memory and evaluation effects in competitive advertising environments

15
Journal of Consumer Research, Inc. Memory and Evaluation Effects in Competitive Advertising Environments Author(s): Kevin Lane Keller Source: Journal of Consumer Research, Vol. 17, No. 4 (Mar., 1991), pp. 463-476 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/2626840 . Accessed: 17/02/2015 14:31 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . The University of Chicago Press and Journal of Consumer Research, Inc. are collaborating with JSTOR to digitize, preserve and extend access to Journal of Consumer Research. http://www.jstor.org This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PM All use subject to JSTOR Terms and Conditions

Upload: gautam-bothra

Post on 17-Nov-2015

231 views

Category:

Documents


5 download

DESCRIPTION

NA

TRANSCRIPT

  • Journal of Consumer Research, Inc.

    Memory and Evaluation Effects in Competitive Advertising EnvironmentsAuthor(s): Kevin Lane KellerSource: Journal of Consumer Research, Vol. 17, No. 4 (Mar., 1991), pp. 463-476Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/2626840 .Accessed: 17/02/2015 14:31

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

    .JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

    .

    The University of Chicago Press and Journal of Consumer Research, Inc. are collaborating with JSTOR todigitize, preserve and extend access to Journal of Consumer Research.

    http://www.jstor.org

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/action/showPublisher?publisherCode=ucpresshttp://www.jstor.org/stable/2626840?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp

  • Memory and Evaluation Effects in

    Competitive Advertising Environments

    KEVIN LANE KELLER*

    A laboratory experiment replicates and extends prior research on how competitive advertising and retrieval cues affect consumer memory and evaluations of brands. The number and valence of competing ads, presence of ad retrieval cues, and valence of target ads were manipulated. A high level of competitive advertising varying in valence produced interference effects for recall and evaluations. Ad retrieval cues offset these effects and enhanced recall and evaluations even when there were no competing ads. Interference effects were more pronounced for recall of brand claims; cue effects were more pronounced for recall of cognitive responses and evaluations of the advertised brand.

    onsumers typically are exposed to advertising for more than one brand in a product category. Be-

    cause many brands within a product category try to reach similar target markets, consumers may even be exposed to these ads in the same media vehicles. This simultaneous exposure may be particularly evident for print ads, given the large number of special interest magazines. Despite this fact, relatively little research has examined the effects of competitive advertising on consumer memory and brand evaluations. The key question is, To what extent is the advertising effec- tiveness for a target brand affected by other advertising, particularly of brands within the same product cate- gory?

    Keller (1987) and Burke and Srull (1988) adopted an information-processing perspective to study com- petitive advertising. They showed experimentally that increases in the amount of competitive advertising produced interference effects and significantly reduced recall of brand claims, given brand name and product category cues. Keller also showed that the decrement in recall of brand claims could be partially offset by

    the presence of advertising retrieval cues (i.e., execu- tional information from the original ad). These studies did not demonstrate, however, that competitive ad- vertising can have any detrimental effect on brand evaluations. Although a wide range of competitive conditions was examined in the Burke and Srull study (i.e., zero to three competing ads), the valence of target and competing ads was not manipulated (i.e., varied in likability or effectiveness). The valence of target and competing ads was manipulated in the Keller study, but only a restricted range of competitive conditions was examined (i.e., one or three competing ads).

    It may be that interference effects for evaluations of a target ad are evident only when comparisons are made between evaluations for a target brand with competing ads that vary in valence and a target brand without competing ads. Greater confusion among ads of a different valence from that of the target brand should produce different evaluations. When there are no competing ads, however, there is no opportunity for such confusion. Thus, the primary purpose of this research is to demonstrate that competitive advertising can influence evaluations of a target brand, and the effects of the number and valence of competitive ads are specifically considered. Moreover, this research examines the effects of ad retrieval cues on memory and evaluations under different competitive ad con- ditions. If interference effects can be produced by competitive advertising, it is important to identify fac- tors that help to eliminate those effects. Finally, this research considers how the effects of competitive ad- vertising and ad retrieval cues depend on the valence of the target ad. Manipulating the valence of the target ad broadens the tests of the hypotheses and increases the robustness of the findings. In short, the goal is to replicate and extend prior research by exploring hoiindarv conditions of the effects of cnmnetitive ad-

    *Kevin Lane Keller is associate professor of marketing, Graduate 'School of Business, Stanford University, Stanford, CA 94305. Dur- ing 1990-1991, he is visiting professor, Australian Graduate School of Management, University of New South Wales, Sydney, Australia. Portions of this research were conducted while the author was on the faculty at the University of California at Berkeley and supported by research grants from the Institute of Business and Economic Research and the Marketing Science Institute. Additional support through a faculty fellowship at Stanford University was made pos- sible by the generosity of James and Doris McNamara. The author thanks Jim Bettman, Dipankar Chakravarti, Joel Cohen, Steve Hoch, Amna Kirmani, John Lynch, Chris Puto, Doug Stayman, Peter Wright, three anonymous JCR reviewers, and participants of seminars at the University of Florida, Pennsylvania State University, and the Stanford marketing camp for helpful comments.

    463 ) 1991 by JOURNAL OF CONSUMER RESEARCH, Inc. 0 Vol. 17 0 March 1991

    All rights reserved. 0093-5301/91/1704-0008$02.00

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

  • 464 JOURNAL OF CONSUMER RESEARCH

    vertising and advertising retrieval cues on consumer memory and brand evaluations.

    CONCEPTUAL BACKGROUND

    Memory

    An associative memory model (Anderson 1983; Wyer and Srull 1986) views memory as a collection of elements that vary in their strength of association. Two main types of communication effects from ad pro- cessing that a person may store in memory as part of an ad memory trace (Hutchinson and Moore 1984; Keller 1987, 199 la) are (1) a representation of the ad itself (e.g., the brand claims expressed in the ad) and (2) responses to the ad (e.g., cognitive responses gen- erated during ad exposure). Accessibility or likelihood of retrieval of communication effects will depend on their organization in memory and the retrieval cues present in the recall setting. This research considers how both of these factors operate.

    Effects of the Number and Valence of Competing Ads. One important factor that has been shown to create interference and affect memory for an advertised brand is competitive advertising (Burke and Srull 1988; Keller 1987). Interference effects in memory occur when the accessibility of communication effects is re- duced by the presence of some other information in memory (Postman and Underwood 1973). The pres- ence of other information in memory may cause target information to be either inaccessible or confused with other information. Thus, if multiple brands advertise in the product category, overlapping ad memory traces may confuse the correspondence of ads and brands in the product category. Moreover, relatively weak as- sociations may be formed between brands and their ad memory traces so that some communication effects in the product category cannot be recalled.

    Keller ( 1987) showed that recall of brand claims was higher when a target ad was competing with one rather than three advertised brands. Recall of cognitive re- sponses, however, was not affected by the number of competing ads. Burke and Srull (1988) showed that recall of information presented in target ads was much higher when subjects saw no competing ads than when they saw one, two, or three other ads. They did not measure recall of cognitive responses. When no other competing brands are advertising, the presence of the product category as a retrieval cue presumably allows access to the ad memory trace and its contents even if the brand name is not strongly associated to the ad memory trace. Moreover, there is no opportunity for confusion. For these reasons, communication effects (e.g., brand claims and cognitive responses stored in memory) should be more accessible with no competing advertising, compared with when competing ads are present. The more competing ads in the product cat-

    egory, however, the more likely an ad will be confused or overlooked.

    Hi: The more competing brands advertising in the product category, the lower the recall of communication effects for a target ad is.

