Peripheral Cues As Sources of Market Inefficiencies in the U.S. and Russia

Timothy B. Heath, University of Pittsburgh
David L. Mothersbaugh, University of Pittsburgh
Michael S. McCarthy, Miami University
Gangseog Ryu, University of Pittsburgh
[ to cite ]:
Timothy B. Heath, David L. Mothersbaugh, Michael S. McCarthy, and Gangseog Ryu (1995) ,"Peripheral Cues As Sources of Market Inefficiencies in the U.S. and Russia", in NA - Advances in Consumer Research Volume 22, eds. Frank R. Kardes and Mita Sujan, Provo, UT : Association for Consumer Research, Pages: 511-517.

Advances in Consumer Research Volume 22, 1995      Pages 511-517

PERIPHERAL CUES AS SOURCES OF MARKET INEFFICIENCIES IN THE U.S. AND RUSSIA

Timothy B. Heath, University of Pittsburgh

David L. Mothersbaugh, University of Pittsburgh

Michael S. McCarthy, Miami University

Gangseog Ryu, University of Pittsburgh

[The authors wish to thank Olga A. Tretjak, Vladimir Tretjak, and Serqey V. Shkurnikov for their help in data collection in Russia.]

While the consumer welfare implications of advertising have been debated extensively, little experimental research has assessed advertising's ability to damage markets by leading consumers to choose objectively inferior brands. The current study examines this issue by assessing the possibility that nonoptimal choices arise from advertising's peripheral cues (e.g., spokesperson fame and liking). Given the expansion of free markets in the Eastern Block, we assessed these effects on both U.S. (Experiment 1) and Russian (Experiment 2) consumers. The results show that in both the U.S. and Russia, well-liked peripheral cues increase the probability of choosing objectively inferior brands even when consumers think hard about their decisions. It appears that consumers are willing to give up certain product features for stronger peripheral cues. These results suggest that peripheral cues can serve as sources of market inefficiencies in various cultures.

The moral and economic implications of advertising have been the subject of protracted debate. Some people credit advertising with improving markets by supplying information and stimulating competition (e.g., Stigler 1961). Others believe advertising damages markets by erecting barriers to entry, providing no information or misinformation, and promoting competition on nonsubstantive features such as an ad's humor (e.g., Norris 1984; Shimp and Gresham 1983). However, little research has experimentally assessed advertising's ability to yield nonoptimal choices. Instead, research typically tests advertising's effects on potential mediators such as beliefs and attitudes. Better ads (Kirmani 1990) and higher prices (Monroe 1973) increase perceived quality, and popular advertising techniques easily mislead consumers (Gaeth and Heath 1987). The current study extends this research by testing advertising's ability to directly bias choice toward nonoptimal brands.

We assess the influence of advertising's peripheral or heuristic cues, nonproduct features such as a spokesperson's fame and liking (Chaiken 1980; Petty and Cacioppo 1986). It is generally believed that central cues (e.g., product features) dominate and peripheral cues wield little influence when consumers think hard about their evaluations (i.e., engage in issue-relevant thinking). However, this belief stems largely from research in noncompetitive settings where only a single brand or issue is evaluated. In the context of competition, recent research shows that peripheral cues can alter attitudes and choices when product attributes are either traded-off or constant across brands, regardless of issue-relevant thinking (Heath, McCarthy, and Mothersbaugh 1994; Miniard, Sirdeshmukh, and Innis 1992). Although these effects extend popular multi-process theories of persuasion, they need not reflect detrimental advertising effects since chosen brands were not necessarily inferior to unchosen brands. The current study extends this research by testing the ability of peripheral cues to damage consumer welfare by generating choices of brands inferior on some attributes but superior on none (dominated or objectively inferior brands).

At least three definitions or indicators of market inefficiency exist: (1) the existence of dominated brands, (2) prices higher than the point where marginal cost equals marginal revenue (supernormal profits), and (3) lower price-quality correlations. Each indicator faces validity questions. Characterizing brands as dominating or dominated requires precise information on objective product attributes and assumes that consumers weight the attributes on which the dominated brand is inferior. Whether profits are considered supernormal can depend on the accounting methods used. And price-quality correlations assume that quality is unidimensional and can be measured exactly, assumptions that have been criticized as untenable (see Curry and Faulds 1986; Hjorth-Anderson 1984). The current study uses a modified version of the dominance definition. We operationalize inefficiency as the probability of choosing a brand dominated on objective product features within experimental choice sets. This definition is easily operationalized and reflects what would be considered damaged consumer welfare in real-world markets. We tested peripheral-cue effects on inefficiencies with both U.S. (Experiment 1) and Russian (Experiment 2) consumers. Since persuasive product advertising is relatively new in the former Eastern Block countries, consumers from those countries may be more or less susceptible to advertising's influence relative to Western consumers.

