutebrand processing.

Another main-effect finding of in-terest was that consumers liked the TVcommercial better regardless of in-volvement. This result may simply in-dicate that the commercial (originallydesigned for TV) was better suited tothat medium. However, the commer-cial was specifically chosen because ithad reasonably strong radio presenceand was very effective in the high-in-volvement radio group. Also, the sameeffect was reported by Edeil and Keller(1989) using different commercials. An-other explanation is that in conditionsof low involvement, consumers enjoyprocessing TV commercials more thanradio commercials because: (1) moresensory inputs are stimulated, (2) TVspots are more entertaining, or (3) theTV spots require little cognitive effortto process superficially (even thoughmore information is involved). Thus,when involvement is low, TV is notonly better liked, it is also more effec-tive. This combined effect suggests thatTV is better suited for promoting

15

products of relatively little interest toconsumers.

At higher levels of involvement, ra-dio requires more cognitive effort toelaborate the claims (since no visualcues are provided) and consumers mayfind this task effortful, and thus, lessenjoyable (thereby causing lower A jSfor radio subjects). However, this maynot be a negative effect in the long runsince the diligent cognitive labor maygenerate more relevant and long-last-ing impressions of the brand and/orad. In this study the better liked TVads were no more effective than radioads under conditions of high involve-ment. While these explanations aresomewhat speculative at this point, thisissue is important enough to warrantfuture research.

Conclusion. After finding little pre-vious research on consumer-processingdifferences between radio and TVcommercials, a preliminary model ofhow consumers respond cognitively tobroadcast commercials was developed.A test of the model showed that thelevel of consumer involvement playeda significant role in determining thenumber of personal connections andsubsequent recognition of brand andcopy points. While this is thought tobe a useful point of departure, consid-erably more research is needed to de-velop a complete model of broadcastadvertising differences. Continued re-search along these lines will aid creativeteams and media planners in their ef-forts to produce and deliver the mosteffective marketing communicationspossible.

References

Age Best TV Commercials of theYear—I9S6." Fred Danzig, ed., Chicago:Crain Communications Inc.

Barker, L. L. (1971). Listening Behavior, Pren-tice-Hall: Englewood ClifFs. N.J.

Bostrom. R. N. and C. L. Bryant (1980), "Fac-tors in the Retention of Information Pre-sented Orally: The Roie of Short-TermListening." Thfi Wesurn journal of Speech

Communication, 44 (Spring). 137-145.Bostrom, R. N. and E. S. Waldhart (1980),

"Components in Listening Behavior: TheRoie of Short-Term Memory," HumanCommumcation Research. 6. (3. Spring). 221-227.

Childers, T L. and M. J. Houston (1984)."Conditions for a Picture-Superiority Ef-fect on Consumer Memory," Journal ofConsumer Research. 11 (September), 643-654.

Craik, Fergus I. M. and R. A. Lockhart (1972)."Levels of Processing: A Framework forMemory Research," Journal of VerbalLearning and Verbal Behavior. 11 (6). 671-684.

Edell, Julie A. and Kevin L. Keller (1989), "TheInformation Processing of CoordinatedMedia Campaigns." Journal of MarketingResearch. 26 (May), 149-163.

Greenwald. A. G. and C. Leavitt (1984). "Au-dience Involvement in Advertising: FourLevels." Journal on Conjumer Research. 11(June), 581-592.

Holbrook, Morris B. (1978), "Beyond Atti-tude Structure," Journal of Marlceting Re-search, 15 (November), 546-556.

Houston. M.J.. T L. Childers, and S. E. Heck-ler (1987), "Picture-Word Consistency andthe Elaborative Processing of Advertise-ments," Journal of Marketing Research, 24(November), 359-369.

Krugman, Herbert E. (1965), "The Impact ofTelevision Advertising: Learning WithoutInvolvement," Public Opinion Quarterly. 29(Fall). 349-356.

Laczniak. Russell N., Darrel D. Muehling, andSanford Grossbart (1989), "ManipulatingMessage Involvement in Advertising Re-search," Journal of Advertising. 18 (2). 28-38.

Leigh, James H. and Anil Menon (1987), "Au-dience Involvement Effects on the Infor-mation Processing of Umbrella Print Adver-tisements," Journal o/Advertising, 16 (3), 3-12.

Maclnnis, D. J. and L. L. Price (1987), "TheRole of Imagery in Information Processing:Review and Extensions," journal of Con-sumer Research. 13 (March), 473-491.

Maclnnis, D.J. and Bernard J. Jaworski (1989),"Information Processing from Advertise-ments: Toward an Integrative Framework,"Journal o/Marketing, 53 (4), 1-24.

MacKenzie, Scott B. and Richard J. LutE (1989),"An Empirical Examination of the Struc-tural Antecedents of Attitude Toward tbeAd in an Advertising Pretesting Context,"Journal of Marketing, 53 (April). 48-65.

MacKenzie. Scott B., Richard J. Lutz, andGeorge E. Belcb (1986). "The Role of At-titude Toward tbe Ad as a Mediator of Ad-vertising Effectiveness: A Test of CompetingExplanations," Journal of Marketing Re-search. 23 (May). 130-143.

Marketer^ Guide To Media (1989), Gerri Lee,ed. New York, NY: A/S/M Communica-tions.

