,1 Llllu.,:.l. ..(\58 f it} ......o$u§: ll ll “H“ j 12 3 LlllllllllllljllllllllljwflllL ‘ : THES'. This is to certify that the thesis entitled . THE COMMUNICATIONS EFFECTIVENESS OF COMPARATIVE ADVERTISING UNDER CONDITIONS OF PRODUCT INVOLVEMENT AND BRAND USAGE presented by MARK BECKER TRAYLOR has been accepted towards fulfillment of the requirements for PH.D. degreein MARKETING 5. TRANSPORTA- TION Ma or professor Date 5‘18‘79 9-7539 MN 1 3 1734 __ ‘ arm v as 1" ‘N‘ 1‘ W 53344! ~ OVERDUE FINES ARE 25¢ PER DAY . PER ITEM Return to book drop to remove ‘ this checkout frOm your record. Q “”9433 Pure? "'89 ‘2’, Q U A i)” 2 ? #31? 6 1995 is. \ . “7» ififlflm mm I’, 9 " " Wig-{5‘3 SEP 2.51996 LE8 n 21995 M5991, h THE COMMUNICATIONS EFFECTIVENESS 0F COMPARATIVE ADVERTISING UNDER CONDITIONS OF PRODUCT INVOLVEMENT AND BRAND USAGE By Mark B. Traylor A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Marketing and Transportation Administration 1979 ABSTRACT COMMUNICATIONS EFFECTIVENESS OF COMPARATIVE ADVERTISING UNDER CONDITIONS OF PRODUCT INVOLVEMENT AND BRAND USAGE By Mark B. Traylor This study was an experiment that examined recall and attitude toward one's own brand for comparative, Brand X, and noncomparative advertising. The market context for the study was that a new, unknown brand was entering a mature product market in which a well- recognized brand could be "attacked" in the comparative advertisement. Print advertisements for a fictitious brand of automobile (a relatively high involvement product) and a fictitious brand of non- diet cola soft drink (a relatively low involvement product) were shown to l20 subjects drawn from a residential area of East Lansing, Michigan. Half of those who saw the automobile ads drove an Oldsmobile which the comparative automobile ad attacked; and half of those who saw the cola ads drank Coca-Cola most often which the comparative cola ad attacked. Measurements of the subjects' attitudes toward their own brands were taken in three time periods-~a pretest, an immediate posttest, and a delayed posttest--to examine attitude change over time. Mark B. Traylor Besides analyzing results based on an a priori classificatiorl of advertisement type and product involvement, subjects reported any comparative type advertisement they saw and rated the level of involvement they had with the product advertised. Thus, the analysis; was carried out on a perceptual, subject-defined basis as well as on an a priori, researcher-defined basis of what constituted a compara- tive advertisement and a high involvement product. Results indicate that comparative advertising--regardless of who defines it--is more effective from a recall standpoint, but not in attitude change. Among the subjects who used Coca-Cola or drove an Oldsmobile, the effectiveness of the comparative advertisement was particularly striking when measured by their ability to recall the content of the advertisement if they recognized that the compari- son was taking place. Based on the measures of communications effectiveness used in ‘the study, comparative advertising was consistently the most eaffective or was equally effective relative to Brand X and non- <:omparative advertising. Thus, it may be an effective competitive tool for new or poorly known brands in mature product markets and for low market share brands generally when there is at least one well-recognized brand already in the market. DEDICATION To my parents ii ACKNOWLEGMENTS I owe much appreciation to Dr. Gil Harrell, chairman of my dissertation committee and to Dr. Terry Allen and Dr. Don Taylor, my other two committee members. Their comments resulted in a dissertation that is a substantial improvement over anything I could have produced on my own. iii TABLE OF CONTENTS LIST OF TABLES . LIST OF FIGURES LIST OF APPENDICES Chapter I. INTRODUCTION AND OBJECTIVES II. LITERATURE REVIEW Comparative Advertising. Attitude and Learning Persuasive Communications and Attitude Change Cognitive Dissonance . Social Judgment-Involvement Approach. Involvement. . . Summary of Issues and Substantive Hypotheses . III. RESEARCH DESIGN Basic Design . Measurement of Dependent Variables Recall . Attitude Selection of Products and Brands . Development of Advertisements . Data Collection . . Coding Reclassification of Advertisements and Involvement . Advertisements Product Involvement Reclassification . Statistical Procedures Using the Respondent Classification . . . . . . iv Page vii xi xii Research Hypotheses . Brand Name Recall . Content Recall . Attitude Change Selective Recall Response Rates FINDINGS Sample Characteristics . . Advertisement Recall and Name Recall By Advertisement Type . . . A_Priori Classification . Respondent Classification DeLorian and Renault. . Discussion and Summary of Brand Name Recall Results Confusion of Target Brand with Sponsoring Brand : Advertisement and Brand Name Recall By Usage of the Target Brand . Brand Name and Advertisement Recall By Product Involvement. A Priori Classification . Respondent Classification . Summary of Brand Name Recall Findings Content Recall A Priori Classification of Advertisement Types and Product Involvement . . . . Respondent Classification . Relationship of Content Recall Findings to Earlier Studies . . . Summary and Implications of Content Recall Findings . . Attitude Change . Prior Brand Attitude and Content Recall SUMMARY AND IMPLICATIONS Summary of the Methodology Summary of Results . 58 58 58 64 64 64 72 82 84 84 87 91 91 92 Chapter Page Managerial Implications . . . . . . . . . 95 Needs For Future Research . . . . . . . . . 97 Conclusion . . . . . . . . . . . . . . 99 BIBLIOGRAPHY . . . . . . . . . . . . . . . . lOO APPENDICES . . . . . . . . . . . . . . . . . 106 vi Table 10. ll. 12. l3. 14. LIST OF TABLES Advertisements Perceived as Comparative by A Priori Classification . . . . . Chi- -Square Analysis of Involvement Score by Type of Product . . . T-test of Product Involvement Score by Product Brand Name Recall by Advertisement Type in the Time 2 Immediate Posttest . . Brand Name Recall by Advertisement Type in the Time 3 Delayed Posttest. . . . Brand Name Recall by Advertisement Type Using the Respondent Classification in the Time 2 Immediate Posttest . . . . . . . . . . Brand Name Recall by Advertisement Type Using the Respondent Classification . Brand Name Recall by Usage of Target Brand in the Time 2 Immediate Posttest . . Brand Name Recall by Usage of Target Brand in the Time 3 Delayed Posttest . . . . Brand Name Recall by Product in the Time 2 Immediate Posttest . Brand Name Recall by Product in the Time 3 Delayed Posttest . . Brand Name Recall by Involvement Score in the Time 2 Immediate Posttest . . . . . . Brand Name Recall by Involvement Score in the Time 3 Delayed Posttest . . . . ANOVA Table For Content Recall Using Time Period as a Factor . vii Page 42 45 45 52 53 54 55 59 59 60 6O 62 62 65 Table 15. 16. 17. 18. 19. 20. 21. 22. ANOVA Table For Content Recall in the Time 2 Immediate Posttest. . . . ANOVA Table For Content Recall in the Time 3 Delayed Posttest . . . ANOVA Table For Content Recall Using Time Period as a Factor: Respondent Classification . . . ANOVA Table For Content Recall in the Time 2 Immediate Posttest Based on the Respondent Classi- fications of Advertisement Type and Product Involvement . . . . . . . ANOVA Table For Content Recall in the Time 3 Delayed Posttest Based on the Respondent Classifications of Advertisement Type and Product Involvement . ANOVA Table For Brand Attitudes Using Time Period as a Factor. A Priori Classification ANOVA Table for Brand Attitudes Using Time Period as a Factor. Respondent Classification . Prior Brand Attitude and Content Recall Correlations by Product Advertised . . . . . . . . Prior Brand Attitude and Content Recall Correlations by Product Involvement . . . . . . Data Matrix for Dependent Variable within a Time Period . . . . . . . . . . Data Matrix for Perceptual Classifications within a Time Period . . . . . . Response to Time 1 Time 1 Questionnaires Returned Completion of Interviews in Times 2 and 3 . Reasons for not Completing the Time 2 Interview . Final Sample Size . Marital Status of Sample viii Page 66 67 73 74 75 86 88 89 89 147 148 150 150 150 151 151 151 F-10. F-12. Occupation of Sample . Number in Household for Sample Ages of Sample Respondents . Education of Sample Respondents Sex of Sample Respondents Race of Sample Respondents . Advertisement Recall by Advertisement Type-~Time 2 . Advertisement Recall by Advertisement Type--Time 3 . Brand Name Recall by Advertisement Type--Time 2 . Brand Name Recall By Advertisement Type--Time 3 . Advertisement Recall by Advertisement Type Based on Respondent C1assification--Time 2.. . Advertisement Recall by Advertisement Type Based on Respondent C1assification--Time 3 . . . . Brand Name Recall by Advertisement Type Based on Respondent Classification--Time 2 . . . Brand Name Recall by Advertisement Type Based on Respondent C1assification--Time 3 . . . Brand Name Recall by Usage of Target Brand Among Subjects Who Saw a Comparative Advertisement--Time 2 n = 40 . . . . . . Brand Name Recall by Usage of Target Brand Among Subjects Who Saw a Comparative Advertisement-- Time 3 (n = 40). . . . . Brand Name Recall by Usage of Target Brand Among Subjects Who Reported Seeing a Comparative Advertisement--Time 2 . . . . . . Name Recall by Usage of Target Brand Among Subjects Who Reported Seeing a Comparative Advertisement-- Time 3.. . . . . . . . . . . ix Page 152 152 153 153 154 154 156 156 156 157 157 157 158 158 158 159 159 160 Advertisement Recall by Product--Time 2 Advertisement Recall by Product--Time 3 Brand Name Recall by Product-Time 2 . Brand Name Recall by Product--Time 3 Advertisement Recall by Involvement Leve1--Time 2 Advertisement Recall by Involvement Level--Time 3 Brand Name Recall by Involvement Level--Time 2 Brand Name Recall by Involvement Leve1--Time 3 Cell Means for Content Recall-~Time 2 and Time 3 Cell Means for Content Recall Time 2 and Time 3 using the Respondent Classification . Cell Means for Summed Brand Attitude Scores Cell Means for Summed Brand Attitude Scores: Respondent Classification Cell Means Using Difference Scores for Change in Attitude Toward One's Brand . . . Cell Means Using Difference Scores for Change in Attitude Toward One's Brand-~Respondents Classifi- cation . . . . . . . . . . . Correlation Matrix for Items Measuring Attitude Toward One's Own Brand-~Time l Pretest Correlation Matrix for Items Measuring Attitude Toward One's Own Brand--Time 2 Immediate Posttest Correlation Matrix for Items Measuring Attitude Toward One's Own Brand--Time 3 Delayed Posttest Page 160 160 161 161 161 162 162 162 164 165 166 167 168 169 171 171 172 Figure LIST OF FIGURES Time Period X Product Interaction . Product Advertisement Interaction in the Time 2 Immediate Posttest Advertisement Type X Target Brand Usage Interaction for the Two Time Periods Combined Using the Respondent Classification of Advertisement Type Advertisement Type X Target Brand Usage Interaction in the Time 2 Immediate Posttest for the Respondent Classification of Advertisement Type Advertisement Type X Target Brand Usage Interaction in the Time 3 Delayed Posttest for the Respondent Classification of Advertisement Type . Brand Usage X Involvement Interaction in the Time 2 Immediate Posttest . Brand Usage X Involvement Interaction in the Time 3 Delayed Posttest . xi Page 69 70 76 77 77 81 81 LIST OF APPENDICES Appendix Page A. TEST ADVERTISEMENTS . . . . . . . . . . . 107 B ITEMS FOR CODING CONTENT RECALL . . . . . . . 114 C. QUESTIONNAIRES . . . . . . . . . . . . 124 D DATA MATRICES . . . . . . . . . . . . . 146 E. TABLES FOR RESPONSE RATES AND DEMOGRAPHIC CHARACTERISTICS OF THE SAMPLE . . . . . . . 149 F. ADDITIONAL CHI-SQUARE TABLES OF BRAND NAME RECALL . 155 G. CELL STATISTICS USED IN THE ANOVA TABLES FOR CONTENT RECALL AND ATTITUDE CHANGE. . . . . . 163 H. CORRELATION MATRICES FOR BRAND ATTITUDE ITEMS . . 170 xii CHAPTER I INTRODUCTION AND OBJECTIVES Comparative advertising compares two or more recognizable brands on one or more attributes. While not a new form of advertising, it was seen infrequently before the Federal Trade Commission's encouragement of its use around 1973. Casual observation reveals a spread of comparative advertising from analgesics and a few other product classes shortly after the FTC's support to photocopying machines, automobiles, spray furniture polishes, wine, mini-computers, liquid bleaches, deodorants, and many other products more recently. Despite the increase in comparative advertising, very little published research has reported on its effectiveness. One suspects that, in the course of advertising research and copy testing carried on by advertising agencies and research suppliers, a substantial bank of proprietary data has been built up. That is, the use of a compara- tive advertisement is likely to be based on some evidence accumulated by the advertiser or its agency. Unfortunately, these reports have not been published and, in fact, very little attention in the market- ing literature has been paid to the communications effectiven85§ 9f M. 6.9111923?“ V1.99 vertjsing. The main objective of this research was to test the communica- tions effectiveness of a comparative form of advertising against Brand X or to noncomparative forms of advertising in the context of a new, unknown brand's entering a mature product market. The communica- tions effectiveness of comparative advertising has not been examined in this light; nevertheless, this form of advertising could offer a marketing strategist particular benefits of building awareness, positioning his brand, and establishing a brand concept very quickly by relating the new brand explicitly to those already in the market. The research also offers an opportunity to test one of the bases of the FTC's position: comparative advertising conveys more information to consumers, so more comparative advertisements will mean better_informed‘cpnsumers. Other objectives of this research will be to examine differ- ences in comparative advertising's effectiveness for high and low involvement products; and for users or nonusers of the brand that is being compared. Finally, attention will be paid to the issue of 5919931¥9-1?9E9199 19 an Advertising context. CHAPTER II LITERATURE REVIEW The literature reviewed in this section is organized under four major headings: Comparative Advertising; Attitude_and Learning; Persuasive Communication and Attitude Change; and Involvement. The research will address itself to issues raised within each of these categories. These issues will be discussed in some detail in the literature review and then will be restated in summary form after- wards. Comparative Advertising Several recent articles have treated comparative advertising in an issue-related, non-empirical manner. Chevins (1975) cited a history of comparative advertising and offered several speculations regarding its effectiveness and how it might be used by advertisers. Barry and Tremblay (1975) cite issues raised among ad practitioners, including speculation that comparative advertising gives competitors free time, solidifies the position of the market leader, that consumers are unable to determine the source of the message correctly (i.e., the sponsoring brand), and that the success of the message is a function of source credibility. They also ran a survey in the Dallas area and found that 84 percent of their 3 respondents reported having seen a comparative ad and could give an example. Deodorants, headache remedies, autos, and furniture polish were the examples given most frequently. Wilkie~andwfgnzi§MSI975) discuss