Applications of the &Quot;Normative Belief&Quot; Technique For Measuring the Effectiveness of Deceptive and Corrective Advertisements

Philip G. Kuehl, University of Maryland
Robert F. Dyer, George Washington University
ABSTRACT - The FTC's use of corrective advertising remedies to counteract misinformation in the marketplace has stimulated an emerging consumer behavior research tradition in this public policy issue in recent years. This study extends knowledge about the FTC's corrective advertising policy by reporting experimental findings on the viability of using brand beliefs for measuring the effectiveness of deceptive and corrective advertisements within the context of Gardner's "normative belief technique."
[ to cite ]:
Philip G. Kuehl and Robert F. Dyer (1977) ,"Applications of the &Quot;Normative Belief&Quot; Technique For Measuring the Effectiveness of Deceptive and Corrective Advertisements", in NA - Advances in Consumer Research Volume 04, eds. William D. Perreault, Jr., Atlanta, GA : Association for Consumer Research, Pages: 204-212.

Advances in Consumer Research Volume 4, 1977   Pages 204-212

APPLICATIONS OF THE "NORMATIVE BELIEF" TECHNIQUE FOR MEASURING THE EFFECTIVENESS OF DECEPTIVE AND CORRECTIVE ADVERTISEMENTS

Philip G. Kuehl, University of Maryland

Robert F. Dyer, George Washington University

ABSTRACT -

The FTC's use of corrective advertising remedies to counteract misinformation in the marketplace has stimulated an emerging consumer behavior research tradition in this public policy issue in recent years. This study extends knowledge about the FTC's corrective advertising policy by reporting experimental findings on the viability of using brand beliefs for measuring the effectiveness of deceptive and corrective advertisements within the context of Gardner's "normative belief technique."

INTRODUCTION

The FTC's policy pursuing "corrective advertising" remedies to negate the effects of deceptive advertising practices has stimulated an emerging professional dialogue among consumer behavior researchers. Furthermore the results of the so-called "Listerine case" in December 1975 -- in which the Commission won a cease and desist judgment involving a major corrective advertising order against a respondent firm (i.e., Warner-Lambert) -- suggest that continued empirical and conceptual efforts toward better understanding the effects of this FTC policy are merited in the future.

The Research Tradition in Corrective Advertising

Three major types of empirical and conceptual contributions can be delineated in the published literature on corrective advertising. While a critical appraisal of these efforts is beyond the scope of this paper, the following discussion briefly highlights and summarizes the major research findings and conceptual viewpoints found in the literature to date.

Preliminary Empirical Insights. Four studies in the literature (Hunt, 1972 and 1973; Dyer and Kuehl, 1974; and Kassarjian, Carlson, and Rosin, 1975) established some preliminary insights on the nature of deception in advertising and the communication effects of corrective messages.

Hunt (1972) used student subjects and print media to examine the effects of two versions of corrective advertising copy content: (1) "Explicit" corrective copy --the point-by-point correction of alleged deceptive claims, and (2) "general" corrective copy -- broad statements that the previous advertising was deceptive. The explicit message treatment used in this first experiment tended to produce a negative attitude toward the brand in question, while the general corrective message tended to produce little attitude change. Hunt (1973) also reported findings where no message effect differences between explicit and general content occurred. His second study, furthermore, provided evidence that FTC-source and company-source corrective messages produce parallel results.

Dyer and Kuehl (1974) also employed student subject to research the effects of corrective message (1) "source" (i.e., either FTC or company message source) and (2) "strength" (i.e., high and low intensity of message content -- similar to Hunt's explicitness variable) in print and broadcast media. In general, this research produced results similar to Hunt since high strength copy tended to produce decreases in brand attitude and purchase intentions. These findings, however, also supported the view that FTC-source corrective messages were more effective than their company-source counterparts in print media (but not in broadcast versions).

The Kassarjian, Carlson, and Rosin study (1975) used realistic newspaper advertisements to examine the impact of corrective advertising copy on the product in question as well as the retailer that carried such an item. While the study used college student subjects and only one major corrective advertising content variable (i.e., a relatively general message sponsored by the retailer acknowledging that accredited research was not available to support the product claim), findings indicate that the overall brand image profile was decreased by corrective ad exposure but that the retailer profile remained relatively stable.

