Accessibility Effects on the Relationship Between Attitude Toward the Ad and Brand Choice

Cynthia B. Hanson, Greensboro College
Gabriel J. Biehal, University of Maryland
ABSTRACT - This study examines direct and indirect effects of attitude toward the ad (Aad) on brand choice, and the impact of brand information accessibility on the role of Aad in brand choice. The results show no direct effect of Aad on brand choice: the effect of Aad was completely mediated by brand attitude. Consistent with theoretical predictions, ad effects were more dramatic when brand information was less accessible. In fact, when brand information was highly accessible, Aad affected brand attitude but not choice outcomes.
[ to cite ]:
Cynthia B. Hanson and Gabriel J. Biehal (1995) ,"Accessibility Effects on the Relationship Between Attitude Toward the Ad and Brand Choice", in NA - Advances in Consumer Research Volume 22, eds. Frank R. Kardes and Mita Sujan, Provo, UT : Association for Consumer Research, Pages: 152-158.

Advances in Consumer Research Volume 22, 1995      Pages 152-158

ACCESSIBILITY EFFECTS ON THE RELATIONSHIP BETWEEN ATTITUDE TOWARD THE AD AND BRAND CHOICE

Cynthia B. Hanson, Greensboro College

Gabriel J. Biehal, University of Maryland

ABSTRACT -

This study examines direct and indirect effects of attitude toward the ad (Aad) on brand choice, and the impact of brand information accessibility on the role of Aad in brand choice. The results show no direct effect of Aad on brand choice: the effect of Aad was completely mediated by brand attitude. Consistent with theoretical predictions, ad effects were more dramatic when brand information was less accessible. In fact, when brand information was highly accessible, Aad affected brand attitude but not choice outcomes.

INTRODUCTION

Over a decade of research supports the importance of a consumer's attitude toward an ad (Aad) in the formation of brand attitude and purchase intention. Brown and Stayman (1992) summarized the results of much of this research in their meta-analysis of Aad studies. Their analysis provides a comprehensive comparison of various models which have been proposed in the literature and identifies several moderating factors. It also serves to illuminate some important gaps in the Aad research both in terms of modeling the effects of Aad and identifying moderating factors. Specifically, most studies have used brand attitude or purchase intention as the outcome variable and, consequently, the effect of Aad on choice has not been fully explored. The distinction between brand attitude, purchase intention and choice is important when studying Aad, as Aad may have an effect on brand attitude without affecting choice behavior (Chattopadhyay and Nedungadi 1990); or, Aad may have a direct effect on choice in addition to the indirect effect through brand attitude (Biehal, Stephens, and Curlo 1992). Furthermore, although the consumer behavior literature shows that memory-based choice processes are very different from externally-based choice processes (Biehal and Chakravarti 1982), very few studies have looked at the impact of differences in brand attribute information accessibility on the role of Aad. Therefore, the purpose of this study is twofold. First, it investigates the direct and indirect effects of Aad on choice, attempting to replicate previous research findings. Second, it extends previous work by examining the impact of brand information accessibility on the role of Aad in brand choice.

CONCEPTUAL BACKGROUND

Attitude Toward the Ad

The most widely supported model of the role of Aad is the "Dual Mediation Hypothesis," first tested by MacKenzie, Lutz, and Belch (1986). This model proposes that both brand cognitions and attitude toward the ad directly influence brand attitude. In addition, Aad is shown as having an indirect effect, through brand cognitions, on brand attitude. This model has received extensive support in the literature (MacKenzie et al. 1986, Homer 1990, Brown and Stayman 1992).

