Advances in Consumer Research Volume 18, 1991 Pages 202-209
BRAND FAMILIARITY AND PRODUCT INVOLVEMENT EFFECTS ON THE ATTITUDE TOWARD AN AD - BRAND ATTITUDE RELATIONSHIP
Joseph Phelps, University of Alabama
Esther Thorson, University of Wisconsin-Madison
This study builds upon earlier Attitude-toward-an-ad (Aad) research. It adds empirical evidence to the current debates about the effects of brand familiarity and product involvement on the Aad-Ab (brand attitude) relationship and makes suggestions for future research.
The results indicate that Aad significantly affects Ab for unfamiliar brands. More important for theory development, Aad affects Ab for familiar brands, even after controlling for prior brand attitude. However, product involvement does not seem to affect the Aad-Ab relationship. It is suggested that product involvement's influence should be reexamined in a more natural viewing environment.
Attitude toward an ad (Aad) can be thought of as a viewer's general liking or disliking of an advertisement. Recently Aad has become a topic of considerable research interest (e.g. Mitchell and Olson, 1981; Shimp, 1981; Shimp and Yokum, 1982; Moore and Hutchinson, 1983; Lutz and Belch, 1983; Gardner, 1985; Batra and Ray, 1986; Muehling, 1987).
One reason these researchers are interested in Aad is the evidence that suggests Aad is one of the factors that influences brand attitude (Ab) and intent to purchase (PI) (Gresham and Shimp, 1985; Cox and Locander, 1987; Moore and Hutchinson, 1983).
Along with these theoretical relationships come some very practical consequences. With the ever increasing cost of media, more effort has been put into the development of commercial pretesting methodology. Pragmatically, understanding the role of Aad may increase the efficiency of designing and pretesting advertisements (MacKenzie, Lutz and Belch, 1986; Gardner, 1985).
Please note that given the limitations of space, this discussion will limit itself to the empirical question that asks: What effects do product involvement and brand familiarity have on the Aad-Ab relationship. For a discussion of the theoretical underpinnings see Thorson and Page (1990).
Product Involvement and the Aad-Ab Relationship
In the complex world of a viewer processing an advertisement, many variables may intervene in the Aad-Ab relationship. Two such variables are product class involvement and brand familiarity.
Although the effects of product involvement and brand familiarity in Aad-Ab relationship have been studied, the effect these variables have is still under debate. In Aad-Ab research, Thorson and Page (1989); Rossiter and Percy (1984); and others have studied product involvement. These studies suggest conflicting observations of product involvement's role in the Aad-Ab relationship. Before discussing the product involvement studies, it is important that a distinction be made between product involvement and message response involvement. Although the two involvement constructs are related they are not the same.
Recently a number of studies have examined message response involvement's role in the Aad-Ab-PI relationship. (Batra and Ray, 1985; Muehling and Laczniak, 1988; Park and Young, 1986), to name just a few. Batra and Ray (1985) define message response involvement as that which:
refers conceptually to the "depth" of processing for a particular message, by a particular recipient at a particular time.... this construct is not purely motivational in origin. Such processing "depth," however operationalized, would be a function not just of the viewer's motivation to respond "deeply," but also his/her ability to do so, as a function of prior usage and knowledge, and of the opportunity to do so, caused by message pace and distraction, for instance.
Laczniak, Muehling, and Grossbart (1989) defined message involvement as the motivational state of an individual induced by a particular advertising stimulus or situation.
Product involvement has also been conceptualized as a motivational construct (Batra and Ray, 1985), where the amount of motivation may depend on the relevance of the product (Zaichowsky, 1985). Thus, previous research has conceptualized product involvement as more purely a motivational construct, while including motivation, ability, and opportunity in the conceptualization of message involvement.
Product involvement was included in this study to examine its potential as a mediator in the Aad-Ab relationship. Message response involvement was not examined in this study. Given the distinctions noted above it is time to return to the product involvement studies. These studies suggest competing observations of product involvement's role in the Aad-Ab relationship.
Rossiter and Percy (1984) reported that most of the published studies where Aad had been found to contribute significantly to Ab used low involvement products, and hence they suggest the Aad-Ab relationship may be stronger for low involvement products than it is for high involvement products. Nevertheless, Thorson and Page (1989) report that the level of product involvement had no impact on the Aad-Ab relationship. This study will examine these conflicting observations.
