The Effects of Message Modality on One- and Two-Sided Advertising Messages

George E. Belch, San Diego State University
ABSTRACT - A laboratory experiment is used to examine the effects of one- and two-sided advertising messages presented through either a print or television medium. Measures of advertiser credibility, cognitive response, and message acceptance are used to examine the impact of message sidedness and medium. The results indicate some negative effects resulting from the use of a two-sided message, although no significant interactions with modality were found.
[ to cite ]:
George E. Belch (1983) ,"The Effects of Message Modality on One- and Two-Sided Advertising Messages", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 21-26.

Advances in Consumer Research Volume 10, 1983      Pages 21-26

THE EFFECTS OF MESSAGE MODALITY ON ONE- AND TWO-SIDED ADVERTISING MESSAGES

George E. Belch, San Diego State University

ABSTRACT -

A laboratory experiment is used to examine the effects of one- and two-sided advertising messages presented through either a print or television medium. Measures of advertiser credibility, cognitive response, and message acceptance are used to examine the impact of message sidedness and medium. The results indicate some negative effects resulting from the use of a two-sided message, although no significant interactions with modality were found.

INTRODUCTION

Traditionally advertisers have presented only the positive arguments or attributes associated with their product in attempting to influence consumer decision making. However, the merits of using a form of a two-sided message structure, whereby the advertiser recognizes attributes where his product may not be superior, has received considerable attention recently. A number of advertising studies have examined the effects of message sidedness by using varied and nonvaried message claims (Belch 1981; Etgar and Goodwin 1982; Mazis 1976; Settle and Golden 1973; Smith and Hunt 1978; Swinyard 1981). In these studies a one-sided message structure is represented by nonvaried message claims whereby superiority is claimed for the advocate brand on all attributes mentioned. Two-sided message structures are represented by varied appeals whereby the advertiser recognized one or more product attributes where the advocate brand is inferior.

Most of the extant research on advertising message sidedness has been conducted using print advertisements as the message modality. Exceptions are the studies by Mazis (1976) which. used audio messages, and Belch (1981) which used television commercials as message stimuli. None of these studies has examined the impact of message modality on the effectiveness of one- and two-sided messages. As noted by Belch (1981), the effectiveness of one- and two-sided advertising messages may be moderated by the message environment. Factors such as the information processing rate and response opportunity afforded by different message modalities vary and may influence the effectiveness of one- and two-sided messages (cf. Wright. 1980).

The purpose of this study is to examine the effects of message modality on one- and two-sided advertising messages. Cognitive responses along with message acceptance and source credibility measures are used to assess the differential effects of one- and two-sided messages presented in a print versus television modality.

RELEVANT LITERATURE

In considering theoretical explanations for the differential effects of message sidedness, two approaches are examined here. First attention is given to competing explanations for the differing effects of message sidedness offered in the social psychology literature. The second perspective considers the attribution theory approach to explaining the effects of message sidedness which has been taken in the marketing literature

Social Psychology: Theoretical Explanations

Early research by Hovland, Lumsdaine and Sheffield (1949) found no overall difference in the effectiveness of one- versus two-sided messages. Further analysis, however, indicated that a two-sided communication was more effective for those message recipients who were relatively more educated and for those who opposed the position of the persuasive communication. When the audience was initially favorable towards the position advocated, however, the two-sided message was found to be less effective than the one-sided communication.

Several explanations have- been offered for the differing effects of one- and two-sided messages. One explanation explains the differential effects in terms of perception of bias. According, to this position the one-sided message would be seen as more biased by those who are aware of opposing arguments (e.g., those who oppose the position advocated in the message) or those who would generally be aware of arguments on both sides of the issue (e.g., well-educated).

In a study by Chu (1967), subjects initially favorable to the position advocated in a message showed no difference in attitude change as a function of prior familiarity with pro and con arguments or ore-versus two-sided communications. Among subjects who were initially unfavorable, however, the two-sided message produced more agreement with the advocated position in comparison to the one-sided message for those familiar with the opposing and support arguments. Chu concluded that the differential responses to the one- and two-sided messages were due to perceived bias in the communications and not to the rehearsal of counterarguments.

