The Moderating Effect of Disclaimer Importance on the Effectiveness of Two-Sided Advertising

ABSTRACT - This study investigates the moderating effect of disclaimer importance on the effectiveness of two-sided advertising. It was found that, relative to the one-sided ad, there was no significant difference in message believability when unimportant or moderately important attributes were disclaimed. However, when highly important attributes were disclaimed there was a significant decrease in believability. Furthermore, attitude toward the ad, toward the brand and toward purchase were not significantly affected when unimportant attributes were disclaimed, but appeared to decline monotonically as increasingly important attributes were disclaimed.

Citation:

Chike Okechuku, Michele Frost, and Melissa Porchuk (1993) ,"The Moderating Effect of Disclaimer Importance on the Effectiveness of Two-Sided Advertising", in E - European Advances in Consumer Research Volume 1, eds. W. Fred Van Raaij and Gary J. Bamossy, Provo, UT : Association for Consumer Research, Pages: 168-175.

European Advances in Consumer Research Volume 1, 1993      Pages 168-175

THE MODERATING EFFECT OF DISCLAIMER IMPORTANCE ON THE EFFECTIVENESS OF TWO-SIDED ADVERTISING

Chike Okechuku, University of Windsor, Canada

Michele Frost, University of Windsor, Canada

Melissa Porchuk, University of Windsor, Canada

Heather Stavenow, University of Windsor, Canada

ABSTRACT -

This study investigates the moderating effect of disclaimer importance on the effectiveness of two-sided advertising. It was found that, relative to the one-sided ad, there was no significant difference in message believability when unimportant or moderately important attributes were disclaimed. However, when highly important attributes were disclaimed there was a significant decrease in believability. Furthermore, attitude toward the ad, toward the brand and toward purchase were not significantly affected when unimportant attributes were disclaimed, but appeared to decline monotonically as increasingly important attributes were disclaimed.

INTRODUCTION

Researchers have recognized for some time that the effectiveness of two-sided advertising is moderated by the perceived importance of the disclaimed attribute (Kamins and Marks 1987; Settle and Golden 1974; Smith and Hunt 1978). A one-sided appeal presents only claims that are supportive of the advertised product while a two-sided appeal not only presents supportive claims but also disclaims superiority on some product attributes in order to enhance credibility. Researchers hypothesize, for example, that if the disclaimer is made on too unimportant an attribute it will typically not be sufficient to enhance credibility. On the other hand, disclaiming a too important an attribute will severely reduce the probability of purchase. Some have found two-sided advertising to be more effective than one-sided advertising for promoting credibility (e.g. Earl and Pride 1980; Etgar and Goodwin 1982; Golden and Alpert 1987; Smith and Hunt 1978) while others have found no difference (e.g. Belch 1981; Hastak and Park 1990; Settle and Golden 1974; Swinyard 1981). The effects on attitudes and purchase intentions have generally been nonsignificant (e.g. Belch 1981; Hastak and Park 1990; Kamins and Marks 1987; Golden and Alpert 1987). While some positive results have been reported (Etgar and Goodwin 1982; Golden and Alpert 1987) some have even found the two-sided ad to be inferior to the one-sided ad particularly when very important attributes are disclaimed (e.g. Stayman, Hoyer and Leon 1987). It is very likely that the equivocal results are attributable, in part, to differences among the various studies in the level of importance of the attributes disclaimed.

This study examines the moderating effect of the importance of the disclaimed attribute on the effectiveness of two-sided advertising. It is important for advertisers who use, or who are contemplating using, a two-sided advertising strategy to understand, in terms of the importance of the attribute(s) to be disclaimed, if and when a two-sided format is likely to be most appropriate. Specifically, we examine the effects of disclaiming attributes of varying importance on advertising credibility, attitude toward the ad, attitude toward the brand, purchase intention, and on the perceived value of the advertised brand. In this study, as in most of previous research in two-sided advertising, no attempt is made to refute the negative product information within the text of the advertisement. Thus, our focus is on two-sided nonrefutational appeal strategies (Kamins and Assael 1987).

PREVIOUS RESEARCH

The seemingly contradictory findings on the effectiveness of two-sided advertising point to the fact that it is mediated by a host of factors. These include audience factors such as the audience's level of education (McGuire 1964; Golden and Alpert 1982), usage status of the advertised brand (Sawyer 1973), and level of involvement (Swanson 1987); media factors such as type of media used (Faison 1971), media frequency (Sawyer 1973) and time (Belch 1981); message factors such as refutation or nonrefutation of the negative information in the ad (Kamins and Assael 1987) and the perceived importance of the negative information (Kamins and Marks 1987; Pechmann 1990; Stayman, Hoyer and Leon 1987).

While many of the above factors have received much empirical research attention, the moderating role of the perceived importance of the disclaimed attribute has been largely ignored. In fact, many researchers, following the prescription of inoculation theory (McGuire 1964), have simply disclaimed unimportant attributes in their studies, ignoring the consideration that this may not be sufficient to enhance credibility (e.g. Etgar and Goodwin 1982; Golden and Alpert 1987; Hastak and Park 1990). At the other extreme, some advertisers disclaim attributes that may be very important to some of their segments, ignoring the consideration that this may diminish purchase intentions. For example, Volkswagen disclaimed appearance and styling in its Volkswagen Beetle ads while Listerine disclaimed taste in its mouthwash ads.