In summary, the four studies just discussed established a body of preliminary findings related to understanding the FTC's corrective advertising policies. However, two major qualifying factors should be noted when these contributions to knowledge about corrective advertising are evaluated in general terms. From a methodological perspective, three types of limitations should be noted. First, all of these authors used only college student subjects in their experiments so that the generalizability of the study findings to other populations is questionable. Second, none of the studies examined the effects of corrective advertising over time. In other words, the "one shot" effect of corrective advertisements does not provide insights useful to FTC policy-makers for judging whether the long-term effect of the remedy is appropriate to fulfill the nonpunitive regulatory mandate that is required in the law. Third, the variety of experimental conditions used by the authors (i.e., variations in media treatments, product categories, brands, message formats, etc.) make it difficult to judge these contributions from a replication perspective.

The second major qualifying factor associated with these studies concerns the narrow conceptual bases used by these authors to support their study designs. In essence, the conceptual bases and measurement for these experiments tended to reflect the "case-by-case" policymaking actions and evolving corrective advertising legal doctrine at the FTC that developed through such cases as Wonder Bread, Hi-C, Firestone, and Chevron. AS a result, the corrective message format and content variables used in these studies tended to describe, though transformed into traditional measurement concepts like brand attitude, company image, etc., current FTC policy actions (i.e., the focus of the early corrective advertising research was descriptive in nature since existing FTC policymaking tendencies were examined.) While such a focus may result in accurate statements of the implications inherent in current policies (if one overlooks some of the obvious external validity limitations in the studies), it does little to advance knowledge in a normative sense. In summary, then, these studies were oriented more toward investigating existing legal theory and FTC policymaking actions vis-a-vis constructing and examining innovative behavioral science concepts, theories, and variables related to corrective advertising.

Conceptualizing Behavioral Research in Corrective Advertising. Concurrent with the empirical work just discussed, other consumer behavior researchers began to conceptualize corrective advertising, on a normative basis, in two distinct ways. First, some scholars began to develop general typologies for classifying and defining deceptive advertising practices (Aaker, 1974; Armstrong and McLennan, 1973; Cohen, 1969 and 1972; Dillon, 1973; Gardner, 1975; Haefner, 1972; Haefner, Birchmore, and Permut, 1974; and Rosch, 1975). While the research cited in the previous discussion focused mainly on the effects of corrective advertisements, these authors were conceptualizing the features and characteristics of alleged deceptive advertising practices.

While this shift in emphasis added an important new conceptual understanding in the corrective advertising research tradition (i.e., all of the early empirical efforts were based on "global" types of deceptions that emerged from FTC case actions), many of these typologies tended to overlap in a definitional sense and were not mutually exclusive or collectively exhaustive. Furthermore, these efforts were not used as theoretical frameworks for empirical research purposes until a study by Ford, Kuehl, and Reksten (1975).

These authors provided additional insights into the nature of deception by first developing a comprehensive typology for identifying and classifying types of deception (i.e., 12 categories of deception in their typology evolved from their reconciliation of similarities found in the work of others and two other categories of deception emerged from their content-type analysis of nearly 200 print media advertisements). This typology was then used in an experiment that examined student subjects' abilities to rank order (from "most" to "least" deceptive) (1) descriptive statements and (2) actual advertisements representing the 14 types of deception in the typology. While high rank order correlations were found when treatment and control group subjects ranked the statement versions of the deception typology, nonsignificant relationships occurred when actual advertisements were rank ordered, showing that respondents had difficulty in correctly perceiving and consistently evaluating deception represented in the advertisements.

In summary, the development of conceptual typologies for defining deception in advertising, while adding new dimensions of consumer behavior thought about the subject area, did not per se stimulate significant empirical research. It does appear, however, that these initial attempts to classify and define deception in advertising did foreshadow scholarly interest in conceptualizing consumer behavior inputs to knowledge about corrective advertising -- the development of normative roles and operational procedures for integrating behavioral science research frameworks and policymaking processes. The early foundations of this type of contribution were established by Jones (1971) and Wilkie and Gardner (1974). Based on these authors' views of the way in which behavioral science inputs to deceptive advertising must be related to policymaking characteristics at the FTC, a series of articles (Armstrong and McLennan, 1975; Jacoby and Small, 1975) were published using modified forms of Gardner's (1973) "normative belief technique." In essence, these authors began to develop conceptual frameworks for precisely defining deceptive advertisements in an operational context that was consistent with FTC legal requirements.

