Message Sidedness At the Brand and Product Form Levels: Overcoming the Shortcomings of Two-Sided Messages?

Bodo Lang, Deloitte & Touche Consulting Group
Christina Kwai-Choi Lee, University of Auckland
Rami Zwick, The Hong Kong University of Science and Technology
ABSTRACT - This paper investigates how messages that contain only positive information vary in their effectiveness from messages that portray both positive and negative arguments. The authors propose that, in a low price condition, a two-sided message with comparison at the product form level has the ability to simulataneously retain the benefits and overcome the weaknesses of two-sided communication, increased source credibility and decrease in purchase intention, respectively. Results from our experiment indicate a strong interaction between price and type of message. The two-sided comparative message at product form level outperforms both the one-sided message and the classical two-sided comparative message at brand level.
[ to cite ]:
Bodo Lang, Christina Kwai-Choi Lee, and Rami Zwick (1999) ,"Message Sidedness At the Brand and Product Form Levels: Overcoming the Shortcomings of Two-Sided Messages?", in NA - Advances in Consumer Research Volume 26, eds. Eric J. Arnould and Linda M. Scott, Provo, UT : Association for Consumer Research, Pages: 485-490.

Advances in Consumer Research Volume 26, 1999      Pages 485-490

MESSAGE SIDEDNESS AT THE BRAND AND PRODUCT FORM LEVELS: OVERCOMING THE SHORTCOMINGS OF TWO-SIDED MESSAGES?

Bodo Lang, Deloitte & Touche Consulting Group

Christina Kwai-Choi Lee, University of Auckland

Rami Zwick, The Hong Kong University of Science and Technology

ABSTRACT -

This paper investigates how messages that contain only positive information vary in their effectiveness from messages that portray both positive and negative arguments. The authors propose that, in a low price condition, a two-sided message with comparison at the product form level has the ability to simulataneously retain the benefits and overcome the weaknesses of two-sided communication, increased source credibility and decrease in purchase intention, respectively. Results from our experiment indicate a strong interaction between price and type of message. The two-sided comparative message at product form level outperforms both the one-sided message and the classical two-sided comparative message at brand level.

INTRODUCTION

Every brand in the consumer's mind is associated with both positive and negative realities. If the brand is, for example, considered high quality, it will most likely be associated with high price in which case a tradeoff is made. Such duality represents a dilemma for marketing communication where a decision has to be made as to the content of persuasive messages. Should it contain both positive and negative information (i.e. be a two-sided message) or should it present only positive information (i.e. be a one-sided message)? Therefore the central question is typically whether, and under what conditions, a two-sided message can enhance the persuasiveness of marketing communication (Cho 1995; Faison 1961; Smith and Hunt 1978).

Previous Findings

There are conflicting results from studies on the effectiveness of two-sided messages. Some studies have found that the inclusion of negative information about a product enhances the persuasiveness of a message compared to a 'positive-only', one sided message (Etgar and Goodwin 1982; Golden and Alpert 1987; Kamins and Assael 1987a; Allen 1991; Pechmann 1992; Allen 1993). This effect is typically explained by reference to inoculation theory (Etgar and Goodwin 1982), attribution theory (Crowley and Hoyer 1994), correspondence theory (Smith and Hunt 1978), and assimilation - contrast theory (Kamins and Assael 1987b; McGinnies 1966). In all cases, the enhanced effectiveness of two-sided messages is due, in part, to reinforced source credibility. Other studies, on the other hand, have failed to replicate these results (Kanungo and Johar 1975; Belch 1981; Hastak and Park 1990). One of the reasons for the conflicting results is the way message persuasiveness is measured.

Two-sided messages are consistently found to enhance the perceived credibility of claims (Kamins and Assael 1987a; Smith and Hunt 1978; Swinyard 1981), and the perceived trustworthiness of the advertisement (Smith and Hunt 1978; Kamins 1989). Some studies found that two-sided messages lead to greater purchase intention (Etgar and Goodwin 1982; Golden and Alpert 1987; Kamins 1989), others reported that two-sided messages lead to lower purchase intention (Kanungo and Johar 1975; Swinyard 1981), and yet others have concluded that there is no significant difference in purchase intention between one- and two-sided communication (Sawyer 1973; Belch 1981; Swanson 1987). Past findings also suggest that purchase intention is particularly sensitive to the strength of the negative information and to the exact dimension on which it is reported.