Contradictory results have also been obtained even when attributes of similar importance were disclaimed, suggesting that other moderating factors were at work. Settle and Golden (1974) and Smith and Hunt (1978) disclaimed attributes that were perceived by consumers to be relatively important and found that their two-sided ads were perceived to be more credible than the one-sided ads. However, Belch (1981) and Stayman, Hoyer and Leon (1987), also disclaiming attributes that were perceived to be important, found no such superiority of the two-sided ad with regard to brand beliefs, attitudes and purchase intentions. Golden and Alpert (1987), disclaiming attributes that were perceived to be unimportant, found that credibility was enhanced for both deodorant and mass transit but behaviourial intentions were enhanced for deodorant only. Hastak and Park (1990), also disclaiming relatively unimportant attributes, found no effects of sidedness on message believability and brand evaluations.

Very little empirical research has systematically manipulated the importance of the disclaimed attribute in order to investigate its impact on advertising effectiveness. One such study was done recently by Stayman, Hoyer and Leon (1987). Based on the results of that and other studies Pechmann (1990) concluded that any increase in credibility gained by using two-sided advertising appears to be equal to or less than the costs incurred by promulgating negative information about the brand. Stayman et al. found that a two-sided ad stating that a record store did not sell concert ticketsCan unimportant attribute to their respondentsCwas not perceived to be any more truthful than a one-sided message. Respondents exposed to this two-sided ad were no more favourably disposed toward patronizing the store than those exposed to the one-sided ad. A two-sided ad stating that the record store scored poorly on store atmosphereCa moderately important attributeCwas perceived to be significantly more truthful than the one-sided message. However, respondents exposed to this two-sided ad were also no more favourably disposed toward patronizing the store than those exposed to the one-sided ad. Finally, a two-sided ad stating that the record store performed poorly on priceCa very important attributeCwas also perceived to be significantly more truthful than the one-sided message. In this scenario, respondents were less favourably disposed toward patronizing the store than those exposed to the one-sided ad. These results suggest that any gain in credibility through the use of two-sided advertising does not necessarily translate into a gain in behaviourial intention but may, in fact, result in diminished behaviourial intention as the disclaimed attribute becomes more important. This study uses an attribution theory framework to investigate the moderating effects of disclaimer importance not only on credibility and behaviourial intentions but also on ad and brand attitudes.

CORRESPONDENCE THEORY AND DISCLAIMER IMPORTANCE

Correspondence theory is a specific model of attribution theory (e.g. Kelley 1973) that is well-suited to the one-sided versus two-sided advertising paradigm (Smith and Hunt 1978). It attempts to explain the conditions under which an individual will make either 'internal' (i.e., correspondent) or 'external' (i.e., noncorrespondent) attributions to a specific event. A correspondent attribution is made when the observer attributes the motive underlying the event to the true feelings or disposition of the actor. A noncorrespondent attribution is made when the motive underlying the event is attributed to environmental factors, such as when the observer feels that the actor is being pressured to act in a certain manner (Jones and Davis 1965; Smith and Hunt 1978; Tannenbaum 1967). Applied to two-sided advertising, correspondence theory suggests that presenting both sides of an issue is likely to enhance the communicator's credibility if the disposition of the communicator that is revealed is that of truthfulness. Since one-sided appeals are the more traditional and commonly used advertising strategy, they often lead to noncorrespondent attributions on the part of the target audience. The audience is likely to attribute selfish motives to the advertiser, such as a desire to promote and sell a product. In contrast, two-sided appeals are more novel and less probable to occur. Consequently such appeals often lead to correspondent attributions. That is, the audience is likely to attribute unselfish motives to the advertiser, such as a desire to be truthful. This supposedly leads to source bolstering and, hence, increase in message credibility (Anderson and Golden 1984; Kamins and Assael 1987; Settle and Golden 1974).

Research evidence suggests, however, that the mere fact of presenting opposing arguments in a message may not be sufficient for correspondent attributions to be made, as attested by the conflicting results discussed earlier. It is possible that whether correspondent or noncorrespondent attributions are made to the advertiser depends on how important the attribute disclaimed is (Pechmann 1990; Stayman, Hoyer and Leon 1987). In particular, Pechmann (1990) notes that the effectiveness of two-sided advertising is moderated by the perceived motivation underlying the potentially unfavourable disclosure. The advertiser's motivation in disclaiming an unimportant attribute may be quite different from that in disclaiming a very important attribute. Kamins and Assael (1987) contend that correspondence theory is concerned primarily with attributions to the source (i.e., bolstering or derogation) based on the communication stimuli. A discussion of the probable consumer attributions to the advertiser in disclaiming attributes of varying importance follows.