Brand Beliefs: Concepts and Research. The initiation of the third major dimension of the corrective advertising research tradition in consumer behavior must be credited to Wilkie (1973) and Gardner (1973). These two scholars suggested that the belief component of the traditional Fishbein attitude model is appropriate for measuring the extent of deception represented in an offending advertisement. In essence, these scholars recognized that precision in defining deception in advertising requires measurement on a brand belief, multi-attribute basis. In justifying the use of brand belief variables vis-a-vis brand attitudes, Wilkie notes that:

Brand attitudes are overall measures of like-dislike . . . and are too general to be attributed to exposures to specific messages . . . (the) measure is too aggregative for the FTC to rely upon (since) they measure consumer's net responses to previous experiences with brand usage as well as various stimuli quite distinct from advertising.

(Brand beliefs) . . . represent precisely the level at which deception is presumed to occur and persist. These measures are specific to each brand attribute and reflect a consumer's expectations as to the degree that a given brand offers a particular benefit.

Wilkie, 1973, pp. 12, 17

This initial rationale for brand belief measures in deceptive advertising was updated by Gardner (1975) in an article receiving widespread acclaim. In this article, he synthesized the major conceptual dimensions of consumer behavior thought on deceptive advertising in three ways. First, he provided a definition of deception that is based on the belief component of the Fishbein model. Second, he developed a three-part typology of deception (i.e., the unconscionable lie, claim-fact discrepancy, and claim-belief interaction). Finally, he developed a systematic framework (i.e., the "normative belief technique'') that can be implemented to examine empirically the existence of alleged deception.

It is important to note, however, that both Wilkie's and Gardner's views related exclusively to conceptualizing deception in advertising (i.e., neither author included a discussion of the applicability of' belief measures for assessing the effectiveness of corrective advertising). However, two recent empirical studies cited below (Mazis and Adkinson, 1976; and Kuehl and Dyer, 1976) report findings illustrating that belief measures are sensitive to both deceptive and corrective advertising claims. In this way, the FTC policy-makers could isolate specific product features that were affected by an alleged deceptive advertising claim and, in addition, ensure that proposed corrective messages also affected only these same specific attribute claims -- while not damaging other product attribute beliefs and thereby fulfilling the nonpunitive intent of the law.

The conceptual view that brand belief measures are appropriate for deceptive-corrective advertising measurement has spawned three empirical studies. Dyer and Kuehl (1974) reported that multiple corrective advertising exposures directed toward college students resulted in a more negative belief regarding the "moisturizing" attribute of a suntan lotion product (while not affecting other attributes such as "produces a good tan," "prevents burning," and "high quality/high price"), which was misrepresented in an experimental print advertisement. This finding suggested that deceptive claims can be countered on a selective basis as Wilkie and Gardner suggested and, it is important to note, these authors found also that overall brand attributes were not significantly affected.

Mazis and Adkinson (1976) also used college student subjects to examine brand beliefs in an experiment that was based on the Listerine case. Findings in this study support the notion that specific attribute features that are misrepresented in a deceptive advertisement are affected by corrective messages (in this case, as Hunt found earlier, there was no significant difference in corrective message effectiveness between the FTC-source and company-source messages). The study results also indicate that the evaluation element of the Fishbein model was affected also by the corrective advertisement that was specifically directed toward the belief element. As a result, the authors conclude that evaluation and belief components of the Fishbein model -- which were assumed to be independent by Gardner -- may influence each other in a corrective advertising context.

Kuehl and Dyer (1976) conducted an experiment on a sample of adolescent-aged children that examined the "normative belief technique" and three-part deception typology developed by Gardner in two product categories (i.e., AM radio stations and blemish creams). Results in this study produced three major findings: (1) it appeared that Gardner's "normative belief technique" can be operationalized, (2) respondent beliefs associated with specific functional attributes were affected by the deceptive advertisements used in the study, and (3) FTC-source and FTC-source high strength corrective messages were more effective in lowering specific functional attributes than company-source messages.

In assessing the impact of the "brand belief" contributions to the literature, four major summary conclusions appear appropriate. First, Wilkie recognized that the belief element of the Fishbein-type attitude model was an essential measurement variable in deceptive advertising vis-a-vis brand attitude, company image, etc. This conceptual viewpoint -- in terms of grouping the essential aspects of FTC corrective advertising policies --represented a critical contribution to consumer behavior thought on the subject area. Second, Gardner made a significant contribution to the literature by (1) developing an operational framework for implementing brand belief research in deceptive advertising which (2) incorporated a conceptual typology for defining deception. Third, the studies by Dyer and Kuehl (1974) and Mazis and Adkinson (1976) support the view that brand beliefs are sensitive variables for measuring deceptive and corrective advertising claims. Finally, the Kuehl and Dyer study (1976) provides empirical evidence supporting Gardner's operational and conceptual framework.