It is quite clear from the above findings that, for a two-sided message to be effective (in terms of purchase intention), the disadvantages incurred by the negative information should be over compensated by the benefits granted via source credibility.

STUDY OBJECTIVES

The objective of this study is to overcome the commonly observed tradeoff between the enhanced perception of source credibility and the possible sacrifice in beliefs about quality evaluation. This objective is achieved by comparing two types of two-sided messages and a one-sided message. The two types of two-sided messages are described below.

Message Sidedness at the Brand and Product Form Levels

It has been reported that the combination of two-sided messages with a comparative format has benefits, such as reducing the respondent's skepticism, enhancing the credibility of the advertisement (Belch 1981), and reducing the number of counter arguments and the amount of source derogation (Kamins and Marks 1988). Therefore, a comparative two-sided format was used in this study.

In this format the negative information is indicated by an inferior attribute in the advertised brand to a reference point. An essential point for this study is that there is more than one way to implement this comparison. In almost all previous studies, the negative information was provided by a comparison to another brand within the same product form. Other alternatives that were seldom tested indicated the inferiority of the advertised brand to a brand within another product form (e.g. an advertised brand of wrist watch calculator is compared to a generic handheld calculator on the variable keyboard size), or within another product category (e.g. an advertised brand of wrist watch calculator is compared to a generic desk top computer on keyboard size).

The level (i.e. product brand, product form, or product category) with which one compares a brand in a two-sided message may have a significant impact on the persuasiveness of the communication because the negative information may be treated differently by consumers. Initial evidence supporting this possibility was provided by Alden and Crowley (1995) who found that a two-sided message with negative information at the product category level performed better than its one-sided counterpart, however they did not compare it to the other two types (i.e. brand and form levels).

Hypotheses

The underlying rationale for our hypotheses is based on the consumer's perception of value and use of expected and unexpected information. The more probable a message is, the less information it conveys. For example, an advertisement conveying a highly probable message, such as in a puffery advertisement, is probably judged to be very uninformative about the actual quality of the brand. On the other hand, an unexpected message, such as when negative information about the brand is provided, would be judged as transmitting more meaningful information.

We pose that the further away the superior reference point is to the target brand, the less informative and relevant the information in the advertisement is considered to be. Likewise, the less informative the advertisement is, the less credible it is judged to be. Consequently, the further away the superior comparative standard is to the target brand, the less damaging the negative information is to the overall evaluation of the target brand's quality. At some point, of course, the comparative standard can be so far away that the comparison becomes trivial or even absurd and is likely to be perceived as not containing any useful information. It is also possible that at this extreme case, message credibility will be damaged, but in this study we will stay within the boundary of a plausible comparison. Hence, a comparison at the product category level was dropped. This notion is supported by studies that found that the more trivial the negative arguments become, the less effective they are in generating source credibility (Hass and Linder 1972).

Consider a brand A , that is evaluated on n attributes {A1, . An} and an advertisement for brand A which claims superiority on attributes A1 to An-1 and inferiority on attribute An. The more relevant the negative information about attribute An is considered to be, the more credible the advertisement is and hence the more believable the superior (and the inferior) claims. At the same time, however, the more relevant the negative information about attribute An is, the more damaging this information becomes. This is especially consequential because we know that negative information influences choice in a disproportional way compared to positive information (Kanouse and Hanson 1972).

This study is motivated by the search for the optimal (from an advertiser's point of view) balance between reducing the relevancy of negative information, on one hand, and increasing message credibility (by providing the same negative information) on the other hand. A tradeoff is unavoidable because both depend on the information value of the message.

First, we expect to replicate the commonly experienced 'tradeoff' effect at the brand comparison level (Kamins and Marks 1987; Crowley and Hoyer 1994), consequently we hypothesize that:

H1a: A two-sided message with a brand comparison (TSb) is more credible than a one-sided message (OS).

H2a: A two-sided message with a brand comparison (TSb) would lead to lower purchase intention compared to a one-sided message (OS).