    Besides quantitative aspects of competitive adver- tising, qualitative aspects also affect recall. The dis- tinctiveness of a piece of information in memory pos- itively influences its retrievability and resistance to interference (Eysenck 1979; Moscovitch and Craik 1976). As Craik notes (1979, p. 97), "The effectiveness of a specific cue will depend on its similarity to the encoded trace but also on its similarity to other traces that are then said to interfere with successful retrieval of the wanted event." Distinctiveness is not an absolute property of a stimulus, however, and can be defined only relative to some background or set of salient stimuli (Jacoby and Craik 1979). For example, re- searchers have shown that associative interference in- creases as semantic similarity between items increases (Dempster 1985; Underwood 1983a), as similarity in input modalities increases (Glenberg 1984; Under- wood 1983b), and so on.

    In the study reported here, input modality and other structural ad elements are held constant, along with the basic product positioning of the brand, and the evaluative congruency of competing advertised brands is manipulated. This strategy recognizes that a partic- ularly important aspect of competing brands may be their overall valence, calculated in terms of consumers' overall ad and brand evaluations. Brands of roughly equal quality are likely to be strongly associated in memory and identified as high- or low-quality brands. Moreover, consumers may distinguish advertised brands on the basis of their reactions to ads. For ex- ample, in reviewing the affect-and-memory literature, Isen (1989, p. 99) states that "not only do people re- member how things made them feel, but. . . they of- ten organize cognitive material in terms of how it made them feel." Thus, consumers may have strong asso- ciations in memory regarding the overall valence of an advertised brand, in terms of the appeal of an ad's execution or the persuasiveness of its claims.

    An advertised brand may be more distinctive and stand out as relatively better or worse if its valence differs from that of others. Ad distinctiveness should play an important role in moderating the effects of competitive advertising. In particular, associative in- terference should be greater when the target and com- peting ads are evaluatively congruent. For example, assume that two brands are advertising in a product category. If consumers evaluate both brands positively and like both ads, then both ad memory traces would likely be associated with "good brands in the product category" and stored close together in memory. Al- though consumers would presumably not actually overlook an ad memory trace with only two ads, they

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

  • COMPETITIVE ADVERTISING ENVIRONMENTS 465

    may have difficulty distinguishing which ad corre- sponded to which brand. All of these assumptions sug- gest the following hypothesis.

    H2: A competing ad of the same valence lowers recall of communication effects for a target ad, compared with a competing ad that evokes different responses.

    Effects of Ad Retrieval Cues. Another key factor affecting memory accessibility is whether external re- trieval cues are available. Information is effective as a retrieval cue to the extent that it is stored in the mem- ory trace with the to-be-remembered information (Craik 1979). The encoding specificity principle (Tulving 1974, 1979) holds that memory accessibility is a function of the compatibility between trace and cue information (see also Morris, Bransford, and Franks 1977). Along these lines, Keller (1987) argues that, in many settings, ad execution information may be more uniquely and strongly related in memory to brand claims in the ad than to the brand name or iden- tification of the advertised product.

    The brand name may have relatively weak associ- ations with elements of the ad memory trace as a result of a variety of factors (Keller 199 la), such as the nature of the ad itself (either in its structure or content), the nature of the surrounding ad environment, and char- acteristics of the person processing the ad. As noted, competitive advertising can increase the likelihood that an ad memory trace is confused or overlooked. Because the brand name is a less effective retrieval cue, exe- cutional information from the ad (e.g., the photo, headline, or some other key graphic from a print ad) may be more strongly linked to the experience of and knowledge acquired from ad exposure. Yet, with no competing ads, as noted above, it is unlikely that the ad memory trace would be confused or overlooked. In such cases, an ad retrieval cue should have little effect on memory because recall is already high.

    It would seem that the effect of an ad retrieval cue on recall should be proportionate to the number of competing brands advertising in the product category. Yet, Keller (1987) found that an ad retrieval cue had roughly equal effects on recall of brand claims with either one or three competing ads. He interpreted this finding as evidence that competitive advertising can produce overlapping memory traces such that there is confusion not only as to which -ad corresponds to which brand name but also regarding which ad elements are associated (e.g., which brand claims with which ad ex- ecution). Thus, competitive advertising can result in weaker links between ad retrieval cues and other com- munication effects in memory. Because this effect could easily be present with one competing ad of the same valence, the following hypothesis distinguishes only between the effects of ad retrieval cues in the presence or absence of competing ads.

    H3: An ad retrieval cue has a greater facilitating effect on recall of communication effects for a target ad in the presence rather than ab- sence of competing ads in the product cate- gory.

    Evaluations

    Effects of the Number and Valence of Competing Ads. Prior research on the relationship of memory and evaluations can be used to develop hypotheses about how consumer memory will affect evaluations of target ads in competitive advertising environments. One well-accepted view of the role of memory is that evaluations are a function of the accessibility and va- lence of information retrieved from memory (Kisielius and Sternthal 1986; Reyes, Thompson, and Bower 1980). It has also been argued that the actual infor- mation retrieved from memory for evaluations de- pends on its relevance (Baker and Lutz 1987) or di- agnosticity (Feldman and Lynch 1988) as well as its accessibility. Because the memory hypotheses devel- oped in the preceding section indicate the accessibility of information in memory, they can be used as a start- ing point for developing evaluation hypotheses.

    Since the product category itself is as an effective retrieval cue when there are no competing ads, eval- uations in this condition should reflect the general va- lence of the target ad. With competing ads, commu- nication effects in the ad memory trace are more likely to be overlooked or confused and thus may not be accessible as input to evaluations. Accessibility prob- lems should affect evaluations, however, only when the valence (i.e., the persuasiveness and likability) of the target ad differs from the valence of compet- ing ads.

    When competing ads have the same valence as the target ad, consumers may have difficulty identifying which ad corresponds to which brand in the product category (per Hypothesis 2), but evaluations of the tar- get brand should not be affected by accessibility prob- lems because (1) regardless of any confusion, the result would still be retrieval of an ad memory trace, the con- tents of which were roughly of the same valence as that of the target ad and (2) consumers presumably could recall that all advertised brands had roughly similar valence. Thus, even if consumers were not able to ac- cess the particular ad memory trace associated with the target brand, their overall product category eval- uations could be viewed as relevant or diagnostic for inferring its quality. In accordance with this reasoning, Baumgardner et al. (1983) show that messages advo- cating similar evaluative responses to similar products attentuate decay of persuasion.

    When competing ads are of a valence different from that of the target ad, however, accessibility problems should lead to different evaluations of the target brand than would occur if there were no competing ads. To

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

  • 466 JOURNAL OF CONSUMER RESEARCH

    illustrate, consider two brands advertised in a product category; one ad is either well or poorly done, and the other has the opposite valence. If a good target ad is confused with a bad competing ad, then lower evalu- ations of the target brand should result, compared with when there are no competing ads and thus no confu- sion. Similarly, if a bad target ad is confused with a good competing ad, higher evaluations of the target brand should result. In both cases, overall product- class evaluations would be less relevant or diagnostic for evaluations of the target brand. If overall product- category evaluations are of roughly neutral valence and if consumers rely on these to evaluate the target brand when they have accessibility problems, they should again give lower evaluations for a good target ad and higher evaluations for a bad target ad. These assump- tions can be summarized by the following hypothesis.

    H4: Evaluations for a good (bad) target ad will be lower (higher) in the presence of compet- ing ads of different valence in the product category.

    Effects of Ad Retrieval Cues. When ads in a prod- uct category have the same valence, the product cat- egory itself functions as an effective retrieval cue. If such ads vary in valence, however, the product cate- gory fails to function as an effective cue. In such cases, correct recall of communication effects from the ad memory trace should have a considerable effect on brand evaluations, and the ability of cues to help access such elements becomes more important.