However, using purchase intention rather than choice as the final outcome measure leaves open two possibilities. First, it is possible that Aad effects on brand attitude and purchase intention do not carry through to choice. The fact that attitudinal indicators, such as brand attitude and purchase intentions, do not always predict choice is well-documented in the consumer behavior and social psychology literature (Smith and Swinyard 1983, Fazio, Chen, McDonel, and Sherman 1982). Attitudes may not be reflected in choice for a variety of reasons, including situational factors, constraints, and lack of product involvement. Research by Fazio et al. (1982) suggests that attitudes which are not formed through direct experience may be more weakly associated with the object, less accessible in memory, and thus less likely to carry through to choice. Furthermore, attitude changes may not be reflected in choice behavior if they are not large enough to affect relative preferences among brands in the choice set. Research by Chattopadhyay and Nedungadi (1990) and Miniard, Sirdeshmukh, and Innis (1992) illustrates that advertising manipulations which significantly affect brand attitudes may not affect choice. Since most advertising managers are ultimately concerned with how their advertising affects choice behavior, it is important to measure choice in addition to attitudinal effects of advertising.

The second reason for extending the Aad models to choice concerns the direct versus mediated effects of Aad on choice. Baker and Lutz (1988) suggest that when two brands are similar, Aad may become a "tie-breaker." This implies that Aad has a direct effect on choice, not mediated by brand attitude. Although previous analyses have found no direct path from Aad to purchase intention, Biehal et al. (1992), in a study which included two very similar brands, did find a significant direct effect of Aad on choice. Therefore, the first research question to be addressed in this study is: does Aad have a direct or mediated effect on brand choice?

Accessibility as a Moderator of Aad Effects

Much of the research on the role of Aad has focused on identifying factors affecting the strength of the various model relationships, particularly the relative influence of brand beliefs and Aad (MacKenzie and Lutz 1989, Brown and Stayman 1992). One factor which has not received much attention, in spite of theoretical and empirical research showing its importance in affecting choice, is brand information accessibility. Biehal and Chakravarti (1982, 1986) showed that choice processing when all information is externally available differs in several respects from choice processing when some information is in memory. Chattopadhyay and Nedungadi (1992) did investigate the effect of accessibility on attitude toward the ad. However, their manipulation weakened the accessibility of ad cognitions while brand cognitions remained stable, resulting in a decline in the impact of Aad over time. In contrast, this study investigates the role of Aad when brand attribute information is less accessible.

Several theoretical frameworks of consumer judgment and choice highlight the importance of an input's accessibility in determining its impact on consumer judgment and choice (Baker and Lutz 1988, Feldman and Lynch 1988, Kisielius and Sternthal 1984). The Relevance-Accessibility Model (Baker and Lutz 1988) is particularly useful for examining the effect of accessibility on Aad, since it is specifically designed as a model of advertising effectiveness. According to the Relevance-Accessibility Model, given a sufficient level of involvement, the primary determinant of a consumer's brand response will be his or her cognitive structure, which is comprised of brand beliefs. However, if the brand attribute information is less accessible in memory, the importance of (stored) evaluative reactions to the ad increases. This implies that Aad will have a greater impact on brand choice when brand attribute information is less accessible in memory. Therefore, the second research question addressed in this research is: does the role of Aad in choice depend upon the accessibility of brand attribute information?

METHOD

The findings reported in this paper are part of a larger study with additional conditions. Only the manipulations and procedures for the four conditions reported in this paper are described below.

Experimental Design

The experiment included two levels of brand attribute accessibility (high accessibility, i.e., information externally available during choice; and, low accessibility, i.e., information on some brands available only in memory) and two levels of ad picture (good and bad) for a target brand. Therefore, the study was a 2 X 2 between-subjects design.

Stimuli

Print ads for eight hypothetical brands of running shoes, labeled A through H, were constructed. Ads contained a headline at the top, a four-color photograph (approximately 6 inches square), and a short paragraph of ad copy at the bottom. The copy described the brands on four attributes, determined through pretesting to be salient features of running shoes: cushioning, durability, support, and color/style. Each attribute could have one of three performance levels. Exhibit 1 shows the brands and their features. The order in which features were presented in the ad copy was changed for each brand, to encourage subjects to read the entire ad copy. Brands were defined so that brands B (the target brand) and D would be the most likely choices. These two brands performed very well on cushioning and color/style, but subjects would need to trade-off B's superior durability against D's higher comfort level. Brands E through H were designed to perform worse than at least one of the brands A through D, so that in the high accessibility condition they were unlikely to be chosen.