EXPERIMENT 1

Experiment 1 tests the impact of consumers' evaluations of peripheral cues on choice. When a famous spokesperson or vivid ad copy is used by a dominated brand, consumers who like that particular person or copy are expected to choose the dominated brand more than those who do not. The implication is that liked peripheral cues can make up for deficits on objective product features (i.e., a compensatory model with peripheral and central cues).

Moreover, this effect should become stronger as information on other product attributes on which the brands are comparable is added, because consumer perceptions of the relative superiority of the dominating brand should decrease. This is because the relative influence of an attribute is typically reduced as others are added. For example, the relationship between price and perceived quality erodes as information on other product features is added (Monroe 1973). Therefore, two information environments were tested: one in which an overall quality rating was not supplied and one in which it was. The two target brands, one dominating the other, were equated on the overall quality rating when it was supplied. Experiment 1 tests the following hypotheses:

H1: Positive evaluations of a dominated brand's peripheral cues will (a) increase the probability that the dominated brand is chosen (increasing inefficiencies), and (b) improve attitudes toward the dominated brand.

H2: Hypothesis 1's peripheral-cue effects on (a) choices of dominated brands and (b) attitudes toward dominated brands will be stronger when dominated and dominating brands are said to be of equal overall quality.

Method

Subjects, Stimuli, and Design. Fifty-six undergraduate marketing students participated in the experiment as part of an in-class exercise for which they received extra-credit points. Subjects were randomly assigned to experimental conditions. Stimuli are presented in the appendices. Two product classes (cameras and cruises) and two types of peripheral cue (spokesperson and ad copy) were tested to enhance generalizability. Pretests indicated that Jay Leno (talk-show host) and Michael J. Fox (actor) were more famous and better liked than other fictitious spokespeople such as Alex Tyler. Pretests also indicated that ad copies were rated as more exciting and more liked when they used slightly more vivid language (e.g., "stroll" vs. "walk"; see Heath et al. 1994).

Four brands were presented within a given choice set. Fictitious brand names were used to eliminate effects of brand familiarity and liking (Brands J, L, N, P). Brands L and N dominated Brands J and P. The latter were inferior on multiple dimensions and superior on none, and were included to reduce experimental demand. Only Brands L and N, the target brands, were expected to be chosen (and, in fact, were the only ones chosen). Within the target brands, Brand N dominated Brand L. Brand N was comparable to Brand L on all dimensions except one. In cruises, Brand N's cabin was 185 square feet whereas Brand L's was 140 square feet. In cameras, Brand N's zoom ratio was 4 to 1 whereas Brand L's was 3 to 1 (see Appendix B).

Experiment 1 manipulated peripheral-cue strength and product dominance within subjects, and presence of overall quality ratings (absent vs. present) between subjects. Type of peripheral cue (spokesperson fame vs. ad-copy vividness) and product class (cameras vs. cruises) were counterbalanced between subjects. There was no spokesperson indicated in the ads testing vividness, and no vividness manipulation in the ads testing spokesperson fame. Brand L, the dominated target brand, always had vivid ad copy or the famous spokesperson (Jay Leno), whereas Brand N had less vivid ad copy or a nonfamous spokesperson (Alex Tyler). To reduce demand, peripheral cues were varied across nontarget brands as well (J and P), where Michael J. Fox served as the famous spokesperson.

Procedure. Each subject received a booklet containing ads for each of four fictitious brands, brand information, and dependent measures. Ads consisted of scripts for radio ads to reduce confounding influences from typical advertising features (e.g., layouts, music, etc.). Brand information was summarized in a table following the four ads (see Appendix B). After making their choices, subjects responded to a series of questions. Subjects were asked to indicate (1) their attitudes toward each of the brands on nine-point bad/good and unfavorable/favorable scales, (2) how much they liked Brand L's famous spokesperson on a nine-point dislike/like scale, or how much they liked Brand L's more vivid text on nine-point dull/exciting and dislike/like scales, (3) the importance of price, quality, and cabin size (cruise conditions) or zoom ratio (camera conditions) on nine-point unimportant/important scales, and (4) their degree of issue-relevant thinking.