    H5: An ad retrieval cue will have a greater effect on evaluations of a target ad in the presence of competing ads of differing valence.

    In summary, competitive advertising is not hypoth- esized to have identical effects on memory and eval- uations. Although a competing ad of the same valence as a target ad should be particularly confusing and may lower recall of the target ad's contents, evaluations should not be similarly affected because (1) any infor- mation recalled incorrectly should have roughly the same valence as information recalled correctly and (2) overall product category evaluations should also be accessible and relevant or diagnostic. Figure 1 sum- marizes the memory and evaluation predictions.

    METHOD The research methodology of the present study fol-

    lowed closely that of Keller (1987) in terms of stimuli, procedure, and measures. Subjects were 145 English- speaking undergraduates at a large West Coast uni- versity, and participation in the study fulfilled part of an introductory marketing course requirement.

    Design and Manipulations The experimental design was a four (competitive ad

    interference) by two (advertising retrieval cue) by two

    (target-ad valence) design. The interference factor was a mixed within- and between-subject manipulation that combined (1) a target ad with no competing ads (1-0), (2) a target ad and one competing ad of the same valence (I-l-s), (3) a target ad and one competing ad of different valence (I-i -d), and (4) a target ad with one competing ad of the same valence and two of dif- ferent valence (1-3). The last two conditions represent the low and high levels of competitive advertising in the design of Keller (1987). Their inclusion creates conditions conducive to interference and permits comparisons with the earlier study. The first two con- ditions represent extensions of the 1987 study.

    The retrieval-cue factor was a between-subject ma- nipulation that had two levels: ad cue absent and ad cue present. This factor also figured in the Keller (1987) design. The target-ad-valence factor was a within-sub- ject manipulation that had two levels: good target ad (G) and bad target ad (B). Manipulating the valence of the target ad broadens the tests of the hypotheses and increases the generalizability of the findings.

    Stimuli Ad Stimuli. A pool of 16 print ads was used for

    the study, four in each of four product categories: ce- real, laundry detergent, pain reliever, and toothpaste. The ads were basically informational and had identical structures. The top half of each ad consisted of a photo and the bottom half contained copy-a headline, an introductory transition paragraph, and two additional paragraphs that described one main claim each. A heading summarized the claim in bold type above the paragraph. Brand names, pretested to be evaluatively neutral and nonsuggestive, appeared in capital letters four times in each ad.

    Brands within product categories were basically unique in their positioning in that claims across ads were different. To implement the target-ad and com- petitive-ad-interference manipulations of valence, two good and two bad ads were made up for each product category. Good ads combined an ad photo that re- flected favorably on a brand in the product category with persuasive claims in the ad copy. Bad ads com- bined an ad photo that reflected unfavorably on a brand in the product category with unpersuasive claims. It was not the case that the claims themselves differed in valence. Pretest subjects found the attributes described by the claims to be equally desirable (i.e., good or had). Stronger arguments were associated with good ads, however, so that the belief that the brand possessed the attribute was manipulated presumably by the ad-valence factor (see Areni and Lutz [1988] for a discussion of the distinction between argument strength and valence). Pretests also established the fa- vorableness of the pictorial information, as well as the overall valence of the ads and brands. An analysis of data from a control group that further assesses the va- lence manipulation is reported subsequently.

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

  • FIGURE 1

    SUMMARY OF MAIN HYPOTHESES OF STUDY

    la. MEMORY PREDICTIONS

    o \~ " Ad Cue Present

    E E 0

    0

    X Ad Cue Absent

    I I I

    o 1 1 Numberof Competing Ads (different (same valence) valence)

    lb. EVALUATION PREDICTIONS

    Ad Cue Present (good target) -> ,,' Ad Cue Absent (good target)

    /- "\ Ad Cue Absent (bad target) o' ' ~~~~~Ad Cue Present (bad target)

    I I I

    0 1 1 Number of Competing Ads (different (same valence) valence)

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

  • 468 JOURNAL OF CONSUMER RESEARCH

    To implement the competitive-ad-interference ma- nipulation, the two good and two bad ads for each product category were placed in one of four groups: G 1, good target ads; B 1, bad target ads; G2, good com- peting ads; B2, bad competing ads. To ensure that sub- jects viewed and responded to a balanced set of ads in terms of valence, the target-ad valence and competi- tive-ad-interference factors were crossed between sub- jects by exposing subjects to one of two possible ad combinations, as follows. (A, B, C, and D represent the four product categories; the target ad in the product category is underlined.)

    A B C D

    Ad combination 1 GI G, BI B, GI B1,G1,B2,G2 Adcombination2 BI Bi1B2 Gi,G2 Gl,Bl,G2,B2

    Thus, subjects in both groups evaluated brands for two good and two bad target ads. In total, subjects in the first ad combination saw five good and four bad ads, and sub- jects in the second saw four good and five bad ads. The competitive-ad-interference factor was mixed within and between subjects in the sense that subjects in the first group saw target ads for one brand in condition 1-0, two brands in I-i -d, no brands in I-i -s, and one brand in 1-3; subjects in the second group saw target ads for one brand in 1-0, no brands in I- 1-d, two brands in I-1-s, and one brand in 1-3.

    Package Stimuli. The mock package fronts were the same size as the print ad. The brand name and product category appeared in large black letters at the top of each mock package front. The lower right-hand corner contained the cue manipulation. If present, the ad retrieval cue consisted of a reduced (roughly by half) black and white reproduction of the ad photo and headline. Because of this alteration of the ad infor- mation, the cue itself evoked little affect and contained little persuasive information, as confirmed by data from the control group reported below.

    Procedure Subjects' first booklet began with a description of

    an electronic information and shopping service, fol- lowed by instructions explaining that a series of ads would be shown from such a service. Subjects were told to judge the ads in terms of how good they felt the advertised brands were, and they were given 40 seconds to view each of the nine ads presented. The order of ads in the first combinalion was as follows. Again, target brands are underlined.

    Ad position

    1 2 3 4 5 6 7 8 9

    Product category D B C A D B C D D Ad type GI BI GI GI BI GI BI G2 B2

    Subjects viewed the ads in the second combination in the following order.

    Ad position

    1 2 3 4 5 6 7 8 9

    Product category D B C A D B C D D Ad type BI B2 G2 BI GI BI GI B2 G2

    The ads for the target brands in positions 4-7 were counterbalanced across subjects to minimize any serial position effects. Note that their interior position should have minimized any primacy or recency effects. Moreover, product category was counterbalanced with the interfer- ence factor so that each product category was associated with each interference level an equal number of times.

    After examining all of the ads, subjects responded to a four-item, 11 -point scale that asked for their overall re- action to all of the ads. Next, subjects completed a seven- minute distractor task by listing what they felt were ad- vantages and disadvantages of an electronic shopping service. Both of these tasks were intended to minimize the possibility of rehearsal and maximize the probability that later responses about the ads would involve retrieval of information from long-term memory.

    At this point, the ad booklets were collected and subjects were given, unexpectedly, a second booklet, which con- tained mock package fronts for the four target brands. Subjects were instructed that one goal of the study was to learn what consumers remembered about advertised brands when exposed to these mock package fronts while shopping. Subjects were asked to describe all they could remember about the ads for the target brands, in as much detail as possible, on the mock package front panels pro- vided. Then subjects wrote down on another set of panels what they remembered about their initial reactions to the ads (i.e., any positive or negative thoughts or feelings they recalled from ad exposure). Finally, subjects responded to a series of ad and brand evaluation measures based on the information on the mock package front. The order of panel exposure was the same for the memory and eval- uations measures and corresponded to the order in which the target ads had been shown. The order in which the memory and evaluation measures were taken, however, was counterbalanced across subjects.