This explanation has been questioned, however, by Jones and Brehm (1970), since the one-sided message was perceived as biased in 311 conditions, yet resulted in message acceptance only among initially unfavorable message recipients who were previously exposed to relevant pro and con arguments. Hass and Linder (1972) have also questioned Chu's perception of bias explanation since "bias" was operationalized as the detection of omission of arguments unfavorable to the communicator's position. Hass and Linder suggest that awareness of arguments opposing the speaker's view is more of a measure of counterargument production than bias. Counterargument production as an explanation for the effect is discussed in more detail below Jones and Brehm (1570) argued for a reactance theory (Brehm 1966) explanation of the effects of one- and two-sided message structures. The logic here is that a one-sided communication will be seen as exerting greater pressure to adopt a particular position than will a two-sided communication. Jones and Brehm, tested the hypothesis that the persuasiveness of a one-sided message would be reduced more than that of a two-sided message when the audience was aware that there were plausible positions on both sides of an issue. According to this position it is not necessary that the message recipient know any specific pro or con arguments or facts on the issue.

Jones and Brehm found the predicted interaction between awareness of plausible positions and message sidedness and concluded that the effectiveness of a one-sided communication is reduced to the extent that the audience 15 aware that there are two plausible sides to the issue. While the results of this study supported the reactance theory prediction, the reactance explanation is based on several untested assumptions concerning the perceived freedom-threatening character of the various message structures.

An explanation based on counterarguing has been suggested by Hass and Linder (1972) who argue that a one-sided message is more effective than a two-sided message for message recipients lacking any counterargument defenses. The two-sided message, by acknowledging valid arguments for the opposition which would be otherwise unavailable to the message recipient, inhibits persuasion by providing counterarguments which might be used to reject the communicator's position.

In a series of experiments, Hass and Linder found support for their counterargument explanation. A two-sided message led to greater acceptance of a communication than a one-sided message when counterarguments were available and less attitude change when counterarguments were not available. They also found that. the sequence of argument presentation was an important factor. When subjects were provided with counterargument defenses on a novel issue, acknowledging counterarguments at the beginning of the message resulted in more persuasion than acknowledging them at the end or not mentioning them at all. However, when subjects were not provided with counterarguments on the issue, more persuasion occurred when opposing arguments were acknowledged at the end of the message than at the beginning.

Hass and Linder also found that when a two-sided message inadequately refutes the counterarguments possessed by a message recipient it may actually strengthen the cognitive defenses of the message recipient and reduce persuasion. This finding runs counter to the earlier theorizing on two-sided message structures (Hovland et al. 1949; Chu 1967) which assumed that the implicit refutation of counterarguments through the acknowledgment of their existence would lead to greater attitude change than would a one-sided message.

Thus, while the general assumption has been that the use of a two-sided message may enhance persuasion through the implicit refutation of counterarguments a message recipient may possess, it appears that the presentation of an inadequate refutation may actually strengthen the recipient's cognitive defenses and inhibit attitude change. Hass and Linder's findings suggest that the availability of counterarguments to a message recipient has an impact on the effectiveness of two-sided message structures.

Marketing Studies: Attributional Explanations

Several studies in the marketing literature have shown positive effects resulting from the use of a two-sided or varied product claim message versus a one-sided or nonvaried product claim approach. Settle and Golden (1973) found that advertising claims which vary over product characteristics resulted in higher confidence rating for each claim and concluded that disclaiming superiority on at least one product attribute is a way of increasing the perceived credibility of the source. However, Settle and Golden's interpretation of their findings, as well as their attribution process explanation, has been criticized by Hansen and Scott (1976) and Burnkrant (1974).

Smith and Hunt (1978) also used the varied/nonvaried product claim paradigm to examine whether attributional inference processes are evoked by advertising messages and whether these attributions mediate perceived credibility of the source. Attributional processing was assessed by having message recipients provide cognitive response measures and by asking them to explain their responses to a direct measure of perceived truthfulness. They found that the varied product claim messages resulted in significantly more correspondent attributions (i.e ., attributions concerning the dispositional properties of the advertiser) and higher perceived truthfulness scores than did the nonvaried claim messages. Smith and Hunt offered an attributional explanation for these findings based on Jones and Davis' (1965) correspondence theory. This explanation suggests that the use of varied product claims moves the message recipient toward a dispositional attribution of truthfulness or honesty which leads to greater source credibility. [Other explanations of how attributional processing might effect message persuasiveness have been offered by Hansen and Scott (1976) and Eagly, Wood and Chaiken (1978).]

Several studies have examined the effects of one- and two-sided messages in the context of comparative and noncomparative advertisements. Using audio messages placed in an actual radio programming environment, Mazis (1976) found no significant differences between one- and two-sided messages on recall, attitude, and purchase intention measures. He did find that more counterarguments were generated for comparative, one-sided messages used by a leading brand than for two-sided, comparative messages used by a less popular brand. Mazis also found a higher level of source derogation for the two-sided message which he suggested was an indication of the subjects responding negatively to the advertiser's disparagement of his own brand.