The preceding discussion has highlighted the scope and thrust of previous conceptual and empirical work on the FTC's corrective advertising policy. It has been shown that the consumer behavior research tradition on this public policy issue contains three major dimensions: (1) a series of four early empirical studies (which incorporated legal doctrine conceptual frameworks and used global attitudinal measures); (2) numerous efforts to conceptualize typologies for defining deception and develop frameworks for operationalizing corrective advertising research from an FTC policy perspective; and (3) the recognition of brand beliefs as a major measurement variable in deceptive advertising that was incorporated into an operational or procedural framework and three empirical studies.

Objectives

The present study extends the consumer behavior/corrective advertising research tradition in two ways by reporting additional findings on the sensitivity of brand beliefs to deceptive and corrective advertising claims.

First, the study design -- which is based on Gardner's "normative belief technique" -- provides data that illustrate the way in which brand beliefs are affected when exposure to three types of deceptive advertisements occurs (i.e., the unconscionable lie, claim-fact discrepancies, and claim-belief interaction). Such data extend the preliminary insights on Gardner's approach reported by Kuehl and Dyer (1976).

Second, data in the study also expand Gardner's original views by reporting on the effects of corrective advertisements on attribute beliefs. In other words, Gardner's original statement of his "normative belief technique" focused only on composing "deceptive" brand beliefs and "normative" brand beliefs and did not include provisions for incorporating corrective advertisements into the framework. In the present study, the effectiveness of corrective claims in equilibrating previously established levels of belief deception and existing norm group beliefs can be evaluated.

METHODOLOGY

A sample of 405 adolescent-aged respondents from the student body of a large suburban junior high school was used. Subjects were told that they were participating in an "Advertising Creativity Study" being sponsored by the University of Maryland.

To ensure that the product category and Fishbein belief scales used in the study were appropriate for the unique market segment represented in the study sample, a set of developmental activities was performed. These activities are highlighted in Figure 1. As indicated in Figure 1, the six steps needed to complete this phase of the study are developmental in nature. After selecting adolescent-aged students as the subjects for the study, preliminary data were collected to ascertain 10 product categories that are most frequently used and purchased by the respondent group. Then, a subsample of 30 respondents was used to identify the major functional and nonfunctional attributes that are associated with the 10 product categories being screened for the study. In the fourth developmental task, the product category "blemish creams" was selected from the original list of 10 product categories considered for the study. To develop realistic print advertisements for the study, a content analysis-type review of existing "blemish creams" was undertaken and deceptive advertisements (for the unconscionable lie, claim-fact discrepancy, and claim-belief interaction) and their corrective counterparts (with source and strength effects) were prepared.

The content and wording of the deceptive advertisements were based on the examples Gardner (1976) and Dyer and Kuehl (1974) used in their discussions. For example, the unconscionable-lie deceptive advertisement for the AM radio station included the following claim: "Blemish-Off clears pores better than any other skin blemish cream. Similarly, the company source -- high strength corrective advertisement used to refute this deceptive claim stated that: "Blemish-Off has engaged in deceptive advertising in a recent campaign -- the claim by Blemish-Off that Blemish-Off clears pores better than any other skin blemish cream is untrue." The fifth developmental task was the development of the study questionnaire that, along with all deceptive and corrective advertisements, was pretested with a second group of 30 respondents.

Activities in this phase of the study, which are summarized in Figure 2, obtained normative product category belief measures for a subsample of study respondents. After selecting a sample of 45 respondents (none of whom were in developmental phase subsample groups), the study questionnaire was used to obtain (a) socio-demographic data and (b) belief measures on the major functional product attributes for the skin blemish product category. All questionnaires were administered by the same survey administrator during regular classroom sessions. Approximately 15 minutes, including instructing the respondent% were needed to complete the questionnaires.

The first two steps contained in Figure 2 consisted of calculating product category attribute belief and attitude scores for blemish creams. According to Gardner (1075), these product category scores establish a "benchmark range" of acceptable belief scores against which alleged deceptive advertisements should be compared. In other words, if brand attribute beliefs exceed, in a statistically significant fashion, the product category attribute beliefs, the advertisement for that specific brand can be viewed as deceptive -- unless a respondent firm can provide supporting factual evidence for the attribute claim. Conversely, brand attribute beliefs that do not significantly exceed product category attribute beliefs would not be considered deceptive.