    After the second booklet was collected, a third was distributed. In this final stage, subjects answered sev- eral additional memory and evaluation questions and some covariate questions. The entire experiment lasted less than an hour.

    Measures Subjects' recall protocols were coded into categories

    associated with the brand claims from the ad and with reported cognitive responses. The brand claims that subjects attributed to the target brand were coded into the total number of correct claims (i.e., the two main claims explicitly made in the target ad) and confused

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

  • COMPETITIVE ADVERTISING ENVIRONMENTS 469

    claims (i.e., claims from a competing ad mistakenly identified as being in the target ad). The coding scheme for the cognitive responses captured three different di- mensions (i.e., focus, level of abstraction, and valence) and was similar to coding schemes used in other ad- vertising studies (e.g., MacKenzie 1986; Mitchell 1986). It included three categories: specific ad reac- tions, specific brand-claim reactions, and overall ad or brand evaluations. The valence of the cognitive re- sponse, either positive or negative, was also coded. Two coders categorized the recall protocols. The estimate of reliability of the recall measures reported below, as measured by coefficient alpha, exceeded .70, which satisfies the conventional standards for basic research.

    Evaluation measures for attitude toward the brand, Ab, and toward the ad, Aad, were collected during ex- posure to the mock packages on four-item, seven-point scales. (For Ab, the items were bad-good, unpleasant- pleasant, dislikable-likable, low quality-high quality; for Aad, dislikable-likable, uninteresting-interesting, bad-good, unappealing-appealing.) The coefficient al- pha reliability estimates of Ab and Aad were .90 and .89, respectively. Scale measures were formed by sum- ming the items, such that larger numbers reflected more positive evaluations. Purchase intention toward the brand assumed a product category need and was measured by a single item on an 11-point scale (ex- tremely unlikely-extremely likely).

    A self-report measure in booklet 3 had subjects in- dicate on a five-point scale, "In considering ads in each product category, how confusing was it to 'keep the information straight' as to which ad went to which brand?" Larger numbers indicated greater confusion. The additional evaluation measure in booklet 3 in- structed subjects to indicate their overall evaluation of all brands for which they saw ads in each product category (1 = very bad, 5 = very good).

    Covariates

    Four covariates captured differences that might oth- erwise have obscured experimental effects. First, a se- ries of measures from booklet 3 assessed subjects' knowledge of the products in the various categories, their frequency of purchase and use, the importance of brand selection for subjects, and perceived quality differences among brands. As in previous research (see, e.g., Batra and Ray 1986; Keller 1987), these five mea- sures of processing motivation and ability were sig- nificantly, positively correlated. When combined, they had a satisfactory coefficient alpha of .74 and thus were averaged for the analysis, as in other studies, and broadly defined as subjects' "product category in- volvement." Second, to account for more general ex- perimental differences between product categories (such as those due to different ads), three dummy vari- ables were constructed. The final two covariates were constructed to account for effects resulting from coun-

    terbalancing orders. A variable for task order was coded 0 if the memory measures preceded the evalu- ation measures and 1 if the evaluation measures pre- ceded the memory measures. A variable for ad order was coded according to the relative position of the tar- get brand in the ad sequence and the memory and judgment tasks (i.e., first, second, third, or fourth).

    RESULTS

    Manipulation Checks For the manipulation checks, the responses of the

    control group were analyzed to ensure that (1) the tar- get ads and competing ads had the correct valence and (2) the advertising retrieval cue did not contain per- suasive information that might bias the interpretation of the evaluation measures.

    The 30 control subjects were given the same shop- per-of-the future cover story as the experimental sub- jects. In order to confirm that the cues on the mock package front panels did not contain any persuasive information independent of prior advertising expo- sure, the control group's first task was to evaluate the target brands on the basis of the mock package panels with the ad retrieval cue present. Subjects evaluated all eight target brands from 1 = very bad to 7 = very good. The mean Ab based on the cued package panels for the four good target brands, 4.06, was significantly less (but marginally so) than the mean Ab obtained from the same subjects later for the corresponding ads for these brands, 4.33 (t = 1.7, p < .10). The mean Ab based on the cued package panels for the four bad tar- get brands, 4.15, was significantly greater than the mean Ab that the same subjects later gave the corre- sponding ads for these brands, 3.37 (t = 5.4, p < .0 1). Not surprisingly, there was no significant difference (t < 1) between the mean Ab based on the cued package panels for the good and bad target ads. Thus, it appears that neither the panels nor the cue information alone predisposed subjects to make either favorable or un- favorable evaluations.

    Next, control subjects evaluated each of the 16 ads according to ad likability (1 = very unlikable, 7 = very likable) and brand goodness (1 = very bad, 7 very good). The ratings from the latter task were broken down into the four groups of intended valence and yielded the following means:

    GI B] G2 B2

    Brand goodness 4.33 3.37 4.61 4.30 Ad likability 3.96 3.18 4.19 3.99

    There were significant differences between the ratings for the brand (t = 5.5, p < .01) and the ad (t = 4.1, p < .01) for the eight target ads (i.e., the four G1 and four B 1 ads), as desired for the manipulation of target-ad va- lence. The control subjects rated the four B2 ads more positively, however, than did the original pretest sample

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

  • 470 JOURNAL OF CONSUMER RESEARCH

    that calibrated them. There are three implications of these higher ratings: (1) the interference condition with two bad ads (I- I -s) in the second ad combination was not achieved; (2) the interference condition with three competing ads (1-3) for both ad combinations had only one bad ad; and (3) both ad combinations had six good and three bad ads. Because of the first of these implications, tests of Hy- pothesis 2 for the bad target ad could not be conducted, and tests of Hypotheses 1, 3, 4, and 5 for the bad target ad could be based on only three competitive-ad-interfer- ence levels (1-0, I-1-d, and 1-3). However, the I-1-d con- dition for the bad target ad now has twice the number of observations as planned.

    Memory Effects of the Number and Valence of Competing

    Ads. Analysis of memory performance was con- ducted for (1) correct recall of the two main brand claims from the ad execution and (2) the total number of recalled cognitive responses reported from ad ex- posure (i.e., specific and overall ad and brand cognitive responses). An aggregate measure of cognitive re- sponses was used to maximize variance because no predictions were made for different types of responses. Hypothesis 1 stated that recall would be lower with more competing advertising in the product category. Hypothesis 2 stated that recall would be lower when the target brand was competing with an advertised brand of the same valence, compared with one of dif- ferent valence. Hypotheses 1 and 2 suggest the follow- ing order for recall of claims (from highest to lowest) for the four levels of interference when no ad retrieval cue is present: no competing ads (1-0), one competing ad of different valence (I-1-d), one competing ad of same valence (I- 1-s), three competing ads of differing valence (1-3). Recall of cognitive responses may not exhibit the same interference pattern, however, given that Keller (1987) found no difference in I-1-d and I- 3 for recall of specific ad reactions. A series of planned contrasts compared these four interference levels in the ad-cue-absent conditions to test Hypotheses 1 and 2.' Because these effects were not hypothesized to vary as a function of the valence of the target ad, the data were aggregated over this factor. The variables of product-category type, subjects' product-category in- volvement, memory-judgment task order, and target- ad order were included directly as covariates in the analysis. (Table 1 contains the means of the memory measures.)