Manipulations

Accessibility. One factor which has been shown to affect subsequent accessibility of brand information in memory is the amount of cognitive elaboration it receives at the time of encoding. The amount of cognitive elaboration, and thus the accessibility of brand attribute information, can be reduced by limiting the amount of time a consumer is given to examine an ad (Kisielius and Sternthal 1984). Therefore, we manipulated accessibility using a speeded learning task designed to represent a situation in which some brand attribute information is encoded fleetingly, without being evaluated or integrated to form attitudes. Pretesting showed that with thirty seconds to learn each ad, subjects were able to encode its information, but it was not very accessible. Further, self-report measures indicated subjects did not form overall brand evaluations.

Ad Picture. The quality of the ads was manipulated by varying ad pictures and headlines. The target brand had either a good ad or a bad ad (between subjects), and all other brands' ads were average. One good, one bad, and seven average ads were selected based on pretesting. The bad ad contained a picture of a man and a woman running. The woman was dressed as a leprechaun and the headline was "Go for the Gold." (This ad is reproduced in Biehal et al. 1992.) The good ad contained a picture of a lone man running on a rural road. The headline was "There is No Finish Line." All ad pictures were taken from magazines, either from old running shoe ads or from photographs of running events. Most headlines were written by the experimenter, although some were taken from the magazine picture captions. References to brand names were removed. Pretesting showed that subjects did not associate any of our fabricated ads with real brands.

Procedure

Figure 1 provides a flowchart of the experimental procedure. Subjects were given a package of eight ads, identified by the letters A-H. Subjects indicated their attitudes toward the eight ads. The ad copy at the bottom of the eight ads was replaced by x's during the ad rating task, to keep subjects from forming brand attitudes or engaging in choice processing. [Since the text of all ads was almost identical (only the attribute performance and statement sequence was varied) it should not have affected relative ad attitudes among brands. Therefore, although the Aad measure was based on the picture and headline, this should not have affected the results, and was felt to be necessary given the study design.] Half the subjects saw a good ad for brand B (the target brand) and half saw a bad ad. After the ad rating task, the subjects returned the ads to the package and closed it. Then they read some background information on running shoes, explaining the four attributes used to describe the running shoes (cushioning, support, style/color, and durability).

The procedure then differed depending on the subject's accessibility condition. Subjects in the high accessibility condition were given a package of eight ads, which now contained brand attribute information on four features, in the form of ad copy, at the bottom of the ad. They then chose the brand they preferred. The eight ads were externally available, i.e., in front of the subjects, when they made their choices.

Subjects in the low accessibility condition were given a limited amount of time to learn the brand attribute information contained in the ads for the first four brands, A-D. After learning the ads the subjects returned them to the envelope and closed it. Then, low accessibility subjects were given recall and recognition tests to check the manipulation. After this, the low accessibility subjects received the ads for the second set of four brands and were asked to choose their most preferred brand from all eight brands, i.e., the memory brands, A-D, and the four external brands, E-H. Thus, low accessibility subjects had a mixed choice task.

As Figure 1 shows, after subjects made their choices, the procedure was the same for both groups. Subjects first returned all the ads to the envelope, and then indicated their brand attitudes toward their chosen brand and toward the target brand, B. Finally, subjects completed a questionnaire designed to measure task reactions and subject characteristics.

Measures

Brand choice was operationalized by having subjects circle the brand label (A through H) of their chosen brand. For purposes of this study, choice was dichotomized into chose/did not choose brand B. Attitude toward the ad was derived from an average of subjects' ratings on four 10-point scales: bad/good, dislike/like, uncreative/creative, boring/interesting. Brand attitude was derived from an average of subjects' ratings on four 7-point scales: bad/good, dislike/like, unpleasant/pleasant, poor quality/good quality. These scales were based on a review of existing research. Coefficient alpha was .94 for Aad and .95 for brand attitude, indicating acceptable levels of reliability for these scales (Nunnally 1978).