comparative advertising in a general sense and from the point of view of regulators, media (especially the three commercial networks), and advertisers. For the last, they discuss tactical questions in terms of effectiveness, building brand awareness, and comprehension. Most importantly, they use the issues developed in their tactical considerations to generate a set of hypotheses to be tested empirically. These are based in the three broad areas of attention, comprehension, and message retention. The substantive hypotheses A and B in this research are adaptations - of two of their hypotheses: Comparison ads will receive more aiigfllionfifrom users of competing brands mentioned than from users of brands not mentioned. Aggregate recall levels of comparison ads will be higher than those for standard appeals (p. 12). Nilkie and Farris derive the first hypothesis from the concepts of selective attention and selective exposure (p. 11). Brand usage as a potentially distinguishing factor in audience response will be elaborated further. Ulanoff (1975) discusses comparative advertising from an historical perspective. He found that comparative advertising was used for razor strops, morning gowns, and other products at least as far back as 1710 (pp. 8, 10). Two empirical studies on comparative advertising have been published recently and were used to develop some of the issues addressed in this research; both relate specifically to the communications-effectiveness of comparative ads. In a laboratory experiment, Prasad (1976) used a durable consumer product (movie camera) a well-known target brand (Kodak), a fictitious sponsoring brand (Ronar) and a sample of 202 students. He tested four hypotheses and found clear evidence for one. The hypothesis that was supported was: The perceived credibility of the claim of a comparative advertisement will be lower among consumers who prefer the competitors brand (i.e., target brand) explicitly named in the advertisement than among others (i.e., other 55 who did not state their brand preference for Kodak) (p. 130). Prasad measured recall by both brand name recall and content recall and found no relationship among S's stated brand preference (Kodak or not Kodak), type#gfifladfi(comparative vs. Brand X advertis- ing), and either of the recall measures. Nor did he find evidence for a concern in the advertising industry that a comparative advertiser could "'trade on the reputa- tion'" of the target brand, thereby enhancing consumers' perceptions of the competitive position of his own brand (p. 130). Prasad found mixed support for his hypothesis that 55 would show higher recall for the comparative ad. This hypothesis was supported using the content recall measure, but not for the brand name recall measure. And these results held in both the immediate posttest and a delayed (one week) posttest. 6 /~- The-second empirical study on comparative advertising 15*«“ $6 Ogilvy and Mather study using TV ads and reported by Levine (1976). The products tested were a beverage, four health and beauty aids, two drug products and a household product. Comparative ads for these products were evaluated against noncomparative ads. Respondents held more negative attitudes toward the comparative ads and, by self-report, found them more confusing and less believable than the noncomparative ads. The comparative ads yielded no greater awareness of the sponsoring brand and, in one case, showed less awareness than the noncomparative advertise- ment. There was greater sponsoring brand misidentification for the comparative ads. A comparative ad was found to be more persuasive (as measured by brand choice) when it was shown among noncomparative ads and Levine suggests this may be the result of noveltyfieffects. Surprisingly, awareness was not significantly greater for comparative ads under these conditions. Levine concludes from his study that: . . there is little to be gained from this type of advertising for the advertising industry, the advertiser, or the consumer. The only one who may benefit is the competitor who is named in the advertising . . . (p. 14). As was the case with Prasad's study, some question remains regarding the generalizability of these findings to other products and media. Levine himself points out this limitation. There is also some suspicion, based on the argumentative tone of the findings, that Ogilvy and Mather may have had something at stake in the outcome of the research. Nevertheless, the empirical research conducted by Prasad and by Ogilvy and Mather has not shown comparative advertising to be particularly effective. Some of its effects were also found to be detrimental to the advertiser. Neither of these studies, however, consciously approached comparative advertising from the standpoint of a new brand entering a mature product market. Comparative advertising offers considerable potential for building awareness and positioning a new brand relative to one or more market leaders (Ulanoff, 1974; Wilkie and Farris, 1975). The basic tactic is relating and contrasting something unfamiliar to something familiar thus building a brand concept in a direct and efficient way. While Prasad used a figtjtgus_(and, therefore, "new") brand name againsta market leader, neither his study nor the Ogilvy and Mather study attacked the question from precisely this stand- point. While both studies have made important contributions to the understanding of comparative advertising, a number of important empirical issues yet remain. About midway through the data collection for this research, two more empirical studies on comparative advertising were published. While they arrived too late to be used in the develop- ment of research issues and hypotheses, they reinforce the notion that additional research on this topic is needed. Jain & Hackleman (1978) used twelve products that they categorized into three goods classifications--convenience, shopping, and specialty--and tested brand name recall for comparative versus noncomparative advertisements. All the brand names used in the study were fictional which obviated any differfiflfjal effects of novelty, familiarity, usage, or brand loyalty. They found, in \‘H _. w contrast to Prasad and Levine, that "brand names appearing in comparison ads were recalled better immediately" but not in a 24 hour delayed posttest (p. 24). These brand names were both_the sponsoring and target brands and they concluded that "comparative advertising would work best for less-well-known brands" (p. 24). The comparative adveFtiggagntgwforconvenience and specialty goods were more effective than for shopping goods. They did not qualify this finding with the caveat that other components of the advertise- ment--graphics, copy, headline, etc.--might also account for differences in addition to the product class or type of good being advertised. A more ambitious study was reported by Shimp and Dyer (1978) who controlled for positioning a fictitious fast food brand (Big Steer) against a market leader (McDonald's) and a less dominant brand (Burger King). Their findings contradicted Prasad's since they found that copy recall was greater for a noncomparative than for a comparative form of advertising. The sponsoring brand was correctly identified more often for the comparative than for the noncomparative advertisement, although the difference was not statistically significant. However, this finding contradicts that of Levine (1976) who found that there was greater misidentification for the comparative advertise- ment. Prasad did not report the direction of his findings--only that there was no statistically significant greater correct brand name recall for the sponsoring brand among subjects exposed to the comparative advertisement. The noncomparative advertisements, it was also found, resulted in greater credibility, were reported as being more informative, and were viewed more favorably. The comparative advertisements were judged as more interesting. Attitude and Learning The relationship between a person's prior attitude toward a topic and his subsequent learning of material related to that topic is a complex one. An early study among pro and anti-communists (Levine and Murphy, 1943) apparently established that subjects will learn information that is consistent with their attitudes better than information that is discrepant. Forgetting of the discrepant information over time, it was found, was also greater than for the consistent information. Partly, perhaps, because these findings made so much intuitive sense, social psychologists accepted this relationship between attitude and learning and subsequent studies focused on variations of the basic theme of selective learning. In a study among thirty black and thirty white juvenile delinquents, Taft (1954) administered a communication containing 10 information that was favorable, unfavorable, ambiguous, and neutral toward blacks. He found that the black 55 recalled both the favorable and unfavorable material better than white 55 in an immediate recall. But in a delayed (3 day) recall measure, the black subjects recalled less of the racially unfavorable material than white Ss while recalling more of the favorable material. Since the material was highly ego-involving for the blaCks, Taft hypothesized that the black 55 learned all_the material better than the whites at the outset, but then repressed the unfavorable material as time passed. Neutral statements in the communication were recalled least by both racial groups. Taft's immediate findings contradict the Levine and Murphy findings of selective learning since the pro and anti-communist communication used by Levine and Murphy was very likely highly ego-involving for their subjects as well. This was not noted specifically by Taft, although he did state that one cannot predict " . . . for any group of Ss . . . whether positive and negative ego involvement will lead to sensitization or repression (of information)" (p. 27). Jones and Aneshansel (1946) tested and found support for the hypothesis that 55 will actually learn discrepant information better than 55 to whom the same information is "covaluant" (i.e., consistent with their values) if they know they will later be asked to formulate counterarguments. They concluded that it is easier to learn contravaluant (i.e., discrepant) material in high 11 motivation settings and that persons ". . . tend to be autistic in the retention of congenial material under conditions of low motivation . . ." (p. 31). They further stated that: The assumption that we . . . learn statements that we agree with better than statements with which we disagree must be placed in the broader context of the learner's over-all purpose in the task (p. 32). And: When the contravaluant material is given a new utility for the learner, the wish that is otherwise autistically pampered is supplanted by more powerful motives of ego enhancement (p. 31). This, of course, modifies the Levine and Murphy findings by introducing a task (counterargument) that requires absorbing discrepent information. Jones and Kohler (1958), in two separate experiments on racial segregation, found "striking" evidence that persons holding extreme positions not only learn plausible statements supporting their view, but implausible (outlandish and ridiculously absurd) statements against their views. Persons with neutral attitudes on the subject matter learn plausible and implausible statements about equally well. So it would seem that this study too modified the Levine and Murphy finding that persons learn information consistent with their attitudes better than discrepant information. Jones and Kohler explain it this way: 12 While the results show that learning of a controversial statement cannot be predicted solely from the direction of the argument, they still support the functionalists' basic assumption that cognitive processes operate . . to promote the constancy of attitide and belief (p. 32D). So it is not so much the "objective" position advocated by the communication that determines learning as it is the receiver's perception of that position. This perception, which is itself a function of attitude, (Hovland, Harvey, and Sherif, 1957) thus assumes a mediating role in the relationship. Nevertheless the basic relationship, modified in perceptual and functionalist terms, is still consistent with the original selective learning phenomenon of Levine and Murphy. While ego involvement, motivation to learn, and plausibility of the discrepant information may modify the relationship, the basic attitude-learning model still holds. Even this relationship, however, was seriously questioned by Waly and Cook (1966) who replicated the Jones and Kohler study for purposes of developing attitude measures on race. When they failed to duplicate the Jones and Kohler results, Waly and Cook ran two more experiments and still failed to confirm the findings. In two of the three studies, Waly and Cook also failed to confirm the Levine and Murphy findings. Waly and Cook's major finding was that statements evoked equal learning difficulty regardless of prior attitudes. In trying to explain the discrepancies among the studies, Waly and Cook also found no consistent relationship between prior familiarity and 13 ease of learning. This was rather surprising in itself, since it seems intuitively reasonable that familiarity should positively affect recall. Another blow to the original attitude-learning relationship came with a set of three studies conducted by Greenwald and Sakumura (1967) who found that S's attitudes did not affect his recall of statements on the Vietnam War. They also found, like Waly and Cook, no relationship between S‘s familiarity with a statement and his ability to recall it correctly subsequently. However, regardless of S's prior attitude, learning increased for "novel" information-—in this case, the more novel anti-war state- ments were recalled better by both anti-war and pro-war 55 than were pro-war statements. This too is a surprising finding, since one would expect that familiarity and novelty would be direct opposites of each other and thus two measures of the same attribute. But in a posttest check using a separate sample, Greenwald and Sakumura found an r of -.73 between familiarity and novelty which yields an r2 of .52 (p. 397). Thus, they concluded variations in one account for only about half the variation in the other. Empirically, novelty and familiarity seem to be two different components of a communication. Novelty as an aid in learning is also supported by work on the Von Restorff Phenomenon (Wallace, 1965) as Greenwald and Sakumura (1967) point out. The Von Restorff Phenomenon states that "isolating an item against a crowded or homogeneous background facilitates the learning of that isolated item" (Wallace, 1965, p. 410). 