FIGURE 1

DEVELOPMENTAL ACTIVITIES

FIGURE 2

MEASURING "NORMATIVE BELIEFS"

The final set of methodological activities -- designed to measure the impact of deceptive and corrective advertisements -- are presented in Figure 3.

FIGURE 3

MEASURING "DECEPTIVE" AND CORRECTIVE" BELIEFS

As indicated in Figure 3, a total of 300 respondents were randomly assigned to experimental groups in a 2 x 2 factorial experimental design used in the study to measure source (i.e., FTC and company) and strength (i.e., "high" and "Low") effects for each of the three types of deception (i.e., the unconscionable lie, claim-fact discrepancy, and claim-belief interaction) described in Gardner's typology. It should be noted that findings related to the source-strength impact of corrective advertising found in this study are reported in another article (Kuehl and Dyer, 1976). In general, these results illustrated that print FTC-source and FTC-high strength messages were more effective than other message treatments in reducing deceptively-based beliefs. Corrective advertising findings reported in this article contain only main treatment effects. Complete data were obtained from 256 respondent subjects -- 78 in the unconscionable lie category, 88 in the claim-fact discrepancy category, and 90 in the claim-belief group. In the second task, respondents were asked to read the deceptive advertisements developed for the study and, using only the information contained in the advertisement, provide belief data on a seven-point scale for each of the brand attributes listed for the skin blemish product. After a two-week interval, similar data were collected from the respondents after exposure to one of the four corrective advertising treatments. As in the previous two phases of the study, all advertisement exposure and data collection activities were conducted by the same survey administrator during regular classroom sessions. Based on the researcher's observations, it appeared that none of the respondents had any prior knowledge of the deceptive-corrective advertising issue or had, in any way, correctly perceived the true intent of the study. The fourth and fifth tasks were similar to those performed on norm group respondents (i.e., brand relief and brand attitude scores were calculated).

ANALYSIS

As stated in the study objectives, findings discussed in this paper provide two types of insights about brand belief measures in deceptive-corrective advertising. First, the study findings provide product attribute belief data to test Gardner's "normative belief technique" for identifying deceptive messages. Data related to this objective are point and interval range estimates of brand attribute beliefs for a "norm" group of respondents and similar data for experimental subjects who were exposed to messages representing the unconscionable lie, claim-fact discrepancy, and claim-belief interaction forms of deception. Second, to extend Gardner's suggested approach, the same "norm" group data are compared to brand attribute belief data obtained after other experimental groups were exposed to corrective advertising messages.

Normative and Deceptive Group Beliefs and Attitudes

In reviewing the results of this study, it should be noted that each of the three hypothetical deceptive advertisements included an explicit claim regarding a single product attribute. The explicit claims varied in the unconscionable lie, claim-fact, and claim-belief conditions, however. In the unconscionable lie treatment, for instance, the message stated that the product was available in a range of skin-tone colors (attribute B2) whereas the claim-fact and claim-belief treatments were devoted to "clears blemishes" (attribute B7) and "clears pores" (attribute B3), respectively.

Table 1 illustrates that hypothetical messages for the three categories of deception produced belief levels that exceeded the point and interval ("acceptable") range estimates of the norm group respondents. In Gardner's terms, the direction of the findings indicates that the functional attribute received more favorable belief scores after exposure to a single, hypothetical deceptive ad. Table 1 also demonstrates that the upper confidence limit for the norm group was exceeded by interval estimates of test groups and that t-tests yielded significantly higher belief levels for the deceptive groups' functional attributes in question.

Table 1 also suggests some further complexities that may be encountered in attempts to implement Gardner's methodology. In the unconscionable lie category, a single message statement of "availability in various skin tones" resulted in an upward shift in belief levels on this functional attribute but generated significantly lower belief ratings than norm respondents on five of the remaining six attributes. Predictably, the overall attitude toward the skin blemish cream (A0 = a x B) was significantly poorer. It is important to note that attribute importance scores or evaluations (ai) did not differ from norm to deceptive group.

The results for the claim-fact discrepancy group parallel the above interpretation with one notable exception. In this version, an explicit claim was made regarding the "clears blemishes" attribute only. Again, point and interval estimates and significance tests illustrated a higher belief for the ad exposure group but lower belief levels for attributes Bi, B2, and B6. Conversely an upward shift of the "non-allergenic" belief occurred despite careful pretest efforts that determined that the "clears blemishes" and "non-allergenic" beliefs were non-overlapping. One possible interpretation is that an explicit claim (deceptive or nondeceptive) for a given attribute interacts with beliefs regarding other product qualities. Using a recent FTC case as an analogy, an explicit claim stating that brand A "fights colds" may not raise belief levels on a "pleasant taste" attribute but may result in higher beliefs for "kills germs/bacteria'' and "recommended by doctors" (Docket No. 8891, Final Order, 1975).