Subjects

The study involved 97 undergraduate students from the business subject pool at a large state university. Their ages ranged from 19-40 years, with a mean of 22, and 56% were female. The subjects received extra-credit in exchange for their participation.

EXHIBIT 1

BRAND FEATURES

RESULTS

Manipulation Checks

Attitude Toward the Ad. Attitude toward ad B was significantly higher when ad B had the good picture versus the bad picture (7.44 versus 5.05, p<.001). A 2 X 2 ANOVA of attitude toward ad B with ad picture and accessibility showed only the ad picture main effect: accessibility had no main (p>.50) or interaction (p>.40) effect on Aad (see Exhibit 2). Attitude toward the other ads ranged from 5.22 to 7.29, with a mean of 6.19. Most importantly, Aad for brand D, the other likely choice candidate, was 5.99, which is significantly lower than the good brand B ad (p<.001), and significantly higher than the bad brand B ad (p<.001).

A series of regression equations (Baron and Kenny 1986) showed that Aad completely mediated the effect of the ad picture manipulation on brand attitude: 1) Ad picture significantly affected Aad (beta=.46, p<.01), 2) Ad picture significantly affected brand attitude (beta=.21, p<.10), and 3) The effect of ad picture on brand attitude disappeared (p >.70) when Aad was included as a predictor of brand attitude.

FIGURE 1

EXPERIMENTAL PROCEDURE

EXHIBIT 2

SUMMARY RESULTS

Accessibility. First, recall was measured by giving low accessibility subjects four minutes to write down everything they could remember about each memory brand's features. Then, recognition was measured by presenting subjects with two descriptions of each memory brand's features (the text at the bottom of the ad), one correct and one incorrect, and having them indicate the correct one. The incorrect description was altered by increasing the level of one attribute and decreasing the level of another attribute. Ad picture recall was checked by having subjects write a brief description of each ad's picture.

Although 76% of the subjects recognized the correct description of brand B, none could correctly recall all four attributes. The average number of attributes correctly recalled was 1.82. The difference between recognition and recall scores indicates that the target brand information was available but not accessible (Thomson and Tulving 1970). The fact that 87% of the subjects were able to correctly describe the ad picture in an uncued recall task suggests that the ad picture was more accessible than the brand attribute information.

Overview of Experimental Outcomes

Exhibit 2 contains summary results for the experiment. Both the brand attitude and choice results indicate that the ad picture had more of an effect when brand attribute information for the target brand was not externally available. A 2 X 2 ANOVA of brand attitude showed a main effect of ad picture (p<.001), a main effect of accessibility (p<.01) and a marginally significant accessibility by ad picture interaction (p=.10), indicating that the ad picture manipulation had a greater effect on brand attitude in the low accessibility condition. Similarly, a logistic regression of brand B choice showed a significant effect of accessibility (p<.01), a significant effect of ad picture (p<.01) and a significant ad picture by accessibility interaction (p<.05). When accessibility was high, about 60% of subjects chose target brand B regardless of ad picture treatment. However, when brand attribute information was less accessible, 47.6% chose brand B when it had the good ad picture and 4.5% chose brand B when it had the bad ad picture. The following analyses investigate more precisely the role of Aad in these outcomes.

Direct Versus Mediated Effects of Aad

The first research question was: does Aad have a direct or mediated effect on brand choice? In order to address this question, brand attitude was regressed on Aad, brand B choice was regressed on Aad, and brand B choice was regressed on Aad and the mediator (brand attitude) simultaneously (Baron and Kenny 1986). [Logistic regression was used to analyze the choice equations in this section.] The results showed that although Aad significantly affected brand attitude (beta=.42, p<.001), and Aad significantly affected choice (beta=.54, p<.05), Aad did not effect choice when the mediator, brand attitude, was included (beta=-.06, p>.80). Thus, the effect of Aad on choice was compeletely mediated by brand attitude.