14 Finally, the absence of a significant relationship between familiarity and recall is indirectly supportive of work reported by Krugman (1972; 1975) that repeated exposures of the same advertisement, beyond two or three, results in a detachment of the audience and a withdrawal of their attention. To the extent that learning is a necessary (but not sufficient) condition of attitude change, these results are also supportive of Winter's (1973) findings of decreasing attitude change as a function of exposure to advertising. Persuasive Communications and Attitude Change Although it has been suggested that comparative advertising is based more on cognitive dissonance than social judgment-involve- ment (Barry and Tremblay, 1975), this seems a somewhat specious argument since it is possible to borrow from either theory and since one can easily entertain the two notions simultaneously. But the failure to confirm a positive relationship between prior attitude and subsequent learning of information (as reported in the previous section) bears disturbing implications for one of the main correlates of cognitive dissonance theory, especially as it relates to advertising. Cognitive Dissonance Cognitive dissonance theory states that when information is received that is incompatible with a person's cognitive set-- his beliefs and information about the world--a state of psychological 15 discomfort will exist. The individual will engage in behavior intended to reduce the discomfort and re-establish consistency in his cognitive system. This behavior can take several forms, such as screening out discrepant information, distorting the information to make it consonant with his existing attitudes, or changing his existing attitudes. In the more recent studies conducted on attitude and learn- ing (Waly and Cook, 1966; Greenwald and Sakumura, 1967), the depen- dent measure was a range of correct recall of the information--a measure that should have picked up screening and distoring behavior among the 55 receiving dissonant material. However, it did not. Other specific exceptions in the theory of cognitive dissonance have been reported including voluntary exposure to dissonant information that is novel (Sears, 1965). The Sears finding is consistent with the Greenwald and Sakumura finding on the relationship between novelty and learning. A review of the selective exposure literature (Freedman, 1965) indicates the empirical tests have not been conclusively in support. There is just as much evidence that individuals prefer non-supportive information, depending upon the situation. For example, the usefulness of the infbrmation may override any discrepancy, as when a staunch environmentalist reads oil company literature in order to counterargue more effectively. Thus, preference may not be exhibited in exposure, but in the individual's evaluation of the material. 16 Social Judgment-Involvement Approach The social judgment-involvement approach makes predictions that are more precise than cognitive dissonance theory (Sherif, Sherif and Nebergall, 1965). Cognitive dissonance theory always allows attitude change toward a discrepant communication as a possibility; the social judgment-involvement approach specifies conditions in which such an attitude change cannot be a consequence: when the position advocated in the message falls within the individual's latitude of rejection. Assimilation-contrast theory grew out of observations of a "boomerang effect" in communications studies. Hovland and Pritzker (1957) found a direct relationship between the amount of attitude change advocated in a communication and the amount of change in the direction advocated for lower-involvement issues communicated by highly respected sources. So the greater the discrepancy, the greater the attitude change. The attitude change taking place among the subjects increased, but at a decreasing rate. It was also found that 5-8% of the responses in each amount-of—change-advocated category were in a negative direction, i.e., opposite that advocated in the communication. This is the "boomerang effect." Two studies conducted almost simultaneously and published shortly after this one began refining the concept of assimilation- contrast . 17 In a study in which 55 evaluated the weights of stimuli relative to an "anchoring" stimulus (i.e., reference point), Sherif, Taub, and Hovland (1958) found that: . . . introduction of anchoring stimuli immediately adjacent to the stimuli being judged will cause displace- ment of judgments in the direction of the anchor. This would constitute an "assimilation effect." Anchors considerably beyond the stimulus range . . . will produce a displacement of judgments away from the anchor ("contrast effect") (p. 154). In other words, S perceived discrepant information as being closer to his reference point than it really was, as long as it was not tpp_discrepant. If the stimulus was discrepant beyond a certain limit, it was perceived as being farther from his anchor than it really was. Hovland, Harvey, and Sherif (1957) obtained supportive findings for the assimilation-contrast effect in a study on prohibition--a controversial topic where the research was carried out--and introduced the terms "latitude of acceptance" and "latitude of rejection" (p. 250). The latitude of acceptance consisted of ". . . S's own position and other acceptable positions" and the latitude of rejection consisted ". . . of all unacceptable positions" (p. 250). If the communication was within the subject's latitude of acceptance, he: 1) perceived the message to be closer to his own position than it "really" was; 2) judged the communication to be "fair and factual" (p. 251), and 3) moved toward the position advocated by the communication. If the communication was within 18 the subject's latitude of rejection, he: 1) perceived the message as farther from his position than it "really" was; 2) judged it as "propagandistic and unfair" (p. 251), and 3) remained unchanged in his initial attitude. The authors did not find significant occurrences of a "boomerang effect" in this study. The sources of the communications were unidentitied and there was no manipulation of issue involvement. Sherif, Sherif and Nebergall (1965) introduced the concept of latitude of noncommitment which included all the positions the individual judged to be neither acceptable nor objectionable. Thus, it could be described as the range of positions left over, after the latitudes of acceptance and rejection were defined. It is intuitively appealing to make a direct analogy between the latitudes of acceptance, rejedtign:andwn6ncommitment on issues and the concept of evoked set in consumer behavior (Campbell, 1973). The evoked set could be defined analogously to the latitude of acceptance as the individualds favorite brand of a product plus all other brands he would consider acceptable. Those brands the consumer would absolutely reject would form the analogy of his latitude of rejection. Finally, there may be some brands he considered neither acceptable nor objectionable (e.g., new brands he has never confronted) which would comprise his latitude of noncommitment. The analogy can be carried only so far however. Sherif, Sherif and Nebergall (1965) used political issues, such as those 19 of the 1960 presidential election, that clearly had two diametrically opposing positions that could be adopted by a voter. Other, more moderate positions, could then be arrayed in a continuum between the two extremes. Such an a priori continuum is not possible where four, five or more brands can be defined in advance for consumers. The latitudes of acceptance, rejection and noncommitment also relate closely to the concept of involvement which has direct bearing on this research. Involvement The concept of involvement has a fairly long history among social psychologists and has more recently become a prominent topic in consumer behavior. Sherif & Cantril (1947) described the ego as a bundle of attitudes ". . . that determines the more or less enduring character of one's personal identity . . ." (p. 4). They go on to state that "all attitudes that define a person's status or that give him a relative role with respect to other individuals . . . are ego- involved" (p. 98). That is, the closer something impinges upon an individual's sense of identity, the more ego-involved that individual becomes--the more he perceives to be at stake in the way he relates to the rest of the world. The authors then state, ". . . the intensity with which we hold these attitudes found in relation to social values . . . depends on the degree to which those attitudes are 20 ego-involved" (p. 151). In other words, the more ego-involving something is to an individual, the more firmly held that attitude is. A person to whom the issue of segregation is highly ego- involving will be less likely to change his attitude toward segrega- tion than someone to whom segregation is not so involving. In a later study relating involvement to opinion change, Zimbardo (1960) found that high response-involvement acted to increase opinion conformity. Response-involvement, as distinct from ego-involvement: . . denotes the instrumental relationship of a given opinion or response to the achievement of a desire goal. The opinion becomes a means of securing reward, approval, or recognition . . . (p. 92). Since response involvement ties the opinion of an individual to a reward, it can be manipulated more readily in an experimental setting than can ego-involvement. For example, Zimbardo manipulated the response-involvement of his subjects by telling half that the opinions they formed from reading a passage would indicate their social values and personalities; the other half were told that the passage was too short, so the opinions they formed would tell nothing about their values or personalities. The first group thus formed their opinions assuming their personalities and values could be evaluated by these opinions; this group was Zimbardo's high response-involvement group. Zimbardo's findings can be summarized in the following graph: 21 Opinion Change High High Response Involvement ‘——‘—”""’flfiflppa,, Low Response Involvement Low L . Latitude Latitude Acceptance Rejection These results, at first glance, seem inconsistent with the results obtained in earlier studies on assimilation-contrast. Highly involved individuals, according to assimilation-contrast, should show no opinion change or even boomerang if the message falls within their latitude of rejection. Zimbardo's findings clearly show that opinion changed even mppg_when the message fell within the individual's latitude of rejection, whether S was highly involved or not. It should be specified, however, that response-involvement and involvement in an issue may have entirely different effects on opinion change. When Zimbardo manipulated his high response-involve- ment 55 by increasing the instrumentality of their responses, he may have overcome any previous involvement in the issue. Issue involvement was replaced by what was at stake in S's response. Thus, he automatically "widened" S's latitude of acceptance. 22 If the latitude of acceptance for the high response-involve- ment condition is indeed wider, this would contrast with earlier suggestions that the latitude of acceptance is wider for low involvement issues (Hovland and Pritzker, 1957, p. 260; Hovland, Harvey and Sherif, 1957, pp. 250-251). It is interesting to speculate on situation-specific effects as determinants of the acceptability of a message by S. A high issue involvement that S brings with him to the research setting may easily be moderated by manipulating his response- involvement. This would be accomplished by changing the perceived instrumentality--changing his perception of the benefits to be received by his response to the questions posed by the researchers so that S's "true“ feelings are overcome, not by the position advocated in the communication, but by what rewards he thinks his answer will get him. Response-involvement, in this sense, is simply another term for reactivity in behavioral measures. Zimbardo's study is important in illustrating the effects involvement might have on the results of certain empirical tests. A possible explanation for the failure to confirm earlier relation- ships between prior attitudes and learning (Waly and Cook, 1966; Greenwald and Sakumura, 1967) is the involvement of S with the issue at hand. Segregation may have been a much "hotter" topic in the mid-1960's than it was in studies conducted a decade earlier. One also suspects that the Vietnam War was a highly involving topic on the college campus where Greenwald and Sakumura conducted their study. 23 Triandis (1971), after reviewing the literature, tried to tie selective exposure (and, indirectly, selective learning) together with involvement when he concluded: . . . people in everday life are exposed to dispropor- tionate amounts of supportive information. On the other hand, laboratory results do not support the view that people prefer to be exposed to supportive . . . informa- tion. Only under certain conditions (i.e., high ego- involvement) will they prefer supportive arguments . . . (p. 64). Freedman (1964) related issue involvement to attitude change in a study among college freshmen and high school students. He found that attitude change is moderated by concern (high involvement) or unconcern (low involvement) with an issue. S's latitude of acceptance was wider for the low involvement condition. This result was basically supportive of conclusions drawn by Sherif, Sherif and Nebergall (1965) who found that the latitude of acceptance did not vary with the degree of issue involvement, but the latitude of rejection grew wider as involvement increased. Naturally, the latitude of noncommitment grew narrower as involvement rose. Miller (1965) found that "high issue involvement consistently reduced the persuasive effects of a discrepant communication . . .“ among high school students (p. 130). This finding was consistent with Freedman (1964). Miller also found that issue involvement had no relationship to S's latitude of acceptance which supported Sherif, Sherif and Nebergall (1965). At this point, the evidence in social psychology indicated that attitude change was inversely related to the degree of 24 ego-involvement in the issue. However, there was conflicting evidence on the relationship between issue involvement and the size of the individual's latitude of acceptance. It was around this time that the concept of involvement was picked up and applied to marketing or, more specifically, to advertising. Interestingly, Sherif and Cantril had noted almost two decades earlier (1947) that advertising tries to associate stated. were composed of esozinyolving,values (p. 352). Nevertheless, serious interest in involvement by the market- ing discipline stems primarily from Krugman (1965) who first suggested that television advertising might be an example of uninvolved learning: the same U curves were showing up in recall tests of TV advertising that social psychologists had shown for recall of nonsense material. Ignoring the possibility that this similarity may be due to nonsense in TV ads (a plausible hypothesis), Krugman used the concept of involvement to account for the similarity. That is, a relative lack of involvement among subjects characterized both nonsense material and television advertising stimuli. Krugman defined involvement in terms of the number of personal connections the viewer made with the stimulus: the more connections, the greater the involvement. In three studies of TV vs. print advertising, Krugman (1966- 1967) found that consumers' involvement with advertised products was higher for magazines than for TV when high involvement products 25 were used. There was no difference for the two media when low involvement products were used. Krugman spoke in terms of a general stimulus involvement at this time, rather than making clearcut distinctions among medium, content, and product involvement. In the course of this work, somehow, he became associated primarily with involvement in media and the more general topic of learning in low involvement conditions. For example, he demonstrated with brain wave measures that relatively high involvement learning took place with print media while low involvement learning occurred with television (Krugman, 1971). Print advertising he concluded, sparked active learning while TV advertising produced