Similarly, although the claim-belief treatment was designed to affect the "clears pores" attribute only, significantly higher belief levels were also uncovered for the "clears blemishes" and "non-allergenic" attributes. Although the sample groups represented in this study were drawn from an adolescent population, it is doubtful that their semantic capabilities are poorer than the general population's. A single claim may (without the sponsor's intent) result in higher belief levels for other attributes. It is interesting to note that overall attitudes were not significantly different for the norm and deceptive ad exposure groups even when the "clears pores" belief carryover extended to two attributes.

The authors agree with Gardner's view that the findings in Table 1 merely demonstrate that the deceptive advertisements were successful in shifting specific belief levels upward and are not per se evidence that deception has taken place, in order for deception to have occurred, several additional conditions must be present. For instance, Gardner concludes that the norm group must have "adequate information" about the product category in question but does not provide for operationalizing this requirement. Since respondents in this experiment frequently used and purchased the products and were randomly assigned to the norm and test groups, it is reasonable to assume that the groups possessed adequate stored product information. In addition, other factors may generate brand beliefs that do not fall within norm group belief estimate ranges. In both the lie and claim-fact categories of deception, for instance, it must be shown that scientific studies or evidence do not support the alleged deceptive attribute claim (as in the Listerine case). The blemish remover cream utilized in this study, for example, would require chemical analysis or clinical studies to invalidate a claim that (a) a certain ingredient is contained in the product's formulation (i.e., lie) or (b) that the product is effective in "clearing blemishes" (i.e., claim-fact discrepancy).

TABLE 1

ATTRIBUTE BELIEFS FOR NORM AND DECEPTIVE GROUPS

Normative and Corrective Group Beliefs and Attitudes

The preceding discussion supports the view that brand beliefs are appropriate response measures for assessing the impact of deceptive advertising messages. In addition to measuring the impact of deceptive claims, it appears logical to view the correction of such "residual misinformation" also as a task related to changing consumer beliefs (i.e., deceptive attribute beliefs should be equivalent to normative attribute beliefs after exposure to corrective advertisements). Unfortunately, there is no discussion in the literature that suggests the degree or amount of correction or belief change that must be undertaken by a respondent firm when a corrective advertising order is implemented. Gardner's methodology appears applicable for the development of "belief change" criteria. For example, just as potential deception exists when specific brand beliefs are higher than "acceptable" or "normative" beliefs for a specific at- tribute, the role of corrective messages should be to restore these beliefs to an acceptable range. Thus, if 20 percent of a national projectable sample of consumers feel that "mouthwash in general" is effective in preventing colds, this would be the lower limit for this belief category after remedial advertising.

A comparison of norm and corrective group beliefs is provided in Table 2. The single exposure to a corrective message was designed to counter the following claims: (1) unconscionable lie -- available in skin-tone colors (B2), (2) claim-fact discrepancy -- clears blemishes (BT) and (3) claim-belief interaction -- clears pores (B3).

A comparison of the deceptive and corrective belief scores for the relevant attributes is shown in Table 3.

Table 3 demonstrates that the corrective messages did, in fact, result in significant downward shifts in attribute beliefs. The question remains, however, "Has enough correction taken place?" This question is partially answered by Table 4, which compares the norm and corrective belief estimates.

The findings shown in Table 4 demonstrate that the mock corrective messages were generally effective in accomplishing the belief reduction task. The unconscionable lie category of deception was countered with a message that restored beliefs within the norm group's range. The sharp drop in beliefs regarding this attribute is demonstrated by the fact that mean "corrected" beliefs (3.71) were significantly lower than the norm group's beliefs (5.35).

The "clears blemishes" claim was countered with a message noting the claim-fact discrepancy, which illustrates a case where further corrective messages may be required. Although no significant differences existed between norm and corrective means, the upper boundary of the 95-per-cent confidence interval around the nore group mean belief score was exceeded by the group exposed to corrective advertising.

Finally, the "clears pores" claim was effectively countered. Beliefs were reduced to an acceptable range and no significant difference between norm and corrective means was found.

Carryover Effects of Corrective Advertising to Other Beliefs About the Brand

Firms who are required to utilize correction advertising should be interested in the potentiality of a "reverse halo effect" with remedial messages. Rather than surgically removing a falsely-based belief about a specific attribute, corrective messages may result in a series of beliefs falling below the norm (although not mentioned in the corrective copy). Further, downward shifts in overall attitude toward the brand may result.