Swinyard (1981) examined the effects of one- and two-sided, comparative and noncomparative print messages in a controlled field experiment. He found that two-sided product claims resulted in less counterarguing and higher evaluations of advertiser truthfulness than did one-sided claims. However, the two-sided message did not result in greater behavioral intention or more favorable behavior (as measured by coupon redemption) than the one-sided message. Etgar and Goodwin (1982) examined the effects of one- and two-sided comparative print ads and found that the two-sided appeal yielded significantly higher brand attitudes than the more traditional one-sided appeal.

Belch (1981) found no differences between one- and two-sided, comparative and noncomparative television commercials seen over three levels of message exposure (one, three and five times) on cognitive response, message acceptance and advertiser perception measures. In this study the two-sided message manipulation was not successful in the noncomparative one and three exposure conditions. However, in the comparative conditions, where the disclaimer was perceived, the two-sided appeals did not generate more favorable cognitive responses or attitudes and purchase intention than the one-sided message.

Reasons for Inconsistent Findings

The studies reviewed above offer some support for the notion that a two-sided advertising message may enhance the recipient's perceptions of advertiser credibility and may even reduce counterarguing against a message. There is, however, less support that the effectiveness of two- sided messages will carry beyond these mediating variables and result in favorable affective and cognitive responses. Favorable attitudinal responses for two-sided versus one-sided messages were found only in the study by Etgar and Goodwin (1982). [In this study price was used as one of the disclaimed attributes. The mentioning of a higher price might not be Perceived as a disclaimer but rather might be perceived as part of a price-quality appeal (cf. Jacoby, Olson and Haddock, 1971). This could result in more favorable attitudes toward the brand in the two-sided condition.] The mixed results found in these studies may be due to several factors. First, the lack of an effect of the two-sided messages on the affective and conative variables suggests that enhanced credibility does not always result in more favorable perceptions of the product or service. Unless the mediating effect of enhanced credibility is fairly powerful, a significant effect on affect and conation should not be expected. Thus enhanced perceptions of advertiser trust and/or credibility may be necessary but not sufficient in determining the effectiveness of a two-sided versus one-sided advertisement.

It should also be noted that a two-sided message may actually cause message recipients to form different beliefs about the advertised brand than will a one-sided message. These beliefs may in turn mediate attitudinal and behavioral differences. This may have been the case in the Etgar and Goodwin (1982) study as the acknowledgment of a higher price in the two-sided message may have resulted in more favorable beliefs regarding product quality as compared to the one-sided appeal. The perception of higher quality would then mediate the favorable attitudes toward the brand found in the two-sided condition.

When no differences are found due to message sidedness, the two-sided message may have failed to create different beliefs or the beliefs may not mediate message acceptance. This latter condition is likely given the fact that the disclaimed attribute is often trivial or nonsalient and is unlikely to mediate any attitudinal or conative changes.

Finally, the studies which have found favorable effects for two-sided versus one-sided messages have all used a print message modality. The studies by Mazis (1976) and Belch (1981), which have used radio and television as the respective message mediums, found few significant differences in effects for the two types of messages.

The lack of significant findings in these studies may be clue to several factors including the processing limitations imposed by the externally paced audio and television presentations and the message environment. For audio and television commercials presented in a compressed time period, the information processing rate is not under the receiver ' s control . Limitations of the recipient's ability to process the message would make it difficult to engage in the processing procedure that might produce a more favorable impact for the two-sided versus one-sided message. Print may be more effective than audio or television as the medium for a two-sided appeal since a print message affords the recipient greater opportunity to process the message and to dwell on the credibility of the advertiser who admits inferiority on some attributes. Modality differences in information processing were found by Chaiken and Eagly (1976) who found that when a message is difficult to understand, the ability to process self-paced printed information makes it easier to recall than externally paced broadcast information.

It should al so be noted that in both the Belch and Mazis studies, the messages were embedded in television and radio programs respectively. Since the messages were presented in the context of a program, receivers could devote attention to the program or other commercials rather than engage in any in-depth processing for the two-sided message.

This study extends the research on advertising message sidedness by not only examining differences in one- and two-sided messages with respect to a number of communication variables, but by also considering whether the effects of message sidedness are influenced by message modality.