Unimportant Attributes. If the unfavourable disclosure is of a minor nature, such as disclaiming too unimportant an attribute, such action is likely to lead to little correspondent attributions. In fact, it can be argued that two-sided advertising, in general, is no longer as novel as it once was. Consumers have become sufficiently exposed to this type of advertising such that saying something mildly derogatory about, or 'poking fun' at, one's self or product has lost its novelty or surprise value. Pechmann (1990, p. 339) also notes that "Consumers are quite accustomed to marketers discounting their brands on relatively unimportant attributes so as to position their brands as being superior on more important, highly correlated attributes". As a result the two-sided message may not be perceived as any more trustworthy or credible than the one-sided message. Any attribution of truthfulness to the advertiser would be too weak to enhance message credibility. Furthermore, attitudes toward the ad and the brand and purchase intentions will not be significantly affected by an unfavourable disclosure about a nondeterminant attribute.

Moderately Important Attributes. If the attribute disclaimed is one that is moderately important, the audience is likely to make correspondent attributions. Stayman et al. (1990) argue that since it is unusual for an advertiser to disclaim relatively important attributes the audience may conclude that only an honest advertiser would do this. Thus, the audience would attribute to the advertiser a motive of truthfulness which, in turn, will enhance the credibility of the message. Kamins and Assael (1987), Kamins and Marks (1987) and Settle and Golden (1974) used a similar reasoning to select disclaimers that were perceived to be of moderate importance in designing the two-sided ads in their studies.

While a strategy of disclaiming an attribute of moderate importance is hypothesized to enhance advertiser credibility, the question remains as to what effect it will have on attitudes and purchase intentions. Using a multi-attribute attitude reasoning, it is expected that consumers' overall evaluation of the advertised brand would be lower when the advertiser disclaims one or more moderately important attributes than when all attributes relevant to choice are positively claimed. Thus, it is hypothesized that attitudes and purchase intentions will be adversely affected when moderately important attributes are disclaimed. In particular, admitting weakness on relatively important attributes will diminish brand attitudes and purchase intentions due to the effect it will have on the perception of the quality of the advertised brand and on functional or performance risk. New products or brands will particularly be vulnerable in this regard since use experience and prior beliefs are absent. While the ad will be seen as more believable, it will not be seen to be as persuasive as a one-sided ad.

Highly Important Attributes. The disclosure of highly negative product information by a company may be occasioned by one of two reasons - it may be required by law or it may be voluntary. Pechmann (1990) notes that if consumers perceive that a marketer is required by law to make a potentially unfavourable disclosure, such as health warnings in cigarette ads, they are not likely to infer that the two-sided ad is more credible than the one-sided ad. We are concerned here with the voluntary disclosure of highly negative information as a tactic for enhancing advertiser credibility.

It may at first seem that studying a scenario involving an advertiser disclaiming highly important attributes is unrealistic since no advertiser is likely to use such a tactic. This is probably so for the most part. However, a number of well-known instances in the past in which advertisers have done exactly that, and the likelihood that it may happen again in the future, make such a scenario worthy of consideration. In long-running ad campaigns, Volkswagen disclaimed styling and appearance in its Volkswagen Beetle ads and Listerine disclaimed taste in its mouthwash ads. Obviously, styling and appearance in automobiles and taste in mouthwash are very important attributes to some consumer segments. For example, a study by Chrysler Corporation of consumer perceptions of automobile nameplates showed that appearance is one of the most important dimensions that consumers use to differentiate automobiles (Koten 1984). The same study also showed that Volkswagen was rated poorly on that dimension.

It may well be that Volkswagen disclaimed styling and appearance as a way of increasing the belief strength regarding its relative advantages in price and reliability while Listerine disclaimed taste as a way of enhancing its claims of germ killing effectiveness. But good styling is not incompatible with low price and reliability, as Japanese imports showed, nor is good taste is incompatible with germ killing effectiveness, as Scope repeatedly emphasized in its ads. Consumers do not have to give up one important benefit to obtain another important benefit when the benefits in question are uncorrelated. Volkswagen's subsequent loss of marketshare to Japanese imports and Listerine to Scope emphasize this point. Thus, disclaiming a highly important attribute, no matter the motivation, may be a risky strategy in that such action may not have the intended inoculating effect. Instead, it may serve to draw consumer and competitor attention to the weakness. If the weakness is important enough, competitors can exploit it to their advantage, resulting in diminished purchase intentions of the advertised brand.

Since voluntary disclosure of highly negative information is the exception rather the rule, such disclosure will likely be met with derogation of the sponsor. The resulting unfavourable attributions will diminish credibility and attitude toward the ad. Again, for a new product, the ad will be unpersuasive because of the very unfavourable information. The perception of the advertised product as a low quality offering will diminish attitude toward the brand and toward purchase. Thus, compared to the one-sided ad, a two-sided ad disclaiming highly important attributes will be less credible and will lead to diminished ad and brand attitudes and purchase intentions.

Perceived Value of Advertised Brand

A final issue of concern is whether a two-sided advertising strategy diminishes the perceived value or worth of the advertised brand. In other words, does the promulgation of negative information about the advertised brand diminish its value in consumers' eyes?

Viewed as a bundle of benefits, a product's utility to the consumer is the sum total of the part utilities of its attributes. Any attribute deficiencies, real or perceived, will tend to diminish the perceived utility of the product. Therefore, two-sided advertising, which by its very nature involves the marketer admitting to attribute deficiencies, will tend to reduce the perceived utility of the advertised brand to the consumer. This will be the case regardless of whether the attribute disclaimed is important or not. If established competitive options are claiming all attributes positively, a new brand disclaiming one or more of those attributes, without offering some offsetting advantage, will be perceived as offering less value for money. Thus, it is hypothesized that the perceived value of the advertised brand, measured in terms of the amount that consumers would be willing to pay for it, will be lower with two-sided advertising than with one-sided advertising.