Table 2 illustrates significant downward shifts in beliefs for four of six attributes (excluding the attribute for which a deceptive claim was made) for the blatant lie category (i.e., beliefs B4, B5, B6, and B7). Examination of Table 1, however, suggests that most of this performance occurred after exposure to the deceptive ad, not its corrective version. Subjects who examined a deceptive ad stressing "availability in skin-tone colors" had no reason to believe the "low price" attribute, for example. A case of a corrective message resulting in further eroding of beliefs is illustrated by the "clears pores" attribute. Norm mean ratings were 4.77 versus the deceptive group's rating of 3.91 (p<.05). Exposure to the corrective message further reduced mean beliefs to 3.28 (p<.001), even though there was no mention of this belief in the corrective messages. Overall attitude ratings also fell from 144 to 131 after exposure to a corrective ad.

Similarly, in the claim-fact treatment group, "clears pores" beliefs (B3) dropped from a norm rating of 5.06 to a deceptive rating of 4.34 (no significant difference) to a corrective belief of 4.18 (significantly lower than the norm mean, p<.05). Again, overall attitudes fell significantly from 164 to 154.

No carryover of corrective advertising to other beliefs was encountered with the claim-belief interaction treatment group. Additionally, no significant shifts in overall attitudes occurred.

Considered collectively, these findings imply that a certain amount of "overkill" will result when a firm is required to publicly correct an outright lie or suggest that their previous claims were not supported by fact. Both reduction in belief regarding nondeceptive attribute claims and overall attitudes toward the brand may occur.

TABLE 2

ATTRIBUTE BELIEFS FOR NORM AND CORRECTIVE GROUPS

TABLE 3

DECEPTIVE VERSUS CORRECTIVE BELIEF SCORES

TABLE 4

NORM VERSUS CORRECTIVE BELIEF SCORES

SUMMARY

Findings from the experimental research reported in this paper support the two study objectives in three major ways. First, results reinforce the view that brand beliefs are sensitive response measures for gauging the impact of both deceptive and corrective advertising claims -- this finding confirming the results of earlier studies reported by Mazis and Adkinson (1976) and Kuehl and Dyer (1976). Second, although brand beliefs are a sensitive response measure, it is improbable that a corrective advertisement can "surgically" affect in the

attributes affected by the alleged deceptive advertisements. In other words, some carryover effect in correction occurs (i.e., from the attributes affected by the deceptive/corrective advertisements to "other" attributes) and does result in lower belief for these other attributes. Third, the measurement of deceptive-corrective brand beliefs within the context of Gardner's "normative belief" technique appears to provide an appropriate framework for obtaining empirical input on two major public policy issues:

1. Is an advertisement "potentially" deceptive (i.e., are brand beliefs higher than normative, product category beliefs without documented evidence to support the alleged deceptive claim)?

2. Has corrective advertising restored a falsely-based belief to an "acceptable" level (i.e., has an equilibrium been reached between normative and brand beliefs which were previously regarded as deceptive)?

Results from this study, when interpreted within the context of the emerging conceptual and empirical research tradition in deceptive-corrective advertising, yield several implications for future research. First, this effort and previous studies have examined deceptive-corrective advertising on a single exposure basis. While this approach has produced a useful body of preliminary knowledge on this public policy issue, the long-run impact of advertising deception and correction -- represented in longitudinal studies using multiple message exposures -- needs to be studied. Second, the present literature contains many other conceptual typologies for defining deception in addition to Gardner's three dimensional view. These other typologies, as well as other "systems" for operationalizing research on deceptive-correction found in the literature, should be incorporated in future research. Third, most of the previous studies have been conducted in laboratory settings using college students. Obviously, future research efforts need to expand knowledge about the effects of advertising deception and correction to other population segments in field study contexts. Fourth, as other governmental agencies begin to examine the potential need for corrective advertising within their regulatory mandates (i.e., the FDA in physician drug advertising), there will be a need for future research to determine if deceptive-corrective "universals'' exist. In any event, it appears that the deceptive-corrective research tradition represents a subject area for continued consumer behavior inputs to public policy decision processes.

REFERENCES

D. A. Aaker, "Deceptive Advertising," in D. A. Aaker and G. S. Day, eds., Consumerism: Search for the Consumer Interest, 2nd ed. (New York: The Free Press, 1974), 137-156.