    The analysis of recall of correct claims indicated that (1) recall in the 1-0 condition was significantly greater

    TABLE 1

    MEAN (SDs) OF MEMORY MEASURES

    Total Correct cognitive

    Sample claims responses Cell size

    Ad cue absent: Good target ad:

    1-0 1.47 (.62) .95 (.74) 38 1-1 -d 1.20 (.82) .89 (.61) 38 1-1-s .89 (.81) .84 (.59) 35 1-3 .80 (.83) .60 (.56) 35

    Bad target ad: 1-0 1.33 (.70) 1.17 (.63) 35 1-1-d 1.10 (.78) 1.00 (.72) 73 1-3 .65 (.75) .79 (.71) 38

    Ad cue present: Good target ad:

    I-0 1.46 (.65) 1.22 (.68) 38 1-1 -d 1.49 (.72) 1.28 (.70) 38 1-1 -s 1.25 (.69) 1.06 (.62) 36 1-3 1.21 (.64) 1.07 (.73) 36

    Bad target ad: I-0 1.53 (.63) 1.32 (.68) 36 1-1-d 1.27 (.60) 1.25 (.61) 74 1-3 1.29 (.60) 1.26 (.61) 38

    NOTE.-Correct claims are the number of the two main claims recalled. Total cognitive responses are the total number of specific and overall ad and brand reactions. 1-0 is no competitive ad condition. 1-1-s is one competitive ad of same valence condition. 1-1 -d is one competitive ad of different valence condition. 1-3 is three competitive ad condition.

    than in I-1-d (F = 6.5, p < .0 1), (2) recall in I-1-d was significantly greater (marginally) than in I- I-s (F = 3, p < .10), but (3) recall in I-I-s was not significantly greater than in 1-3 (F < 1). The analysis of recall of cognitive responses, however, indicated that: (1) recall in 1-0 was not significantly greater than in I-1-d (F < 1), (2) recall in I-1-d was not significantly greater than in I-I-s (F < 1), and (3) recall in I-I-s was not significantly greater than in 1-3 (F < 1). Recall of cog- nitive responses in 1-3 was lower than in 1-0 (F = 9.9, p

  • COMPETITIVE ADVERTISING ENVIRONMENTS 471

    ended self-report in the third booklet many subjects reported having seen fewer ads than they had actually been exposed to in the product category with three competing ads (X = 3. 1). Also, subjects reported claims that were from competing ads (i.e., confused claims) significantly more often with three competing ads compared with one competing ad (1-3 = .50, I-1 = .3 1; F = 4.8, p < .05).

    The valence of the competing ad also affected recall for claims, as predicted, but did not affect recall for cognitive responses. It appears that ads were not over- looked in this case as subjects reported, on average, that they had seen two ads in the product category for both interference conditions. Greater confusion was evident, however, with two evaluatively congruent ads, as indicated by analysis of subjects' self-reported ability to associate ads with brands. Subjects reported signif- icantly greater confusion (F = 9.0, p < .01) when the target ad was competing with an ad of the same valence (X = 3.4) than when it was competing with an ad of different valence (X = 2.9). Conversely, recall of cog- nitive responses exhibited weaker interference effects than recall of brand claims, consistent with prior re- search (see, e.g., Chattopadhyay and Alba 1988), which suggests that a person's evaluative responses to infor- mation are more durable and accessible in memory than the actual information itself.2

    Ad-Retrieval-Cue Interactions. Hypothesis 3 stated that an ad retrieval cue has a greater facilitative effect on recall in the presence of competing ads. A planned contrast compared the ad-cue-present and ad-cue-ab- sent conditions across all four levels of the competitive- ad-interference factor. The ad retrieval cue signifi- cantly increased recall of both claims (F = 10.5, p

  • 472 JOURNAL OF CONSUMER RESEARCH

    effects are expected to depend on the valence of the target ads, however, the analysis was done separately for good and bad ads. Results are reported only for the Ab measure, as tests of the purchase intention and Aad measures yield essentially the same results. (Table 2 contains the means of the evaluation measures.)

    Tests of Hypothesis 4 for a good target ad indicated the following. In accordance with Hypothesis 4, brand evaluations were higher for the 1-0 condition than for the 1-3 condition (Ab: F = 5.2, p < .0 1). Contrary to Hypothesis 4, however, brand evaluations were not significantly different for the 1-0 condition than for that of I-i -d (F < 1). Also contrary to Hypothesis 4, brand evaluations in the I- I-s condition were not sig- nificantly different from those in the I- 1-d or 1-3 con- ditions (F < 1). Tests of Hypothesis 4 for a bad target ad indicated the following. As hypothesized, evalua- tions for a bad target ad were lower in the 1-0 condition than in the 1-3 condition (Ab: F = 7.5, p < .0 1). Con- trary to Hypothesis 4, however, evaluations were not significantly different for the 1-0 condition than for the I- 1-d condition.

    The findings provide only mixed support for Hy- pothesis 4. Interference effects for brand evaluations were clearly present for both good and bad target ads with three competing ads, compared with no compet- ing ads, but were not present with one competing ad. Thus, the interference effects for brand evaluations closely resemble the interference effects for recall of cognitive responses.4

    TABLE 2

    MEANS (SDs) OF EVALUATION MEASURES

    Brand Purchase Ad Sample attitudes intentions attitudes Cell size

    Ad cue absent: Good target ad:

    1-0 4.72 (1.29) 5.26 (2.67) 4.09 (1.61) 38 1-1-d 4.47 (1.19) 4.87 (2.46) 4.03 (1.45) 38 1-1-s 4.25 (1.17) 4.17 (2.58) 3.76 (1.33) 35 1-3 4.05 (1.33) 4.03 (2.81) 3.51 (1.44) 35

    Bad target ad: I-0 3.46 (1.50) 2.83 (2.55) 2.82 (1.60) 35 1-1-d 3.82 (1.39) 3.89 (2.63) 3.55 (1.58) 73 1-3 4.28 (1.33) 4.42 (2.31) 3.82 (1.43) 38

    Ad cue present: Good target ad:

    1-0 4.78 (1.17) 5.61 (2.49) 4.59 (1.40) 38 1-1 -d 4.86 (.98) 5.70 (2.63) 4.26 (1.47) 38 1-1-s 4.58 (1.13) 4.86 (2.57) 3.96 (1.59) 36 1-3 4.63 (1.00) 5.25 (2.18) 4.12 (1.33) 36

    Bad target ad: 1-0 2.83 (1.44) 1.72 (2.88) 2.46 (1.53) 36 1-1-d 3.64 (1.48) 3.23 (2.71) 3.26 (1.81) 74 1-3 3.61 (1.39) 3.31 (2.54) 3.06 (1.64) 38

    NOTE.-I-0 is no competitive ad condition. 1-1 -s is one competitive ad of same valence condition. 1-1 -d is one competitive ad of different valence condition. 1-3 is three competitive ad condition.

    Ad-Retrieval-Cue Interactions. Hypothesis 5 stated that an ad retrieval cue has a greater effect on evalu- ations of a target ad in the presence of competing ads of different valence. First, planned contrasts compared

    taken was found in the analysis of correct claims (memory-evalu- ation = 1.26, evaluation-memory = 1.13; F = 4.3, p < .05). Task order did not have any effect on reported recall of cognitive responses but did interact with the ad-retrieval-cue factor, however, to affect ad and brand evaluations (p < .05): Higher evaluations were found when the memory measures were first when subjects were not cued and when the evaluation measures went first when subjects were cued. The order for the ads and corresponding panels for the target brands did not have any effect on the recall measures but did have a significant main effect on ad and brand evaluations (p < .01): Subjects gave higher evaluations when the ad and panel had a later position in the sequence. There were no significant interactions with ad order and the experimental factors.

