Attitude Toward the Ad As a Mediator of Advertising Effectiveness: Determinants and Consequences

ABSTRACT - Attitude toward the ad (AAD) is receiving increasing attention as a mediator of advertising effects. The empirical portion of this paper examined the impact of two moderating variables - product class importance and product class knowledge - on the relationship between A and brand attitude. Following Petty and Cacioppo (1981), peripheral processing appeared to dominate the low knowledge/low importance subsample, but, contrary to predictions, central processing served only as a supplement to peripheral processing in the high knowledge/high importance subsample. Explanations for this deviation from the predicted relationship were discussed, and the paper concluded with some preliminary thoughts about possible determinants of AAD including credibility, ad perceptions, attitude toward the advertiser, attitude toward advertising in general, and mood.


Richard J. Lutz, Scott B. MacKenzie, and George E. Belch (1983) ,"Attitude Toward the Ad As a Mediator of Advertising Effectiveness: Determinants and Consequences", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 532-539.

Advances in Consumer Research Volume 10, 1983      Pages 532-539


Richard J. Lutz, University of Florida

Scott B. MacKenzie, UCLA

George E. Belch, San Diego State University

[Richard J. Lutz is a Professor of Marketing; Scott B. MacKenzie is a doctoral candidate in marketing; and George E. Belch is an Assistant Professor of Marketing. Requests for reprints should be directed to Richard J. Lutz at the College of Business Administration, University of Florida, Gainesville, FL 32611.]


Attitude toward the ad (AAD) is receiving increasing attention as a mediator of advertising effects. The empirical portion of this paper examined the impact of two moderating variables - product class importance and product class knowledge - on the relationship between A and brand attitude. Following Petty and Cacioppo (1981), peripheral processing appeared to dominate the low knowledge/low importance subsample, but, contrary to predictions, central processing served only as a supplement to peripheral processing in the high knowledge/high importance subsample. Explanations for this deviation from the predicted relationship were discussed, and the paper concluded with some preliminary thoughts about possible determinants of AAD including credibility, ad perceptions, attitude toward the advertiser, attitude toward advertising in general, and mood.


The research reported here represents one of a series of studies we have conducted aimed at delineating the possible causal mediating role of the affective reactions to an advertising stimulus. Briefly, it is posited that the recipients of an advertising message develop an attitude toward the ad (hereafter AAD) which in turn exerts an influence on subsequent measures of advertising effectiveness such as brand attitude and purchase intentions. We believe that the effects of A are particularly relevant in a diagnostic commercial AD pretesting setting wherein the communications effectiveness of an ad is being assessed. Due to the heightened salience of advertising stimuli in such a situation, the effects of AAD may be substantial and a hence could lead to artificially inflated effects on other at response measures. Alternatively, as postulated by Shimp (1981) and Mitchell and Olson (1981), A may be an important mediator in naturalistic advertising exposure situations, and not just an artifact of the testing situation. In either case, the role of AAD seems worthy of investigation.




Conceptualization of the Role of AAD

In thinking about AAD as a mediating causal variable, it became apparent early on that there are several possibilities for exactly how it may intervene. Figure 1 portrays four competing explanations for the impact of AAD on Purchase Intentions. While current space constraints preclude a full description of these four competing models, a brief summary is in order. interested reader should consult our earlier working paper (MacKenzie and Lutz 1982) for a complete discussion.

In brief, Figure 1 depicts four possible response sequences which may follow exposure to a persuasive communication. Five constructs form the core of all four models:

1. Ad Cognitions (CAD) - i.e., recipients' perceptions of the ad itself (e.g., its execution);

2. Brand Cognitions (CB) - i.e., recipients' perceptions of the brand being advertised (e.g., brand attributes);

3. Attitude toward the Ad (AAD) - i.e., recipients' affective reactions (e.g., like-dislike) to the ad itself;

4. Attitude toward the Brand (AB) - i.e., recipients' affective reactions toward the advertised brand (or, where desirable. attitude toward purchasing the brand); and

5. Purchase Intention (PI) - i.e., recipients' assessments of the likelihood that they will purchase the brand in the future.

The three solid arrows in Figure 1 represent causal relationships which are common to ali four models. That is, all four models predict that CAD leads to AAD; that CB leads to AB and that AB leads to PI. The remaining relationships, which are depicted by dashed or dotted arrows, are hypothesized to be present in one or more of the models, but not all four.