passive learning. Preston (1970) took exception to Krugman's attributing involvement effects to media in an interesting study: Preston simply monitored the kinds of products advertised on TV vs. magazines. Products with low differentiation among brands and high substitut- ability (soap, aspirin, soft drinks, etc.) are advertised on television. Highly differentiated products (furniture, clothing, etc.) tend to appear more in print. Thus, media may be confounded by what products are being advertised. Hupfer and Gardner (1971) had undergraduates rate the importance of ten products and ten issues. They found that products are a quite trivial matter to consumers, relative to the issues used in the study. Exceptions to this were a house and automobile which were rated more important than membership in a fraternity and the Apollo space flights. They concluded that since much of the 26 social psychology literature on attitude change dealt with relatively \\ ,__.._.. ego:involving issues, it may not be directly applicable to consumer behavior. Bowen and Chaffee (1974) related product involvement to "pertinent" advertising. They defined pertinence as: ". . . the comparative discrimination a person makes between two objects . . . when he is evaluating alternatives" (p. 614). They specifically noted comparative advertising as an example of what could be a pertinent ad. Bowen and Chaffee found a strong interaction between pertinence of ad appeal and product involvement as summarized below: Favorable High Product Involvement Low Product Involvement Unfavorable . . I 1 Pertinent Non-pertinent Appeal Appeal Clearly, these results show that a high involvement product advertisement will be received more favorably if it contains "pertinent" information. .This conclusion suggests that one reason why so much of the empirical work on comparative advertising has been negative is that is has not been used properly. That is, when combined with a highly involving product, comparative advertising may be extremely 27 effective in terms of favorable evaluation of the ads. This is in direct contrast to the study reported by Levine (1976). Bowen and Chaffee conclude: As involvement increases, so does information need, so that the advertiser who provides pertinent information where it is appropriate stands to gain two kinds of rewards. First, he would seem more likely to gain ; . . consumer acceptance of his claims. Second, he should reduce the risk of . . . constraints by consumer protection agencies (p. 615). This rests on the assumption that consumers desire more information about highly involving products--an assumption shared by others (Chaffee and McLeod, 1973; Day, 1973; Robertson, 1976). The relationship between the consumer's involvement in the product and the type of ad has also been considered by Rothschild (1977) and Tybejee (1977). Both make the much-needed distinction between product involvement and message involvement and suggest that a highly involving message may compensate for a low involvement product. If this is the case, and if a comparative ad is indeed more pertinent, its effectiveness should increase regardless of the type of product being advertised. One final issue is the relationship between involvement and attitude change. The conventional wisdom has been that the more ego-involved an individual is in something, the less likely he will be to change his attitude (Sherif and Cantril, 1947; Sherif, Sherif, and Nebergall, 1965; et. a1.). There is only a short step from this assumption to that of a direct relationship between product involve- ment and brand commitment. That is, the more involved someone is in a product class, the more firm he will be in his commitment to a brand. 28 This is not conceptually reasonable, since it is possible to be highly involved in a product class and completely uncommitted to a particular brand (e.g., when evaluating all different brands of automobiles in anticipation of a purchase), or to be completely uninvolved in a product class but highly committed to a brand because it simplifies the purchase decision (e.g., always buying Skippy Peanut Butter because the consumer feels a more thorough evaluation is a waste of time). Lastovicka and Gardner (1977) found what they termed two components of product involvement: normative importance and brand commitment. But since they defined normative importance more or less identiCally to the traditional definition of involvement, they probably did not discover two components of involvement, but, rather, rediscovered involvement plp§_another construct, namely, brand commitment. Perhaps with the benefit of hindsight, it seems strange that it required a piece of empirical work to uncover something so conceptually apparent--doubly so, since this difference had been discussed over a decade earlier as it related to issue involvement and commitment to a stand on an issue (Freedman, 1964). Once the distinction is made, it calls into question much of the work performed on involvement. Lastovicka and Gardner, for instance, question whether the latitudes of acceptance, rejection and noncommitment really refer to the individual's involvement on an issue or his commitment to a position on that issue (p 17). 29 The analogous concepts in marketing would, of course, be evoked set and product involvement. Does high product involvement necessarily mean a small evoked set size? Perhaps, but high brand commitment certainly implies a small evoked set size. The relationship between brand commitment and attitude change seems fairly straightforward: the higher the commitment, the smaller the attitude change. The relationship between product involvement and attitude change, however, is not so obvious. Part of this research will be to test that relationship. Summary of Issues and Substantive Hypotheses Based on the literature reviewed in the previous section, the major research issues can be summarized in this way: 1. While some empirical work into the communication effectiveness of comparative advertising has been done, the effects of product involvement and brand usage have not been explicitly taken into account. 2. There is contradictory evidence on the relationship between prior attitude and learning (selective learning). This may relate to the level of involvement the communica- tion has for the receiver. 3. Intuitively, it seems the less involved the individual is in the product class, the more readily he will change his attitude toward a particular brand. This research offers an opportunity to test this empirically. The substantive research hypotheses flow directly from these issues. A. Comparative advertising is a more effective form of advertising than Brand X or noncomparative advertising. B. The effectiveness of comparative advertising will be greater among users than nonusers of the target brand. 30 The effectiveness of advertising will be greater if a high involvement product is being shown than if a low involvement product is. There is no relationship between one's brand attitude and recall of a high involvement product advertisement. There is a positive relationship between one's brand attitude and his recall of a low involvement product advertisement. CHAPTER III RESEARCH DESIGN Basic Desigp_ The research design was an experiment with repeated measures of the dependent variables. Chi-square analysis and analysis of variance (ANOVA) were the major statistical tools used in the analysis. The independent variables were: 1) product type (high versus low involvement); 2) brand usage (users versus non-users of the target brand); 3) and advertisement (comparative versus Brand X versus noncomparative). The dependent variables were advertising recall and the subject's attitudes toward the brand he used most often. ‘ To look at changes in attitude and recall over time, the study was conducted in three time periods: a pretest (also used to screen subjects and randomly assign them to see the different advertisements): a posttest administered immediately after the subject was shown the advertisement; and a delayed posstest. The data matrix for the study is given in Appendix D. 31 32 A control group was not used since the study aims at differences based on which type of advertisement is seen rather than differences between seeing an advertisement and not seeing one. Unwanted effects that carry over from one time period to the next are assumed to be uniform since the advertisement a subject sees is assigned at random; if this assumption is met, the differences that are observed can be attributed to which advertisement the respondent saw. Besides being of interest in itself, the usage factor was intended to serve as a blocking variable. That is, it was thought .that users of the target brand would be more homogeneous in characteristics that correlated with their attitude toward the target brand and their ability to recall advertisements explicitly mentioning the target brand. To the extent that this relationship did exist, it would reduce the error terms of the F-tests, thus increasing the power of the overall design. Measurement of Dependent Variables Resell. Recall of the ads was determined using unaided brand name recall and unaided content recall. Name recall was scored into three major categories: no recall of the ad; incorrect name recall; and correct name recall. In addition, two dichotomous methods of scoring name recall were used: 1) recall versus no recall of the advertisement; and 2) correct versus incorrect name recall. 33 Content recall was scored by the percentage of items of information the subject correctly recalled from the ad. Since any advertisement normally consists of several pieces of information (e.g., a headline, a price, a list of attributes, a photograph, etc.). a respondent who could recall more of these pieces of information was so recognized in the measurement and differentiated from one who recalled, say, only a single item of information. The items of information were defined ahead of time thus generating an p_ppippj_system of coding the open-ended responses of content recall. (See Appendix B for a list of these items.) Since content recall depends on the subject's ability to remember the advertisement, each subject was reminded of the sponsoring brand when asked to recall what the advertisement said. Recall was measured twice: in the immediate posttest (Time 2) and in the delayed posttest (Time 3). Attitude The subject's attitude toward his own brand was determined with three items: 1) brand preference; 2) intention to buy his brand again; and 3) general liking for his brand. These brand attitudes were measured three times: during the pretest, in the immediate posttest, and in the delayed posttest. All the attitude items were measured using 7—point bipolar scales. Identical bipolar semantics were used for each of the three time periods to ensure consistency. (See the questionnaires in Appendix C). 34 Selection of Products and Brands The two products used for the study were a non-diet cola soft drink and an automobile. The cola was found to be a relatively low involvement product and the automobile a high involvement product in an earlier published survey and in a survey conducted by the researcher in the local area (Hupfer and Gardner, 1971; Traylor, 1978). In this latter study, it was also determined that an automobile yielded a good brand split on Oldsmobile versus non- Oldsmobile drivers (roughly, 40%-60%); the cola had a good brand split on Coca-Cola versus non-Coca-Cola users (roughly 45%-55%). The brand split was operationally important because brand usage was an independent variable in the study: half the automobile users had to drive a particular brand and half the cola drinkers had to use a particular brand. Development of Advertisements Advertisements were developed by a senior student in the Graphic Arts Department at Michigan State University with advertis- ing experience. Copy was written by the researcher. The ads were developed according to the following guidelines: 1. Gain the attention and interest of the reader. 2. Instill awareness of the brand name. 3. Communicate the brand's major attributes and benefits for the consumer; that is, the reader should remember not only the brand, but its superiorities as well. 4. Shift the consumer's liking, preference, and buying intentions away from his or her current brand to the new brand. 35 5. Create a liking for the advertisement itself. These guidelines were developed to simulate those that might actually occur for a new brand entering a mature product market and to facilitate conmunication of objectives between the graphics designer and the copywriter. Coca-Cola was the target brand for the soft drink compara- tive advertisement. Vesta Cola was the name given to the fictitious brand. Oldsmobile was the target brand for the automobile ad and DeOrlean was the name given to the sponsoring brand. To lend credibility to the automobile ad, the DeOrlean was represented as a division of Renault. In drawing up the ads, an approach considered was to make the advertisements completely comparable except for the product advertised. That is, the same brand name, amount of copy, graphics, etc. would have been used for both the cola and the auto- mobile. However, this was not possible if the advertisements were to be at all credible to the subject. For example, all automobile advertisements contain EPA mileage estimates which would not be appropriate for a cola. Therefore, different types of advertisements were developed for the two types of products. The automobile advertisements were relatively copy-heavy and used a testimonial. The cola advertisements relied more on graphics and stressed only a few points in the c0py. The attempt was to make the automobile advertisement a plausible representation of a print advertisement 36 for a fairly expensive consumer durable; and to make the cola advertisement representative of an ad for an inexpensive, nondurable, convenience item. Within the automobile advertisements and within the cola advertisements, the only differences were in the comparative, Brand X, or noncomparative approaches. This can be seen in Appendix A which shows the six advertisements developed for the study. Data Collection Data collection took place in three phases. The question- naires are given in Appendix C. Time 1 was a self-administered questionnaire used for screening subjects and assigning them to one of the twelve cells in the design. The questionnaire, a cover letter, and a self-addressed stamped envelope were left at the door or by the mailbox of 600 residences in East Lansing. The residences were chosen for convenience and were primarily apartments. Some single family houses and duplexes were also included. Instructions stated that the questionnaire was to be filled out by the person who did most of the household shopping. Items in the questionnaire asked for the brand used most often for five products: a non-diet cola soft drink, an automobile, and three other products intended only to divert attention from the products of interest. For each product, the respondent completed the three attitude questions about his most-used brand using the 7-point scaled items listed earlier. 