METHOD

Study Design

The data for this study were collected as part of a laboratory experiment examining the effects of message structure and message modality on comparative and noncomparative messages. A 2 x 2 x 2 factorial between subjects design was used with type of message (comparative or noncomparative), message sidedness tone or two-sided), and message modality as the factors. [Three-way ANOVAs performed for each of the dependent measures revealed that the only variables affected by the comparative/noncomparative factor were source derogations and source bolstering. For these cognitive response measures the interaction involving message-type and message sidedness was significant. The interactions were the result of more source derogations and source bolsters being generated in response to a two-sided message in the noncomparative versus comparative conditions. Since the message-type factor did not have a significant effect on any other measures and is not of interest in this study, the data were analyzed by collapsing across the comparative/noncomparative message factor. This results in a 2 x 2 design with message sidedness and medium as the factors and 50 subjects per cell.] Four print advertisements and four television commercials for a new, fictitious brand of toothpaste were used as the message stimuli for the study. The copy used in the print message was identical to the message presented by the announcer in the television commercial. All four commercials made the same superiority claims for the new brand. The comparative messages made direct comparisons with a leading brand of toothpaste on several attributes whereas the noncomparative messages did not mention any other brand. The two-sided appeals disclaimed superiority on the whitening ability attribute while the one-sided message presented only positive claims. The decision to use whitening ability as the disclaimed attribute was based on pretesting of the importance of various product attributes. The results of this pretest indicated that while all of the toothpaste attributes were of at least some importance to consumers, whitening ability was rated lowest.

Subjects and Procedure

Two hundred students enrolled in business administration classes were used as subjects with 25 students randomly assigned to each cell of the design. Subjects were told they would be participating in a study on consumers' reactions to advertising for a new product. After receiving the instructions and completing several demographic measures, the stimulus message was presented by either showing a videotape of the commercial or having the subjects read the print advertisement. Subjects in the print conditions were instructed not to refer back to the advertisement after they completed reading it. Immediately after the message was read or viewed, the subjects were given the cognitive response instructions and were given two minutes to list their thoughts. They were then asked to complete the dependent measures pertaining to their evaluation of the commercial and product.

Dependent Measures

In addition to cognitive responses, a number of dependent measures were taken, including attitude toward the commercial, attitude toward using the new brand, purchase intentions for the new brand, and measures of advertiser credibility. Attitude toward the commercial was measured on three semantic differential scales ( favorable/unfavorable; interesting/ boring; original/ unoriginal). Attitude toward the brand was measured on two semantic differential scales ( favorable/unfavorable; good/bad) while purchase intention was measured by two semantic differential items (likely/ unlikely; possible/ impossible). The dependent measures used in the analyses were calculated by averaging across the appropriate scales for these measures.

Perceptions of advertiser credibility were assessed by having the subjects indicate their impression of the advertiser based on the advertisement they had just viewed or read. These five-point scales included measures of truthfulness, believability, and credibility.

The cognitive response categories used were counterargument, support argument, source derogation, and curiosity thoughts defined by Wright (1973), as well as the categories of simple affirmations and simple disaffirmations described by Beaber (1975). An additional category, source bolstering, was also used. This categorization is the positive counterpart of source derogation in that the thought is positive in valence and is directed toward the advertiser or the approach taken by the advertiser in presenting the message . The cognitive response protocols were coded by three judges who were given operational definitions of the response categories and were trained in the application of these definitions. Unanimous agreement was achieved among the three judges for 63% of the cognitions. Two of the three judges agreed on another 34% of the cognitions. Thus, 97% of the cognitions could be classified using a modal scoring convention. The remaining responses were classified after some discussion among the judges.

Manipulation Checks

The effective use of a two-sided message requires that the product attribute( s) for which superiority is disclaimed actually be perceived as inferior by message recipients. To determine whether the two-sided manipulations were successful, subjects were asked to indicate their impression of the new brand's whitening ability (the disclaimed attribute). In the comparative treatments, subjects were asked to indicate their impression of the whitening ability of the new brand relative to that of the comparison brand; in the noncomparative treatments, subjects were asked to indicate their impression of the new brand's whitening ability in comparison with that of other brands of toothpaste. A five-point scale ranging from definitely has more whitener to definitely has less whitener was used to measure these perceptions.

The results of the manipulation checks for the various treatments are reported in Table 1, which shows the mean scores on the whitening ability perception scale for each experimental treatment. Higher scores on this measure reflect a more negative impression of the whitening ability of the new brand. To determine whether the two-sided message was effective in conveying a negative impression of the new brand's whitening ability, t- tests were conducted comparing the perceptions of whitening ability in the two-sided and one-sided treatments. Separate comparisons were made in the comparative and noncomparative conditions in both the print and television conditions.