Brand Familiarity and the Aad-Ab Relationship
The role brand familiarity plays in the Aad-Ab relationship is also controversial. Although there is general agreement that Aad affects Ab when unfamiliar brands are tested, there is conflicting evidence on whether this Aad-Ab relationship occurs with familiar brands.
In theory, if a person is unfamiliar with a brand, the information they get from the ad and their Aad should have a relatively strong influence on their Ab. However, if the person is familiar with the brand, they may have already formed an Ab and their Aad should not have as strong of an effect.
Most of the early studies that found a relationship between Aad-Ab used either unfamiliar or hypothetical brands. However, studies using familiar brands have also found Aad-Ab effects (Thorson and Page, 1989; Batra and Ray, 1986; Edell and Burke, 1984, 1986; Messmer, 1979; Gresham and Shimp, 1985).
In contrast, Machleit and Wilson (1988), hypothesized there is no Aad-Ab relationship for familiar brands and suggested the studies that showed a direct effect of Aad on Ab for familiar brands were flawed because of the researchers' failure to account for prior brand attitude.
In their study, Machleit and Wilson controlled for prior brand attitude. Without such a control, the Aad-Ab relationship, under familiar brands, may be inflated. They indicate that Aad did not significantly affect Ab for familiar brands.
However, Edell and Burke's (1986) results conflict with what Machleit and Wilson found. Edell and Burke reported that both prior brand attitude and Aad affected Ab even when brands were familiar. The present study also controlled for prior brand attitude. Given the conflicting results noted above, this study has a unique opportunity to test the competing hypotheses.
In light of the literature on brand familiarity and product involvement, four hypotheses were articulated. Because the preponderance of studies support Aad influence on Ab for both unfamiliar and familiar brands (Thorson and Page, 1989; Batra and Ray, 1986; Edell and Burke, 1984; Messmer, 1979; Gresham and Shimp, 1985) it was hypothesized that this will be true even when prior brand attitude is controlled. In accordance with these assumptions, the following hypotheses were made:
H1: Prior brand attitude will significantly affect (explain a significant portion of the within-subjects variance) Ab under the familiar brands condition.
H2: After controlling for prior brand attitude, Aad will significantly affect Ab under the familiar condition.
H3: Aad will significantly affect Ab under the unfamiliar brands condition.
Given the conflicting evidence reviewed above, it is harder to predict product involvement's role in the Aad-Ab relationship. However, it seems likely that as motivation (product involvement) increases, Aad's influence should decrease.
H4: The Aad-Ab relationship under high product involvement will be significantly weaker (Aad will explain less of the Ab variance) than under low product involvement.
A pretest was designed to determine which product categories were high or low-involvement for college students and which brands were familiar and unfamiliar to them. The pretest was also designed to measure a subject's prior brand attitude.
The pretest was administered four weeks prior to the main experiment. Undergraduate students in an introductory marketing course were told that the test was designed to find out how they felt about various products. The respondents filled out the questionnaire in class. Sixty-five questionnaires were distributed and 63 were returned in useable condition.
The construct of involvement has been conceptualized and operationalized in many different ways. These discrepancies have led to much confusion when trying to interpret results across studies. Salmon (1986) examined the various conceptualizations of involvement in consumer and communication research. He states:
Differences in perspectives on involvement reflect the extent to which involvement is seen primarily as a characteristic of the individual or of the stimulus
In this study product involvement is viewed as an interaction between the individual and the stimulus. It includes the salience or relevance of a product and the individuals interest in a product.
To operationalize product involvement, a modified version of the Personal Involvement Inventory developed by Zaichowsky (1985) was used to differentiate high and low-involvement products. The Personal Involvement Inventory is a 20-item seven-point semantic differential scale designed to measure the involvement construct. The modified version used in this study is a 10-item seven-point semantic differential scale (important-unimportant, of no concern for me-of concern to me, irrelevant-relevant, very meaningful to me-means nothing to me, trivial-fundamental, matters to me-doesn't matter, interesting-not interesting, significant-insignificant, vital-superfluous, boring-exciting). This modified version was selected after studies by Nowak (1986) and Nowak and Salmon (1987) found that the chosen 10 items still produced a reliable measure of involvement and had a higher coefficient alpha that the original instrument.
Using the scale, a product that was of no interest to a respondent would theoretically have a total score of 10 points. Each scale item representing the low extreme receives one point, while items rated at the high extreme receive seven points. The-total score for a product can range from 10 to 70. The mean scores for the product categories used in the pretest are given in Table 1. High involvement products were products that had an average score between 48 and 57; and low involvement products had an average score between 23 and 29. Based on this pretest, 8 product categories were chosen for use in the main experiment.