Motel 1 posits a direct one way influence of AAD on AB. This model was proposed by Shimp (1981) and supported empirically by Mitchell and Olson (1981). Model 2 asserts both a direct and direct influence of AAD on AB. The indirect influence operates through CB in the sense that a positive relationship between AAD and acceptance of ad claims regarding the brand is postulated. This model is consistent with one proposed by Lutz and Swasy (1977) and closely related to one supported empirically by Holbrook (1978). Model 3 is essentially a balance theory account of the AAD - AB relationship, with causation flowing in both directions. For a new brand AA should exert relatively more influence on s than vice versa; for an existing brand, the flow from AB to AD would be expected to dominate. Finally, Model 4 represents a modification of Howard's (1977) distinction between Brand Concept (in our terminology AB) and Impersonal Attitude, i.e., attitude toward the "conditions of purchase."

Howard's construct is intended to represent situational pressures on purchase such as availability and deals. It seems reasonable in an advertising exposure setting to construe AAD as the relevant situational variable. As shown in Figure 1, Model 4 posits that AAD influences PI independently of AB, i.e., the two attitudinal constructs are unrelated.

Previous Empirical Research

Our first attempt to sort out the relative merits of the four explanations offered above utilized a data set provided by Ford Motor Company (see MacKenzie and Lutz (1982) for a full description of that study). The results of that study ruled out model 4 as a reasonable account of the role of AAD, due to the fact that there was a strong relationship between AB and AAD. Therefore, the assumption of the independence of those two constructs which characterized Model 4 was clearly refuted. The relative "fits" of the remaining three models were only marginally different, but a cross-validation procedure showed that Model 1 achieved the most stable fit. Hence, there was fairly strong evidence in support of the direct flow of effects from AAD to AB, but less so for the indirect effect postulated by Model 2 or the two-way causation inherent in Model 3.

The results of the first study must be tempered by the fact that some of the measures available to operationalize the five constructs in the competing models were less than satisfactory (see MacKenzie and Lutz (1983) for a discussion o. the measurement shortcomings). Therefore, the conclusions reached regarding the relative efficacy of Models X, 2 and 3 must be regarded as tentative. However, despite the measurement caveats, we were encouraged enough by this initial study to pursue the topic further.

At this point we located two additional data sets both I of which incorporated adequate measures of the five 5 constructs of interest. The first of these data sets formed the basis for some previous advertising research (Belch 1981), while the second, currently under analysis, resulted from a follow up study to the first data set. AAD measures were incorporated into these two studies but were not a central focus of the investigations. These two data sets were analyzed independently, using structural equation procedures to test the relative explanatory strengths of Models 1-4. A manuscript reporting the results is currently in preparation (MacKenzie, Lutz and Belch 1982). Briefly, Models 1 and 2 fit well in both data sets, with Model 3 lagging behind somewhat and model 4, once again, clearly inferior. In the interest of parsimony, Model 1 (with one less relationship than Model 2) was judged to be slightly superior. There is apparently fairly strong support for the causal influence of AAD on AB in all three data sets tested to date.

Purpose of the Present Study

The purpose of this study is to explore one of the existing data sets a bit further in order to uncover some potential moderating variables influencing the AAD - AB relationship. Following the empirical portion of the paper, some preliminary thoughts will be offered with the goal of moving toward a somewhat richer conceptualization of the determinants of AAD.

Petty and Cacioppo (1981) have proposed that there are two basic routes to changing attitudes: central and peripheral. Briefly, the central route refers to attitude change resulting from recipients' actively thinking about the content of a persuasive message.

In contrast, the peripheral route to persuasion reflects attitude change due to ancillary aspects of the message such as its source. [Chaiken (1980) has made a similar distinction using the term systematic to refer to effortful message-based persuasion and heuristic to represent less effortful processing based on source and other non-content cues.] In the present context, changes in AB governed by AAD would be a peripheral process, while the influence of brand cognitions (CB) on AB would be seen as a central process. Most past advertising research has implicitly assumed a central processing route, though Mitchell and Olson (1981) implicitly considered the peripheral route in their study of the mediating role of AAD.