The above discussions about the effects of disclaimer importance on the effectiveness of two-sided advertising, relative to one-sided advertising, are summarized in the following hypotheses:

H1: Advertising credibility will be unaffected when unimportant attributes are disclaimed, increase when moderately important attributes are disclaimed, and diminish when highly important attributes are disclaimed,

H2: Attitude toward the ad will be unaffected when unimportant attributes are disclaimed, but will become less favourable when moderately or highly important attributes are disclaimed,

H3: Attitude toward the brand will be unaffected when unimportant attributes are disclaimed, but will become monotonically unfavourable as increasingly important attributes are disclaimed,

H4: Purchase intention will be unaffected when unimportant attributes are disclaimed, but will diminish monotonically as increasingly important attributes are disclaimed, and

H5: The perceived value of the advertised brand will be lower regardless of the level of importance of the disclaimed attribute.

METHOD

Subjects

Data for this study were obtained from 108 undergraduate students at a medium-sized university in southwestern Ontario. Subjects were randomly assigned to one of four appeal-type treatment groups. One group was exposed to a one-sided advertisement while the remaining three were exposed to two-sided ads varying in the levels of disclaimer importance (i.e. low, moderate, or high).

Product Category

A ballpoint pen was selected as the product category of study in this research for a number of reasons. A pen is relevant to a student population, has few hidden qualities, can be evaluated in a short period of time, and is not gender-specific (Kamins and Assael 1987). Further, research suggests that advertisements for low-involvement products produce a greater change in attitude than ads for high involvement ones (Faison 1961). Since a ballpoint pen is generally low-involving, any changes in attitude between treatment groups should be more easily detectable. Also, ads for functionally utilitarian products tend to receive higher attitude scores than those for socially utilitarian ones (Etgar and Goodwin 1982).

A foreign pen, the "Pentel" pen, not available in the local market, was selected for use in this study. An unfamiliar pen was selected so as to minimize preconception bias.

A pilot study was conducted with a sample of thirty-eight students, different from the ones used in the main study, to measure, among other things, the level of familiarity of the sample population with different pen brands. The mean familiarity rating for the "Pentel" pen was only 2.8 on a seven-point scale, compared to 6.6 for "Bic" and 5.8 for "Papermate". Thus, the "Pentel" pen was relatively unfamiliar to the study population.

Selection of Product Attributes

A review of previous studies using a pen as the focal product (Anderson 1973; Hastak and Park 1990; Kamins and Assael 1987; Kamins and Marks 1987; Olson, Toy and Dover 1978) and of pen advertisements had enabled the identification of twenty-three category attributes for initial consideration. The pilot sample of thirty-eight students was asked to rate each attribute on the extent to which it would be important in a purchase decision. Each attribute was rated on a seven-point 'not at all important' to 'extremely important' scale. Following this, two top-rated, two moderately-rated, and two low-rated attributes were selected. Pairwise t-tests were used to ensure that the ratings did not differ significantly between pairs of attributes rated to be of similar level of importance while differing significantly between pairs rated to be of different levels of importance, at p=.05. Using these criteria, 'consistency of ink flow' and 'writing performance' were selected as the two highly important attributes (mean ratings 6.23 and 6.03, respectively); 'quality of construction' and 'writes at any angle' were selected as the two moderately important attributes (mean ratings 4.97 and 4.93, respectively); 'appearance' and 'colours available' were selected as the two low importance attributes (mean ratings 3.30 and 3.27, respectively). These six attributes were used for the construction of the stimulus ads for the main study.

Advertisements

Four different black and white advertisements were commercially developed, one for each of the four appeal-type treatments. All of the ads had the same visual background. The visual portion comprised about seventy percent and the verbal portion the remainder of the nonblank page. The verbal portion described the pen on the six attributes identified above. In the one-sided ad positive claims were made on all six attributes. In the two-sided ads, four of the six attributes (67.7 percent) were claimed positively while two (33.3 percent) were disclaimed. In comparison, 3 of 5 attributes (60 percent) were positively claimed while 2 (40 percent) were disclaimed in the Etgar and Goodwin (1982) and Kamins and Assael (1987) studies.

The first two-sided ad disclaimed the two low importance attributes, the second disclaimed the two moderate importance attributes, and the third disclaimed the two high importance attributes. In each case positive claims were made about the remaining four attributes. Disclaimed attributes were placed early in the message. Sternthal et al. (1978) suggest that placing the disclaimer early in the message further enhances belief change. The one-sided ad said:

You'll appreciate the smooth and consistent ink flow of the PENTEL pen. The overall writing performance of the PENTEL pen is second to none. The PENTEL pen is well constructed and is capable of writing at any angle. It has a pleasant appearance and is available in several colours. GET IT WRITE, BUY A PENTEL PEN!