Results

The data were collapsed across counterbalanced variables (product class and type of peripheral cue) since these variables yielded no effects. The two scales measuring liking for the dominated brand's ad copy were averaged due to their high correlation (r=0.90). Median splits on this average as well as on spokesperson liking divided subjects into those who liked the dominated brands' peripheral cues more and those who liked them less. Even after collapsing across counterbalanced variables, a two-way LOGIT based on peripheral-cue liking and presence of overall quality ratings was not estimable due to small n's when the two main effects and interaction were included. Therefore, two models were run separately: one with the main effects of peripheral-cue liking and presence of quality ratings, and the other with the main effect of peripheral-cue liking and the interaction between peripheral-cue liking and presence of quality ratings. The probability of choosing dominated brands across experimental conditions is summarized in Table 1.

Hypothesis 1a predicted that peripheral-cue liking would increase choice probabilities of dominated brands. Hypothesis 1a was supported. Peripheral-cue liking generated greater shares for dominated brands (36% vs. 3%; c2(1)= 6.80, p <.01). [This is the X2 from the main effects model and is virtually identical to that from the model in which the other term wa the interaction of peripheral-cue liking with quality rating (X2(1) = 6.62).]

Hypothesis 2a predicted that reporting equivalent overall quality across the dominated and dominating brands would increase the peripheral-cue effect on choices of the dominated brand. Hypothesis 2a was not supported. The interaction between peripheral-cue liking and presence of quality ratings was not statistically significant (c2(1)= 0.02); nor was the main effect of presence of quality ratings (c2(1)= 0.79).

Hypotheses 1b and 2b were tested by subjecting attitudes toward dominated brands to a two-way between-subjects ANOVA: peripheral-cue liking by presence of quality ratings. The two attitude scales were combined due to their high correlation (r=0.92). Hypothesis 1b predicted that peripheral-cue liking would increase attitudes toward dominated brands. The significant main effect of peripheral-cue liking supported Hypothesis 1b (F(1,51)= 8.42, p <.01). Attitudes toward dominated brands were higher when peripheral cues were more liked (M=7.62) than when they were less liked (M=6.83).

Hypothesis 2b predicted that the peripheral-cue effect would be stronger when the two brands were said to have the same overall quality. Hypothesis 2b was not supported. The interaction between peripheral-cue liking and presence of quality ratings was not significant (F(1,51)= 1.61); nor was the main effect of presence of quality ratings (F(1,51)< 1.00).

Discussion

The results of Experiment 1 demonstrate that peripheral cues can lead consumers to choose objectively inferior brands. Liking for peripheral cues led over 30% of the people to choose a brand dominated on objective product features. Although it is possible that subjects did not recognize Brand N's superiority over Brand L, this seems unlikely for three reasons. First, the attribute information was clearly presented in table form so that cross-brand differences were quite salient. Second, Brand N commanded a large share which it would not have commanded had subjects been unaware of its superiority. Third, two items measuring issue-relevant thinking suggest that the subjects considered their decisions carefully. The first measured how carefully subjects evaluated the experimental brands relative to how carefully they evaluate household products such as paper towels, laundry detergents, and cleansers. The second measured how carefully subjects evaluated the experimental brands relative to how carefully they evaluate consumer electronics products such as televisions, stereo equipment, and microwave ovens. The two scales ranged from -8 (much less carefully) to +8 (much more carefully). Subjects reported evaluating experimental brands more carefully than household products (M=4.14; t(55)= 9.07, p <.01) and about as carefully as consumer electronics products (M=-0.64; t(55)= 1.71, ns). Therefore, it appears that subjects were cognizant of Brand N's objective superiority, and that some were willing to trade that superiority for stronger advertising features such as vivid copy or famous spokespeople.