A repeated measures anova indicated that the lowest high involvement mean (for gasoline) was significantly higher than the highest low involvement mean (for candy bars). [P( 1 ,56)=49.28].
The most straightforward conceptualization of brand familiarity is used here. Baker, Hutchinson, Moore and Nedungai (1986) define brand familiarity as a unidimensional construct that is directly related to the amount of time that has been spent processing information about the brand, regardless of the type or content of the processing that was involved.
The respondents indicated on a single bipolar scale whether they were familiar or unfamiliar with the brand. The brands were selected for the pretest based on the following criteria: There would be at least two brands for each of the product categories in the survey. Brands that did not advertise in this area of the country were included to increase the likelihood that at least one brand in each product category would be unfamiliar the respondents.
Thus the pretest designated eight product categories, four each for high and low involvement products. It also designated two brands for each of the products, producing a total of 16 brands for which ads would be shown in the main experiment.
Prior Brand Attitude
The pretest was also designed to measure the respondent's prior brand attitude. The respondents were asked to complete a four-item seven-point scale (dislike very much - like very much, bad - good, unpleasant - pleasant, worthless - valuable).
A 2 (product involvement level) X 2 (brand familiarity) factorial design was used to test the hypotheses. Each subject saw 24 commercials; 16 of which were test commercials.
The stimulus materials consisted of 24 television commercials, 21 of which were 30-seconds long. All the test ads were 30-seconds long. Of the filler ads, two ran 60 seconds and one ran 15 seconds.
Three stimulus tape orders were prepared to randomize the effects of order. The only deviation from randomization was that brands from the same product category were separated to minimize interference. Each commercial was followed by 10 seconds of black to provide the moderator time to pause the VCR between ads.
Subjects: The subjects were undergraduate students enrolled in an introductory marketing course at the University of Wisconsin-Madison. A total of 63 students participated in the pretest and 58 of these same students also participated in the main experiment. Seven questionnaires were not useable. This left a total of 51 completed questionnaires. Students taking part in the pretest and the main experiment received extra credit points for their participation.
The moderator told the subjects they would be watching commercials and answering questions about them. The first ad was shown, the tape stopped, and the moderator went through the questionnaire instructions with the respondents. After each ad was shown, 90 seconds were allotted for answering the relevant questions and then the next ad was shown. No subject had difficulty responding in this time. The experiment took approximately 50 minutes.
After watching an ad, the subjects completed the section of the questionnaire for the specific ad and brand. The questions involved three categories: questions regarding Aad; Ab; and a manipulation check concerning brand familiarity.
Attitude Toward an Ad
To operationalize Aad, the mean of a 5-item seven-point semantic differential scale was used. The 5-item scale (favorable-unfavorable, boring interesting, dislike very much-like very much, not irritating-irritating, holds attention-does not hold attention) was based on the work of many researchers (e.g. Lutz, MacKenzie and Belch, 1983; Mitchell and Olson, 1981; Gardner, 1985; MacKenzie, Lutz and Belch, 1986). A check on the reliability of the items indicated a Chronbach's alpha =.91
The operationalization of Ab in this study is based on past research (Shimp, 1981; Park and Young, 1986; Gardner, 1985; Gardner, Mitchell and Russo, 1985; Muehling, 1987). Ab is defined as the mean of four seven-point bipolar scales (dislike very much - like very much, bad - good, unpleasant pleasant, worthless - valuable). A check on the reliability of the items indicated a Chronbach's alpha =.82
To test the hypotheses, separate within subjects repeated measures hierarchial regressions were run for high product involvement, low product involvement, familiar brands, and unfamiliar brands. These four regressions were run using Ab as the dependent variable. The basis for this statistical technique can be found in Cohen and Cohen (1983) and Pedhazur (1977,1982).
The design used here involved multiple observations from each subject. With a repeated measures design it is necessary to partition out the variance attributable to a subject's unique bias on the dependent variable (subject effect). To remove this subjects variance, in each regression reported below the mean for each subject was first calculated and then entered as the first independent variable. After the subjects variance was removed it was possible to examine the residual effects of the other variables. Variance attributable to subjects is referred to below as bmean. Therefore, for the regressions with Ab as the dependent variable, the independent variables were bmean, prior brand attitude, and Aad.
Table 2 shows the semipartial (sr2) and partial (pr2) regression coefficients. The semipartial and the partial coefficients indicate the contribution of each independent variable to the multiple correlation ( see Cohen and Cohen, 1983). Figures 1 and 2 show the significant paths demonstrated by the regression results.