The effect described in H2a is typically due to insufficient compensation for the damaging effect of the negative information by improvement in the advertisements credibility and hence to the believability of the positive information.

Based on the rationale discussed above we hypothesize that:

H1b: A two-sided message with a product form comparison (TSf) is more credible than a one-sided messages (OS).

H2b: A two-sided message with a product form comparison (TSf) would lead to higher purchase intention compared to a one-sided message (OS).

H2b is based on our conjecture that the negative information will not be regarded as highly relevant to quality evaluation (since the information is not unexpected), but yet will be sufficient to enhance source credibility (because the information is not trivial) and hence the believability of the positive information.

An advertisement is considered to be less informative the further away the superior comparative standard is to the target brand. Consequently, we hypothesize that:

H1c: A two-sided message with a brand comparison (TSb) is more credible than a two-sided message with a product form comparison (TSf).

In other words, the more informative the advertisement the more credible it is.

H2a and H2b imply that a two-sided message with a product form comparison (TSf) would lead to a higher purchase intention compared to a two-sided message with a brand comparison (TSb). This is based on the conjecture that the damage due to relevant negative information provided by a brand comparison is not sufficiently offset by the enhanced source credibility (we will refer to this statement as H2c, although it is not an independent hypothesis).

The hypotheses stated thus far are expected only when consumers are willing to accept negative aspects of the brand. It is clear that the more expensive the brand is, the less tolerant consumers will be to negative information. Consequently, past research has found that two-sided messages are more effective at low/moderate price levels than at a high price level (Chebat and Picard 1985).

We hypothesize that a high price will not change the effect of message sidedness on the perceived credibility of the advertisement but will directly affect purchase intention by reducing the consumers' willingness to accept a quality tradeoff. Consequently, we expect the one-sided message to dominate both versions of the two-sided message on purchase intention at the high price level. Further, the negative information in a two-sided message with a product form comparison is not as relevant as in a two-sided message with a brand comparison. Thus, we hypothesize that purchase intention will be higher when consumers are exposed to the first rather than the latter version.

H3: A two-sided message with a brand comparison (TSb) is more credible than a two-sided message with a product form comparison (TSf), which is more credible than a one-sided message at a high price levels.

H3 states that H1a-c will hold at both price levels.

H4: For high priced products, a one-sided message (OS) would lead to higher purchase intention compared to a two-sided message with a product form comparison (TSf), which will lead to a higher purchase intention compared to a two-sided message with a brand comparison (TSb).

H4 states that H2a and H2c will hold at a high price level but H2b will be reversed.

The next section describes the experiment that was used to test the above hypotheses.

METHOD

A study with a 3 (Message Sidedness; OS, TSf, and TSb) x 2 (Price; Low=HK$199.95, and High=HK$999.95 [The exhcange rate between Hong Kong and US dollars is approximately 7.8 to 1.]) between-subjects design was conducted to examine these hypotheses. A total of 215 undergraduate students participated in the study. Subjects were tested in groups of 30 to 40 people during 20-minute sessions. Subjects in the same session were assigned to the same experimental treatment.

TABLE 1

KEYBOARD SPECIFIC TEXT USED IN THE THREE VERSIONS

The Product

An advertisement for an imaginary brand of wrist watch financial calculator, called Calcu SX-2000 was used as the stimulus. The product was selected because it is a product with which the students are reasonably familiar but for which they hold a somewhat unfavorable attitude. [See description of the pre-test below.] Both of these conditions were found to facilitate the effectiveness of two-sided communication (Faison 1961; McGinnies 1966). Furthermore, the nature of this investigation requires that there is a product form (e.g. handheld financial calculators) that is superior with regard to a relevant attribute (e.g. keyboard size).

The Advertisement

Full-page black and white print advertisements were constructed. All advertisements were identical except for the price (and the claim made about the price) and the information provided about the keyboard. Information about the keyboard appeared in the second paragraph. [The negative information was presented early in the message, but not first as this has been found to be the most beneficial place for negative information in two-sided messages (Hass and Linder 1972; Hunt, Smith and Kernan 1984; Crowley and Hoyer 1994).] The different versions of this text defining the three message sidedness levels are presented in Table 1. Note that the same physical information about the keyboard size (1x2 cm) is provided in all three versions. No picture was included in the advertisement since we wanted all evaluations to be based on only the verbal information provided.