TABLE 1

PROBABILITY OF CHOOSING THE DOMINATED BRAND IN EXPERIMENT 1

The findings are subject to three alternative explanations. First, subjects may have chosen the dominated brand because they did not consider the attribute on which it was inferior to be important (cabin size or zoom ratio). This explanation can be tested by examining reported attribute weights of subjects choosing Brand L. If subjects choosing dominated brands did not weight cabin size or zoom ratio, then their reported importances for these attributes should be near 1 on the 1-9 scales. However, this test is conservative because attribute weights were measured after choice and, therefore, may be deflated by the desire to justify their decision and/or reduce dissonance. Despite this downward bias, people choosing dominated brands rated the importance of cabin size and zoom ratios significantly higher than 1.0 (M=4.30; t(9)= 5.36 , p <.01) and not significantly lower than the scale's mid-point of 5.0 (t(9)= -1.13, ns). It appears, therefore, that Brand N dominated Brand L on objective attributes even in the eyes of those choosing Brand L. These consumers, however, were willing to trade off objective inferiority for stronger peripheral cues.

Second, and relatedly, consumers may generally regard the difference between Brands N and L to not matter. However, this explanation is untenable since it implies that consumers would be perfectly indifferent between, for example, cruise cabins of 140 sq. ft. and 185 sq. ft. (in the absence of peripheral cues). A modified version of this explanation, however, is plausible, although it is consistent with the conclusion that peripheral cues led consumers to nonoptimal choices. Peripheral cues may have signalled that the attribute deficits were not important. For example, consumers may have reasoned that if the smaller cabin was large enough for a given spokesperson, then it was large enough for them. Although the current study did not assess such effects, they seem far less probable when vivid text served as the peripheral cue. Therefore, it is likely that Brand L's weaknesses mattered to consumers in the absence of peripheral cues, although in the context of peripheral cues some consumers (1) were willing to trade off stronger peripheral cues for weaker product features, and/or (2) assumed that the weaknesses were inconsequential when associated with stronger peripheral cues. Both effects reflect the ability of peripheral cues to lead consumers to choose objectively inferior brands.

Third, it might be argued that the peripheral cues conveyed product information which then eliminated dominance. However, this is unlikely since, in an earlier study (Heath et al. 1994), comparable cues had no effect on brand attitudes when subjects thought hard about their evaluations and (1) evaluated a single product (Experiment 1) or (2) evaluated multiple products with small trade-offs across attributes (Experiment 3). Therefore, it appears that consumers make decisions in part on the basis of persuasive marketing messages that then lead them to choose objectively inferior brands, brands they would not choose otherwise. This effect may help to account for the existence of dominated brands in the marketplace (see Hjorth-Anderson 1984).

The peripheral-cue effects reported here run counter to multi-process theories claiming that peripheral cues are ineffective in the context of issue-relevant thinking (see Petty and Cacioppo 1986; Chaiken 1980). However, the results are consistent with recent demonstrations that peripheral cues wield considerable influence in competitive settings despite issue-relevant thinking (cf. Heath et al. 1994; Miniard et al. 1992). They extend these findings by showing that the effects are not limited to situations in which trade-offs exist across brands.

Contrary to Hypothesis 2, the peripheral-cue effect on choices of dominated brands was not moderated by additional information on the brands. One potential explanation is that there was enough comparable information across the two target brands that adding one more piece of information was relatively unimportant.

EXPERIMENT 2

Experiment 2 tests the impact of peripheral cues on choices of dominated brands by Russian consumers. Russian markets are increasingly open to Western goods and advertising is increasing rapidly. However, it is not clear whether Russians will react to persuasive advertising messages in the same way Americans do. For example, Russia's anti-capitalistic heritage might leave Russians less susceptible to advertising's persuasive efforts. Therefore, Experiment 2 reassesses Hypotheses 1 on Russian consumers.

Since the presence of overall quality ratings had no effect in Experiment 1, it was excluded from Experiment 2. Instead, we assessed two other variables that might moderate the effects of peripheral cues on inefficiencies: (1) the importance of the product attribute on which the dominated brand is inferior relative to the dominating brand, and (2) the size of the difference on that attribute between dominated and dominating brands. Experiment 2, therefore, also tests the following hypothesis:

H3: The positive effects of peripheral-cue liking on (a) choice probabilities and (b) brand attitudes will increase as the dominated attributes become less important and/or the discrepancy between dominating and dominated brands decreases.