Petty and Cacioppo went on to specify two factors which influence the occurrence of the two routes to persuasion. Motivation to process message information is expected to vary directly with the degree of central processing. This factor has also been labeled involvement by Petty and Cacioppo and by Chaiken (1980). Ability to process message information is also expected to vary directly with the degree of central processing; determinants of a recipient's ability to process a message include things such as message cogency, the recipient's prior knowledge of the topic, and response opportunity. [Chaiken (1980) did not include an ability factor in her conceptual model; however, in the empirical portion of her paper she used topics about which the subject population had sufficient prior knowledge. Hence, she implicitly included processing ability in her work.]

Petty and Cacioppo's view of persuasion is a potentially useful one for examining the role of AAD in response to advertising communications. Specifically, one would anticipate that consumers low in processing ability and motivation would be characterized by peripheral processing (and, hence, a strong AAD - AB relationship). In contrast, consumers high in both ability and motivation should proceed via the central route, exhibiting a strong CB - AB relationship. The purpose of the present study was to investigate these predictions, which can be stated more formally as research hypotheses.

H1: In an advertising pretesting context, consumers low in both motivation and ability to process information exhibit a relatively strong influence of AAD on AB and a relatively weak influence of CB on AB.

H2: In an advertising pretesting context, consumers high in both motivation and ability to process information exhibit a relatively weak influence of AAD on AB and a relatively strong influence of CB on AB.


[See Belch (1981) for a more thorough description of the data collection procedure.]


Arrangements were mate with two church groups whereby a donation was given to the church in exchange for each participant the church recruited. Two hundred sixty participants were obtained in this way. The subjects ranged in age from 18 to 75 with a mean age of 45. Sixty-nine percent were women and thirty-one percent men. All subjects were blind to the experimental treatments and hypotheses.


Upon arrival at the research setting subjects were given a brief description of the reason for their presence and were informed that they would be viewing a one-hour S episode of a television show (Quincy). In addition, they were told that following the show they would be asked some questions about both the show and the commercials they had seen.

After listening to the instructions, subjects were asked to fill out the first part of the questionnaire which included demographic questions, a television viewing profile, and premeasures on issues addressed in the program. Then the one-hour episode of Quincy was shown. Immediately after the program ended and the last commercial had been shown, the subjects were told that their reactions to the commercials seen in the show were of interest. The subjects were then given two minutes to list what they were thinking (i.e., cognitive responses; cf., Wright 1980) while viewing the last commercial shown. Next the subjects were asked to complete a program evaluation form and a set of post measures about the issues addressed in the program. The subjects were then asked to respond to the dependent measures pertaining to their evaluations of the last commercial shown, the brand advertised in this commercial (Shield toothpaste) and their purchase intentions with respect to Shield. The final measures taken were two 7-point scales used to assess how knowledgeable (not very - 1/very = 7) they were about the product class and how important (very important = 7/very unimportant s 1) they perceived the purchase decision regarding toothpaste to be.

Categorization and Coding of Cognitive Responses

The cognitive responses listed were categorized according to whether they were directed toward the brand or the ad (e.g., statements about the physical quality of the ad), and according to whether they were evaluatively positive or negative. The coding of the cognitive responses according to object was tone independently by three judges, while the coding of the evaluative direction of the cognitive responses was done by the subjects after completing the experimental task. Overall, the coding definitions and procedures used were quite reliable (see Belch 1981).

Splitting the Sample

One way to examine the moderating effect of product knowledge (i.e., processing ability) and importance (i.e., processing motivation) on the hypothesized relationships is to split the sample into groups that are more homogeneous with respect to these factors and test the hypotheses in the groups separately. In the present study, this involved splitting the sample into four distinct groups according to whether subjects were above or below the median on each of these two factors. A chi-square test revealed that subjects who were very y (not very) knowledgeable about the product class (i.e., toothpaste) were significantly more likely to perceive the product class as being very (not very) important (x2=15.8, df=1, p<.01). Because knowledge and importance were strongly related, only the high knowledge (K=5.1)/high importance (I=4.6) group and the low knowledge (K=2.3) low importance (I=1.4) groups were used in the analysis. Of the original sample of 260 subjects, 82 were in the former group and 80 were in the latter

Construct Operationalization

Of the six constructs shown in Figure 2, three (AAD, AB, and PI) were operationalized as latent variables and three (CAD, CB, and Credibility) as manifest variables.