37 Returned questionnaires were classified into one of four groups: 1) Cola users who drank Coca-Cola most often; 2) Cola users who drank a brand other than Coke most often; 3) Automobile drivers who drove an Oldsmobile most often; 4) Automobile drivers who drove a car other than Oldsmobile most often. Each respondent who filled out a Time 1 questionnaire fit into one of these categories. Within the first two categories, subjects were randomly assigned to see one of the three Vesta Cola ads; in the other two categories, subjects were randomly assigned to see one of the three DeOrlean ads. Time 2 was the treatment phase during which subjects were shown one of the six test advertisements. Trained interviewers arranged appointments with subjects from Time 1 who indicated a willingness to participate in the research. These face-to-face interviews usually took place in the subject's home, although several took place at work. Each interview took place at least seven days after the Time 1 questionnaire was received. Although the interviewers were experienced and had been trained on the Time 2 questionnaire, they were kept naive regarding the study's objectives. During Time 2, the subject was shown a booklet of thirteen black and white full page advertisements. The advertisements always appeared on the right-hand page facing; innocuous editorial appeared on the left. Each subject was instructed to flip through 38 the booklet as though he were doing so with a magazine. The interviewers were instructed to make the setting relaxed and informal. The order of the advertisements was the same and the test advertisement was always the seventh to appear in the booklet; thus, it was always in the middle, whether the respondent went through the booklet front-to-back or vice-versa. After the respondent had finished looking through the booklet or after a maximum time of four minutes, the booklet was removed from his sight and the brand name and content recall questions were asked. The booklet was then returned to the respondent and the attitude and credibility questions were asked while the respondent had an opportunity to look at the advertisement being discussed. Thus, each subject saw the test advertisement twice: once when he flipped through the booklet and again when the attitude questions were asked. A "hand card" with the 7-point scale and the appropriate polar items was given to the subject to use in answering these questions. In addition to the test ad, four of the twelve filler advertisements were also used for the same series of recall, attitude, and credibility questions. The purpose of asking questions about these advertisements was solely to disguise the purpose of the study which the respondent and interviewer might have guessed had all the questions focused only on the test advertisement. The interviewers were systematically reassigned to interview respondents in different cells to reduce interviewer bias. 39 Nevertheless, a chi-square test of interviewer by advertisement type was highly significant (p < .01) because two interviewers completed a disproportionate number of interviews within certain cells.* The measure that would most readily indicate interviewer bias would be the open-ended content recall question of Time 2. A significant difference across interviewers would indicate differences in probing and recording open-ended responses. Fortunately, this difference is not significant (p = .43) nor is the interviewer x advertisement interaction (p = .54). Thus, although the systematic rotation of the interviewers was not effective in "spreading them out" over the treatment conditions, significant interviewer bias probably did not occur. At the end of Time 2, the interviewers were debriefed and the study was explained to them. At that time, it was learned that none of the interviewers had guessed that the study compared the effectiveness of different advertisements. While they were aware of the car versus cola advertisements, none had noticed the comparative versus Brand X versus noncomparative advertisements on each product. Nor had any interviewer realized that half the "cola respondents" were users of Coca-Cola or that half the "auto *This was accounted for by two "real world" problems. Some of the cells of the design were easier to fill than others; for example, it was easier to find an automboile driver than a cola drinker and easier to find a non-Oldsmobile driver than an Oldsmobile driver. Interviewers initially assigned to the "easy" cells had a much larger proportion of the cell quota to draw from in a given time period than would interviewers assigned to the "hard" cells. Complicating this were differences in scheduling and interviewer productivity. 40 respondents" drove Oldsmobiles. Thus, the interviewers remained naive throughout the study. Time 3 was a telephone interview conducted by the researcher and took place at least ten days following the Time 2 interview. Calls were made to the respondent's home or, in some instances, to his place of work, depending on his preferences stated during Time 2. The attitude questions about the subject's own brand were the same as those in Times 1 and 2. By this time, the respondent had had considerable experience with the 7-point sEETEd“items‘afid'thEFEifl was no difficulty in applying these over the phone. Demographic information was obtained at the end of the Time 3 interview and the respondent was asked if he knew the nature of the study. Most knew it "was about advertising" or "how much people remember from advertising," but none guessed the experimental nature of the study or the comparisons of communication effectiveness. Surprisingly, none guessed that the test focused on the DeOrlean advertisement or the Vesta Cola advertisement, even though Time 3 did not refer specifically to any phony advertisement. Thus, the subjects remained naive. Respondents were debriefed and asked not to discuss the research with anyone else. Coding Coding of the open-ended questionnaires was done in a three- step process: each questionnaire was coded by the researcher; each was then checked independently by another coder; any disagreement 41 was then resolved by discussion until agreement was reached between the researcher and the coder. The content recall items were coded using the p_ppippi codes in Appendix B. The percentage of content recall was computed by counting the number of items of information reported by the subject and dividing by the total number of items in the advertise- ment. In the DeOrlean comparative advertisement, for example, there were 25 items of information; so, a subject who recalled, say, three items received a score of 12. Of the 120 questionnaires coded, there were seven disagree- ments between the researcher and the coder on the correct value for the content recall variable; and since content recall was measured twice (once in Time 2 and once in Time 3), there were actually 240 potential disagreements. Each of the seven disagreements was resolved by discussion. Reclassification of Advertisements and Involvement Advertisements Although the advertisements were defined p_pgjppi_as being comparative, Brand X, or noncomparative, question #33a in the Time 2 questionnaire was used to check this classification on a perceptual basis; that is, each subject was asked to identify any comparative ads he had seen in the booklet. Table 1 shows the number of respondents who named the Vesta or DeOrlean advertisement as a comparative advertisement broken down by the p_ppjppj_type of advertisement they "actually" saw. It is obvious from this table 42 TABLE l.--Advertisements Perceived as Comparative by A_Priori Classification. Comparative Brand X Noncomparative f % f % f % 32 80 17 43 3 8 that, while most of the respondents agreed with the p_prippi classi- fication, some did not. In particular, 12 of the 20 subjects who saw the Vesta Cola Brand X advertisement reported this as a compara- tive advertisement. Of these 12, five reported that the target brand was Coca-Cola. Another two said the target brand was Pepsi. This discrepancy raises the question of construct validity in determining types of advertising. Wilkie and Farris (1975) understood the need for a perceptual emphasis when they mentioned the criterion of recognizability of a target brand or brands in a comparative advertisement. But none of the empirical literature has approached this problem and, in fact, all the studies have used an 2 ppippi_classification developed by the researcher. This makes sense from an advertiser's point of view, since information about the relative efficacy of comparative advertising is useless if the advertiser cannot "make" a comparative ad. Nevertheless, the problem ought to be approached from a perceptual standpoint as well, simply in the interest of generating consumer-based information. 43 The issue here is similar to that regarding a marketer's versus consumer's basis for classifying goods. Within the respondent-based method of classification, there were three approaches that could be taken for purposes of this research. The first was that a comparative advertisement could be defined as such only if the consumer correctly identified the brand, or brands being compared; the second required the consumer to name some brand, whether or not it was the correct target brand. Third, the consumer had merely to recognize that the sponsoring brand was being compared to a competitor's product, whether or not he could assign a brand name to this product. It is this last, most general approach that was adopted for this research. In fact, of the twenty subjects who saw a Brand X or noncomparative advertisement, but reported it as making a compari- son against a competitor's product, only seven explicitly named another brand as the target brand. Other responses were "all other brands," "similar brands," etc. Those thirteen respondents who did not explicitly name a target brand could have been classified as having seen a Brand X advertisement or grouped into the noncomparative perceptual category. Any way of classifying them in a group would have been partially judgmental. Besides the problems of sample size, the theoretical rationale for assigning them to the comparative category was that they recognized that some type of product comparison was taking place. Whethercn*not they could explicitly name a target brand was III II‘IVIIIIIIVII 11141 .11 x Ill-TI III: 1 . (III I .II . I 44 of less interest than the dynamics that may have caused them to report a comparison. In this sense, a more appropriate term for the comparative advertisement category from the respondent's standpoint might be "pertinent advertising." This term was used in a study cited earlier to indicate ". . . the comparative discrimination a person makes between two objects in a situation where he is evaluating alternatives" (Bowen and Chaffee, 1974, p. 614). This implies that all comparative advertising is pertinent, but the converse does not necessarily hold. To maintain consistency, the term comparative advertising was used. However, it should be kept in mind that 25% of these subjects did not explicitly name a target brand. Product Involvement Although there was previous support for using an automobile as a high involvement product and a soft drink as a low involvement product (Hupfer and Gardner, 1971; Traylor, 1978), this too was checked in Time 2. The results are given in Tables 2 and 3 which confirm using the two products to represent the two ends of the involvement spectrum. However, some subjects rated the cola at a high level of involvement while others rated the automobile at a low level of involvement.* * The actual measurement asked the subjects to rate the importance of the product to themselves. This measurement of involve- ment was used by Hupfer and Gardner (1971) and Traylor (1978). See the Phase II questionnaire in the Appendix. 45 TABLE 2.--Chi-Square Analysis of Involvement Score by Type of Product. Low Involvement Neutral High Involvement l 2 3 4 5 6 7 Auto 4 2 O l 6 20 27 Cola 10 13 7 11 ll 3 5 x2 = 55.13 d.f. = 6 p < .01 TABLE 3.--T-Test of Product Involvement Score by Product. X' 52 s n Automobile 5.85 2.84 1.69 60 Cola 3.48 3.41 1.85 60 t = 6.23 d.f. = 118 p < .01 (1 - tailed test) NOTE: The difference between the two variances is not significant at a = .26. 46 Reclassification As a result of these discrepancies in the classifications of the type of advertisement seen and in product involvement, the analysis was conducted using both the p_ppjppi_categories and the respondent categories. For the respondent categories, the advertise- ments were split into two levels: comparative and noncomparative. Involvement was dichotomized into high involvement and low involve- ment.* The g_ppjppi classification uses the 3 x 2 x 2 factorial design within each time period as discussed earlier; the respondent classification uses a 2 x 2 x 2 design within each time period. For illustration, Appendix D contains the data matrix for the respondent classifications with the cell sample sizes filled in. Stastical Procedures Using the Respondent Classification An important question related to the unbalanced design obtained from the respondent reclassification concerns the nonortho- gonality of the main effects with each other and of the interaction effects with the main effects. Unless corrected, this could distort the statistical tests. There are two ways to handle the ANOVA if cell sizes are unequal and disproportional. *The median score for the involvement measure was 5.2 when the cola and automobile responses were combined. This statistic was the basis for dichotomizing the involvement variable. 47 One method is an unweighted means ANOVA which treats each cell mean as a single observation and uses the harmonic mean to compute the sums of squares for each effect (Keppel, 1973, pp. 347- 362; Glass and Stanley, 1970, p. 440). The other method is the regression approach which is a least squares technique.* Authors who propose the unweighted means ANOVA emphasize the importance of the orthogonality among the sources of variation. The unweighted means ANOVA "forces" orthogonality into the design by using the harmonic mean to compute the sums of squares. Thus, the sums of squares for each effect add to the total sum of squares. Proponents of the regression approach emphasize the impor- tance of performing an exact F-test. Exact F-tests are performed by the regression method since it partitions the individual effects while adjusting for all the other effects: the F-test is performed for the unique, incremental contribution of each effect, having already taken all the other effects into account. In the regression approach, however, the sums of squares will, in general, not sum to the total if cell sizes are unequal: since the effects are not orthogonal, some of the variation can be attributed to no single source . *Neter and Wasserman (1974). In addition, two popular computer packages of statistical routines, SPSS and SAS, both provide the regression technique, but not the unweighted means technique. Neither "side of the controversy" mentions the other, at least not in the above literature. From a phone conversation, James H. Goodnight, who wrote the General Linear Model (GLM) set of routines for the SAS package, was unaware of the unweighted means approach. 48 If all cell sizes are equal, the regression approach will yield exactly the same results as the standard ANOVA computations. In this study, both methods were used on several tables and yielded nearly identical results both in computing the sums of squares and when the final F was computed. This was because the cell sample sizes were only slightly disproportional and the disproportionality was not statistically significant. For this reason, the regression approach was used, since it was readily available on computer statistical packages. Research Hypotheses Given the issues and substantive hypotheses developed for the research, and the dependent measures discussed at the beginning of this section, eleven research hypotheses have been formulated. These correspond to the substantive hypotheses listed earlier, but are here operationalized using the dependent measures of the study. Brand Name Recall H1 Brand name and advertisement recall will be highest among those who are exposed to the comparative advertise- ment. H2 For those who are exposed to the comparative advertise- ment, brand name and advertisement recall will be higher if they are users of the target brand than if they are nonusers. H3 Brand name and advertisement recall will be higher among those who are exposed to an advertisement for a high involvement product than among those exposed to a low involvement product. 