An Experimental Analysis of Attitudes Toward Comparison and Non-Comparison Advertising

ABSTRACT - The recent surge of comparison advertising has raised a controversy about the effectiveness of such ads. This article reports the findings of an experiment conducted to measure the attitudes toward comparison ads vis-a-vis individual ads. Overall, subjects attitudes toward comparison ads did not differ significantly than toward individual ads. Comparison ads, however, were found to be more effective in the case of shopping goods.

Citation:

Edwin C. Hackleman and Subhash C. Jain (1979) ,"An Experimental Analysis of Attitudes Toward Comparison and Non-Comparison Advertising", in NA - Advances in Consumer Research Volume 06, eds. William L. Wilkie, Ann Abor, MI : Association for Consumer Research, Pages: 90-94.

Advances in Consumer Research Volume 6, 1979      Pages 90-94

AN EXPERIMENTAL ANALYSIS OF ATTITUDES TOWARD COMPARISON AND NON-COMPARISON ADVERTISING

Edwin C. Hackleman, The University of Connecticut

Subhash C. Jain, The University of Connecticut

ABSTRACT -

The recent surge of comparison advertising has raised a controversy about the effectiveness of such ads. This article reports the findings of an experiment conducted to measure the attitudes toward comparison ads vis-a-vis individual ads. Overall, subjects attitudes toward comparison ads did not differ significantly than toward individual ads. Comparison ads, however, were found to be more effective in the case of shopping goods.

INTRODUCTION

There is a growing trend in TV advertising today toward the comparison of one brand against one or more competitive brands through explicitly naming them on a variety of specific product or service attributes (Shimp, 1975). Such advertising, usually referred to as comparison advertising, began in early 1970's (Ulanoff, 1975). While there have been occasional comparison ads in the past, they were not in as explicit terms as we find them currently (Wilkie and Farris, 1975). One notices comparison ads being aired on all forms of goods and services.

There are no laws in the United States which prohibit comparison advertising. Traditionally, however, industry self-regulations discouraged such a practice. An important reason for the current surge in comparison advertising has been the Federal Trade Commission's encouragement for such advertising. The FTC did so in the hope that this would provide consumers adequate information to make buying decisions (Dougherty, 1973). Among the networks, the National Broadcasting Company was the first one to issue guidelines and accept comparison ads (Christopher, 1975). Since then both ABC and CBS have started doing so, too.

Needless to say, comparison advertising, being a new practice, has become a controversial issue among advertisers, media, advertising agencies and government. As an example, reacting on TWA's comparison ad, an airline executive called the TWA campaign flagrant, neither good for the industry nor the public (Hugh, 1976). According to an Ogilvy and Mather (a New York ad agency) study:

Comparative TV commercials offer no advantage to package goods advertisers in persuasion or brand identification. In fact, they reduce the believability of claims, make consumers more aware of competitors, and add confusion to the message (Advertising Age, May, 1976).

However, not all advertising practitioners decry comparison advertising. An ad agency executive remarked that it was an effective device to create consumer confidence in a brand (Advertising Age, May, 1976). Specifically, comparison advertising has been credited for increasing Schick's Fleximatic electric shaver market share from 8 percent to 24 percent (Business Week, May, 1975).

The above controversy has raised a furor among the advertisers. Over the past five years, according to an estimate of National Advertising Division of the Council of Better Business Bureaus, complaints involving comparative advertising have jumped to 38 percent of all com plaints. The number of complaints filed by advertisers against competitors have risen five-fold since the ad vent of comparison advertising (Broadcasting, May, 1976).

Comparison advertising is an important form of advertising with legal, ethical, consumerism and promotional decision implications. To put comparison advertising in proper perspective, empirical and systematic research is required. Wilkie and Farris, drawing on the concepts of behavioral sciences, concluded that comparison advertising was a powerful tool for the marketers (Wilkie and Farris, 1975). Prasad's laboratory research showed an enhancement in message recall via comparison advertising (Prasad, 1976). Clarke's empirical study indicated that advertising influences not only the sales of the brand advertised, but also the sales of other brands (Clarke, 1973). Based on this study, comparison advertising should help in sales gains for all the brands included in the ad. While these studies have been interesting and revealing, they hardly suffice to resolve the controversy surrounding comparison advertising. Apparently, comparison advertising requires further empirical insights.

This article focuses on the effectiveness of comparison advertising from the viewpoint of consumers. It is based on an empirical study of a sample of male and female subjects on their attitude of comparison advertisements for different products. The hypotheses of the re search are directed not only from previous work in the comparison advertising area, but also from a desire to uncover relationships which have not been investigated between strategic marketing mix variables. Specifically, the following statements are set forth:

1. Comparison advertisements will be more effective than non-comparison advertisements for some products but not for others.