The two-sided ad disclaiming the two highly important attributes said:

Although the PENTEL pen is not the best pen available in terms of smooth and consistent ink flow, and it may not give the user the best overall writing performance, it does have several appealing characteristics. The PENTEL pen is well constructed and is capable of writing at any angle. It has a pleasant appearance and is available in several colours. GET IT WRITE, BUY A PENTEL PEN!

The other two-sided ads differed only in the attributes claimed negatively or positively.

Experimental Procedure

The main study was carried out in a classroom setting. Data collection began with the establishment of a guise unrelated to the real purpose of the study. They were told that there were no right or wrong answers to the questions, just their honest opinions. Finally, they were asked not to discuss their opinions with others while completing the questionnaires. Subjects first completed a questionnaire measuring the importance to them of the six attributes used in the ads. Upon collecting this questionnaire, the experimental ads were handed out, rotating the four ads among subjects. Subjects were instructed to read the ads for three minutes after which they were instructed to turn them face down on their desks. They then completed a questionnaire which measured their attitudes toward the ad they had just read, their attitudes toward the brand advertised, their purchase intentions, and how much they would be willing to pay for the pen advertised. Finally, respondents rated the pen on how they thought it would perform on each of the six attributes.

Of the 108 usable questionnaires, 27 evaluated the one-sided ad, 27 evaluated the two-sided ad disclaiming attributes of low importance, 26 evaluated the two-sided ad disclaiming attributes of moderate importance, and 28 evaluated the two-sided ad disclaiming attributes of high importance.

Dependent Variables

The dependent variables measured were respondent's opinion as to the believability of the ad, respondent's attitudes toward the ad, toward the advertised brand, purchase intention, and amount willing to pay for the advertised brand.

Advertising believability was measured by a single item - the extent to which the respondent felt the ad he or she just read was believable.

Attitude toward the ad was measured by the extent to which the respondent felt the ad was interesting, appealing, pleasant, believable, impressive, informative, clear, inoffensive, persuasive, effective, tasteful, likeable, and truthful.

Attitude toward the brand was measured by the extent to which the respondent felt the brand advertised was good, of superior performance, of better quality than leading brands, likeable, distinctive, not risky, and reliable.

Purchase intention was measured by the extent to which the respondent would 'buy' and 'seek out' the advertised brand.

Each item was measured on a seven-point semantic differential scale.

Finally, the perceived value of the advertised brand was measured using an open-ended question that asked how much the respondent would be willing to pay for it.

Following data collection, item-to-total correlations and reliability analyses suggested that the items 'inoffensive' and 'truthful' should be dropped from the pool measuring attitude toward the ad. Cronbach's alpha for the remaining eleven items in this pool was 0.86. Likewise, the item 'distinctive' was dropped from the pool measuring attitude toward the brand. Cronbach's alpha for the remaining six items in this pool was 0.90. Finally, the correlation between the two items measuring purchase intention was 0.65 (p <.01).

DATA ANALYSIS AND RESULTS

Manipulation checks

The first manipulation check was conducted to ascertain that the study sample perceived the importance of the six attributes used in the ads similarly to the pilot sample.

EXHIBIT

MEAN SCORES AND ANALYSES OF VARIANCE BY APPEAL TYPE

The ratings of the importance of the six attributes by the study sample were compared with the ratings provided by the pilot sample, using multivariate analysis of variance. Not only were no differences in mean ratings significant at p <.05 but also the rank-ordering of the attributes according to the level of importance was preserved.

A second manipulation check was conducted to ascertain that the operationalization of the two-sided appeals was perceived as intended. For each two-sided appeal, t-tests were used to compare the ratings of the expected performance of the pen on the disclaimed attributes between subjects exposed to that appeal and those exposed to the one-sided appeal. The results showed that subjects exposed to each two-sided appeal consistently rated the pen lower on the disclaimed attributes than subjects exposed to the one-sided appeal (p <.02).

Effects of Appeal Type on the Dependent Variables

The scores on the items measuring attitude toward the ad, attitude toward the brand and purchase intentions were compared among appeal types using MANOVA. The main effect of appeal type was significant (Wilk's lambda =.496, d.f.=12.246, p=.000). Univariate analyses of variance were then used to understand better the nature of the effects of appeal type. The exhibit shows the mean scores for believability, attitude toward the ad, attitude toward the brand, purchase intentions, and the perceived value of the advertised brand in terms the amount respondents are willing to pay for it in dollars, by appeal type. It also shows the results of univariate F-tests and contrasts using the Tukey multiple comparison procedure.

Believability. The believability scores show that the one-sided ad in this study was perceived to be somewhat believable, with a mean score of 4.96 (out of 7). When unimportant attributes were disclaimed believability was not significantly affected (mean score 4.85). When moderately important attributes were disclaimed mean believability score, at 5.00, did not change significantly. However, when highly important attributes were disclaimed believability decreased to 3.75. Contrasts show mean believability to be significantly lower when highly important attributes are disclaimed than when no attributes or when moderately important attributes were disclaimed. These results partially supports H1. Attitude toward the ad. Overall, respondents exposed to the one-sided ad were somewhat favourably disposed toward it, as indicated by the mean score of 4.78 (out of 7) across the eleven items measuring this variable.