Choice Outcomes

The second research question was: does the role of Aad in choice depend on the accessibility of brand attribute information? To answer this question we first consider the correlation coefficients for each accessibility group (Exhibit 3). It is clear that brand attitude (Ab), Aad, and target brand choice are all significantly related to one another, with one exception: the Aad-chose B correlation (r=.14) is not significant for high accessibility subjects. Thus, when subjects made their brand choice with all ad information externally available, Aad affected brand attitude but not choice (cf. Chattopadhyay and Nedungadi 1990). To see if the interaction between Aad and accessibility is significant, we ran an OLS regression model with target brand choice as the criterion, and brand attitude, Aad, accessibility, and the two-way interaction between accessibility and brand attitude and Aad as predictors (Exhibit 4). The overall model was significant (F(5,91)=16.1, p<.0001). The accessibility by brand attitude interaction was also significant (beta=-.65, p<.05): its negative sign reflects the fact that as accessibility decreases, so does the correlation between brand attitude and choice. The accessibility by Aad interaction is marginally significant (beta=.44, p<.10): the relationship between Aad and choice is stronger when brand attribute information is less accessible. [Use of logistic regression increases the significance level of the accessibility by Aad interaction to p < .14.]

EXHIBIT 3

CORRELATION MATRICES

EXHIBIT 4

RESULTS OF CHOICE REGRESSION MODEL

DISCUSSION

The following presents a summary and brief discussion of the key findings of our research on accessibility and Aad.

1. Under high brand information accessibility, Aad affected brand attitude but not choice outcomes. This finding underscores the importance of measuring choice as an outcome variable. Chattopadhyay and Nedungadi (1990) present a similar outcome (Aad affecting attitude but not choice) under different circumstances. As discussed previously, attitudinal indicators are not always reflected in choice, for a variety of reasons.

2. Aad had no direct effects on choice, under high or low accessibility. This finding contrasts with and represents a failure to replicate Biehal et al. (1992). One possible explanation for the different finding is that the current research involved eight hypothetical brands instead of four. With only four brands in the Biehal et al. (1992) study, the ad seemed to work as a tie-breaker, tipping the probability of choice for the target brand when it had the good ad. However, with eight brands in the current study, the bad ad may have drawn attention to the target brand, which was a logical choice based on features. Thus, the attention factor may have outweighed the bad ad factor.

3. Though the effect was not strong, Aad had more of an effect on choice under low accessibility, when subjects had to try to recall ad and brand information in order to make their choice. This is consistent with several theoretical frameworks which predict that the likelihood that ad cognitions and ad attitude will be used in judgment and choice increases when the accessibility of alternative inputs, e.g., brand-related information, decreases (Feldman and Lynch 1988, Baker and Lutz 1988). Our findings can be contrasted with those of Chattopadhyay and Nedungadi (1992), who found that the effects of ad attitude dissipated after a delay. The key seems to be the relative accessibility of ad versus brand information. In our research, ad information (i.e., ad picture) was more accessible than brand information, as may be expected when the brand information is fairly complex and there is not much variance among brands. Therefore, the ad information had a significant effect on brand attitudes and choice.

In summary, it has been well-established through consumer research that externally-based choice processes are very different from memory-based choice processes. The research presented here shows that accessibility also affects the role of Aad in choice. Ad effects on choice were more dramatic when brand information was less accessible. In fact, in our research, ad attitude affected brand attitude but had no effect on choice outcomes when the subjects had all information in front of them when making their choices. Given the existence of a number of situations (e.g., catalogs, Yellow Pages) where choices are made based on externally available ad information, this finding deserves further attention.

REFERENCES

Baker, William E. and Richard J. Lutz (1988), "The Relevance-Accessibility Model of Advertising Effectiveness," in Nonverbal Communication in Advertising, Sidney Hecker and David W. Stewart, eds., Lexington, MA: Lexington Books, 59-84.