Effects of Brand Familiarity
In the pretest, subjects indicated whether they were familiar or unfamiliar with each brand. As a manipulation check subjects were also asked to indicate whether the brand was familiar or unfamiliar in the main experiment. A comparison of the responses indicated that the brands which were identified as unfamiliar/familiar in the pretest were also identified as unfamiliar/familiar in the main experiment.
AMOUNT OF WITHIN SUBJECTS VARIANCE EXPLAINED
AMOUNT OF WITHIN SUBJECTS VARIANCE EXPLAINED
Hypothesis 1 suggested that prior brand attitude would significantly affect Ab when brands were familiar. After subjects variance was removed, prior brand attitude explained over 22 percent of the Ab variance. [F(1,168 =21.35, p<.05]. Hypothesis 1 was therefore supported. Not only was prior brand attitude significant, it was also the best predictor of Ab under the familiar brands condition.
Hypothesis 2 suggested that after controlling for prior brand attitude, Aad will significantly affect Ab under the familiar brand condition. This hypothesis was also supported. After controlling for prior brand attitude, Aad still explained over 12 percent of the remaining Ab variance. [F(1,168)=11.54, p<.05].
Hypothesis 3 suggested that Aad will significantly affect Ab under the unfamiliar brands condition. As expected, Hypothesis 3 was supported. After removing the between subjects variance, Aad explained over 27 percent of the remaining Ab variance for unfamiliar brands. [F(1,168)=23.50, p<.05] .
Effects of Product Involvement
Hypothesis 4 suggested the Aad-Ab relationship under high product involvement will be significantly weaker than under low product involvement. Hypothesis 4 was not supported. Under both high and low product involvement, Aad explains just over 18 percent of the Ab variance. There is not a significant difference. See Figure 2.
The results of this study provide a better understanding of the Aad-Ab relationship under various conditions of brand familiarity and product involvement. This study suggests that Aad can influence Ab for unfamiliar and familiar brands. It also suggests that the Aad's affect on Ab is significant even when one controls for prior brand attitude.
The results also suggest that the level of product involvement does not influence the Aad-Ab relationship.
As the results indicate, Aad affects Ab for familiar and unfamiliar brands. Finding Aad explains a significant portion of the Ab variance for unfamiliar brands is not surprising. This result has been shown often in the literature. However, finding Aad explains a significant portion of the Ab variance for familiar brands, after controlling for prior brand attitude is important. This result contradicts the Machleit and Wilson (1988) study which reported Aad did not significantly affect Ab if prior brand attitude was controlled, and supports the Edell and Burke (1986) results. This result is also consistent with other studies (Batra and Ray, 1986; Edell and Burke, 1984) that found Aad significantly affected Ab for familiar brands, but did not control for prior brand attitude.
Although the results show that Aad affects Ab for familiar and unfamiliar brands, it must be noted that the strength of the Aad-Ab relationship is influenced by brand familiarity. The Aad-Ab relationship was much stronger for unfamiliar brands than for familiar brands. In fact, for familiar brands, prior brand attitude explained more of the Ab variance than did Aad. This makes sense considering that people have more information on which to base their Ab than just the ad when the brand is familiar.
Product involvement did not significantly affect the Aad-Ab relationship. Aad explains approximately an equal amount of the Ab variance for low and high involvement products. This is consistent with the results of Thorson and Page (1989). However, it is inconsistent with reports that the Aad-Ab relationship is stronger for low involvement products than for high involvement products (Rossiter and Percy, 1984).
Limitations of the Study
As with all studies, aspects of the design may limit its generalizability. A first concern is the ecological validity of the viewing situation. Subjects viewed commercials in a large room, with one group as large as 25. Subjects knew they would be answering questions after each commercial they viewed. Obviously the unnatural viewing environment may have altered the way the subjects processed the commercials, perhaps increasing the individual involvement with watching the ads.
Artificially heightened involvement is likely most serious in looking at the effects of product involvement on the Aad-Ab relationship. The laboratory setting may force a deeper processing of commercials for low involvement products than would otherwise be found. If this is the case, then a difference in the Aad-Ab relationship might actually exist for low and high involvement products, but that difference would not be apparent. It is therefore suggested that future research use a more natural viewing environment when examining product involvement role in the Aad-Ab relationship.
Several measurement issues must be addressed. Subjects completed the Aad measures just prior to the Ab measures. Edell and Burke (1984, 1986) speculate that taking similar measures of attitudes proximally may overstate their actual correlation. Future research should separate the measures.