The contrasts show that attitude toward the ad was unaffected when unimportant attributes were disclaimed relative to the one-sided ad. It declined significantly when moderately or highly important attributes were disclaimed. Thus the one-sided ad was significantly better than the-sided ads disclaiming moderately or highly important attributes. The figure shows that the decline in attitude toward the ad appears to be monotonic. These results support H2.

Attitude toward the brand. Respondents exposed to the one-sided ad were somewhat favourably disposed toward the advertised brand as indicated by the mean score of 5.04. The contrasts show that attitude toward the brand was unaffected when unimportant attributes were disclaimed. It declined significantly when moderately important attributes were disclaimed to a mean score of 4.08 and and to 3.04 when highly important attributes were disclaimed. The contrasts show that attitude toward the brand appears to become increasingly less favourable as the disclaimed attributes become increasingly important. These results support H3.

Purchase intentions. Respondents exposed to the one-sided ad displayed neutral purchase intentions toward the advertised brand (mean score 4.13). When unimportant attributes were disclaimed purchase intentions were not significantly affected. However, when moderately and highly important attributes were disclaimed purchase intentions declined significantly. Contrasts show that purchase intentions were significantly higher when no attributes were disclaimed or when unimportant attributes were disclaimed than when moderately or highly important attributes were disclaimed. There were no significant differences in purchase intentions between disclaiming moderately important attributes and disclaiming highly important attributes. These results largely support H4.

The figure summarizes the above results graphically. It shows plots of the overall mean scores of the dependent variables by appeal type. Believability showed a small increase when moderately important attributes were disclaimed and a severe plunge when highly important attributes were disclaimed. Attitude toward the ad, attitude toward the brand and purchase intentions all appear to decline monotonically as the importance of the disclaimed attributes increased.

FIGURE

PLOT OF DEPENDENT VARIABLES BY APPEAL TYPE

Perceived value. Finally, respondents exposed to the one-sided ad were willing to pay $3.11 for the advertised pen. The amount they were willing to pay decreased significantly when unimportant, moderately important and highly important attributes were disclaimed. Although the amount willing to pay is somewhat higher when moderately important attributes were disclaimed than when unimportant attributes were disclaimed the difference is not statistically significant and could have been simply due to chance variation. As expected, however, respondents would be willing to pay the least amount when highly important attributes were disclaimed. The contrasts show that the perceived value of the advertised brand was significantly higher with the one-sided ad than with any of the two-sided ads. This result supports H5.

DISCUSSION

This study investigated the moderating effect of disclaimer importance on several measures of advertising effectiveness using a two-sided nonrefutational appeal.

The results showed that when unimportant attributes were disclaimed, believability, attitude toward the ad, attitude toward the brand and purchase intentions were not significantly affected relative to the one-sided ad. It may be that because an unfamiliar brand was used in this study there were no pre-existing beliefs about the brand advertised and, consequently, no basis for counterarguing. Furthermore, it appears the one-sided ad that was used was perceived to provide factual information, without puffery. Since it was already quite believable, as indicated by the score of 4.96 out of 7, there was little room for improvement using a two-sided format. In fact, the two-sided ad disclaiming unimportant attributes led to a small decrease in believability. In any case, the present results are not surprising. They support past research which, disclaiming unimportant attributes, also found no significant effects on advertising evaluations (Belch 1981; Golden and Alpert 1978; Hastak and Park 1990; Stayman, Hoyer and Leon 1987).

When attributes of moderate importance were disclaimed, believability increased slightly but not significantly while attitude toward the ad, attitude toward the brand and purchase intentions diminished significantly relative to the one-sided ad. Thus, this result only provides directional support to other studies that have reported increases in message credibility when moderately important attributes were disclaimed (Kamins and Marks 1987; Settle and Golden 1974).

When highly important attributes were disclaimed, believability, attitude toward the ad, attitude toward the brand and purchase intentions all diminished significantly relative to the one-sided ad. Stayman et al. also found a significant decrease in behavioral intention toward the record store in their study when highly important attributes were disclaimed. As predicted using correspondence theory, disclaiming highly important attributes resulted in source derogation. Many of the respondents exposed to the ad disclaiming highly important attributes, in answer to open-ended questions, found it unbelievable and stupid that a marketer would "say such bad things about his product". These results suggest disclaiming highly important attributes is a risky strategy particularly in a new product situation. The negative information promulgated about the new product will have adverse effects on brand attitudes and purchase intentions. When highly important attributes have been disclaimed in practice it has typically been in situations where negative brand beliefs already exist, such as Volkswagen (appearance) and Listerine (taste). By admitting to already perceived deficiencies, the marketer hopes to diffuse their effects. Additional research is required, but it may be that a two-sided message strategy is better used as a reactive, rather than a proactive, strategy.

The finding that the amount respondents are willing to pay for the advertised brand declines when attributes are discounted has important implications for pricing strategy. The promulgation of negative information about a brand through two-sided advertising apparently diminishes its perceived value. This will be particularly so for a new brand being introduced into a market with acceptable competitive options. The present results suggest that the sidedness of the message should be consistent with the pricing strategy. If a premium price is being charged the advertiser may benefit most by using a one-sided appeal, making only positive claims about the brand. On the other hand, if a discount price is being charged, a two-sided appeal may be more effective, such as in comparative item pricing across food stores. However, a two-sided appeal may still be effective where a premium price is being charged if it is the price of the product that is being disclaimed as in, for example, 'This brand costs more but its worth it'.