Calcu SX-2000 was described using six attributes selected on the basis of two articles in Consumer Reports (Henle 1982; Denver 1991) ensuring that these attributes were meaningful. Keyboard size was selected as the attribute for the negative information because, first, past studies have shown that two-sided messages work better when respondents are somewhat aware of 'the negative side' of the product (Jones and Brehm 1970; Hass and Linder 1972; Crowley and Hoyer 1994), and in a pre-test we found that students were aware of the potential drawback of a small keyboard on a wrist watch calculator. Second, past research suggests that a two-sided message is most effective when a moderately important attribute is used in including the negative information (Stayman, Hoyer and Leone 1987; Crowley and Hoyer 1994). The two Consumer Reports mentioned above suggest that 'keyboard size' is neither the most important attribute nor a trivial one. The other attributes were overall size, weight, price, memory management, and screen design.

Procedure and Measures

Respondents were randomly assigned to one of the six experimental conditions and correspondingly exposed to one of the advertisements. At the beginning of the session subjects received a packet of material and were asked to read the attached instructions and to follow the instructions at their own pace. Subjects were instructed to examine the advertisement as if they were reading it in a local newspaper. After they had done so, they were asked questions about the advertisement, about the brand, and about their attitude toward financial calculators in general.

All responses were measured on 7-point scales. The advertisement's credibility was measured by asking whether the advertisement was not credible/credible and untrustworthy/trustworthy, and by asking how likely is it that the brand has the advertised properties (cheapest, [Only in the $199.95 version.] easy to read display, smallest, lightest, superior memory capabilities, and uses latest technology in screen design). The overall attitude toward the brand was also measured using three items: low quality/high quality, dislike it very much/like it very much, poorly designed/well designed. Next, purchase intention was measured in two ways. First, respondents were asked to indicate the likelihood that they would purchase the brand (very unlikely/very likely). Second, subjects were informed in the instructions that as a reward for their participation, one of them would be randomly chosen to receive either the advertised watch or an equivalent amount of money ($199.95 or $999.95). On the questionnaire they were asked to indicate what they would rather receive (the watch or the money) if they were to be chosen for the reward. [The nature of the study was spelled out in the debrief and the chosen subject received the money.] Finally, subjects' involvement with financial calculators were measured along the following dimensions: important, interesting, relevant, exciting, appealing, fascinating, valuable, involving, and needed (all on 7-point scales with endpoints of the adjective and its antonym).

TABLE 2

MEANS AND PERCENTAGES FOR THE MAJOR DEPENDENT VARIABLES

In addition, a separate pre-test group of 60 subjects (10 in each cell) was asked how informative and relevant they consider the information in the advertisement to be. Secondly, they were asked how well they liked financial wristwatch calculators in general, if they thought that a wristwatch calculator had a small keyboard, what an average price for such a product would be, and, finally, to rank the six attributes of keyboard size, overall size, weight, price, memory management, and screen design from the most to the least important.

RESULTS

To simplify the analysis, all multiple-item measures (advertisement credibility, overall attitude toward the brand, and involvement) were averaged, and the resulting composites were used in all subsequent analyses. This is justified because reliability analyses showed that all multiple-item measures achieved high levels of internal consistency, as indicated by coefficient alphas greater than 0.75.

Table 2 presents the means (percentages where appropriate) and standard deviations for all the important dependent variables. The dependent variables were subjected to a two-way ANOVA with Price and Version as independent variables.

Manipulation Checks

Analysis revealed that there is a significant interaction effect between price and version on subjects' involvement levels (F(2, 212)=10.25, p<0.01). As can be seen in Table 2, subjects expressed higher involvement in version OS in the low price, whereas the highest involvement was expressed in version TSf in the high price.

The causes of this interaction and the reason why the involvement level is not the same across all conditions remains unclear. Nevertheless, this pattern does not repeat itself in any of the other manipulation checks and does not correspond to any of the major findings.