Mitchell, Andrew A. and Jerry C. Olson(1981). "Are Product Attribute Beliefs theonly Mediator of Advertising Effects onBrand Attitude?" Journal of Marketing Re-search. 18. (August), 318-332.

Murdock. B.B.. Jr. (1967). "Auditory and Vis-ual Stores in Short Term Memory," ActaPyychologica. 27 (Topic 4). 316-324.

Murdock, B. B., Jr. (1968), "Serial Order Ef-fects in Short Term Memory," Journal ofEjcperimental Ps5icholog:y Monograph Supple-ment, 76 (April), M5.

Murdock, B. B., Jr. (1969) "Where or When:Modaliry Effects as a Function of Temporaland Spatial Distribution of Information,"Joiirnal of Verbal Learning and Verbal Be-havior, 8 (June), 378-383.

Nichols. R. G. (1957), "Listening is a Ten PartSkill." Nationi Business. 45 (July), 4.

Nichols, R. G. and L. A. Stevens (1957), "Lis-tening to People," Harvard Business Revieu',35 (September-October), 90-97.

Pauk, Walter (1984). "Forgetting and Remcm-benng," Hou; To Study In College, Hough-ton Mifflin Co., 82-113.

Petty. Richard E. and John T Cacioppo (1983),"Central and Peripheral Routes to Persua-sion: Application to Advertising" in Ad-vertising and Consumer Psychology, LarryPercy and Arch G. Woodside, eds., Lex-ington, MA: Lexington Books.

Petty, Richard E., John T Cacioppo, and Dav-id Schumann (1983), "Central and Periph-eral Routes to Advertising Effectiveness:The Moderating Role of Involvement,"Journal of Consumer Research. 10 (Septem-ber), 135-146.

Siegel, A. W. and J. P Allik (1973), "A De-velopmental Study of Visual and AuditoryShort-Term Memory." Journal of VerbalLeaming and Vferbal Behavior, 12 (Augusr),409-418.

Smith, Robert E. and William R. Swinyard(1983), "Attitude-Behavior Consistency: TbeImpact of Product Trial Versus Advertis-ing," Journal of Marketing Research. 20,(August), 257-267.

Smith, Robert E. and William R. Swinyard(1988), "Cognitive Response to Advertisingand Trial: Belief Strength. Belief Confi-dence and Product Curiosity," Journal ofAdvertising 17 (3), 3-14.

Spearritt, Doriald (1962). Listening Compre-hension—A Factorial Analysis. Melbourne:G. W. Green &. Sons PTY. LTD.

Wright. Peter L. (1973), "The Cognitive Pro-cesses Mediating Acceptance of Advertis-ing," Journal of Marketing Research, 10,(February), 53-62.

Wright, Peter L (1974), "Analyzing Media Ef-fects on Advertising Responses," PublicOpinion Quarterly, 36 (Summer), 192-205.

16

Wrigbt,PeterL.(1975),"FactorsAflFectingRe- Wrigbt, Peter L. (1980), "Message Evoked Received Nowmber 22.1989. Revision acceptedsistance to Advertising," JoHmalo/Comum- Tboughts; Persuasion Research Using for publication February 7. 1990.er Research, 1 (June), 1-9. Thought Verbalizations," journal of Con-

sumer Rexarch, 7 (September), 151-175. ^

APPENDIXCoding Definitions for Cognitive Responses

I. Type of Thought:

A. Product-Related Thoughts. These thoughts refer to the brand or product class (computers).TTiey include:

1. Identification and /or evaluation of product attributes.2. Statements related to the performance of the product.3. Statements related to the consequences of using the product.4. Questions about the brand or product class.5. Statements indicating how the product could solve a problem.

B. 'hAessage-Related Thoughts. Any thought that identifies or evaluates execution aspects of the advertising message.

1. Statements regarding the effectiveness of the ad.2. Statements expressing interest in the ad.3. Questions about the ad.4. Statements regarding attributes of the ad.

C. SouTcc Related TKoughts. The source is defined as the perceived purveyor of the message. This category includes any thoughtthat relates to the credibility and/or effectiveness of the source of product information (references to the "advertiser" or the"source"). In this study the source would be the sponsor/manufacturer (Apple referred to as a company rather than as aproduct).

1. Statements regarding the perceived expertise of the source (i.e., the ability of the source to make accurate asscrtations).2. Statements regarding the perceived trustworthiness erf the source (i.e., the willingness of the source to make accurate

assertations).3. Statements regarding the effectiveness of the source (i.e., the source's likability, similarity, confidence, status, etc.).

D. Unrelated Thoughts. All thoughts not fitting the above categories.

II. Intent of Thought:

A. Positive Statements. Any statement that is in favor of or otherwise supports the product/message/source.B. Neutral Statements. Declarative statements regarding the product/message/source that do not indicate a favor^le or

unfavorable intent.C. Negative Statements. Any statement that is unfavorable toward the product/message/source. Any question that derogates or

challenges assertions made about the product/message/source.D. Curiosity Statements. Any statement that expresses a desire for additional information about the product/message/source.

These statements are distinguishable from negative statements based on your judgment of the subject's intent.

III. Presence of Personal Connections:Thoughts are considered personal connections if respondents connect the brand, product class, or elements in the advertisementto aspects of their own life (e.g., "I thought about using the computer," or "This computer would be great for my accountingproblems)."

17