G. M. Armstrong and J. P. McLennan, "The FTC and the Investigation and Regulation of Deception in Advertising,'' in Thomas V. Greet, ed., Increasing Marketing Productivity/Conceptual and Methodological Foundations of Marketing (Chicago: American Marketing Association, 1973), 430-434.

G. M. Armstrong and F. A. Russ, "Detecting Deception in Advertising," MSU Business Topics, 23, No. 2(Spring 1975), 21-31.

D. Cohen, "The Federal Trade Commission and the Regulation of Advertising in the Consumer Interest," Journal of Marketing, 33(January 1969), 40-44.

D. Cohen. "Surrogate Indicators and Deception in Advertising,'' Journal of Marketing, 36(July, 1972), 10-15.

T. Dillon, "What is Deceptive Advertising?" Journal of Advertising Research, 13(October, 1973), 9-12.

R. F. Dyer and P. G. Kuehl, "The 'Corrective Advertising' Remedy of the FTC: An Experimental Approach," Journal of Marketing, 38(January, 1974), 48-54.

R. F. Dyer "The Effectiveness of the FTC's 'Corrective Advertising' Policy: New Empirical Insights," in Sharilyn K. Zeigler, ed., Proceedings of the 1974 National Conference of the American Academy of Advertising, Boston: American Academy of Advertising, 166-177.

G. T. Ford, P. G. Kuehl and O. Reksten, "Classifying and Measuring Deceptive Advertising: An Experimental Approach,'' in Proceedings: 1975 Fall Educators' Conference of the American Marketing Association, Rochester, New York. (pages unassigned)

D. M. Gardner, "Deception in Advertising: A Conceptual Approach," Journal of Marketing, 39(January, 1975), 40-46. (Based on "Can Deceptive Advertising Be Measured?'' Working Series in Marketing Research, No. 18. The Pennsylvania State University, College of Business Administration, June 1973. 22 pages.)

J. E. Haefner, "The Legal Versus the Behavioral Meaning of Deception," in M. Venkatesan, ed., Proceedings of the Association for Consumer Research, Fall, 1972, 356-360.

J. E. Haefner, M. T. Birchmore and S. E. Permut, "Applications of Belief Measures to Specific and Overall Advertising Claims: A Program for Research," in S. K. Zeigler, ed., Proceedings of the 1974 National Conference for University Professors of Advertising, American Academy of Advertising, 311-317.

J. E. Haefner and S. E. Permut, "An Approach to the Evaluation of Deception in Television Advertising," Journal of Advertising, 3, No. 4, 40-45.

H. Keith Hunt, "Source Effects, Message Effects, and General Effects in Counteradvertising," in M. Venkatesan, ed., Proceedings, 3rd Annual Conference, Association for Consumer Research, 1972, 370-381.

H. Keith Hunt, "Effects of Corrective Advertising," Journal of Advertising Research, 13(October, 1973), 15-24.

J. 3acoby and C. Small, "The FDA Approach to Defining Misleading Advertising," Journal of Marketing, 39(October, 1975), 65-68.

M. G. Jones, "The FTC's Need for Social Science Research," Proceedings, 2nd Annual Conference, Association for Consumer Research, 1971, 1-9.

H. H. Kassarjian, C. J. Carlson and P. E. Rosin, "A Corrective Advertising Study," in Mary Jane Schlinger, ed., Advances in Consumer Research, Vol. 2, 1975, 631-642.

P. G. Kuehl and R. F. Dyer, "Brand Belief Measures in Deceptive-Corrective Advertising: An Experimental Assessment," in Kenneth L. Bernhardt, ed., Proceedings of the 1976 American Marketing Association Fall Conference, 1976. (forthcoming)

Michael B. Mazis and Janice E. Adkinson, "An Experimental Evaluation of a Proposed Corrective Advertising Remedy," Journal of Marketing Research, XIII (May, 1976), 178-183$

J. T. Rosch, "Marketing Research and .the Legal Requirements of Advertising," Journal of Marketing, 39(July, 1975), 69-79.

"United States of America Before Federal Trade Commission in the Matter of Warner-Lambert Company," Docket No. 8891, Final Order, December 9, 1975 (103 pages).

W. L. Wilkie, "Research on Counter and Corrective Advertising,'' paper presented at the American Marketing Association Conference on Advertising and the Public Interest, Washington, D.C., May 9-11, 1973, 16 pages.

W. L. Wilkie and D. M. Gardner, "The Role of Marketing Research in Public Policy Decision-Making," Journal of Marketing, 38(January, 1974), 38-47.

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