Method

Subjects consisted of 134 students at Russia's Railway Institute in St. Petersburg who were paid $2 (U.S.) for their participation (the institute is similar to an engineering college in the U.S.). Three native Russians briefed us on various facets of Russian culture prior to the experiment's design. Based on their recommendations, Experiment 2 excluded the vividness manipulation due to translation problems, and excluded cruises due to their limited exposure among most Russians. Our consultants assured us that Russians were knowledgeable about automatic focus cameras and that they had various cultural heros that could serve as spokespeople. Therefore, Experiment 2 assessed cameras and the impact of Vladislav Tretjak as a spokesperson. Tretjak was a famous goalie on the Russian hockey team for many years. The camera ads from Experiment 1's spokesperson conditions were translated into Russian. To reduce demand, one of the nontarget brands used a famous spokesperson (Yuri Senkevich) while the other did not.

TABLE 2

PROBABILITY OF CHOOSING THE DOMINATED BRAND IN EXPERIMENT 2

The structure of Experiment 2's brand attributes was comparable to that of Experiment 1's camera conditions with quality-ratings except that two attributes were dropped (weight and zoom ratio) and one was added (rewind speed; see Appendix C). As seen in Appendix C, Brand L was inferior to Brand N on either an unimportant attribute (rewind speed) or an important attribute (overall quality). Moreover, Brand L's weakness was relatively small or relatively large. The dominated attribute and the size of the dominance were varied between subjects. The procedure was comparable to that of Experiment 1.

Results and Discussion

A median split on subjects' reported liking of the spokesperson was performed and then this variable plus the dominance attribute (less important rewind speed vs. more important overall quality) and dominance size (small vs. large) were submitted to LOGITs. However, the higher-order interactions were not estimable due to small n's. Interactions added one at a time proved to be nonsignificant. Therefore, we report results from the main effects model. Choice probabilities across experimental conditions are summarized in Table 2.

Hypothesis 1a predicted that peripheral-cue liking would increase choice probabilities of dominated brands. Hypothesis 1a was supported. Subjects rating the spokesperson more positively were more likely to choose the dominated brand (23.7%) than those who rated him less positively (10.7%; c2(1)= 4.00, p <.05). Hypothesis 3a predicted that this effect would depend on (1) the importance of the attribute on which the dominated brand was inferior and (2) the size of the dominated brand's deficiency. However, Hypothesis 3a was not supported. The only significant effect on choice was that of spokesperson liking. As in Experiment 1, this effect cannot be attributed to no weight being placed on the attributes on which the dominated brand was inferior. Subjects choosing the dominated brand rated the importance of quality and rewind speed significantly above the scale's mid-point (M=7.45; t(21)= 5.24, p <.01). Therefore, it appears that they were willing to trade off quality or rewind speed for a spokesperson's recommendation.

Hypothesis 1b predicted that peripheral-cue liking would increase attitudes toward the dominated brand. The two measures of attitudes toward the dominating brand were averaged due to their high correlation (r=.79). A three-way ANOVA on these measures failed to support Hypothesis 1b. Peripheral-cue liking had no effect on attitudes toward the dominated brand. The only significant effect was the main effect of the attribute on which dominance occurred (F(1,90)= 5.52, p <.05). The dominated brand was rated lower when it was inferior on overall quality (M=2.07) than when it was inferior on rewind speed (M=2.68).

Issue-relevant thinking was again prevalent. Subjects reported evaluating experimental brands more carefully than household products (M=4.91; t(134)= 16.56, p <.01) but about as carefully as consumer electronics products (M=1.74; t(134)= 1.74, ns). As in Experiment 1, therefore, the peripheral-cue effects reported here run counter to the popular belief that peripheral cues yield little influence in the context of issue-relevant thinking (see Petty and Cacioppo 1986; Chaiken 1980).

GENERAL DISCUSSION

Prior experimental research shows that advertising has the potential to bias consumer beliefs and alter choice probabilities when trade-offs exist across brands, even when consumers engage in issue-relevant thinking (e.g., Heath et al. 1994). The current study extends this research by showing that peripheral cues can lead consumers to choose objectively inferior brands. Despite careful evaluations, consumers are willing to trade off certain product features for strong advertising cues such as famous spokespeople or more vivid copy. This effect occurred in both U.S. and Russian consumers. The implication is that peripheral cues can generate market inefficiencies and potentially damage consumer welfare. Although it is sometimes argued that an advertiser can get a consumer to purchase an inferior product only once, the evidence suggests otherwise (see Norris 1984). Consumers often have limited information on other brands, and product experiences are often ambiguous and subject to advertising-based biases. For example, Levin and Gaeth (1988) found that ground beef tasted better when consumers were told beforehand that it was 85% lean than 15% fat.