Baron, Reuben M. and David A Kenny (1986), "The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic and Statistical Considerations," Journal of Personality and Social Psychology, 51 (6), 1173-82.

Biehal, Gabriel, Debra Stephens, and Eleonora Curlo (1992), "Attitude Toward the Ad and Brand Choice," Journal of Advertising, 21 (3), 19-36.

Biehal, Gabriel and Dipankar Chakravarti (1986), "Consumer's Use of Memory and External Information in Choice: Macro and Micro Perspectives," Journal of Consumer Research, 12 (March), 382-405.

Biehal, Gabriel and Dipankar Chakravarti (1982), "Information-Presentation Format and Learning Goals as Determinants of Consumers' Memory Retrieval and Choice Processes," Journal of Consumer Research, 8 (March), 431-41.

Brown, Stephen P. and Douglas M. Stayman (1992), "Antecedents and Consequences of Attitude Toward the Ad: A Meta-analysis," Journal of Consumer Research, 19 (June), 34-51.

Chattopadhyay, Amitava and Prakash Nedungadi (1990), "Ad Affect, Brand Attitude and Choice: The Moderating Roles of Delay and Involvement," in Advances in Consumer Research Volume 17, Marvin E. Goldberg, Gerald Gorn and Richard W. Pollay, eds., Provo, UT: Association for Consumer Research, 619-620.

Chattopadhyay, Amitava and Prakash Nedungadi (1992), "Does Attitude Toward the Ad Endure? The Moderating Effects of Attention and Delay," Journal of Consumer Research, 19 (June), 26-33.

Fazio, Russell, Jeaw-Mei Chen, Elizabeth McDonel, and Steven Sherman (1982), "Attitude Accessibility, Attitude-Behavior Consistency, and the Strength of the Object-Evaluation Association," Journal of Experimental Social Psychology, 18, 339-57.

Feldman, Jack M. and John G. Lynch (1988), "Self-Generated Validity and Other Effects of Measurement on Belief, Attitude, Intention, and Behavior," Journal of Applied Psychology, 73 (3), 421-35.

Homer, Pamela M. (1990), "The Mediating Role of Attitude Toward the Ad: Some Additional Evidence," Journal of Marketing Research, 27 (February), 78-86.

Kisielius, Jolita and Brian Sternthal (1984), "Detecting and Explaining Vividness Effects in Attitudinal Judgments," Journal of Marketing Research, 21 (February), 56-64.

MacKenzie, Scott B. and Richard J. Lutz (1989), "An Empirical Examination of the Structural Antecedents of Attitude Toward the Ad in an Advertising Pretest Context," Journal of Marketing, 53 (April), 48-65.

MacKenzie, Scott B., Richard J. Lutz, and George E. Belch (1986), "The Role of Attitude Toward the Ad as a Mediator of Advertising Effectiveness: A Test of Competing Explanations," Journal of Marketing Research, 23 (May), 130-43.

Miniard, Paul W., Deepak Sirdeshmukh, and Daniel E. Innis (1992), "Peripheral Persuasion and Brand Choice," Journal of Consumer Research, 19 (September), 226-39.

Mitchell, Andrew A. and Jerry C. Olson (1981), "Are Product Attribute Beliefs the Only Mediator of Advertising Effects on Brand Attitude?" Journal of Marketing Research, 19 (August), 318-32.

Nunnally, Jum C. (1978), Psychometric Theory 2nd edition, New York: McGraw-Hill Book Company.

Smith, Robert E. and William R. Swinyard (1983), "Attitude-Behavior Consistency: The Impact of Product Trial Versus Advertising," Journal of Marketing Research, 20 (August), 257-67.

Thomson, Donald M. and Endel Tulving (1970), "Associative Encoding and Retrieval: Weak and Strong Cues," Journal of Experimental Psychology, 86 (2), 255-62.

----------------------------------------