49 Content Recall H4 Content recall will be highest among those exposed to the comparative advertisement. 5 Content recall will be higher among those who are exposed to an advertisement for a high involvement product than among those exposed to a low involvement product. H6 For those exposed to the comparative advertisement, content recall will be higher if they are users of the target brand than if they are nonusers. Attitude Change H7 Attitude change away from their own most-used brands will be greatest among those exposed to the comparative advertisement. H8 For those exposed to the comparative advertisement, attitude change away from their own most-used brands will be greater among users of the target brand than among nonusers. H9 Attitude change away from their own most-used brands will be greater among those exposed to an advertisement for a low involvement product than among those exposed to an advertisement for a high involvement product. Selective Recall H10 For those exposed to a high involvement product advertise- ment, there will be no relationship between one's atti- tude toward his own most-used brand and content recall. “11 For those exposed to a low involvement product advertise- ment, there will be a positive relationship between one's attitude toward his own most-used brand and content recall. Response Rates The response rate to the Time 1 self-administered question- naire was 35%. Tables describing the response rates are given in Appendix B. 50 Of the 209 respondents who completed and returned the Time 1 questionnaire, eight indicated they would be unwilling to continue in the study, leaving 201 questionnaires. Of these 201 respondents, 125 completed Time 2. Most of the other 76 respondents were never contacted for a Time 2 interview since they were not needed to complete the study. Thus, the "dropout rate" was quite low. Of the 125 respondents who completed Time 2, 120 completed Time 3 which was the sample size required. Of the five who finished Time 2, but were excluded from the final sample, two could not be contacted for Time 3 by the end of the data collection; and the other three had switched brands of automobiles. Since the attitude measures were based on the respondent's same brand over all three phases of the study, brand switching rendered these results incomparable. These respondents were, therefore, deleted from the final sample. CHAPTER IV FINDINGS Sample Characteristics Tables showing the demographic characteristics for the sample are given in Appendix E. The sample was biased toward whites, professionals and managers, the well educated and young adults (205 and 30s) who lived by themselves or with one other person. A large number were also single. This demographic profile is not surprising given the apartment area near the university from which most of the sample was drawn. In the random assignment of subjects to the three different types of advertisement, there was no significant bias by any demo- graphic group. However, bias in the other independent variables occurred in two ways: (1) younger respondents were more likely to have been cola users and tended to be assigned to see one of the three cola advertisements; and (2) older respondents were more likely to have been Oldsmobile drivers, so were assigned to see one of the three automobile advertisements. There was no other statistically significant bias with the independent variables by the demographic variables measured in the study. In particular, age was not confounded with the type of advertisement that was seen. 51 52 Advertisement Recall and Name Recall By Advertisement Type A Priori Classification Table 4 shows the results of the unaided brand name recall for the total sample based on the p_ppippi_classification for the Time 2 immediate posttest. While high statistical significance was not obtained, the results were in the direction predicted. In particular, nearly twice as many subjects who saw the comparative advertisement recalled the brand name correctly. Table 5 shows the results for Time 3 delayed posttest. TABLE 4.--Brand Name Recall by Advertisement Type in the Time 2 Immediate Posttest. Comparative Brand X Noncomparative Ad Ad Ad T°ta' Correct Recall 18 10 10 38 Incorrect Recall 12 17 12 41 No Recall 1Q l_3_ 18 Q Total 40 40 40 120 x2 = 6.98 d.f. = 4 p = .14 These results are generally consistent with those of Time 2. Again the direction was as predicted, although high statistical significance was not obtained. The big change from Time 2 to Time 3 was that more subjects could recall the advertisement in Time 3. This was probably because the advertisement had been discussed and 53 TABLE 5.--Brand Name Recall by Advertisement Type in the Time 3 Delayed Posttest. Comparative Brand X Noncomparative Ad Ad Ad T°ta' Correct Recall 16 ll 7 34 Incorrect Recall 19 27 27 73 No Recall _5 _2 _6 13 Total 40 4O 40 120 x2 = 7.34 d.f. = 4 p = .12 questions asked about it during the Time 2 interview. That is, subjects who were asked the recall question in Time 2 were responding to one exposure of the advertisement; in Time 3, they had seen the advertisement twice and had discussed it with the interviewer during the second exposure. Therefore, it is not surprising that recall of the advertisement went up from one time period to the next. The total sample was broken down by product type and by brand usage to see if any notable differences emerged. In general, the results were consistent with those for the total sample. In addition, the dependent variable was rescored in two ways: (1) As a dichotomous variable: ad recall vs. no recall, in which correct and incorrect versions of the brand name were combined. A mention of "a soft drink" or "some car" was also counted as ad recall. (2) As a dichotomous variable: correct name recall vs. incorrect name recall in which a failure to recall the ad was counted as incorrect. The tables thus generated are given in Appendix F and are also generally consistent with the results presented here; that is, the 54 results were usually in the direction predicted, but were not statistically significant. Respondent Reclassification The same analysis was done based on the respondent reclassi- fication in which the subjects defined what advertisements made comparisons and whether the cola or automobile were reported as high or low involvement products. Results for the total sample in the Time 2 immediate posttest are given in Table 6. TABLE 6.--Brand Name Recall by Advertisement Type Using the Respondent Classification in the Time 2 Immediate Posttest. Comparative Noncomparative Ad Ad Total Correct Recall 21 17 38 Incorrect Recall 17 24 41 No Recall 14 _2_7_ 41_ Total 52 68 120 x2 = 3.67 d.f. = 2 p = .16 The results are consistent with those when the p_ppippi classifications were used. The direction is as predicted, but does not reach high statistical significance. The Time 3 results from the delayed posttest are given in Table 7. These results are in the direction predicted and reach a higher level of statistical significance. 55 Chi-square tables that used the Recall-No Recall method of scoring and the Correct-Incorrect method of scoring are given TABLE 7.--Brand Name Recall by Advertisement Type Using the Respondent Classification. Compaxgtive Noncompgrative Total Correct Recall 21 13 34 Incorrect Recall 26 47 73 No Recall __5_ __8__ _l_3_ Total 52 68 120 x2 = 6.60 d.f. = 2 p = .04 in Appendix F. These are consistent with Tables 6 and 7 shown here. As with the p ppjpgj_method of classifying the advertise- ments, additional subsample breakdowns were run to see if any patterns emerged based on brand usage and the level of product involvement. These subsample results generally reflect the results for the total sample. While the direction of results was as predicted, statistical significance was usually not reached. DeLorian and Renault A number of respondents confused the name DeOrlean with DeLorian.* Interviewers were instructed to count DeLorian as an incorrect response. Although, given the random assignment of respondents, the DeOrlean-DeLorian confusion should not have * A man named DeLorian, it was learned, left General Motors recently and is organizing to manufacture a car bearing his name. Ironically, he is supposedly negotiating with Renault to supply the chassis for his car. 56 mattered, the data were reanalyzed counting DeLorian as a correct response. This had the effect of lowering the X2 values by moving more subjects from the incorrect to the correct category. In addition, counting Renault as a correct response had the 2 same effect of lowering the X values and raising the significance levels. Discussion and Summary of Brand Name Recall ResuTts The results on correct brand name recall and recall of the advertisement are partially consistent with those of Prasad* who found no significant differences in correct brand name recall by type of advertisement. But unlike his research, this evidence is not entirely clearcut. In the 72 Chi-square analyses performed using the breakdowns by product type, by usage, and for the alterna- tive methods of scoring name recall, the direction predicted in H1 was found in all but a few circumstances--especial1y when using the respondent method of classifying the advertisements--although the accepted level of statistical significance was usually not there. For the most part, the significance levels hovered in the .12 to .16 range. Stated one way, there was insufficient evidence that, given an unknown brand in a mature product market, a comparative advertise- ment will result in a higher level of correct brand name recall. *Prasad (1975) and Levine (1975) also report no advantage in using comparative advertising; Jain and Hackleman (1978) found a significant difference in favor of comparative advertising. 57 Stated in a less statistically rigorous, but perhaps more practical way, the preponderance of this evidence points to greater name recall effectiveness of comparative advertising in the aggregate, although not under specific conditions of brand usage, the type of product advertised, or the level of involvement. Regardless of which of these two interpretations is favored, the evidence, on the whole, certainly does not show comparative advertising to be lg§§_effective than Brand X or noncomparative advertising. A few tables were produced in which the direction was not as predicted, but the sample sizes were low, the results were not statistically significant, and such results may be produced by chance. The effectiveness of comparative advertising on recall of the brand name may be an area for additional research using larger sample sizes. Confusion of Target Brand with Sponsoring Brand Respondents recalled the target brand name as the sponsoring brand twice, both times in Time 3. One recalled Oldsmobile and the other Coca-Cola, and both saw the comparative type of advertisement. This is to be expected since the target brand is mentioned only in the comparative advertisement, and the number who confuse the target brand with the sponsoring brand would likely be greater than or equal to the number who confuse the two in another kind of advertisement. However, the contention that comparative advertising 58 gives "free advertising" to competitors does not seem to be supported by this research. Advertisement and Brand Name Recall By Usage,of the Target Brand Tables 8 and 9 are the Chi-square contingency tables for advertisement and sponsoring brand name recall broken down by usage or non-usage of the target brand named in the comparative advertisement. It can be seen that, for the total sample, usage or non-usage of the target brand made little difference in the impact of the advertisement on recall of the sponsoring brand name. When the analysis was restricted to those subjects who saw the comparative advertisement, usage was not a significant factor in affecting brand name recall or recall of the advertisement in general. Whether the comparative advertisement was defined as such on the p_ppip[i_or the respondent basis did not change the results. Thus, H2 which hypothesized that recall would be greater for users of the target brand than for non-users was not supported. Usage of the target brand had little effect on the ability to recall the sponsoring brand's name and the results were contrary to the direction predicted. Brand Name and Advertisement Recall By Product Involvement A Priori Classification Tables 10 and 11 are the Chi-square contingency tables that break recall down by the product advertised. The results were 59 TABLE 8.--Brand Name Recall by Usage of Target Brand in the Time 2 Immediate Posttest. Brand Users Brand Non-Users Total Correct 17 21 38 Incorrect 22 19 41 No Recall 21_ 29_ .31 Total 60 60 120 x2 = .66 d.f. = 2 p = .72 TABLE 9.--Brand Name Recall by Usage of Target Brand Delayed Posttest. in the Time 3 Brand Users Brand Non-Users Total Correct 19 15 34 Incorrect 34 39 73 No Recall _2_ _p_ 13_ Total 60 60 120 x2 = .89 d.f. = 2 p = .64 60 TABLE lO.--Brand Name Recall by Product in the Time 2 Immediate Posttest. Automobile Cola Correct Recall 19 19 Incorrect Recall 28 13 No Recall 13 28 x2 = 10.98 d.f. = 2 p < .01 TABLE ll.--Brand Name Recall by Product in the Time 3 Delayed Posttest. Automobile Cola Correct Recall 16 18 Incorrect Recall 41 32 No Recall 3 10 x2 = 5.00 d.f. = 2 p = .08 61 significant in both time periods. The results were also significant when recall was scored as recall vs. no recall of the advertisement. The cola name, Vesta, was recalled about equally correctly as the name for the automobile, DeOrlean. But the automobile advertisement was recalled better than the cola advertisement. This would imply that while the automobile advertisement was more readily remembered, the name Vesta registered about as well as the name DeOrlean. The product in the advertisement, however, was confounded with the graphics and copy. That is, the differences should not be attributed to the products, but to the advertisement in toto. No attempt was made to render the automobile advertisements comparable to the cola advertisements in any respect, except that both were in black and white. To attribute differences in recall to the product advertised would, therefore, be misleading. Respondent Classification When the involvement scores, rather than the product type were used, the results were not significant except in Time 3 as shown in Tables 12 and 13. The direction was as predicted by H3: recall was higher for high involvement products than for low involvement products. Thus, it would seem that no difference in the ability to recall a sponsoring brand name or advertisement can be attributed to differences in how involving the advertised product is for the reader in an immediate recall setting; however, involvement with 62 TABLE 12.--Brand Name Recall by Involvement Score in the Time 2 Immediate Posttest High Involvement Low Involvement Total Correct 22 16 38 Incorrect 26 15 41 No Recall gfl_ 11 51. Total 72 48 120 x2= 31 d.f.=2 p= 86 TABLE l3.