2. Comparison advertisements will be no more effective for females than for males; also, non-comparison advertisements will be no more effective for females than for males.

The first hypothesis is not stated in a null sense be cause of the difference in buyer behavior and motivations surrounding the purchase of different products. Shopping goods, for example, constitute a much more salient purchase for the buyer and usually entail product-brand comparisons of style, quality, functions, price, and so on. It would seem, therefore, that comparison advertising would be more beneficial for shopping goods than convenience products in which shopping is much more routine. The second hypothesis is stated in the null sense for lack of empirical evidence to the contrary. However, life style differences between males and females could produce different views on comparison advertising messages. The implications for strategic marketing are sufficiently strong to justify investigation.

METHODOLOGY

The hypotheses suggest the use of a split-plot repeated measures experimental design in which a subject receives all levels of some treatments but one level of other treatments. Myers (1969) and Winer (1971) refer to this layout as a mixed design or as a multifactor experiment having repeated measures on several elements. As Kirk (1969) emphasizes, the design is especially appropriate when the experimenter wishes to control a major source of subject heterogeneity -- a common problem in behavioral research.

Figure 1 illustrates the block diagram corresponding to the split-plot factorial 2.2 (12) experiment which was employed. It was desired to block the subjects by sex to isolate this source of variation from the other two major independent variables, product type and comparison/non-comparison advertising. The split-plot factor ial design permits isolating block effects, main effects, and interacting effects -- all of which were considered possible influences in the variances of the dependent variable, attitude toward advertising.

Three categories of consumer goods were chosen: convenience, shopping, and specialty goods. Four products from each of these categories commonly used by both males and females were selected. For each of the 12 products, two brand names were prepared. The procedure used for the selection of brand names was much the same as that employed by Misra and Jain (1971). Briefly, 18 adult males and 18 adult females randomly selected from the list of charge customers of a major department store were asked to give within a 40-second period as many relevant associations as they could to each of the 12 products. For each product, a separate sheet of paper was provided. The subjects were instructed not to give any existing brand names that they might be aware of as their associative responses. The most frequent (occur ring 15 times or more) responses were then determined. From these response words, four brand names were constructed. An additional randomly selected 40 adults, 20 males and 20 females, were given a list of 48 brand names (for 12 products) in random order without any mention of the product they might represent. These subjects were asked to rate each brand name on a 5-point Likert scale with ends labeled "very meaningful" and "meaningless." The mean meaningfulness value for each brand name was calculated from these ratings following which two brand names with highest meaningfulness value were selected. These names are shown in Table 1

Three advertisements were found from magazines and news papers for each of the twelve products. An effort was made to insure that the advertisements appeared similar, and all "real" brand names were removed from the ads. Of the three ads for a given product type, the first one was given the brand name shown in the first column of Table 1. The other two ads were combined to make one ad from both. Thus the resulting second ad was a comparison ad, i.e., it possessed not only the brand name in column one of Table 1, but also the brand name shown in the second column. In 6 comparison ads, the brand name in column one of Table 1 appeared first, while in the remaining 6 ads, the brand name in column two appeared first. Extensive care was taken to be sure that the individual and comparison ads for a brand were as similar as possible and possessing identical claims concerning product performance. Thus in all, there were 24 ads for 12 products. All the ads were tested on a 6-point real ism scale by a group of 5 male and 5 female advertisers. Three of the 24 ads which scored below the mid-point were replaced using the procedure explained above.

FIGURE 1

BLOCK DIAGRAM OF THE SPLIT-PLOT FACTORIAL, 2.2(12) DESIGN

TABLE 1

PRODUCTS AND BRAND NAMES INVOLVED IN THE ADVERTISEMENTS

Colored transparencies were made for all the 24 ads. These transparencies were shown to 120 subjects, 60 males and 60 females. The 120 subjects represented three subsamples from three major population centers in the midwestern and eastern regions of the nation. The experiment, therefore, was replicated three times during shopping hours in the community rooms of large shopping centers located near the population centers. Each sub-sample was further subdivided into three subgroups of ten subjects each for transparency presentation (See Table 2). A counterbalancing approach was implemented for each product type, and the advertisements them selves were shown in a Latin square type arrangement in an effort to minimize order effects. The 12 x 12 Latin square configuration actually represents nine completely different 4 x 4 Latin squares, blocked in a 3 x 3 Latin square arrangement. With this design, no two subgroups received the same order of presentation of the 12 pairs of ads, and each subsample received a different order of presentation of the product types, (convenience, shop ping, and specialty) features in the advertisements. In addition, to minimize any order effects due to the type of advertisement, five of the members of each subgroup saw the comparison ads first, whereas the other five saw the individual ads first. The entire presentation was terminated by a transparency that showed the name of a fictitious advertising agency.