As predicted (in the auxiliary to the main hypotheses), the TSb version was considered to be the most informative and relevant of the three versions at both price levels, TSf was next, and the OS version was considered the least informative and relevant (see Table 2). The 2-way ANOVA revealed an overall main effect for Version (F(2, 212)=5.63, p<0.01) and no Version by Price interaction. A further one-way analysis within Price level revealed a significant Version effect (F(2, 106)=4.60, p<0.01; F(2, 105)=7.23, p<0.01, for low and high price respectively). A Scheffe's multiple-comparison procedure indicated that all means were significantly different from each other at each Price level (except TSb and TSf in the high price).

Advertisement Credibility

Version significantly affects the credibility of the advertisement (F(2, 212)=10.99, p<0.001). Both Price and the interaction between Price and Version are not significant (F(1, 212)=1.02, p>0.3; F(2, 212)=0.26, p>0.7, for Price and Price X Version, respectively). As can be seen in Table 2, mean credibility for Version TSb is the highest followed by TSf. OS has the lowest mean credibility of the three versions. A Scheffe's multiple-comparison procedure indicated that in each Price level only OS is significantly less credible than the other two versions supporting H1a and H1b but not H1c, and partially supporting H3.

Brand evaluation

Version and Price interacted to affect brand perception (F(2, 212)=10.78, p<0.001). As can be seen in Table 2, the most favorable attitude toward the brand was expressed in version TSf in the low price condition, and in version OS in the high price condition. A Scheffe's multiple-comparison procedure indicated that in the low price condition, brand's attitude in version TSf is significantly higher than in the other two versions but TSb and OS do not differ significantly from each other. In the high price condition, on the other hand, the brand's attitude in version OS is significantly higher than in the other two versions but TSb and TSf do not differ significantly from each other. These findings mirror the results of purchase intention.

Purchase Intention

Expressed. Version and Price interacted to affect purchase intention (F(2, 212)=2.81, p<0.06). As can be seen in Table 2, the highest intention was expressed in version TSf in the low price condition, and in version OS in the high price condition. There is also a main effect for Price, where in general, as expected, the higher the price the lower the expressed intention to buy. A Scheffe's multiple-comparison procedure failed, however, to detect significant differences between the means in the low price condition. In the high price condition only the difference between OS and TSb is significant. Nevertheless, the behavioral data reported below support H2a-b and H4, that is, the ordinal relationships between all means are in the predicted directions. Also note that the failure to detect significant results in the low price condition can be attributed in part to the much higher standard deviation in the TSf condition (3.24), compared to all other standard deviations in this study. We can not offer any explanation for this fact except to point out that the directional order among the means supports our hypotheses, and that these results are strongly confirmed by the behavioral measure.

Behavioral. Subjects were asked to choose between receiving the brand or an equivalent amount of money. Table 2 shows that, first, the higher the price the lower the preference for the brand (except in the OS version). Second, and more importantly, Price and Version significantly interacted to affect the revealed preferences for the brand. In the low price condition, most subjects (60%) preferred the brand to the money in the TSf version, whereas the money was preferred to the brand in both OS and TSb versions (brand's preference of 30.6% and 12.9% respectively). Further, more subjects preferred the brand in the OS version compared to the TSb version. All three pair-wise comparisons (of proportions) are statistically significant at the 5% level (t=1.73, -2.57, -4.03 for OS vs TSb, OS vs TSf, and TSb vs TSf, respectively). These findings support H2a-c. In the high price condition revealed preferences are different. 37.1% of the subjects in the OS version preferred the brand to the money. This preference declines to 18.5% in the TSf version and further decline to 9.1% in the TSb version. The preference to the brand in the OS version is significantly higher than in the other two versions (t=3.01, 1.60 for OS vs TSb and OS vs TSf, respectively) but the proportions in TSb and TSf do not differ significantly from each other (t=-1.16). These findings partially support H4.

DISCUSSION

The study confirms our hypotheses that the level at which the negative information is provided will significantly affect the effectiveness of two-sided messages. We have shown that the TSf version can attenuate the negative implications of presenting negative information (as compared to the TSb version) and, at the same time, enjoy the benefit of strengthening a message's credibility due to the presentation of the same negative information.