Researchers have found that both brand beliefs and Aad are mediators of Ab (Mitchell and Olson, 1981; Mittal, 1990). Brand beliefs were not investigated in this study. According to Mittal (1990) the size of the Aad-Ab relationship may be overestimated when brand beliefs (both utilitarian and image beliefs) are not included. Future research should include brand beliefs. However, it should be noted that Mittal's results show that the Aad is still a significant predictor of Ab even after controlling for the contribution of brand beliefs.
Another measurement issue concerns the measure of brand familiarity used here. Brand familiarity can be conceptualized and operationalized in many different ways. The most straightforward conceptualization of brand familiarity is used here. Brand familiarity was conceptualized to be unidimensional, and therefore a single-item bipolar scale (unfamiliar - familiar) was used as the index of the concept. However, brand familiarity has also been conceptualized as a more complex concept. Questions have been raised about when someone is truly familiar with an object and whether it may be more useful to conceptualize familiarity as a continuous rather than a dichotomous variable. Using a single-item scale to measure a construct always increases the possibility of not tapping into all the facets of the construct. Although the potential problem is lessened for unidimensional constructs, if familiarity is conceptualized as multidimensional, then the risk involved in using a single-item scale increases tremendously. Therefore, future research should explore the use of multiple item measures of brand familiarity.
Conclusions and Practical Implications
The results of this study may be particularly important for introducing a new brand to the market. For brand introductions, the ad is often the first information about the brand for the consumer, and is very important to help insure the consumer will form a favorable Ab. This study suggests that one way to increase the probability of obtaining a favorable Ab is by having ads that elicit a favorable Aad. Therefore, measures of Aad should be included when pretesting commercials.
Although the study indicates that the Aad-Ab relationship is considerably weaker for familiar brands than unfamiliar brands, it is still a significant relationship. That is important because most of the advertisements which consumers are exposed to are for existing brands (Edell and Burke, 1987).
Baker, W., J. W. Hutchinson, D. Moore and P. Nedungadi (1986) Brand Familiarity and Advertising: Effects on the Evoked Set and Brand Preference. Advances in Consumer Research, Vol. 9, Ann Arbor: Association for Consumer Research, pp. 637-642.
Batra, R. & Ray, M.L. (1985) How Advertising Works on Contact. In Psychological Processes and Advertising Effects: Theory and Research Applications, L.F. Alwitt & A.A. Mitchell (Eds.), Hillsdale, N.J.: Erlbaum, pp. 1344.
Batra, R. & Ray, M.L. (1986) Affective Responses Mediating Acceptance of Advertising. Journal of Consumer Research, Vol.13, pp. 234-249.
Cohen, J. & Cohen, P. (1983) Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. 2nd edition. Hillsdale, New Jersey: Lawrence Erlbaum Associates
Cox, D.S. & Locander, W.B. (1987) Product Novelty: Does it Moderate the Relationship Between Ad Attitudes and Brand Attitudes? Journal of Advertising, Vol. 16(3), pp. 39-44.
Edell, J.A. & Burke, M.C. (1984) The Moderating Effect of Attitude Toward an Ad on Ad Effectiveness Under Different Processing Conditions. Advances in Consumer Research, Vol. 11, Ann Arbor: Association for Consumer Research pp. 644-649.
Edell, J.A. & Burke, M.C. (1986) The Relative Impact of Prior Brand Attitude and Attitude Toward the Ad on Brand Attitude After Ad Exposure. In Advertising and Consumer Psychology, Jerry Olson & Keith Stentis, (eds), Vol. 3, pp. 93-107.
Gardner, M.P. (1985) Does Attitude Toward the Ad Affect Brand Attitude Under a Brand Evaluation Set?. Journal of Marketing Research, Vol. XXII, pp.192-198.
Gardner, M.P., Mitchell, A.A., & Russo, J.E. (1985) Low Involvement Strategies for Processing Advertisements. Journal of Advertising, Vol. 14(2), pp. 4-12.
Gresham, L.G. & Shimp, T.A. (1985) Attitude Toward the Advertisement and Brand Attitudes: A Classical Conditioning Perspective. Journal of Advertising, Vol. 14(1), pp. 10-17.
Laczniak, R.N., Muehling, D.D. and Grossbart, S. (1989) Manipulating Message Involvement in Advertising Research. Journal of Advertising, Vol. 18(2), pp. 28-38.