In summary, this study found the one-sided ad to perform generally better than the two-sided nonrefutational ad, regardless of the level of importance of the attribute disclaimed. However, it does more than show that the one-sided message strategy works better than the two-sided strategy when the audience is initially predisposed toward the advertiser's position, as appeared to be the case in this study. The results of this study also support the contention that disclaiming an unimportant attribute may not be sufficient to enhance message credibility while disclaiming too important an attribute may penalize brand attitudes and purchase intentions (Kamins and Marks 1987; Kamins and Assael 1987).

The main contribution of this study, apart from confirming or raising questions about previous findings in this topic area, is that it delineates the effects of disclaiming attributes of varying importance on different measures of advertising effectiveness. Although some of the findings are intuitive, such as the negative impacts on attitudes and purchase intentions of disclaiming moderately and highly important attributes, the extent of the impact was not obvious and could not have been anticipated. Some may question the value of, or contribution to knowledge, in researching a message tactic that is unrealistic and unlikely to be used. It is true that disclosing highly negative information about one's product is not commonplace, but it is not as unrealistic as it first seems. The results of this study would be relevant not only to advertisers who, for legal or regulatory reasons, are required to disclose highly negative product information (eg., health warnings, harmful side effects) but also to the few who, perhaps to increase belief strength about some other positive product attributes, may voluntarily discount highly important attributes (eg., Volkswagen, Listerine). This research provides some insights into the consequences of such actions.

This study also addressed the impact on the perceived value of a new or unfamiliar product of using a two-sided nonrefutational appeal strategy, an issue that has not been addressed before. The results suggest that an advertiser may diminish the perceived value of a new product or brand by disclaiming any attributes, whether important or not.

LIMITATIONS AND FUTURE RESEARCH

A number of considerations limit the generalizability of the present results. Only a single product was used in this study. It is quite likely that different results may be obtained for other product categories, as did Stayman et. al. for alarm clock versus record store. Golden and Alpert (1987) also found different results for deodorant and mass transit. They caution that the effectiveness of the one-sided versus two-sided message remains open, depending on the product, the amount of puffery, the target market, the situation and the perceptions of competitive options available to the receivers of the message. Thus, one cannot assume that one type of message will always be better than the other.

Although the product used in this study was relevant to the population studied, the use of a student sample always calls into question the generalizability of findings to the population at large. There is speculation that university students in general tend to be more sceptical about advertising claims than other groups (Baker and Churchill 1977).

The use of only one medium, a forced single exposure in an unnatural environment and at a single point in time also constitute limitations to the generalizability of the present results across media type, media frequency and time.

While all of the above limitations are matters of external validity, the internal validity of the present findings should be unaffected.

Additional work is obviously required to validate these findings since the importance of the disclaimed attribute is a potentially important moderator of effectiveness in two-sided advertising and one that, as noted, has potentially important ramifications for advertisers who use this strategy. In addition to issues regarding importance of the disclaimed attribute, other message factors are potential moderators of effectiveness in two-sided advertising. One factor is the degree of counter- versus pro-attitudinality of the message overall. The more counter-attitudinal a message, the more it will benefit from a two-sided format. For example, comparative appeals are counter-attitudinal when they are made by market followers (Belch 1981). Such messages may benefit from a two-sided format, although the risk remains that any increase in credibility may be offset by a decrease in persuasiveness. Another factor is the degree of correlation between the negatively-claimed attribute and another positively-claimed attribute in the message. Pechmann (1990) notes that when the negatively-claimed attribute is highly correlated with a positively-claimed attribute, consumers may infer that the brand has a more favourable performance on the positively-claimed attribute. For example, suppose price and quality were highly correlated for the situation in question. A negative claim on price will tend to enhance a positive claim on quality. Thus, with a two-sided ad that claimed "It costs a bit more but it's worth it", consumers would be more accepting of the higher quality claim. By the same token, a two-sided claim by a grocery store of lower prices (positive claim) but fewer amenities (negative claim) should be more believable than a one-sided claim of lower prices. However, as Belch (1981) found, if consumers prefer more amenities and higher prices, the increased credibility may not translate into increased behavioral intentions. Nevertheless, the moderating effect of the degree of correlation between the negatively- and positively-claimed attributes on the effectiveness of two-sided advertising needs additional research (Pechmann 1990).

The present results were obviously contingent upon the use of nonrefutational ads. Research is required comparing the effectiveness of two-sided ads with refutation and without as the importance of the disclaimed attribute varies. Kamins and Assael (1987), disclaiming moderately important attributes, found no difference between their refutational and nonrefutational ads on change in belief and purchase intention under an attack condition in the form of a disconfirming product trial. Perhaps the results may differ if unimportant or highly important attributes were disclaimed. An important question to be addressed is whether refutation of negative information in two-sided ads becomes more (or less) necessary as the negative information becomes more severe.