AAD was represented by ratings of how favorable-unfavorable and interesting-uninteresting the ad was. AB was represented by ratings of how good-bad, favorable-unfavorable and wise-foolish subjects felt purchasing the brand to be. PI was measured by subjective estimates (likely-unlikely, probable-improbable and possible-impossible) of the probability of trying the brand.



Communicator credibility (CRED) was represented by an index calculated by averaging each subject's ratings of how biased-unbiased, believable-unbelievable, truthful-untruthful, honest-dishonest qualified-unqualified, and sincere-insincere the communicator was. The cognitive responses directed toward the ad's execution were represented by a single index derived by subtracting the number of negative cognitive responses directed toward the ad's execution from the number of positive ad execution statements. Similarly, the measure of brand-related cognitive responses was computed by subtracting the number of negative brand cognitions from the number of Positive.


The parameter estimates for the model were calculated using LISREL's maximum likelihood technique (Joreskog and Sorbom 1981). The scale of measurement was established for the three hypothetical constructs (AAD, AB and PI) by constraining one of the measurement model coefficients from each of them to be equal to 1.0. The three remaining variables (CAD, CB and CRED) were modeled as manifest variables by fixing their measurement model coefficients at 1.0 and their measurement error terms at 0.

The testing of the proposed theoretical relationships involved two major phases. First, the conceptual model shown in Figure 2 was estimated for each sample (high knowledge/importance and low knowledge/importance) separately. [The model actually tested in the analysis differs from the conceptual model shown in Figure 2. Figure 2 does not include the correlations between the exogenous factors or the correlations between the measurement error terms that were est mated. These relationships, caused by unknown factors exogenous to the system, were estimated so they would not obscure the examination of the theoretical relationships of interest.] At this point the significance of the individual parameter estimates and the adequacy of the overall fit of the model were assessed. The second stage of the analysis examined the effects of the moderating variables (knowledge/importance) on the strengths of the relationships that were found to exist in the first stage.

The standardized parameter estimates for the high knowledge/importance group are displayed in Figure 3. The unstandardized estimates for all the non-zero standardized estimates shown were significantly (p<.05, one-tailed) different from zero. Thus, it appears that both communicator credibility and ad execution cognitions influenced attitude toward the ad; brand cognition and ad attitude influenced brand attitude; and brand attitude affected brand purchase intentions, as predicted.

Overall, the model fit the data in this subsample reasonably well. The fit indices were high: Bentler and Bonett's (1980) fit index was .93, and Joreskog and Sorbom's (19812 fit index was .80. The traditional chi-square test of the overidentifying restrictions also provided some support for the model (x-=53, df=38, p<.05). On the basis of the significance of the parameter estimates and the overall fit of the model, it can be concluded that all the hypothesized relationships were supported when subjects' ratings of knowledge and importance were high.



The standardized parameter estimates for the low knowledge/importance group are shown in Figure 4 (all non-zero coefficients were significantly different from zero; p<.05. one tailed). As in the high knowledge/importance sample, communicator credibility and ad execution cognitions influenced attitude toward the ad; ad attitude influenced brand attitude; and brand attitude influenced brand purchase intentions. However, in this subsample brand cognitions did not influence brand attitudes.

Overall, the model tested fit the data in this subsample fairly well. The fit indices were both high: Bentler and Bonett's (1980) fit index was .91 and Joreskog and Sorbom's (1981) fit index was .76. The chi-square test of the overidentifying restrictions also provided adequate support (X=53, df=38, p>.055. It can be concluded, therefore, that all the hypothesized relationships, except the one between CB and AB, were supported in the low knowledge/importance subsample.

The second phase of the analysis concentrated on examining the effects of the moderator variables (knowledge and importance) on the hypothesized relationships. However, before any statements about the effects of the moderator variables on the structural relationships could be made it was necessary to ensure that the measurement model parameters (i.e., the factor loadings) were the same in both groups. Had these parameters not been the same in the two samples the structural estimates could not have unambiguously compared because they would have represented relationships between slightly different latent constructs.