--Brand Name Recall by Involvement Score in the Time 3 Delayed Posttest High Involvement Low Involvement Total Correct 22 12 34 Incorrect 46 27 73 No Recall _4 _2 E Total 72 48 120 X = 5.22 d.f. = 2 p .07 63 the advertised product may result in recall differences after a period of time has elapsed. This conclusion should be tempered, however, with the pos- sibility that the statistically significant difference in Table 13 occurred by chance. And, it should be recalled that the product type is strongly correlated with the level of product involvement; thus, the same confounding of involvement with the advertisement itself may be operating here. On the whole, the Chi-square analyses of brand name and advertisement recall show a consistent lack of significance regard- less of which criterion variable is used. This may be attributed partially to the low power of the statistical test as well as to the lack of anything significant "really out there." In the course of the brand name recall analysis, a total of 114 Chi-square tables were generated,so there is a 90% chance that eleven or twelve of them will show significance (at a = .10) by chance alone. The lack of any consistent pattern in the results shows this may have happened. That is, the significant Chi-square tables occurred here and there at random. Theoretical speculation about why names for high involvement products will be remembered better than for low involvement products after some time has elapsed is apparent: fodxmt‘ting is less because the product is closer to the individual'5" '2 of self. Whether or not this phenomenon is demonstrated here is questionable, signifi- cant Chi-square results notwithstanding. 64 Summary of Brand Name Recall Findings In general, the hypotheses regarding brand name recall by advertisement type, by usage or non-usage of the target brand, and by level of product involvement are not supported. However, the direction of results indicate that the sponsoring brand name will be recalled correctly more often if a comparative advertisement is seen; or if the product is rated as highly ego involving. Usage of the target brand does not affect the ability to recall the name of the sponsoring brand correctly and, in fact, the results were in the other direction. Content Recall A Priori Classification of Advertisement Types and Product Involvement Overall, the comparative advertisement resulted in higher levels of average percent content recall than either the Brand X or noncomparative types of advertisement. This is shown in the signifi- cant main effect for advertisement type shown in Table 14. Planned comparisons show that these differences are all significant (p < .01). In addition, these results are observed within each of the two time periods when the content recall measure was taken. This is seen in Tables 15 and 16. The cell means and standard deviations used to generate Tables 14 to 16 are given in Appendix G. The significant main effect for the time period in Table 14 occurred because the content recall scores increased from the Time 2 65 .coumcmeocme we» 88 8 x a x < Na» 8888 888888 888 "ammuuu :8 88888888588 8;» 88 a x a x < "m 888:. 88N 88.88N88 8<888 88.N8 88. 88.8888 8 x 8 x < “88 88.88. 88_ 88.88888 N x 8 x < "N 82 8.. 88.8 N 8N.8. a x 8 x < x 8 82 N8. 88.N8N N 88.888 .8 x 8 x < 82 P8. 88.N8 P 88.N8 a x 8 x 8 82 _8. 8N.88 N 88.888 8 x < x A 8: 8N. N8.88 N 88.8N 8 x < x 8 82 _8. 8N.8N F 8N.8N .8 x 8 82 N8.N 88.888 N 88.88N .8 x < 82 88.. 88.888 N 88.N88 .8 x 8 N8. NN.8 88..88 _ 88.888 8 x 8 82 88. 88.88 P 88.88 8 x 8 82 N8. 88.8 N 8N.N 8 x 8 82 N8. 88.8 P 88.8 .A88 88888 88888 888888 82 _8. 88.N8 F 88.N8 .Aav 8888888 P8.v 88.82 N8.888N N 88.88N8 .A88 888k 88 .8.v 88._p 8N.8N8 P 8N.8N8 A88 8888 8 A mamaam :88: .8.8 menacm 88 Sam aucaom .LOpumu 8 mm towsma meh mcmma rpmowm acmucou Lou mFAMF <>ozoz<--.mp m88ozHo>=H Ho.v mH.o~ Ho.HoHH H Ho.H©H¢ Hoz<--.HH u4mHo>=H x aaHH u< mz om. oo.mm H Ho.mm mm: ucacm x Hean>Ho>=H No. mo.m HH.mmm H HH.mmm am: ucmam x maHH u< mz me. om.mm H om.om Heasm>Ho>=H x maHH u< mz Ho. mm.H H Hm.H mm: ucacm mz mo. aw.m H mN.m Hcasm>Ho>=H Ho.v mm.mH wo.mmHN H mo.mmHN aaHH u< a u mgmzcm com: .m.u magmacm we saw wogzom .Hcmsm>Ho>cH Huzcoca use «Haw HcmsmmHHHm>w< Ho mcoHumuHHHmmmHu Hcmvcoammm egg co ummmm Hmmuumom meHumEEH N mEHH msu cH HHmuwm Hcmucou Hog mHamH <>ozHo>=H Ho. No.H oo.MHw H oo.mHm mm: cemem x mHHH u< mz mm. mm.m~ H mm.m~ Hewsm>Ho>=H x maHH u< mz co. mm.¢ H mm.¢ am: ucaam mz Ho.v em. H Hm. Hcmsm>Ho>=H Ho.v oo.¢~ Ho.mmH~ H Ho.mmHN aHHH u< a H ogmsam cam: .H.u mmgmacm Ho saw .ucwso>Ho>:H Husuoca can maaH ucmsmmHucm>n< Ho mcoHumuHHHmmmHu pcmucoamwm on» :o commm ammuumom umxmpmo m mEHH mg» cH HHmumm Hcmucou Hog mHan <>oz<--.mH UHm_h(u<&£OUZO 2 u>=(¢<&!OU u>=<8(&£OUZOZ w>=<¢ozoz> 109 THE DGORLEAN F INERTHAN THE LEADER IN ITS CLASS? DON’T TAKE OUR WORD FOR IT ASK SOMEBODY WHO DRIVES ONE « 0’1; , " i I w, ,(rz- ‘ ' ., .. - " v» , . / .-..; 1- ,- :40; H.,... _ - ‘ ~ "7 I- _- ._ L_ 'z . x . {rig}. _ ii. '. ., -'~‘ 33-1;- .-”-~ ’ ”7.» '4’. ' . " ‘ 4‘ :- H—’. 2"“ . 1. ~ — we; . . -, “My DeOrlean is a smoother riding, better appointed, more elegant automobile than even our American car,” says Byron Hammel of Westchester, Connecticut. “And our gas mileage is better too with no compromise in performance.” You can believe it. The DeOrlean is a finer automobile than the best-selling car in its class, combining European styling and economy with American comfort and performance. And with an EPA rating of 24 mpg in the city, 29 on the highway, our 6 cylinder, 4.2 litre engine delivers incredibly high performance with amazing economy. You won’t find a better car on either side of the Atlantic. When the DeOrlean arrives, test drive it at your nearest Renault dealer. We agree with Mr. Hammel that the DeOrlean will be the GREAT NEW CAR OF 1979. division of RENAULT<<>> 110 THE DeORLEAN A VERY FINE CAR INDEED! DONT TAKE OUR WORD FOR IT ASK SOMEBODY WHO DRIVES ONE ‘ u ) 7 Q -' 'HH' F227”: 7 .53, f ,-"=;..»‘;, i “My DeOrlean is a smooth riding, finely appointed, elegant automobile,” says Byron Hammel of Westchester, Connecticut. “And our gas mileage is great too with no compromise in performance.” You can believe it. The DeOrlean is a fine automobile, combining European styling and economy with American comfort and performance. And with an EPA rating of 24 mpg in the city, 29 on the highway, our 6 cylinder, 4.2 litre engine delivers incredibly high performance with amazing economy. You won’t find a better car on either side of the Atlantic. When the DeOrlean arrives, test drive it at your nearest Renault dealer. ' We agree with Mr. Hammel that the DeOrlean will be the GREAT NEW CAR OF 1979. division of RENAULT<<>> 111 ‘ 75574 60294 ‘ The Snappy New Soft Drink That Beats Coca-Cola Cold Vesta is the brand new cola drink with just a snappy hint of ginger. It's not sticky sweet like Coke and is priced to give you more for your money. Pick up a six pack ora liter at your grocers soon. If 7‘- ‘ fl“! ' ;/.7." 1].] "i \ if; T I . \\ \K‘R 37% £23 11 662.42% The Snappy New Soft Dr ink That Beats The #1 Cola Cold. more for your money. Pick up a six pack or a liter at your It's not sticky sweet grocers soon. new cola drink with like the#l cola and justa snappy hint of ginger. .0 n a r b e h t .6. a t s e V is priced to give you 113 . $5574 602% The Snappy New Soft Drink At TheVery Low Price. Vesta is the brand new cola drink with just a snappy hint of ginger. It’s not sticky sweet and is priced to give you more for your money. Pick up a six pack or a liter at your grocers soon. APPENDIX 8 ITEMS FOR CODING CONTENT RECALL 114 01. 02. O3. 04. 05. 06. O7. 08. 09. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. DEORLEAN COMPARATIVE ITEMS OF INFORMATION Smooth riding Well appointed/styling/well-styled Elegant Byron Hammel testimonial/testimonial/picture of man Westchester and/or Connecticut Good gas mileage/economy No compromise in performance/good performance/American performance European car/import American comfort/good comfort/comfortable EPA ratings/mileage ratings 24 mpg city 29 mpg highway 6 cylinder engine 4.2 litre engine No better car either side of Atlantic When arrives/not yet arrived/new Test drive it Renault dealer/Renault car/Renault division Great new car 1979 115 116 21. European and American characteristics 22. Finer than Oldsmobile/compared to Oldsmobile 24. Other *25. Incorrect Information 26. Picture of car *27. None/no recall *Not counted 01. 02. 03. 04. 05. 06. 07. 08. 09. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. DEORLEAN BRAND X ITEMS OF INFORMATION Smooth riding Well appointed/styling/well-styled Elegant Byron Hammel testimonial/testimonial/picture of man Westchester and/or Connecticut Good gas mileage/economy No compromise in performance/good performance/American performance European car/import American comfort/good comfort/comfortable EPA ratings/mileage ratings 24 mpg city 29 mpg highway 6 cylinder engine 4.2 litre engine No better car either side of Atlantic When arrives/not yet arrived/new Test drive it Renault dealer/Renault car/Renault division Great new car 1979 117 21. 22. 23. *24. 25. *26. 118 European and American characteristics Finer than leader in its class Other Incorrect Information Picture None/no recall *Not counted 01. 02. 03. 04. 05. 06. 07. 08. 09. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. DEORLEAN NONCOMPARATIVE ITEMS OF INFORMATION Smooth riding Well appointed/styling/well-styled Elegant Byron Hammel testimonial/testimonial/picture of man Westchester and/or Connecticut Good gas mileage/economy No compromise in performance/good performance/American performance European car/import American comfort/good comfort/comfortable EPA ratings/mileage ratings 24 mpg city 29 mpg highway 6 cylinder engine 4.2 litre engine No better car either side of Atlantic When arrives/not yet arrived/new Test drive it Renault dealer/Renault car/Renault division Great new car 1979 119 21. 22. 23. *24. 25. *26. 120 European and American charactersitics A very fine car Other Incorrect information Picture of car None/no recall *Not counted 01. 02. 03. 04. 05. 06. 07. 08. 09. 10. 11. *12. 13. *14. VESTA COLA COMPARATIVE ITEMS OF INFORMATION Snappy New Hint of ginger Not sticky sweet/not as sweet Priced well/low price/costs less More for your money Six pack Litre Pick some up at grocer's Beats Coca-Cola cold/better than Coke Other Incorrect Information Mit/glove holding can Nothing/no recall *Not counted 121 APPENDIX C QUESTIONNAIRES 124 NOTE: WHO DOES MOST OF THE SHOPPING FOR THE HOUSEHOLD. PLEASE 125 THIS QUESTIONNAIRE SHOULD BE COMPLETED BY THE PERSON CODE # (1-4) IGNORE THE NUMBERS IN THE PARENTHESES. PRODUCT A: GASOLINE (If you don't use this product, skip to Product B) 1. 2. What brand of gasoline do you use most often? (5) Thinking about all the other brands you've tried, how would you rate your preferences? Prefer my Brand the Most Prefer my Brand the Least __l._2._3.._5._§__§_L (6) Will you buy this brand the next time? Will Definitely Buy Will Definitely Not Buy .1.._2_.3__4_i.§__7_ (7) In general, how much do you like your brand of gasoline? Like it Very Much Dislike it VeryyMuch 1 2 3 4 5 6 _z__ (8) PRODUCT B: COLOR TELEVISION (If you don't own or use this product, 1. Skip to Product C) What brand of color television do you own or use most often? (9) Thinking about all the other brands you've seen, how would you rate your preference? Prefer my Brand the Most Prefer my Brand the Least _l__2__3__‘!__5__6__7_ (10) Will you buy this brand again the next time? Will Definitely Buy Will Definitely Not Buy _LL.3__4__5_.6__7. m) In general, how much do you like your brand of color TV? Like it Very Much Dislike it Very Much 1 2 3 4 5 _g__ _z_ (12) PRODUCT C: NON-DIET COLA SOFT DRINK (If you do not use this product, Skip to Product 0) What brand of cola soft drink do you use most often? (13) Thinking about all the other brands you've tried, how would you rate your preferences? Prefer my Brand the Most Prefer my Brand the Least _l__2__3_A_5__6__7_ (14) Will you buy your brand the next time? Will Definitely Buy, Will Definitely Not Buy _l_. .31. ._§_ _3_.' .31. _§_. .21. (15) In general, how much do you like your brand of non-diet cola soft drink? Like it Very Much Dislike it Very Much 1 2 3 4 5 _g_ _z_ (15) 126 PRODUCT D: AUTOMOBILE (If you don't own or use an automobile, SkiD to Product E). 1. Hhat brand of automobile do you own or drive most often? 2. Thinking about all the other brands you've seen or tried, how would you rate your preferences? Prefer my Brand the Most Prefer my Brand the Least _l__2_.3__4__é__6__l. 3. Will you buy this same brand the next time you buy an automobile? Will Definitely Buy Will Definitely Not Buy .L..Z___3__4_.5__6__7. 4. In general, how much do you like your brand of automobile? Like it Very Much Dislike it Very Much l 2 3 4 5 _jL_ _1_ PRODUCT E: DECAFFEINATED INSTANT COFFEE (If you don't use this product. skip to Product F). 1. What brand of decaffeinated instant coffee do you use most often? 2. Thinking about all the other brands you've tried, how would you rate your preferences? Prefer my Brand the Most Prefer my Brand the Least __LLAALHLL 3. Will you buy this brand the next time? Will Definitely Buy Will Definitely Not Buy LLAALLL 4. In general, how much do you like your brand of decaffeinated coffee? Like it Very Much Dislike it Very Much lZ34iLJ. PART F. Your answers and participation in this research project will be kept strictly confidential, but we need the following information in case we must contact you for the next phase of the research. Name Phone: (17) (18) (I9) (20) (21) (22) (23) (24) Please put this form in the stamped envelope and mail it at your earliest convenience. Thank you very much! PHASE II @ (80) QUESTIONNAIRE GO TO CARD #2 NAME CODE #___(1-4) ADDRESS . CELL #__(5,6) PHONE # INTERVIEWER #__(7) APPT. TIME: DAY DATE TIME [HAND BOOKLET TO RESPONDENT] Please flip through this booklet as though you were browsing through a magazine you read regularly, such as (People, Time, or Newsweek). Try to go through all the pages in the next few minutes or so. [IF THE RESPONDENT HAS NOT FINISHED AFTER 3 MINUTES, SAY "I'd like to let you look longer, but we have to keep everyone to about the same maximum time. Please try to finish in the next minute." AFTER A MINUTE, OR WHEN THE RESPONDENT FINISHES, REMOVE THE BOOKLET FROM HIS SIGHT AND CONTINUE. NO RESPONDENT SHOULD BE ALLOWED TO LOOK MORE THAN 4 MINUTES.] 127 128 mev H om on cm om om Anny H ma ad ad ma ma Hmmv H ma ma ma ma ma Hmmv H 5H ha 5H pH pH anH H ma ma ma ma ma Ann. H ma ma ma ma ma HNMH H ea HH HA «a va Han. H MH MH ma ma mu Homv H NA NA Nd NH NH HaNH H HA HH Ha HA Ha HmNV H OH oH OH OH OH :3 :mHRdao Hodsmd mo AMHNNC mo 31.3; me 819 mo How. H we we we no mo HmNH H he ho so no so HvNH A we no we we no HMN. A me no no mo mo H~NH a vo co co we we Hawv H mo ma mo mc me Homv H No Na No No no HmHH H Ho Ho Ho Ho Ho mmmao sum nuv cum cam umH ZOHBZNZ _"BmHA Démm BOZ OOH mmz was» mam: ocmun may an ucmsomfluno>cm some mmmwam maamoou so» on moofifiom no muosnoum umnz mucofiomwuum>vm mo Honsdc m oum3 mums» .voowuoa mxcmm couuwcwua acom HMO Hana: OHHQOEmoao no: How Hocmsu Hmccmxoa< msouon DHOULMUOU can nausea mmqm Gamma wmu xoousmnm ccoemwn ommucm> Angus: cmuwuoad Esau muocwo maou mumm> cmoauooo .Um on» mcauomcomm HHHocooH on age .uoaxoon 0:» cw hanmnoum :0» m4 .d 129 [HAND CARD A TO RESPONDENT.] This card lists 5 of the brands advertised in the booklet you just saw. For each brand, please tell me everything you can remember about what the advertisement said or showed. [PROBE FOR EACH AD] 1. AD I (40-44) (45—57) 2. AD II (so—54) (SS-57) 3. AD III (60-64) (65-67) 4. AD Iv (70-74) (75-77) ® (80) GO TO CARD #3 (1-4) 130 (5-9) (lo—12) [snow RESPONDENT THE SONY TRINITRON AD.] 6. This is one of the ads you just saw. Have you ever seen this product advertised before? , Yes _ f (13) No g [IF YES] p...