Although subjects in the low price condition evaluated the TSb version to be more informative/relevant than the TSf version, this did not translate into a higher credibility score. [Since TSb and TSf were evaluated the same on the informative/relevant scale in the high price condition, no credibility difference is expected.] Nevertheless, both TSb and TSf versions, in both price levels, were judged more informative/relevant than the OS version, and also more credible. This replicates the typical two-sided message advantage over the one-sided version (Smith and Hunt 1978; Kamins and Assael 1987a).

In both the low and the high price conditions, subjects expressed a lower (or the same) intention to buy after exposure to the TSb version compared to the OS version, illustrating the possible trap of using a two sided message (Kanungo and Johar 1975; Swinyard 1981; Golden and Alpert 1987). The main contribution of our study is that we have shown that, in a low price condition when a tradeoff in quality is tolerated (but only to a degree), it is possible to reduce the cost of using a two-sided message without simultaneously reducing the benefits gained. This can be achieved by providing the negative information at the product form rather than at the brand level.

This study strengthens the applicability of assimilation - contrast theory to the area of message sidedness. The results suggest that the product form a reference point belongs to (i.e. same product form as target brand: TSb, or different product form than target brand: TSf) exerts an influence on how shortcomings are perceived. Even though the negative information provided is the same, a more distant reference point (i.e. TSf) can make shortcomings of a brand appear less severe as consumers infer that the shortcoming may be product form specific rather than brand specific. This effect occurs through lower expectations for the whole product form rather than the product brand resulting in assimilation effects (Kamins and Assael 1987b).

Furthermore, this study takes a novel approach with regard to the use of choice as a surrogate measure for purchase intention. This turned out to be very important in respect to the results of this study. Although the choice in our experiment was not an actual purchase, it seems to be closer to the underlying construct of purchase intention, and was affected to a greater degree than expressed intention by the experimental manipulation.

The practical nature of this study has to be stressed. The type of message used in advertising is under the strategic control of the marketer. Consequently, this study adds one more possibility to the marketer's arsenal. The potential of the TSf are diverse, ranging from applications in the car industry (e.g. comparing a brand of sports sedan to sports cars on acceleration), the fast-food industry (e.g. comparing one hamburger chain with Mexican style chains on diversity of menu items) to the soft-drink industry (comparing a brand of 'energizer/smart drink' to 'coke' style soft drinks on price). The TSf has been shown to be effective in our study only for a low price brand, where the use of comparative tactics is common and quality tradeoffs are accepted. The challenge for future research is to determine if it is possible to reduce the costs of using two-sided messages without simultaneously reducing the benefits gained in the high price range.

Finally, it appears that this study has an answer to Ferguson and Jackson's (1982) inquiry: "Negative information, asset or liability?" Our results suggest that the former might be the case under certain conditions.

REFERENCES

Alden, Dana L. and Crowley, Ayn E. (1995), "Improving the Effectiveness of Condom Advertising: A Research Note," Health Marketing Quarterly, Vol. 12, (4), pp. 25-38.

Allen, Mike (1991), "Meta-Analysis comparing the Persuasiveness of One-Sided and Two-Sided Messages," Western Journal of Speech Communication, Vol. 55, (4), pp. 390-404.

Allen, Mike (1993), "Determining the Persuasiveness of Message Sidedness: A prudent note about utilizing Research Summaries," Western Journal of Communication, Vol. 57, (1), pp. 98-103.

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, Vol. 18, (August), pp. 333-349.

Chebat, Jean-Charles and Picard, Jacques (1985), "The Effects of Price and Message-Sidedness on Confidence in Product and Advertisement with Personal Involvement as a Mediator Variable," International Journal of Research in Marketing, Vol. 2, (2), pp. 129-141.

Cho, Jeon-Keun (1995), "An Examination of the attitudinal Effects of comparative vs. noncomparative Advertising and their Causal Paths in the Context of Message Sidedness and Product Involvement," PhD dissertation, The University of Connecticut.

Crowley, Ayn E. and Hoyer, Wayne D. (1994), "An integrative Framework for understanding Two-Sided Persuasion," Journal of Consumer Research, Vol. 20, (March), pp. 561-574.