As markets open in the former Eastern Block, governments, marketers, and consumers should be sensitive to the costs and benefits of advertising. The current study suggests that consumers in these countries are just as susceptible to advertising-generated biases as consumers in the West. The implication is that markets and advertising should be carefully monitored and, if need be, regulated as they evolve, rather than waiting until advertisers have the chance to establish entrenched beliefs that are not easily changed after the fact. For example, consumers continued to purchase Listerine for its purported cold-prevention properties even after a corrective advertising campaign had been run (see Norris 1984; Wilkie, McNeill, and Mazis 1984).

APPENDIX A

ADVERTISING COPY

In the current study, the effects of peripheral cues were such that they persuaded a consumer to try choose one brand rather than another. However, more insidious effects are possible when we consider choices to engage in a given behavior or not (e.g., smoking). The findings of the current study suggest that relatively innocuous advertising stimuli such as a spokesperson or vivid text might be enough to draw a consumer into detrimental behaviors when s/he is relatively undecided (e.g., Joe Camel). The results support restrictions on advertisements for potentially harmful products.

APPENDIX B

PRODUCT INFORMATION FOR EXPERIMENT 1

APPENDIX C

PRODUCT INFORMATION FOR EXPERIMENT 2

REFERENCES

Chaiken, Shelly (1980), "Heuristic Versus Systematic Information Processing and the Use of Source Versus Message Cues in Persuasion," Journal of Personality and Social Psychology, 39 (November), 752-766.

Curry, David J. and David J. Faulds (1986), "Indexing Product Quality: Issues, Theory, and Results," Journal of Consumer Research, 13 (June), 134-145.

Gaeth, Gary J. and Timothy B. Heath (1987), "The Cognitive Processing of Misleading Advertising in Young and Old Adults," Journal of Consumer Research, 14 (June), 43-54.

Heath, Timothy B., Michael S. McCarthy, and David L. Mothersbaugh (1994), "Spokesperson Fame and Vividness Effects in the Context of Issue-Relevant Thinking: The Moderating Role of Competitive Setting," Journal of Consumer Research, 20 (March), 520-534.

Hjorth-Anderson, Chr. (1984), "The Concept of Quality and the Efficiency of Markets for Consumer Products," Journal of Consumer Research, 11 (September), 708-718.

Kirmani, Amna (1990), "The Effect of Perceived Advertising Costs on Brand Perceptions," Journal of Consumer Research, 17 (September), 160-171.

Levin, Irwin P. and Gary J. Gaeth (1988), "How Consumers are Affected by the Framing of Attribute Information Before and After Consuming the Product," Journal of Consumer Research, 15 (December), 374-378.

Miniard, Paul W., Deepak Sirdeshmukh, and Daniel E. Innis (1992), "Peripheral Persuasion and Brand Choice," Journal of Consumer Research, 19 (September), 226-239.

Monroe, Kent (1973), "Buyers' Subjective Perceptions of Price," Journal of Marketing Research, 10 (February), 70-80.

Norris, Vincent P. (1984), "The Economic Effects of Advertising: A Review of the Literature," in Current Issues and Research in Advertising, Vol. 2, eds. James H. Leigh and Claude R. Martin, Jr., Ann Arbor: University of Michigan, 39-134.

Petty, Richard E. and John T. Cacioppo (1986), Communication and Persuasion: Central and Peripheral Routes to Attitude Change, New York, Springer.

Shimp, Terence A. and Larry Gresham (1983), "An Information Processing Perspective on Recent Advertising Literature," in Current Issues and Research in Advertising, eds. James H. Leigh and Claude R. Martin, Jr., Ann Arbor: University of Michigan, 39-76.

Stigler, George J. (1961), "The Economics of Information," Journal of Political Economy, 69 (June), 213-225.

Wilkie, William L., Dennis L. McNeill, and Michael B. Mazis (1984), "Marketing's 'Scarlet Letter': The Theory and Practice of Corrective Advertising," Journal of Marketing, 48 (Spring), 11-31.

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