To test whether he measurement parameters were the same in the two samples it was necessary to estimate the model shown in Figure 2 simultaneously for both groups and compare the chi-square value associated with that solution to the chi-square value obtained from a solution which constrains the measurement parameters to be equal in the two groups. The former solution had a chi-square value of 106.42 at 76 degrees of freedom, while the latter had a chi-square value of 108.92 at 81 degrees of freedom. thus indicating that the measurement model parameters shown in Figure 3 and 4 were not significantly different.

Since the measurement parameters were equal in the two subsamples, it was possible to test for differences in the structural estimates. One difference suggested by the separate single group solutions was that CB and AB were not related in the low knowledge/importance subsample. When the measurement parameters were constrained to be equal in a two group solution, this finding was confirmed. The coefficient for the path between CB and AB was insignificant.

A second difference in the structural estimates between the two subsamples involved the relationship between AB and PI. It appears that this relationship was stronger in the high knowledge/importance subsample than in the low. To test or a difference between these estimates, the chi-square value associated with a solution in which these parameters were contrained to be equal was compared to the chi-square value derived from a solution which allowed these estimates to be estimated independently. The chi-square difference between these models was 5.4 which was significant (p<.05) at 1 degree of freedom. This means that the coefficient was significantly larger in the high knowledge/importance subsample than in the low, even when the measurement parameters were constrained to be equal in the two.


Several aspects of the above results are noteworthy. As expected, AAD was a significant mediator of AB in both subsamples. As hypothesized in H1 AAD dominated CB in influencing AB in the low knowledge/low importance subsample. However, contrary to H2, A was also a stronger influence than CB in the high knowledge/high importance subsample. In fact, with the exception of the link between CAD and AAD, all relationships were stronger in the high/high subsample than in the low/low. The rejection of H2 calls into question the viability of the central vs. peripheral distinction in an advertising context. Instead of a switch from peripheral to central processing in the high/high subsample, central processing (in the form of the CB - AB relationship) emerged as a supplement to the skill-dominant peripheral processing mode.

There are a number of possible explanations for this pattern of findings. First, the importance of the processing task (end hence motivation to process) may have been limited to relatively low levels. A median split cannot guarantee highly motivated processing, especially for a product like toothpaste. Therefore, motivation may have never reached a level sufficient to induce the message-oriented processing expected under the central route to persuasion.

Second, the relationship he tween CB and AB was unexpectedly low (zero in the low/low subsample). which suggests that the CB measure may have been inadequate in capturing the essence of the central processing route. For instance, Lutz (1977) and Mitchell and Olson (1981) found much stronger relationships between CB and AB, but they used close-ended cognitive structure measures to tap CB instead of the open-ended cognitive response measures used here. It may be that the ability of the cognitive response measure to represent the same set of beliefs as do the traditional cognitive structure measures is more limited than previously assumed. The degree of non-overlap may be exacerbated where message repetitions are involved, as in the present study. Cognitive responses given in response to the third and fifth exposure of an ad may omit responses to earlier exposures (i.e., the cognitive response measure may not be cumulative). Unfortunately, the present sample size is too small to test this explanation within the structural equation system.

Third, the generally weak CB - AB relationship may be accounted for by a statistical artifact of the measurement procedure. The two potential predictors of AB (i.e., CB and AAD) differed substantially in their potential to vary. AAD was measured via the average of two 7-point scales, while CB was represented by an index. constructed by subtracting negative from positive brand-related cognitive responses. Since the total number of responses was typically small (two or three per subject), the variation in the index was also quite low. The small relative variation of CB may have had the effect of suppressing the CB - AB relationship. [However, it should be noted that CAD, which was a similarly constructed index, was a reasonably strong predictor of AAD.]

Fourth, demand characteristics may account for the apparent dominance of the "peripheral" processing mode. Placing subjects in an ad pretesting situation may simply "set" them for a particular mode of response. Drawing more attention than is normally the case to the ad may suppress otherwise naturally occurring brand-related responses. One way to test for this potential biasing effect would be to vary the "set" in the situation, with some subjects being instructed to focus on the ad, some on the brand, some on the program, etc. If the dominance of AAD over CB as a predictor of AB, holds up across sets, then the effect would be considered more robust.

Finally, AAD may in fact be the relatively influential mediating variable which these results suggest. In reality, people may focus more of their attention on evaluating advertising stimuli than on learning about the brands advertised. Generating an attitude toward a commercial may be a more natural and less effortful response than attempting to encode, evaluate and possibly argue against specific brand data (cf., Krugman 1965, Zajonc 1980). Given this possibility, it may well be that AAD is generally an important mediator of AB and not just an artifact of the experimental context.