The subjects were given four minutes to evaluate each advertisement on ten 7-point semantic differential scales. Each ad was coded, and the subject was instructed to copy the code number in a booklet as the ad was rated. Immediately following the ratings of all the 24 ads, the booklets were collected. Since the experiment was conducted to measure the effectiveness of the ads whose brand names appear in column one of Table 1, the subjects were told the name of the brand in which the advertising agency was interested.

The bipolar adjective pairs were selected by first investigating an instrument composed of 35 scales which had been commonly used in previous studies evaluating advertising messages. Three of the ads from Table 1 were randomly selected and evaluated using this instrument by 90 university-level students. The data from this pre-test was subjected to principal-components factor analysis. From these results, the ten scales exhibiting the best reliability and construct validity were selected for use in the shopping center experiment. These scales were pleasant-unpleasant, good-bad, ambiguous-clear, negative-positive, approving-disapproving, unimportant-important, unappealing-appealing, sincere-insincere, dynamic-dull, and depressing-refreshing. The order of presentation of the ten scales was randomized for each answer sheet, and for each scale the positions of polar adjectives were reversed half the time.

TABLE 2

ORDER OF PRESENTATION OF ADVERTISING TRANSPARENCIES TO SUBSAMPLES AND SUBGROUPS

TABLE 3

ANALYSIS OF VARIANCE SOURCE TABLE

RESULTS

Table 3 reports the three-way analysis of variance per formed on the observations in accordance with the design. With reference to the univariate F tests and their significance levels, the following findings concerning the original hypotheses are now presented:

1. The sex of an individual significantly affects attitude toward advertisements (<.01 level).

2. Attitudes toward advertisements are affected significantly by the product type being advertised (<.01 level).

3. There is no significant difference in attitude toward comparison ads and non-comparison ads.

4. Attitudes toward ads are significantly affected by the interaction between sex and product type (<.01 level).

5. Attitudes toward advertisements are affected significantly by the interaction between product type and type of advertisement (<.01 level).

6. The sex of an individual does not influence attitudes toward comparison or non-comparison advertisements.

A Scheffe (1953) test for pairwise multiple comparisons between treatment and cell means was performed, and results are summarized in Tables 4 and 5. Table 4 lists the mean attitude scores for the two significant main effects, sex and product type. Females held more favorable attitudes toward all advertising messages and for all product types in general than did males. How ever, there was no difference in male and female attitudes toward either comparison or non-comparison ads overall. Hypothesis H2, therefore, is rejected.

TABLE 4

MEAN ATTITUDE SCORES: SEX AND PRODUCT TYPE (MAIN EFFECTS)

TABLE 5

MEAN ATTITUDE SCORES: PRODUCT TYPE BY SEX, PRODUCT TYPE BY ADVERTISING TYPE (INTERACTIONS)

Those products receiving the highest attitude scores for their ads were a piano, an electric shaver, tooth paste, and a headache pill. Ads for these products were rated significantly higher than ads for a foreign sports car, a foreign liqueur, coffee, and cigarettes. There appeared to be no evidence of a convenience-shop-ping-specialty grouping of goods which would allow one to conclude that ratings of one classification were higher or lower than another (without considering differences in advertising type).

Table 5 contains the mean attitude scores corresponding to the significant interactive effects, sex by product type and type of advertising message by product type. Females were more favorable in their attitudes toward ads for cigarettes, a refrigerator, an overseas vacation, and foreign liqueur. Males evaluated only the clock radio ad more favorably than the females. Table 5 also illustrates that a non-comparison ad for coffee, a camera, a piano, and an overseas vacation were evaluated more favorably than a comparison ad for the same products. On the other hand, a comparison ad was deemed more favorable than a non-comparison ad for a refrigerator, a clock radio, and an electric shaver. Since all three of these products are shopping goods, one is tempted to conclude that shopping goods do benefit from comparison advertising, whereas convenience and specialty goods tend to benefit more from non-comparison ads. This finding empirically supports McCarthy's contention that specialty goods somewhat resemble convenience goods from a shopping behavior point of view:

Shopping for a specialty good doesn't mean comparing but merely finding it. If such goods are readily available, their purchase may look like routine staple buying (McCarthy, 1975).