Denver, Eileen (1991), "Printing Calculators," Consumer Reports, (August), pp. 527-531.

Etgar, Michael and Goodwin, Stephen A. (1982), "One-Sided versus Two-Sided Comparative Message Appeals for New Brand Introductions," Journal of Consumer Research, Vol. 8, (March), pp. 460-465.

Faison, Edmund W. J. (1961), "Effectiveness of One-Sided and Two-Sided Mass Communication in Advertising," Public Opinion Quarterly, Vol. 25, (Spring), pp. 468-469.

Ferguson, Jeffrey M.and Jackson, Donald W. (1982), "Negative Information Asset or Liability?," An Assessment of Marketing Thought and Practice, Vol. 48, pp. 34-38.

Golden, Linda L.and Alpert, Mark I. (1987), "Comparative Analysis of the Relative Effectiveness of One- and Two-Sided Communication for Contrasting Products," Journal of Advertising, Vol. 16, (1), pp. 18-25, 68.

Hass, R. Glen and Linder, Darwyn E. (1972), "Counterargument Availability and the Effects of Message Structure on Persuasion, Journal of Personality and Social Psychology, Vol. 23, (2), pp. 219-233.

Hastak, Manoj and Park, Jong Wong (1990), "Mediators of Message Sidedness Effects on Cognitive Structure for Involved and Uninvolved Audiences," in Advances in Consumer Research, ed. G. G. Marvin Goldberg, and Richard Pollay, Association for Consumer Research, Provo, UT, pp. 329-336.

Henle, Guy (1982), "Low-cost Calculators," Consumer Reports, (September), pp. 444-448.

Jones, Russell A. and Brehm, Jack W. (1970), "Persuasiveness of One- and Two-Sided Communications as a Function of Awareness there are two Sides," Journal of Experimental Psychology, Vol. 6, (January), pp. 47-56.

Kamins, Michael A. (1989), "Celebrity and Noncelebrity Advertising in a Two-Sided Context," Journal of Advertising Research, Vol. 29, (June/July), pp. 34-42.

Kamins, Michael A. and Assael, Henry (1987a), "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, Vol. 24, (February), pp. 29-39.

Kamins, Michael A, and Assael, Henry (1987b), "Moderating disconfirmation of Expectations through the use of Two-Sided Appeals: A longitudinal Approach," Journal of Economic Psychology, Vol. 8, (2), pp. 237-253

Kamins, Michael A. and Marks, Lawrence J. (1987), "Advertising Puffery: The Impact of using Two-Sided Claims on Product Attitude and Purchase Intention," Journal of Advertising, Vol. 16, (4), pp. 6-15.

Kamins, Michael A. and Marks, Lawrence J. (1988), "An Examination into the Effectiveness of Two-Sided Comparative Price Appeals," Journal of the Academy of Marketing Science, Vol. 16, (2), pp. 64-71.

Kanouse, David E. and Hanson L. Reid, Jr. (1972), "Negativity in Evaluations" in E. E. Jones et al (eds.), Attribution: Perceiving the Causes of Behavior. Morristown, NJ: General Learning Press.

Kanungo, Rabindra N. and Johar, Jotindar S. (1975), "Effects of Slogan and Human Model Characteristics in Product Advertisements," Canadian Journal of Behavioural Science, Vol. 7, (2), pp. 127-138.

McGinnies (1966), "Studies in Persuasion: Reactions of Japanese Students to One-Sided and Two-Sided Communications," The Journal of Social Psychology, Vol. 70, pp. 87-93.

Pechmann, Cornelia (1992), "Predicting when Two-Sided Ads will be more Effective than One-Sided Ads: The Role of Correlational and Correspondent Inferences," Journal of Marketing Research, Vol. 29, (November), pp. 441-453.

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

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

Stayman, Douglas and Hoyer, Wayne D.; Leone, Robert (1987), "Attribute Importance In Discounting Product Features In Advertising," in paper presented at the American Marketing Association Summer Educators' Conference, Toronto, Canada.

Swanson, Lauren A. (1987), "The Persuasive Effect of Volunteering Negative Information in Advertising," International Journal of Advertising, Vol. 6, (3), pp. 237-248.

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

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