Up to this point, the present investigation has been concerned primarily with the mediating role of AAD. We now turn our attention to some of the possible antecedents of AAD. Given the seemingly important role of AAD in mediating advertising response, further consideration of its origins is warranted. The previous research we have conducted has focused on determinants of AAD which, from the Petty and Cacioppo (1981) perspective, would be labeled central processing. That is, both of the underlying factors considered here were of a perceptual nature, suggesting that consumers somehow perceive certain characteristics of commercial stimuli, process them, and then form an attitude toward the stimulus. However, just as AB may be formed on a more peripheral basis, as suggested by our empirical results presented earlier, so might AAD be formed more peripherally. The present discussion addresses possible AAD determinants, both central and peripheral, and speculates with respect to possible interrelationships among them.



Five Antecedents of AAD

As shown in Figure 5, which some of our colleagues have light-heartedly labeled the "Balloon Model," there are at least five potential antecedents of AAD: (a) the credibility of the ad; (b) other perceptions of the ad; (c) attitude toward the advertiser; (d) attitude toward advertising in general; and (e) the recipient's general affective state or "mood" at the time of exposure. Each of these determinants is elaborated u?on in turn below.

Credibility of the ad. Credibility of the ad refers to how truthful or believable the audience perceives the assertions made about the brand to be. Clearly, Ad Credibility is one aspect of Ad Perceptions, which is considered as a separate determinant of AAD. We have chosen to treat Ad Credibility separately for heuristic reasons. There is such a rich tradition of research on credibility in the communications literature that it seemed reasonable to segregate it from other forms of perceptual responses to advertising. As with other forms of ad perceptions, the mechanism by which Ad Credibility is thought to influence AA derives directly from Fishbein and Ajzen's (1975) expectancy-value theory of attitude formation: the subjective likelihood of the ad being credible, multiplied by the subjective value (i.e., goodness-badness) of credibility, would serve as a partial determinant of AAD.

Figure 5 also depicts what might be labeled a Credibility Subsystem which consist of possible underlying determinants of Ad Credibility. Three factors are proposed: (a) the perceived claim discrepancy of the ad; (b) the credibility of the advertiser; and (c) the credibility of advertising in general. Briefly, Ad Claim discrepancy is viewed as the fundamental determinant of Ad Credibility. To the extent that claims made about the brand in the ad are inconsistent with the recipient's prior perceptions of the brand, Ad Credibility will be damaged (cf. Fishbein and Ajzen, 1975).

The effects of Advertiser Credibility and Advertising Credibility are thought to affect the credibility of any particular ad through a force toward logical consistency (cf., Fishbein and Ajzen, 1975). Audience members are likely to perceive a direct relationship between how credible any given advertiser is and the credibility of the ads emanating from that advertiser. Similarly, Bauer and Greyser (1968) found that perceived truthfulness was one of the dominant perceptual dimensions underlying the American public's reactions to advertising in general. It seems reasonable to assume that consumers' general perceptions of credibility would exert at least some influence on their assessments of a particular ad's credibility.

Perception of the ad. As noted above, while credibility has been perhaps the most widely studied ad perception, there is a wide array of other perceptions which have been shown to influence AAD. For example, Bauer and Greyser (1968) found that four perceptual dimensions - annoying, enjoyable, informative, and offensive - related strongly to AAD. As with Ad Credibility, our conception of the mechanism by which Ad Perceptions influence AA is the expectancy-value formulation. Perceptions of ad characteristic, coupled with the desirability of those characteristics, would form a predictive index.

Several studies have suggested the kinds of perceptual dimensions which may be relevant. Most prominent has been the so-called "reaction profile" work (e.g., Wells, Leavitt, and McConville, 1971), which has identified six factors: humor, vigor, sensuousness, uniqueness, personal relevance, and irritation. More recently, Aaker, Bruzzone and Norris (1981) identified the four factors of entertaining, personal relevance, dislike, and warm. And, as shown in Figure 5, ad execution cognitive responses (CAD in the empirical portion of this paper) have been shown to play an important role (see also Belch, 1981).