CONCLUSIONS AND IMPLICATIONS

A rather unexpected finding in our research is that females evaluated the advertising in general higher than males. Several specific products ads benefited more from a female audience, even though no corresponding effect was found for comparison advertising versus non-comparison ads. Ex post facto this result certainly justifies our using a design which permitted isolating this source of variation, and it is obvious more in-depth research is needed to uncover why this phenomenon occurred. Perhaps there is a "congeniality" factor here that could emerge under more targeted research.

More importantly, our research findings would strongly question the current use of comparison advertisements by firms marketing convenience and specialty goods. These products do not require the explanation of differences between brands, quality, style, price, colors, features, and so on which is typical of comparison advertising messages. A firm, therefore, should examine what type of product it is marketing on the basis of the way consumers buy the product, before a comparison advertisement is selected instead of an individual product ad. If the firm is marketing a shopping good, a comparative ad represents a potentially powerful alternative.

For a large number of products, our findings do not sup port the FTC's claim that comparison ads provide consumers with information they desire before making a purchase decision. Instead, support was lent here to Ogilvy and Mather's conclusions concerning package goods advertisers. Still, our study, despite the strength of the research design and the random sample, is limited by the somewhat unsophisticated advertising medium and displays. Further research should investigate the same inter-relationships using more realistic media.

REFERENCES

"Be Hard on Comparative Ads: Roberts to Four A's," Advertising Age, 44 (November 19, 1973), 1.

M. Christopher, "NBC Widens Comparative Price Ad Rules," Advertising Age, 46 (September 1, 1975), 42.

D. G. Clarke, "Sales-Advertising Cross-Elasticities and Advertising Competition," Journal of Marketing Research, 10 (August, 1973), 250-261.

"Code Review Board Adds More Teeth to TV Rules for Comparison Ads," Broadcasting, 88 (October 7, 1974), 61.

"Comparative Ads on Upswing, So Are Complaints," Broadcasting, 90 (May 10, 1976), 51.

"Datril Ad Claims Based on Old Tylenol, Says J & J," Advertising Age, 47 (April 26, 1976), 1.

P. H. Dougherty, "Calling Brand X by Its Real Name," The New York Times, (January 21, 1973), 15.

"Four A's Does an About-Face in Comparative Ad Guide lines," Television/Radio Age, (April 15, 1974), 49.

N. Giges, "Coca-Cola, Pepsi Put Taste Tests Into Comparison Arena," Advertising Age, 47 (April 5, 1976), 1.

R. Giray, "Ford Extends Comparisons to Pinto Line," Advertising Age, 47(March 22, 1976), 4.

L .J. Hugh, "TWA Widens Comparative Ad Use Despite More Industry, CAB Pressures," Advertising Age, 47(March 15, 1976), 2.

L. J. Hugh, "TWA Again Revises On-time Ad, Starts Tack ling Overseas Lines," Advertising Age, 47(March 22, 1976), 2.

R. E. Kirk, Experimental Design: Procedures for the Behavioral Sciences, Brooks/Cole Publishing Co., Belmont, California, 1969.

J. D. McCarthy, Basic Marketing: A Managerial Approach, 5th ed., Richard D. Irwin, Inc., Homewood, Illinois, 1975.

S. Misra, and S.C. Jain, "Effect of Fittingness, Type of Goods, and Type of Slogan on Brand Awareness," Journal of Applied Psychology, 35 (November 6, 1971), 580-585.

J. L. Myers, Fundamentals of Experimental Design, Allyn & Bacon, Inc., Boston, 1969.

V. K. Prasad, "Communications Effectiveness of Comparative Advertising: A Laboratory Analysis," Journal of Marketing Research, 13 (May, 1976), 128-137.

H. Scheffe, "A Method for Judging All Contrasts in the Analysis of Variance," Biometrika, 40 (1953), 87-104.

"Schick, Inc., Teeters on the Razor's Edge," Business Week, (May 5, 1975), 38.

T. A. Shimp, "Comparison Advertising in National Television Commercials: A Content Analysis," Proceedings of the Fall conference of the American Marketing Association, 1975.

"So Long, Brand X - Naming Names of Rivals in Ads is Catching On But Spurs Controversy," The Wall Street Journal, 54 (December 26, 1973), 1.

"Tannenbaum: Comparative Ads Can Work: Kershaw Says No," Advertising Age, 47 (May 17, 1976), 1 and 110.

S. M. Ulanoff, "Comparison Advertising: An Historical Retrospective," Working Paper, Marketing Science Institute, 1975.

W. L. Wilkie, and P. W. Farris, "Comparison Advertising: Problems and Potential," Journal of Marketing, 39 (October, 1975), 7-15.

B. J. Winer, Statistical Principles in Experimental Design, 2nd ed., McGraw-Hill Book Co., New York, 1971.

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