    4Because evaluations were somewhat higher than expected when a good target ad was competing with a bad ad (I- 1 -d) and somewhat lower than expected when a good target ad was competing with another good ad (I-l-s), a follow-up study further examined the effects of evaluative congruency on evaluations for these two in- terference conditions. One possible explanation for the findings is that the competing ad may have created a contrast effect. For ex- ample, social judgment theory (Sherif, Taub, and Hovland 1958) would predict that a persuasive and likable target ad could be viewed more favorably when competing ads were not as good if the com- peting ads were used as a reference point or anchor. The procedure, stimuli, and measures to the follow-up study were identical to the main study, but the design was different. Only good target ads were used, and only two factors were manipulated in a fully crossed within-subject design. These were the competitive ad valence (good or bad) and the competitive ad order (proactive interference if the competing ad preceded the target ad and retroactive interference if the competing ad followed the target ad). It was thought that contrast effects might be particularly evident under proactive interference

    conditions because the competing ad should be more salient if eval- uated before the target ad. To provide further comparisons with the main study, ad retrieval cue was manipulated as a between- subject factor (presence or absence). The findings indicated that competitive ad valence affected recall as predicted (i.e., recall of correct claims was lower and incidence of confused claims was higher when the competing ad was good, compared with when it was bad). No main effects or interactions were evident, however, for evalu- ations. Also, as in the main study, the presence of the ad retrieval cue led to improved recall and higher evaluations. Thus, contrast effects were not evident in the follow-up study. As in the main study, even though subjects appeared to have more memory difficulties when the competing ad was of the same valence, compared with when it was different, the valence of the competing ad had no effect on evaluations of the target brand. Two explanations of this finding seem plausible. First, in both studies, cognitive responses for the target brand were equally accessible in the two interference con- ditions and may have been used as inputs to brand evaluations. Second, in evaluating the target brand, subjects were given the product category as a cue. Even if they could not recall the particular communication effects associated with the target brand, they could have accessed their overall product-category evaluations. Consistent with this explanation, subjects in the follow-up study did provide significantly higher overall product-category evaluations when the good target ad was competing with a good rather than a bad ad (I-I-s = 3.2, I-l-d = 2.9; F = 6.0, p < .01). The overall product- category evaluations in the main study are consistent with those in the follow-up study: Subjects in the main study gave significantly higher overall product evaluations (F = 6.9, p < .01) when the com- peting ad was good rather than bad (I-I-s = 3.4, I-l-d = 3).

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

  • COMPETITIVE ADVERTISING ENVIRONMENTS 473

    the ad-cue-present and ad-cue-absent conditions across all four levels of the interference factor. For the good target ad, the ad cue led to significantly higher evalu- ations (F = 4.6, p < .05). For the bad target ad, the ad cue led to significantly lower evaluations (F = 6.1, p < .05). Next, an interaction contrast compared the cue effects when there were no competing ads of different valence with those when there were competing ads of different valence. The interaction contrast for the good target ad, which compared the cue effects in the 1-0 and I-1-s conditions with those in the I-1-d and 1-3 conditions, was not significant. Thus, these effects did not vary by interference condition. The interaction contrast for the bad target ad, which compared the cue effects in 1-0 with those in I-1-d and 1-3, was also not significant (F < 1). Again, these effects also did not vary by interference condition.

    Hypothesis 5 is not supported because the ad re- trieval cue affected evaluations regardless of the pres- ence or absence of competing ads. Possible inter- pretations of this finding are offered in the general discussion below. The ad retrieval cue did have a sufficiently greater effect on brand evaluations under a high level of competitive advertising, so that cued evaluations for a good target ad did not significantly differ between the 1-0 and 1-3 conditions (F < 1). As with recall of cognitive responses, the presence of the ad retrieval cue fully offset the detrimental effect of competitive advertising on brand evaluations. How- ever, it was associated with fairly sizable negative ef- fects on brand evaluations for a bad target ad even when there were no competing ads. Consequently, cued evaluations for bad target ads were still lower in 1-0 than in 1-3 (F = 6.8, p < .0 1). As Isen (1989) has noted, positive and negative affect can have asym- metric effects on memory. Here, cued evaluations for a bad target ad were very low when there were no com- peting ads. Perhaps subjects found reminders of some- thing that they did not like to be particularly irritating.

    DISCUSSION

    Implications The goal of this research was to examine how quan-

    titative and qualitative aspects of competitive adver- tising and the presence of advertising retrieval cues affect consumer memory for communication effects (i.e., recall of claims and cognitive responses following ad exposure) and, thus, evaluations of an advertised brand. The findings and implications can be summa- rized as follows.

    Effects of Comnpetitive Advertising. This study rep- licated earlier findings that the more competing brands advertising in the product category, the lower recall of brand claims for a target ad is (Burke and Srull 1988; Keller 1987). These interference effects result from confusion with competing ads ("misretrieval") and

    failure or inability to retrieve. This study extended prior research by establishing several boundary con- ditions for the effects of competitive advertising. Compared with when there were no competing ads, a high level of competitive advertising (i.e., three com- peting ads) varying in valence reduced recall of cog- nitive responses and, most important, produced lower evaluations of a good target ad and higher evaluations of a bad target ad. This study also demonstrated that the valence of a competing ad can affect memory per- formance. With one competing ad, recall of brand claims was lower when the competing ad's valence was the same as that of the target ad rather than when it was different. Thus, qualitative aspects of competitive advertising, such as evaluative congruency, can in- crease associative interference and inhibit memory accessibility.

    The valence of a competing ad, however, did not affect recall of cognitive responses or evaluations of the target brand. More generally, brand evaluations exhibited a pattern virtually identical to that for recall of cognitive responses in this study: Interference effects were evident only when there were three competing ads varying in valence and were not evident when there was one competing ad. These memory and evaluation findings reinforce two important points made in prior research. First, the different types of communication effects encoded during ad exposure and stored in the ad memory trace vary in their sensitivity to contextual effects. Interference effects were more pronounced for recall of brand claims than for cognitive responses (see, e.g., Chattopadhyay and Alba 1988). Second, cognitive responses are important mediators of advertising ef- fects on brand and ad evaluations (see, e.g., Hastak and Olson 1989). Failure to recall brand claims from the ad did not affect evaluations of the target brand as long as cognitive responses from ad exposure were ac- cessible.

    The fact that interference effects were evident with recalled cognitive responses and evaluations suggests an important caveat to prior research. It has been ar- gued that, when overall or global evaluations are formed during encoding of stimulus information, these evaluations can be easily accessed to make subsequent judgments even if subjects have difficulty later recalling the specifics of that information (see Hastie and Park [1986] for a review). Yet, the findings of this study suggest that, when competing ads differing in valence are present, interference effects can occur such that prior evaluations themselves may be inaccessible for later judgments. Even though subjects evaluated the target ads during ad exposure, they evidently were un- able to access the relevant communication effects suc- cessfully when they had also seen three competing ads varying in valence. Interference effects for evaluations were not evident with one competing ad, however, be- cause cognitive responses were still accessible. More- over, when a good target ad was competing with an-

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

  • 474 JOURNAL OF CONSUMER RESEARCH

    other good ad, overall product category evaluations were also accessible and presumably could have been relevant for evaluations of the target brand.

    Effects of Ad Retrieval Cues. This study also rep- licated Keller's (1987) finding that the presence of an ad retrieval cue (i.e., a reduced version of the ad photo and headline) can lead to higher recall of brand claims and cognitive responses and more favorable evalua- tions of a good target ad. This study extended prior research by showing that ad retrieval cues can offset the detrimental effects of competitive ad interference: When subjects were given the ad retrieval cue, recall of cognitive responses for and evaluations of a good target ad with three competing ads did not differ from when there were no competing ads. Recall of correct claims was still lower with three competing ads, which suggests that, even if subjects accurately recall what they thought about an ad when given a cue, they still had some difficulty correctly recalling the associated claims.