From the above discussion, it should be clear that both Ad Credibility and Ad Perceptions influence AAD in a central processing mode. AAD is formed on the basis of perception and analysis of the ad. We now turn our attention to more peripheral processes through which AAD may be formed.

Attitude toward the advertiser (AADV). AADV refers to the audience's affective feelings about the advertiser per se. This construct is quite similar to the notion of source attractiveness, which is another fairly heavily studied communications variable. The mechanism by which AADV influences AD is thought to be a rather straightforward generalization of affect: feelings about the advertiser govern feelings about the ad itself. Little active thought is required; the process is more or less automatic.

Figure 5 also depicts the possible determinants of AADV. In other words, AADV is something the audience carries into the ad exposure situation; it was formed previously and serves as a summary judgment of the advertiser. The underlying source of AAD is thought to be perceptions of the advertiser, which say be formed directly, on the basis of past experience with or information about the advertiser, or indirectly, via a process of logical consistency wherein perceptions of advertising in general are used to draw inferences about any particular advertiser. Key perceptual dimensions with respect to the advertiser would include such things as attractiveness, reputability and similarity. Again, credibility would constitute an important dimension, but it has been treated separately in the Credibility Subsystem.

General attitude toward advertising (AAG). AAG is defined as the audience member's affective reaction advertising in general. AAG, like AADV is seen as transferring virtually automatically to AAD. As shown in Figure 5, AAG is determined by perceptions of advertising in general. Bauer and Greyser (1968) identified four perceptual dimensions in addition to truthfulness (previously discussed): essential, raises standard of living, raises prices, persuades people to make unnecessary purchases. These dimensions form a reasonable starting point for explaining AAG, but further exploratory work is needed to ascertain if other dimensions have become salient.

"Mood". Mood refers to the recipient's general affective state at the time of exposure to the commercial message. These feelings, whether positive or negative, are thought to transfer to A . Some indirect support for this process was found by Srull (1983), who demonstrated a direct effect of Mood on AB.

Sources of determinants of Mood, as shown in Figure 5, are individual differences (some people are generally more negative or positive in their outlook on life) and the reception context. Thus, Mood is seen as a "state" variable, resulting from a "trait" (individual difference) and contextual factors. Contextual factors include things such as (a) the nature of the exposure (e.g., and intrusive exposure like television commercials vs. consumer-instigated exposure to ads as a form of information search); (b) "clutter," or the surrounding program or editorial matter, which is a well known rule of thumb among advertiser regarding qualitative factors in media selection.

Each of the last three determinants of AAD - i.e., AADV, AAG, and Mood is seen as influencing AAD through a more peripheral processing route. The chief mechanism is the simple transfer of affect rather than elaborated cognitive processing. In view of recent findings by Csikszentmihalyi and Kubey (1981) which show that people are generally cognitively unaroused while viewing television (consistent with earlier speculation by McLuhan and Krugman), these peripheral processes warrant more investigation in future research on television commercial effectiveness.


The foregoing discussion is not intended to represent a formal theory of AAD; it is too brief and omits many possible inter-relationships for the sake of simplicity. The goal of the discussion was to raise some important issues regarding potential antecedents of A and attempt to provide some sort of organizing framework. Our hope is that the ideas presented here will stimulate further research on the origin and role of AAD.


Aaker, David A., Bruzzone, Donald E., and Norris, Donald (1981), "Viewers' Perceptions of Prime-Time Television Advertising and Characteristics of Commercials Perceived as Informative," Marketing Science Institute Working Paper No. 81-103, Cambridge: MSI.

Bauer, Raymond A., and Greyser, Stephen A. (1968), Advertising in America: The Consumer View, Boston: Harvard University Press.

Belch, George E. (1981), "An Examination of Comparative and Noncomparative Television Commercials: The Effects of Claim Variation and Repetition on Cognitive Response and Message Acceptance," Journal of Marketing Research, 18 (August), 333-49.

Bentler, Peter M., and Bonett, Douglas G. (1980), "Significance Tests and Goodness of Fit in the Analysis of Covariance Structures," Psychological Bulletin, 88, 588-606.

Chaiken, Shelly (1980), "Heuristic Versus Systematic Information Processing and the Use of Source Versus Message Cues in Persuasion," Journal of Personality and Social Psychology, 39 (November), 752-67.