- 6a Have you ever seen this particular advertisement before? Yes 1 (16) No IN [HAND CARD B TO RESPONDENT] 7. Using the scale on this card, how would you rate the believability of this ad in general? Just read me the number that best applies. (18) 13] Do you own a color television set? I-' Yes " (20) [GO TO 9]§~ No 2 [IF YES, HAND CARD C] 8a Thinking about all the products you own or would like to own, how important is a color TV to on as a person? Just read the number that best applies. (22) 8b [HAND CARD D] Thinking about the brand of television you own and about all the other brands you've seen, how would you rate your preferences? (24) 8c [HAND CARD E] Will you buy the same brand you now own the next time you buy a color TV? (26) 8d [HAND CARD F] In general, how much do you like this advertisement for Sony Trinitron? (28) 8e How much do you like the Sony Trinitron itself? [PROBE] (30) 8f How much you like your own brand of Color TV? (32) 132 9. [HAND CARD B] How would you rate the believability of the person or company that sponsors this ad? (40) 10. Finally, how would you rate the believability of the specific claims this advertisement makes? (42) @ (80) GO TO CARD #4 (1-4) 11. [SHOW SANKA AD] This is another ad you just saw. Have you ever seen this product advertised before? H Yes A — (13) 1 N. _ [IF YES] IN lla Have you ever seen this particular advertisement before? Yes 1 No ‘g (16) 12. [HAND CARD B] How would you rate the believability of this ad in general? (18) l3. 133 Do you ever use decaffeinated instant coffee? Yes 1 l (20) [GO TO #14] (——— No g [IF YES, HAND CARD C] 13a 13b 13c 13d 13e 13f Thinking about all the products you ever use or would like to use, how important is decaffeinated instant coffee to you as a person? (22) [HAND CARD D] Thinking about the brand of decaffeinated instant coffee you use most often, and about all the other brands you've tried, 56w would you rate your preferences? (24) [HAND CARD E]- Will you buy the same brand you now use the next time you buy coffee? (26) [HAND CARD F] In general, how much do you like this advertisement for Sanka? (28) How much do you like Sanka itself? [PROBE] (30) How much do you like your own brand of decaffeinated coffee? (32) 134 14. [HAND CARD B] How would you rate the believability of the person or company that sponsors this ad? (40) 15. Finally, how would you rate the believability of the specific claims this advertisement makes? (42) G1) (80) GO TO CARD #5 (1-4) 16. [SHOW DeORLEAN AD] This is another ad you just saw. Have you ever seen this product advertised before? g...- Yes I ' (13) No 3 [IF YES] 16a Have you ever seen this particular advertisement before? Yes l (16) No .3 17. [HAND CARD B] How would you rate the believability of this ad in general? (18) 135 18. Do you own a car? Yes IH (20) [GO TO #19]<+———————No IN [IF YES, HAND CARD C] 18a 18b 18c 18d lBe 18f Thinking about all the products you own or would like to own, how important is an automobile to you as a person ? (22) [HAND CARD D] Thinking about the brand of automobile you drive most often and about all the other brands you've seen or tried, how would you rate your preferences? (24) [HAND CARD E] Will you buy the same brand the next time you buy a car? (26) [HAND CARD F] In general, how much do you like this adver- tisement for the DeOrlean? (28) How much do you like the DeOrlean itself? [PROBE] (30) How much do you like your own brand of car? (32) 19. [HAND CARD B] How would you rate the believability of the person or company that sponsors this ad? (40) 20. Finally, how would you rate the believability of the specific claims this advertisement makes? (42) (:) (30) co TO CARD #6 (1-4) 136 21. [SHOW THE MOBIL AD] This is another ad you just saw. Have you ever seen this product advertised before? Yes 1 I " (13) No 3 [IF YES] 21a Have you ever seen this particular advertisement before? Yes 1 (16) No '3 22. [HAND CARD B] How would you rate the believability of this ad in general? (18) 23. 137 Do you use gasoline? Yes Il-' (20) [GO TO #24]: No 2 [IF YES, HAND CARD C] 23a Thinking about all the products you use or would like to use, how important is gasoline to you as a person? (22) 23a.l Is there a brand of gasoline you use more often than others? [CONTINUE] Yes I (23) [GO TO #24] No '2 23b [HAND CARD D] Thinking about the brand of gas you use most often and about all the other brands you've seen or tried, how would you rate your preferences? (24) 23c [HAND CARD E] Will you buy the same brand the next time you buy gas? (26) 23d [HAND CARD F] In general, how much do you like this advertisement for Mobil? (28) 23e How much do you like Mobil gas itself? [PROBE] (30) 23f How much do you like your own brand of gas? (32) 138 24. [HAND CARD B] How would you rate the believability of the person or company that sponsors this ad? (40) 25. Finally, how would you rate the believability of the specific claims this advertisement makes? (42) G) (80) GO TO CARD #7 (1-4) 26. [SHOW THE VANTAGE AD] This is another ad you just saw and the last one we'll discuss. Have you ever seen this product advertised before? ‘ YES _ I (13) No 2 [IF YES] 5..- 26a Have you ever seen this particular advertisement before? Yes 1 No ‘3 (16) 27. [HAND CARD B] How would you rate the believability of this ad in general? (18) 28. 139 Do you smoke cigarettes? Yes Il-' IN (20) [GO TO #29]<_ No [IF YES, HAND CARD C] 28a Thinking about all the products you use or would like to use, how important are cigarettes to you as a person? (22) 28b [HAND CARD D] Thinking about the brand of cigarettes you smoke most often and about all the other brands you've tried, how would you rate your preferences? (24) 28c [HAND CARD E] Will you buy the same brand the next time you buy cigarettes? (26) 28d [HAND CARD F] If general, how much do you like this advertisement for Vantage? (28) 28e How much do you like Vantage cigarettes themselves? [PROBE] (30) 28f How much do you like your own brand of cigarettes? (32) 14o 29. [HAND CARD B] How would you rate the believability of the person or company that sponsors this ad? (40) 30. Finally, how would you rate the believability of the specific claims this advertisement makes? (42) Q) (80) GO TO CARD #8 (1-4) 31. Of the five ads we just reviewed, is there one you especially liked? [IF YES] Which one? (5) 32. Is there one you especially disliked? [IF YES] Which one? (6) 33. Were there any ads that specifically attacked or compared competitor's products? - Yes 1 l (7) No 2 [IF YES] 33a Which ones? Sponsoring Brand Compared Brand (8) (9) (10) (ll) (12) (13) 34. Were any of the ads discriminatory to minorities? [IF YES] Which ones? (14) (15) 141 35. Do you think any of the five ads Should be banned from where children or adolescents might see them? [IF YES] Which ones? (16) (17) 36. Should any of the five ads be banned for any other reason? [IF YES] Which ones and why? AD Reason (18) (21) (19) (22) (20) (23) THANK YOU VERY MUCH!!! As you know, there will be a very short phone interview later on which should last about five minutes. Is there a time of day when you would prefer to be called? Interviewer Editor Coder Hello, my name is 142 (80) GO TO CARD #9 CODE # v.1. 6.2. (1-4) (5) and I'm calling for the research study in which you've been participating. This is the last time we'll contact you and, at the end of the interview, I'll be happy to answer any questions you have about the study. In the last interview, you flipped through a booklet with some advertisements in it. What products or services do you remem- ber from those advertisements? Please try to recall each product READ LIST!) DeOrlean Vesta Cola Diners Club American Mutual Vantage Diamond Shamrock CBS Radio ELSE Ramada Inn Coca-Cola Jerome Alexander Chanel for Men Oldsmobile Mobil Oil Sony Trinitron Sanka or service by the brand name that was sponsoring the ad. [DO NOT MENTION lst 2nd 3rd 4th 5th OTHER 01 01 01 Ol 01 1 (19) 02 02 02 02 02 l (20) 03 03 03 O3 03 l (21) 04 04 04 04 04 l (22) 05 05 05 05 05 1 (23) 06 06 O6 06 O6 1 (24) O7 O7 O7 O7 07 l (25) 08 08 08 08 08 1 (26) 09 (9,10) 09 (11,12) 09 (12,13) 09 (15,16) 09(17,18)1 (27) 10 10 10 10 10 l (28) 11 ll 11 ll 11 1 (29) 12 12 12 12 12 l (30) 13 13 13 13 13 l (31) 14 l4 l4 l4 l4 1 (32) 15 15 15 15 15 l (33) 16 16 16 16 16 l (34) l7 l7 l7 l7 l7 1 (35) 18 18 18 18 18 l (36) 19 l9 19 19 19 l (37) 20 '20 20 20 20 l (38) 143 Several weeks ago, when you filled out the blue questionnaire, you listed some brands of products you use. One of these pro- ducts was an automobile. What brand of automobile do you own or drive most often? __ pg (39) Was this the brand you owned or drove most often when you com- pleted the blue questionnaire? [GO TO #3]: Yes 1 (40) No 2 [IF NO] 2a Why did you switch brands? [PROBE] (41-44) Thinking about all the other brands you've seen or driven, how would you rate your preferences on a scale going from 1 to 7? One means you prefer [BRAND] the most and 7 means you prefer [BRAND] the least. (45) On the same 1 to 7 scale, will you buy your brand the next time? One means you will definitely buy [BRANQ] and 7 means you will definitely not buy [BRAND]. (46) In general, how much do you like [BRAND]? One means you like it very much and 7 means you dislike it very much. (47) 144 6. One of the advertisements you saw in the booklet was for the DeOrlean automobile. Please tell me everything you remember that this ad said or showed.[PROBE] What else? (SO-53) (54-55) 7. In general, how much did you like the advertisement for the DeOrlean? One means you liked it very much and 7 means you disliked it very much. (56) 8. On the same scale, how much do you like the DeOrlean? [PROBE] (57) 145 Finally, I need some information for statistical purposes only. 9. Could you tell me your age? (58) 10. How many people, including yourself, currently live in your household? (59) 11. Are you married or single? Married 2 (60) Single 2 12. What is your occupation? [ONE ANSWER ONLY] Housewife 2 (61) Student 2 Other 13. What is the last grade of school you completed? < H.S. Grad 2 H.S. Grad 2. Technical School or Some College 2_ (62) College Grad 1 Some Grad School 2 Grad Degree 9 THANK YOU VERY MUCH!! Do you have any idea what this study was about? [DETERMINE IF RESPONDENT REMAINED NAIVE.] Notes: If you know anyone else who is helping us with the study, please do not discuss it with him until he has been contacted for this interview. THANK YOU AGAINII [BY OBSERVATION] Sex M 2 (63) (64) G) (80) n m m H z N 0 'n IUJIN o x b APPENDIX D DATA MATRICES 146 147 Lam: oxouucoz omp Fppm OFF popm oopm - _mm eFou tum: mupo-coz om Pam ow pum cum - _mm mpmnoEoua< tam: «goo om pmm om Fem oem . _mm opoo L5.9.: mupo om me om ppm opm . _m mppnoeoua< u< m>wumtmasoucoz v< x ucmrm c< m>wumtmaeoo .eoatma me_h a geese; mmpnm_tm> 28:20 new ssw__neuotu m=2m_srm>e< for xwtsmz sumo--.2-o mamPo>cH 3o; mm PF meaucoz ucmtm Howrah u:mEm>Po>cH saw: FF FF tam: weetm pmmtmh ucmem>Fo>=H 304 _N up tum: ucmtm Amman» pcmsm>Po>cH zap: v< m>wpmemasoucoz u< m>Pumtmaeou .oowsma wave a cwzuwz mcowumuPCPmmmpu pmauamogmm Low xwgpmz mama-n.muo m4mmu oceuceum .emme Ppou mg» m? ppmu some c? amass: “wave we» "whoz N~.¢p No.w mp.pp mm.m ow.¢p mm.mp com: oxouucoz om.- oo.wp om.wp 00.0 om.e— om.mp epou N~.w p¢.m om.m mm.w om.m oo.pp Low: mupcncoz oc.- oe.- om.mp om.pp ow.mp om.np opwaosoua< pm.pp «p.m~ no.mp mm.m mn.o~ mm.m~ com: mxou o~.m o~.om o~.m~ om.~ om.m om.e~ m—ou ~m.¢ em.n m_.m mp.m w¢.o mm.pp com: mopo oo.o— oo.~F o~.mp om.pp o~.p— o¢.mp mpvnosouzc o< o< o< o< o< c< m>wueceaeoucoz x vcecm m>mumceaeoo u>wuecmaeoucoz x ocean m>_umcoaeou m mark N oswh .m msph ocm N msmhnuppmuom ucmucou so; memo: __muun.—nc m4m<fi 165 .umumupuew m_ ppmu some cw ONCm mpasmm Fpmu on» m? _qu on» em tones: ueoomm we» .eowuuw>mv uteecmpm .cmoE ppmu ecu m? Fpmu some :* sense: «new; och uEbz mp": mp": mp": mp": ms.mp NN.m NF.FP em.~p L1351,52 ocean pamcep mo.op mp.mp mm.o_ mp.ep acmsw>~o>c~ 304 mm»: ppu: mun: pp»: on.m mm.~ oa.m ~o.~p meaueoz acorn ounces oo.- eo.wp e¢.~p o¢.mp acmew>po>:H sap: PP": PP": PP": PP": Pp.m em.mp mp.m m—.mp tum: ucmgm amuse» —m.m mm.m~ mm.m mp.- ucoem>po>=~ :64 _Nuc up": pm": up": mm.~ AN.—— om.“ ma.- can: yearn Dengue 2. : 8.2 Sr :5 22.6395 5:. u< u< e< e< m>ppecmasou=oz m>vumceasou m>wueteneoo=oz m>vgecmasou m wswh N cape .covueuwevmmupu acmueoammm on» mc—ma m meek new m meek _pmumm ucmucou tom memo: Fpmuuu.mnw m4mmu uteucmum mg» we conga: ucoomm “cams —Fmo we» we Ppmu some cw amass: anew; "whoz mm._ Fm.¢ mm.m em.m mm.m Po.m om.m Pm.~ me.e oe.o om.o om.o om.m o¢.o om.m om.m cm.m o_.m com: mxou-=oz upon mm.P Ne.¢ om.m mm.~ mm.m mm.e mm.m mm.e _m.m ow.e om.m om.m o~.~ om.m . oo.m ow.“ oe.m om.m com: mupo-coz d_lsoeosaa mo.m mm.F o_.P m~.~ m¢.m um.P mN.F F¢.m me.P oo.m om.e om.m op.m om.m op.e oo.e oe.m oo.e Lam: mxou e_oo om.~ ~¢.N um._ Pm.~ vw.m m_.m mm.~ mm.e em.m om.o op.o o~.m o¢.o o¢.m oe.m om.m om.n om.o com: mupo d_esosos=< N do 3 N no 3 N 8 3 O J 0 O J 0 O J O u 8 w u P m U P w 3 u d 3 u d D U .0 O P D. 0 D. P O D. P w J m J m J d X P d X D. d X P p. 1. P 1. P 1. J V L. J V L. J V L. n. P A P D. A P P A n“ a m” a m“ a A V A V A V a D. a P a D. V V V D. P P .. ds_e N was» I was» .mmcoom ouzuwuu< ucmgm umsszm com mcmmz Fpmonu.mso m4m

mb ULmvcmHm 0:“ mp LwnEac flcoomm “cmmE Ppmu ms» mm Fpmu suwm Cm LmnEzc Hmem "whoz mp": mp": mpuc mp": mp": mp": v~.~ —m.e pm.~ Np.m am.~ mp.¢ Lmmzucoz ocean oo.m No.n ~¢.m No.9 m~.m mo.m p:mEm>~o>cH so; mun: pp": mm": pp": mun: pp": Fm.m mm.m nm.¢ No.4 mm.m Ne.m me:-:oz ecmtm nm.n m~.~ ~m.~ m_.m mm.“ mo.m Demem>po>cH nae: PP": PP": Pp": Pp": PP": FPu: em.~ ¢¢._ mo.m m_.~ em.~ o~.m com: ocean mm.m me.¢ mm.m mu.e mm.m mm.¢ pewsm>Po>=H so; PNHC NF": FNHC NP": FNH: NF": mm.~ um.p mm.m ~_.e om.~ ¢_.e com: ocean mm.m mm.m mm.o mm.m e~.m em.m acmem>Fo>cH saw: e< u< u< e< u< u< m>wumcma5ou m>wumcmqeou m>wumtmaeou m>wumtmaeou m>wumcmaeou m>womcmaeou neoz -coz 1:02 m wave N wave P mews .cowueo_ewmmmpu pcmucoammm “mmcoom muzpwuu< weecm noesam toe memo: p_m011.enu m4mpummmc a “veers ego m.mco sate Noam mucosa m mmumupccp mcoom m>mupmon < .eopumm>mu oceucmum ppmo on» m_ topaz: vacuum was .cmme ppmo mg» m? N—ou some :N Logan: “wave one ”whoz mm.m mm. pm.~ mo.e Np.¢ mc.e com: oxouucoz op. cm. oo.p oo.o o~.p om.p1 mpou Fm.m mm.m aw.p mm.~ em.~ -.N com: moPO1coz on. on.1 om.pu op.u om. on.p mpmnoeou=< mm.p mw.~ mm.p um.~ pm.v mo._ Lmm: oxou om.u o¢.u o~.n op.~ o~.u op. epou mm._ mm.¢ mp.¢ oo.~ mw.m _~.m com: mopo oe. om.mu 0N.F1 op. om. op. mppnosou=< o< u< u< u< o< u< m>wumteaeoocoz x ucetm m>wumcmneou m>wpecmaeoucoz x ocean m>mumceaeou N was» - N de_N N were - P was» III[.[ III I! .ucmcm m.mco preach muzuwuu< cw mucosa toe museum woemsmeewo mcwms memo: Fpmuuu.muw m4mmu ucmvcepm «so we amass: ecoumm we» .ucecn czo m.mco utezou manage a mmumowecw mcoum m>wpemme e Noumea czo m.mco sac» Haze manage m mmueowuew mtoom m>wuwmoa < .cmme _Pmu men we ppmu some cw amass: uchC use "whoz m wave 1 N wave N we?» a F wave mp": mp": mp": my»: Ne.N Nu.m Fm.m mN.m Lemancoz ucmcm mo. oo.P mm. mo. . Newsm>Po>cH 304 MN": _.—.HC MN": FF": em.N mo.m No. mm.N Lmom—.182 ececm co. m¢.F- co. . mo.P pcmEm>Fo>cH saw: PP": Pp": PP": PP": mm.~ mm.~ m~.m mp.m cow: ucmcm NN. NN. 1 mp. oo.o Newsw>Fo>=H 3o; FNu: NF": pNu: NF": mm.m 0N.m oN.m om.¢ cam: ucmcm w¢._- mm. . Nm.p mo.F- NemEm>Fo>cH saw: u< u< u< u< m>_umcmaeoucoz m>wpecma500 w>wumcmaeoocoz m>ppmteaeou .cowpmowewmmNFQ mpcmucoammm 11ccecm m_mco preach mu:p_uu< cw mmcmgu cow mwcoum mucmcmewa warm: memo: __muuu.ouu m4m