REFERENCES

Anderson, Rolph E. (1973), "Consumer Dissatisfaction: The Effect of Disconfirmed Expectancy on Perceived Product Performance," Journal of Marketing Research, 10 (February), 38-44.

Anderson, W. Thomas, Jr. and Linda L. Golden (1984), "Bank Promotion Strategy," Journal of Advertising Research, 24 (April/May), 53-65.

Baker, Michael J. and Gilbert A Churchill, Jr. (1977), "The Impact of Physically Attractive Models on Advertising Evaluations," Journal of Marketing Research, 14 (November), 538-555.

Belch, George E. (1981), "An Examination of Comparative and Noncomparative Television Commercials: The Effects of Claim Variation and Repetition on Cognitive Response and Message Acceptance," Journal of Marketing Research, 18 (August), 333-349.

Earl, Ronald L. and William M. Pride (1980), "The Effects of Advertisement Structure, Message Sidedness, and Performance Test Results on Print Advertisement Effectiveness," Journal of Advertising, 9 (September), 36-46.

Etgar, Michael and Stephen A. Goodwin (1982), "One-sided Versus Two-sided Comparative Message Appeals for New Brand Introductions," Journal of Consumer Research, 8 (March), 460-464.

Faison, Edmund W. (1971), "Effectiveness of One-Sided and Two-Sided Mass Communications in Advertising," Public Opinion Quarterly, 25 (Fall), 468-469.

Golden, Linda L. and Mark I. Alpert (1982), "The Impact of Education on the Relative Effectiveness of One-sided and Two-sided Communications," in An Assessment of Marketing Thought and Practice, Bruce J. Walker, et al. (eds.), 30-33.

Golden, Linda L. and Mark I. Alpert (1987), "Comparative Analysis of the Relative Effectiveness of One-and Two-Sided Communication for Contrasting Products," Journal of Advertising, 16, No. 1, 18-25.

Hastak, Manoj and Jong-Won Park (1990), "Mediators of Message Sidedness Effects on Cognitive Structure for Involved and Uninvolved Audiences," in Advances in Consumer Research, Marvin E. Goldberg et al. (eds), 17, 329-336.

Jones, E.F. and K.E. Davis (1965), "From Acts to Dispositions: the Attribution Process in Person Perception," Advances in Experimental Social Psychology.

Kamins, Michael A. and Henry Assael (1987), "Two-Sided Versus One-Sided Appeals: A Cognitive Perspective on Argumentation, Source Derogation, and the Effect of Disconfirming Trial on Belief Change," Journal of Marketing Research, 24 (February), 29-39.

Kamins, Michael A. and Lawrence J. Marks (1987), "Advertising Puffery: The Two-sided Claims on Product Attitude and Purchase Intention," Journal of Advertising, 16, No. 4, 6-15.

Kelley, H.H. (1973), "The Processes of Causal Attribution," American Psychologist, 28, 107-128.

Koten, John (1984), "Car Makers Use 'Image' Map as Tool to Position Products," The Wall Street Journal, March 22, 31.

McGuire, William J. (1964), "Inducing Resistance to Persuasion: Some Contemporary Approaches," in Advances in Experimental Social Psychology, L. Berkowitz, ed., 191-229.

Olson, Jerry C., Daniel R. Toy, and Philip Dover (1978), "Mediating Effects of Cognitive Responses to Advertising on Cognitive Structure," in Advances in Consumer Research, H. Keith Hunt (ed.), 5, Ann Arbor, MI: Association for Consumer Research, 72-78.

Pechmann, Cornelia (1990), "How Do Consumer Inferences Moderate the Effectiveness of Two-Sided Messages?", in Advances in Consumer Research, Marvin E. Goldberg et al. (eds), 17, 337-341.

Sawyer, Alan G. (1973), "The Effects of Repetition of Refutational and Supportive Advertising Appeals," Journal of Marketing Research, 10 (February), 23-33.

Settle, Robert B. and Linda L. Golden (1974), "Attribution Theory and Advertiser Credibility," Journal of Marketing Research, 11 (May), 181-185.

Smith, Robert E. and Shelby D. Hunt (1978), "Attributional Processes and Effects in Promotional Situations," Journal of Consumer Research, 5 (December), 149-158.

Stayman, Douglas, Wayne Hoyer and Robert Leon (1987), "Attribute Importance in Discounting Product Features in Advertising," in AMA Educators' Proceedings No. 53, Susan P. Douglas et al. (eds), American Marketing Association Summer Educator's Conference, Toronto, Canada.

Sternthal, Brian, Ruby Dholakia, and Clark Leavitt (1978), "The Persuasive Effect of Source Credibility: Tests of Cognitive Response," Journal of Consumer Research, 4 (March), 252-260.

Swanson, Lauren A. (1987), "The Persuasive Effect of Volunteering Negative Information in Advertising," Journal of Marketing Research, 18 (May), 175-186.

Swinyard, William R. (1981), "The Interaction Between Comparative Advertising and Copy Claim Variation," Journal of Marketing Research, 18 (May), 175-186.

Tannenbaum, Percy H. (1967), "The Congruity Principle Revisited: Studies in the Reduction, Induction, and Generalization of Persuasion," in Advances in Experimental Social Psychology, L. Berkowitz, ed., 3, 271-320.

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