Csikszentmihalyi, Mihaly, and Kubey, Robert (1981), "Television and the Rest of Life: A Systematic Comparison of Subjective Experience," Public Opinion Quarterly, 45 (Fall), 317-28.

Fishbein, Martin, and Ajzen, Icek (1975), Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research, Reading, Mass.: Addison-Wesley.

Holbrook, Morris P. (1978), "Beyond Attitude Structure: Toward the Determinants of Attitude," Journal of Marketing Research, 15 (November), 545-56.

Howard, John A. (1977), Consumer Behavior: Application of Theory, New York: McGraw-Hill.

Joreskog, Karl G., and Sorbom, Dag (1981), LISREL: Analysis of Linear Structural Relationships by the Method of Maximum Likelihood, Chicago: International Education Cervices.

Krugman, Herbert E. (1965), "The Impact of Television Advertising: Learning Without Involvement," Public Opinion Quarterly, 29 (Fall), 349-56.

Lutz, Richard J. (1977), "An Experimental Investigation cf Causal Relations Among Cognition, Affect, and Behavioral Intention," Journal of Consumer Research, 3 (March), 197-208.

MacKenzie, Scott R., and Lutz, Richard J. (1982), "Monitoring Advertising Effectiveness: A structural Equation Analysis of the Mediating Role of Attitude Toward the Ad," Working Paper No. 117, Center for Marketing Studies, UCLA.

MacKenzie, Scott R. (1983), "Testing Competing Theories of Advertising Effectiveness Via Structural Equation Models," in Proceedings of American Marketing Association Research Methods and Causal Modeling Conference, forthcoming.

MacKenzie, Scott R., and Belch, George E. (1983), "The Role of Attitude Toward the Ad as a Mediator of Advertising Effectiveness: A Test of Competing Explanations," unpublished manuscript (in Preparation).

Mitchell, Andrew A., and Olson, Jerry C. (1981), "Are Product Attribute Beliefs the Only Mediator of Advertising Effects on Brand Attitude?" Journal of Marketing Research, 18 (August), 318-32.

Petty, Richard E., and Cacioppo, John C. (1981), Attitudes and Persuasion: Classic and Contemporary Approaches, Dubuque, Iowa: Wm. C. Brown Co.

Shimp, Terence A. (1981), "Attitude Toward the Ad as a Mediator of Consumer Brand Choice," Journal of Advertising, 10(2), 9-14.

Srull, Thomas K. (1983), "Affect and Memory: The Impact of Affective Reactions in Advertising on the Representation of Product Information in Memory," in Richard P. Bagozzi and Alice M. Tybout (eds.), Advances in Consumer Research, Vol. X, Ann Arbor: Association for Consumer Research.

Webb, Peter R. (1979), "Consumer Initial Processing in Difficult Media Environment, "Journal of Consumer Research. 6 (December). 225-36.

Wells, William D., Leavitt, Clark, and McConville, Maureen (1971), "A Reaction Profile for TV Commercials," Journal of Advertising Research, (December), 11-17.

Wright, Peter (1980), "Message-Evoked Thoughts: Persuasion Research Using Thought Verbalizations," Journal of Consumer Research. 7 (September), 151-75.

Zajonc, Robert B. (1980), "Feeling and Thinking: Preferences Need No Inferences," American Psychologist, 35 (February). 151-75.



Richard J. Lutz, University of Florida
Scott B. MacKenzie, UCLA
George E. Belch, San Diego State University


NA - Advances in Consumer Research Volume 10 | 1983

Share Proceeding

Featured papers

See More


Consumers’ Implicit Mindsets and Responses to Cause-related Marketing Campaigns

Meng-Hua Hsieh, Kent State University, USA
Ozge Yucel-Aybat, Pennsylvania State University Harrisburg

Read More


Time-insensitive Budget Tracking: Nudging Consumers to Spread out Spending over Time

Liang Huang, University of Arizona, USA
Anastasiya Pocheptsova Ghosh, University of Arizona, USA

Read More


I11. Self-Presentation in the Mating Market: The Influence of Gender and Sexual Orientation on Profiles on Tinder and Grindr

Chaim Kuhnreich, Concordia University, Canada
Lilian Carvalho, FGV/EAESP
Gad Saad, Concordia University, Canada

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.