    The findings also extended prior research by sug- gesting two other points about how ad retrieval cues work. First, although the ad retrieval cue had no effect on recall of brand claims when there were no com- peting ads in the product category, the cue did have some effect on recall of cognitive responses and ad and brand evaluations. This finding has several interpre- tations. It may suggest that ad retrieval cues are par- ticularly effective at eliciting reactions to ad exposure from memory, which is consistent with Tulving's en- coding specificity principle and suggests that the in- formation that served as the basis for cognitive re- sponses should be a good reminder of the specific nature of those cognitive responses. Besides affecting the ability to access communication effects from the ad memory trace, the ad retrieval cue may have mo- tivated more extensive retrieval than would have oc- curred with no competing ads. For example, if little retrieval effort were forthcoming, the ad retrieval cue may have stimulated deeper retrieval of those com- munication effects from the ad memory trace, such as cognitive responses, most closely linked to it. Thus, ad retrieval cues may affect both retrieval ability and motivation (Keller 199 la, 199 lb). Either way, the ad retrieval cue could have had an effect on recall of cog- nitive responses and evaluations of the brand when no competing ads were present and when the valence of the competing ad was the same as for the target ad, which is consistent with the study findings.5

    The second point is that ad retrieval cues resulted in higher brand evaluations when the target ad was

    good and lower brand evaluations when the target ad was bad. This effect was particularly meaningful at the highest level of competitive advertising. When there were three competing ads and subjects were not given the ad retrieval cue, evaluations for a good target ad were not significantly different from evaluations for a bad target ad (F < 1). When there were three competing ads and subjects were given the ad retrieval cue, eval- uations for a good target ad were significantly higher than evaluations for a bad target ad (F = 13.5, p < .01). The interaction between ad retrieval cues and target- ad valence on evaluations provides a stronger dem- onstration of ad retrieval cue effects than found in Keller (1987).

    Limitations and Future Research The topic of this research-memory and evaluations

    in competitive advertising environments-is clearly complex, and many important issues could not be re- solved or even addressed in this one study. Keller (1987) describes the limitations associated with a number of aspects of a closely related experimental methodology. Two specific limitations of the present study are highlighted here. First, ad valence was ma- nipulated in terms of both how good the advertised brand was and how good the ad itself was. Moreover, the brand-information-valence manipulation involved verbal stimuli (i.e., strong or weak arguments) and the ad-information-valence manipulation involved visual stimuli (i.e., photographs that reflected favorably or unfavorably on the advertised brand). Thus, it is un- clear whether the observed effects were driven by the visual ad-information manipulation, the verbal brand- information manipulation, or both. Second, ads were for unknown, hypothetical brands. Greater knowledge about the brand would imply greater ability to process ad information and could reduce interference effects in memory because stronger associative links could be formed (but see Srull 1983).

    Besides addressing these limitations, future research should explore several general areas that deserve ad- ditional study. Although advertising distinctiveness or differentiation has been stressed as an important man- agerial objective, it has received surprisingly little re- search attention. Ad distinctiveness should depend on the organization of memory according to the dimen- sions used by consumers to categorize or interpret ads. In other words, ad distinctiveness depends on the communication effects that make up the ad memory trace (i.e., consumers' ad representations and ad re- sponses stored during ad exposures) and the location of the ad memory trace in memory. This research ex- amined distinctiveness according to the number and valence of competing advertised brands. Other aspects of ad distinctiveness should also be explored, such as the positioning expressed by competing brands and the distinctiveness of the ad execution itself in terms of style, structure, or modality.

    5It should be noted that subjects may have had additional cog- nitive responses while retrieving the ad representation from the ad memory trace and responding to the cognitive-response probe. With retrospective protocols, even with instructions to subjects to list only responses that occurred during ad exposure, it may be that additional reactions produced during retrieval are reported (see Keller 1987, pp. 326-327).

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

  • COMPETITIVE ADVERTISING ENVIRONMENTS 475

    The effects of competitive advertising during en- coding should also be explored. Are comparisons ever made to other ads during processing? In other words, to what extent do consumers interpret or evaluate an ad in isolation? Two issues are particularly important here. First, if explicit or implicit comparisons are made during ad exposure, what possible standards are used? Second, given that comparisons are made, what effects do they have on evaluations? For example, compara- tive processing with competitive ads may be distracting and may inhibit encoding processes during ad exposure because, as competitive advertising increases, con- sumers may try to distinguish competing ads and brands in memory. In attempting to form separate brand-based memory traces, consumers may have less processing capacity available such that fewer cognitive responses or brand claims related to the target ad are encoded. Understanding factors that predict which frame of reference, if any, consumers invoke and what effects those frames have is an important research priority. Conceptually, such work may build on the model of ad attitude formation suggested by Mac- Kenzie and Lutz (1989). Methodologically, concurrent verbal protocols during ad processing may be neces- sary.

    Finally, a clearly important future research priority is a more complete understanding of the effects of re- trieval cues and strategies on memory and evaluations. In an early application of information-processing the- ory, Tybout, Calder, and Sternthal (1981) showed how retrieval cues could inhibit recall of negative associa- tions and counter the potentially adverse effects of a rumor on brand evaluations. Similarly, this research has shown how retrieval cues can facilitate recall of positive associations to counter the potentially inhib- iting effects of competitive advertising on brand eval- uations. As Biehal and Chakravarti (1983) note, future research should explore how consumers respond to re- trieval difficulties and how different retrieval strategies affect judgment and choice outcomes. The relevance- accessibility model (RAM) put forth by Baker and Lutz ( 1987) makes a number of interesting predictions con- cerning the interactive effects of encoding and retrieval factors on advertising effectiveness. In accordance with that research, this study suggests that the effects of cues on retrieval motivation and ability deserves careful study. Because of the nature of the experimental task in this study, decision involvement and, thus, retrieval motivation were probably high. The motivational property of ad retrieval cues in low-involvement de- cision settings should be examined.

    [Received September 1988. Revised August 1990.]

    REFERENCES Anderson, John R. (1983), The A rchitectucre of Cognition,

    Cambridge, MA: Harvard University Press.

    Areni, Charles S. and Richard J. Lutz (1988), "The Role of Argument Quality in the Elaboration Likelihood Model," in Advances in Consumer Research, Vol. 15, ed. Michael J. Houston, Provo, UT: Association for Consumer Research, 197-20 1.

    Baker, William E. and Richard J. Lutz (1987), "The Rele- vance-Accessibility Model of Advertising Effectiveness," in Nonverbal Communications in Advertising, ed. Syd- ney Hecker and David W. Stewart, Lexington, MA: Lexington, 59-84.

    Batra, Rajeev and Michael L. Ray (1986), "Situational Ef- fects of Advertising Repetition: The Moderating Influ- ence of Motivation, Ability, and Opportunity to Re- spond," Journal of Consumer Research, 12 (March), 432-445.

    Baumgardner, Michael H., Michael R. Leippe, David L. Ronis, and Anthony G. Greenwald (1983), "In Search of Reliable Persuasion Effects. II. Associative Interfer- ence and Persistence of Persuasion in a Message-Dense Environment," Journal of Personality and Social Psy- chology, 45 (3), 524-537.

    Biehal, Gabriel and Dipankar Chakravarti (1983), "Infor- mation Accessibility and Consumer Choice," Journal of Consumer Research, 10 (June), 1- 14.

    Burke, Raymond R. and Thomas K. Srull (1988), "Com- petitive Interference and Consumer Memory for Ad- vertising," Journal of Consumer Research, 15 (June), 55-68.

    Chattopadhyay, Amitava and Joseph W. Alba (1988), "The Situational Importance of Recall and Inference in Con- sumer Decision Making," Journal of Consumer Re- search, 15 (June), 1- 12.