Lutz, R.J. & Belch, G.E. (1983) Attitude Toward the Ad as a Mediator of Advertising Effectiveness: Determinants and Consequences. Advances in Consumer Research, Vol. 10, Ann Arbor: Association for Consumer Research pp. 532-539.
Machleit, K.A. and Wilson, R.D. (1988) Emotional Feelings and Attitude Toward the Advertisement: The Roles of Brand Familiarity and Repetition. Journal of Advertising, Vol. 17(3), pp. 27-35.
MacKenzie, S.B., Lutz, R.J., & Belch, G.E. (1986) The Role of Attitude Toward the Ad as a Mediator of Advertising Effectiveness: A Test of Competing Explanations. Journal of Marketing Research, Vol. XXIII, pp. 130-143.
Messmer, D.J. (1979) Repetition and Attitudinal Discrepancy Effects- on the Affective Response to Television Advertising. Journal of Business Research, Vol.7, pp. 75-93.
Mitchell, A.A. & Olson, J.C. (1981) Are Product Attribute Beliefs the Only Mediator of Advertising Effects on Brand Attitude?. Journal of Marketing Research, Vol. XVIII, pp. 318-332.
Mittal, B. (1990) The Relative Roles of Brand Beliefs and Attitude Toward the Ad as Mediators of Brand Attitude: A Second Look. Journal of Marketing Research, Vol. XXVII, pp. 209-219.
Moore, D.L. & Hutchinson, J.W. (1983) The Effects of Ad Affect on Advertising Effectiveness. Advances in Consumer Research, Vol. 10, Ann Arbor: Association for Consumer Research pp. 526-531.
Muehling, D.D. (1987) An Investigation of Factors Underlying Attitude-Toward-Advertising-In-General. Journal of Advertising, Vol. 16(1) pp. 32-40.
Muehling, D.D. (1987) Comparative Advertising: The Influence of Attitude-Toward-The-Ad on Brand Evaluation. Journal of Advertising, Vol. 16(4), pp. 4349.
Muehling, D.D. & Laczniak, R.N. (1988) Advertising's Immediate and Delayed Influence on Brand Attitudes: Considerations Across Message-Involvement Levels. Journal of Advertising, Vol. 17(4), pp. 23-34.
Nowak, G. (1986) The Effects of Product Involvement, Message Appeal, and Viewing Condition on Memory and Evaluation of Television Commercials. Unpublished Masters Thesis at the University of Wisconsin-Madison.
Nowak, G. & Salmon, C. (1987) Measuring Involvement with Social Issues. Paper presented at the Association for Education in Journalism and Mass Communication, San Antonio.
Park, C.W. & Young, S.M. (1983) Types and Levels of Involvement and Brand Attitude Formation. Advances in Consumer Research, Vol. 10, Ann Arbor Association for Consumer Research pp. 320-323.
Park, C.W. & Young, S.M. (1986) Consumer Response to Television Commercials: The Impact of Involvement and Background Music on Brand Attitude Formation. Journal of Marketing Research, Vol. XXIII, pp. 11-24.
Pedhazur, E.J. (1977) Coding Subjects in Repeated Measures Designs. Psychological Bulletin, Vol. 84, pp. 298-305.
Pedhazur, E.J. (1982) Multiple Regression in Behavioral Research: Explanation and Prediction. Second edition. New York: CBS College Publishing.
Rossiter, J.R. & Percy, L. (1984) Advertising Communication Models. Advances in Consumer Research, Vol. 12, Ann Arbor: Association for Consumer Research, pp. 510-524.
Salmon, C.T. (1986) Perspectives on Involvement in Consumer and Communication Research. In Progress in Communication Sciences, Brenda Dervin & Melvin J. Voight (eds). Ablex Publishing Corporation, Norwood, New Jersey. pp. 243-269.
Shimp, T. (1981) Attitude Toward the Ad as a Mediator of Consumer Brand Choice. Journal of Advertising, Vol. 10(2), pp. 9-15.
Shimp, T. & Yokum, J.T. (1982) Advertising Inputs and Psychophysical Judgments in Vending-Machine Retailing. Journal of Retailing, Vol. 58(1), pp. 95-113.
Thorson, E. and Page, T.J. (1990) On the Ubiquity of Aad-->Ab Effects. Paper presented at the Annual Meeting of the American Academy of Advertising, Orlando, Florida.
Zaichowsky, J.L. (1985) Measuring the Involvement Construct. Journal of Consumer Research, Vol.12(3) pp. 341-352.