    Craik, Fergus I. M. (1979), "Human Memory," Annual Re- view of Psychology, 30, 63-102.

    Dempster, Frank N. (1985), "Proactive Interference in Sen- tence Recall: Topic Similarity Effects and Individual Differences," Memory and Cognition, 13 (1), 81-89.

    Eysenck, Michael W. (1979), "Depth, Elaboration, and Dis- tinctiveness," in Levels of Processing in Human Mem- ory, ed. Laird S. Cermak and Fergus I. M. Craik, Hills- dale, NJ: Erlbaum, 89-118.

    Feldman, Jack M. and John G. Lynch, Jr. (1988), "Self- generated Validity and Other Effects of Measurement on Belief, Attitude, Intention, and Behavior," Journal of Applied Psychology, 73 (August), 421-435.

    Glenberg, Arthur M. (1984), "A Retrieval Account of the Long-Term Modality Effect," Journal of Experimental Psychology: Learning, Memory, and Cognition, 10 (1), 16-31.

    Hastak, Manoj and Jerry C. Olson (1989), "Assessing the Role of Brand-Related Cognitive Responses as Media- tors of Communication Effects on Cognitive Structure," Journal of Consumer Research, 15 (March), 444-456.

    Hastie, Reid and Bernadette Park (1986), "The Relationship between Memory and Judgment Depends on Whether the Judgment Task Is Memory-based or On-Line," Psy- chological Review, 93 (June), 258-268.

    Hutchinson, J. Wesley and Daniel L. Moore (1984), "Issues Surrounding the Examination of Delay Effects of Ad- vertising," in Advances in Consumer Research, Vol. 11, ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 650-655.

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

  • 476 JOURNAL OF CONSUMER RESEARCH

    Isen, Alice M. (1989), "Some Ways in Which Affect Influ- ences Cognitive Processes: Implications for Advertising and Consumer Behavior," in Advertising and Consumer Psychology, ed. Alice M. Tybout and Patricia Cafferata, New York: Lexington, 91-117.

    Jacoby, Larry and Fergus I. M. Craik (1979), "Effects of Elaboration of Processing at Encoding and Retrieval: Trace Distinctiveness and Recovery of Initial Context," in Levels of Processing in Human Memory, ed. Laird S. Cermak and Fergus I. M. Craik, Hillsdale, NJ: Erl- baum, 1-21.

    Keller, Kevin Lane (1987), "Memory Factors in Advertising: The Effect of Advertising Retrieval Cues on Brand Evaluations," Journal of Consumer Research, 14 (De- cember), 316-333.

    (1991 a), "Memory Retrieval Factors and Advertising Effectiveness," in Advertising Exposure, Memory, and Choice, ed. Andrew A. Mitchell, Hillsdale, NJ: Erlbaum, in press.

    (1991b), "Cue Compatibility and Framing in Ad- vertising," Journal of Marketing Research, 28 (Febru- ary), in press.

    Kisielius, Jolita and Brian Sternthal (1986), "Examining the Vividness Controversy: An Availability-Valence Inter- pretation," Journal of Consumer Research, 12 (March), 418-431.

    MacKenzie, Scott B. (1986), "The Role of Attention in Me- diating the Effect of Attribute Importance," Journal of Consumer Research, 13 (September), 174-195.

    and Richard J. Lutz (1989), "An Empirical Exami- nation of the Structural Antecedents of Attitude toward the Ad in an Advertising Pretesting Context," Journal of Marketing, 53 (April), 48-65.

    Mitchell, Andrew A. (1986), "The Effect of Verbal and Visual Components of Advertisements on Brand Attitudes and Attitude Toward the Advertisement," Jolurnal of Con- sumer Research, 13 (June), 12-24.

    Morris, C. Donald, John D. Bransford, and Jeffrey J. Franks (1977), "Levels of Processing versus Transfer Appro- priate Processing," Jolurnal of Verbal Learning and Verbal Behavior, 16 (5), 519-533.

    Moscovitch, M. and Fergus I. M. Craik (1976), "Depth of Processing, Retrieval Cues, and Uniqueness of Encoding as Factors in Recall," Journal of Verbal Learning and Verbal Behavior, 15 (4), 447-458.

    Postman, Leo and Benton J. Underwood (1973), "Critical Issues in Interference Theory," Memory and Cognition, 1, 19-40.

    Reyes, Robert M., William C. Thompson, and Gordon H. Bower (1980), "Judgmental Biases Resulting from Dif- fering Availabilities of Arguments," Journal of Person- ality and Social Psychology, 39 (July), 2-12.

    Sherif, Muzafer, Daniel Taub, and Carl I. Hovland (1958), "Assimilation and Contrast Effects of Anchoring Stim- uli," Journal of Experimental Psychology, 55 (2), 150- 155.

    Srull, Thomas K. (1983), "The Role of Prior Knowledge in the Acquisition, Retention, and Use of New Informa- tion," in Advances in Consumer Research, Vol. 10, ed. Richard P. Bagozzi and Alice M. Tybout, Ann Arbor, MI: Association for Consumer Research, 572-576.

    Tulving, Endel (1974), "Cue-dependent Forgetting," Amer- ican Scientist, 62, 74-82.

    (1979), "Relation between Encoding Specificity and Levels of Processing," in Levels of Processing in Human Memory, ed. Laird S. Cermak and Fergus I. M. Craik, Hillsdale, NJ: Erlbaum, 405-427.

    Tybout, Alice M., Bobby J. Calder, and Brian Sternthal (1981), "Using Information Processing Theory to De- sign Marketing Strategies," Journal of Marketing Re- search, 18 (February), 73-79.

    Underwood, Benton A. (1983a), "Conceptual Similarity and Cumulative Proactive Inhibition," Journal of Experi- mental Psychology. Learning, Memory, and Cognition, 9 (3), 456-461.

    (1983b), "Interference in Memory Produced by the Acoustic Attribute," American Journal of Psychology, 96(1), 113-125.

    Wyer, Robert S., Jr. and Thomas K. Srull (1986), "Human Cognition in Its Social Context," Psychological Review, 93 (July), 322-359.

    This content downloaded from 202.78.175.199 on Tue, 17 Feb 2015 14:31:04 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsp

    Article Contentsp. 463p. 464p. 465p. 466p. [467]p. 468p. 469p. 470p. 471p. 472p. 473p. 474p. 475p. 476

    Issue Table of ContentsJournal of Consumer Research, Vol. 17, No. 4 (Mar., 1991), pp. 375-531Volume Information [pp. 523 - 531]Front MatterA Multistage Model of Customers' Assessments of Service Quality and Value [pp. 375 - 384]Determinants of Household Expenditures for Services [pp. 385 - 397]Artifacts, Identity, and Transition: Favorite Possessions of Indians and Indian Immigrants to the United States [pp. 398 - 411]Selves in Transition: Symbolic Consumption in Personal Rites of Passage and Identity Reconstruction [pp. 412 - 425]Multitrait-Multimethod Matrices in Consumer Research [pp. 426 - 439]Consumer Responses to Advertising: The Effects of Ad Content, Emotions, and Attitude toward the Ad on Viewing Time [pp. 440 - 453]Effects of Word-of-Mouth and Product-Attribute Information on Persuasion: An Accessibility-Diagnosticity Perspective [pp. 454 - 462]Memory and Evaluation Effects in Competitive Advertising Environments [pp. 463 - 476]Inference Effects without Inference Making? Effects of Missing Information on Discounting and Use of Presented Information [pp. 477 - 491]Time-Inconsistent Preferences and Consumer Self-Control [pp. 492 - 507]Consumer Multiattribute Judgments under Attribute-Weight Uncertainty [pp. 508 - 522]Back Matter