TABLE 1

PERCEPTIONS OF WHITENING ABILITY FOR EXPERIMENTAL TREATMENTS

As Table 1 shows, perceptions of the new brand's whitening ability are more negative for the two-sided messages than for the one-sided appeals in 211 conditions. These differences are all highly significant with the exception of the noncomparative television condition where the difference is in the expected direction but is only marginally significant (p < .10). Overall, the results of the manipulation check indicate that the disclaimer was perceived and the two-sided message manipulation was successful.

RESULTS

To determine whether there are differences in the effectiveness of one- and two-sided messages in the context of print versus television commercial medium, a two-way analysis of variance was performed for each of the dependent measures of interest with message sidedness and medium as the factors. While main effects might be expected, of particular interest are the interaction effects involving message sidedness and medium. The results for the various dependent measures are summarized below.

Advertiser Credibility

The results of the ANOVAs for the three measures of advertiser credibility revealed a marginally significant main effect of medium for the truthfulness variable (F = 3.48, 1/187. d.f., p < .064). This main effect was a result of the television commercial messages being perceived as more truthful than the print messages. The only other effect to approach significance was the interaction of message sidedness and medium (F = 2.87, 1/187 d.f., p < .042) for the credibility measure. This effect was due to the one-sided message being perceived as more credible than the two-sided appeal in the print condition while the two-sided message was perceived as more credible in the television condition.

Attitudes and Purchase Intention

The results of the ANOVA for the attitude toward the ad measure yielded a highly significant main effect for medium (F = 12.08, 1/187 d.f., p < .001), as well as a marginally significant interaction of message sidedness and medium (F = 3.57, 1/187 d.f., p < .06). The main effect was the result of the print message being perceived more favorably than the television commercials, while the interaction stems from the two-sided message being perceived much less favorably than the one-sided message in the television condition versus the print condition.

For the attitude toward the brand measure, there was a marginally significant main effect for medium (F = 3.37, 1/187 d.f., p < .07) as the print message produced more favorable attitudes than the television messages. No significant effects were found for the purchase intention measure.

Cognitive Response Measures

Separate ANOVAs were performed for each of the individual cognitive response measures as well as for three summary indices including total thoughts, total positive thoughts, and total negative thoughts. The only individual cognitive response measure for which significant effects were found was source derogations. The main effect of message sidedness was significant (F = 7.76, 1/187 d.f., p < .01) with more source derogations being generated by recipients of two-sided versus one-sided advertisements.

With respect to the summary, cognitive response indices, a highly significant main effect of medium was found for the total thoughts measure (F - 11.24, 1/187 d.f., p < .001). More cognitive responses were generated by recipients of television messages (x = 3.72) than print messages (x = 3.16). No significant effects were found for the total positive thoughts or total negative thoughts measures.

DISCUSSION

The purpose of this study was to determine whether the relative effectiveness of one- and two-sided advertising messages is influenced by the medium in which the messages are presented. It was suggested that a two-sided message might be more effective than a one-sided message when presented through a print versus television commercial medium. However, the study results show few significant differences in the effectiveness of one- and two-sided messages and no significant interactions of message sidedness with message modality.

Two-sided messages were not found to be significantly better than one-sided messages in enhancing advertiser credibility. This finding is not consistent with the results of other studies which have found that two-sided or varied claim appeals evoke more positive perceptions of advertiser credibility than one-sided or nonvaried claim appeals (Settle and Golden 1974; Smith and Hunt 1978; Swinyard 1981). The absence of a significant difference between one and two-sided messages for the attitude and purchase intention measures is consistent with the findings of other studies (Belch 1981; Mazis 1976; Swinyard 1981) which have failed to find differences in affective and conative measures due to message sidedness. The ineffectiveness of the two-sided message is not surprising as the use of the disclaimer did not show a positive effect on any of the mediating variables of advertiser credibility or cognitive response. The inability of the two-sided message to enhance advertiser credibility suggests that the message recipients did not make the dispositional attributions discussed by Smith and Hunt (1978).

The greater number of source derogations generated by recipients of two-sided versus one-sided messages in this study is consistent with the findings of Mazis (1976). Mazis suggested that this finding may have been a reflection of the subjects' responding negatively to the advertiser's disparagement of his own brand. A similar explanation is viable here as many of the source derogations listed by recipients of two-sided messages questioned the wisdom of using the disclaimer (e.g., "I was turned off when they said it had less whitening ability"). The fact that the disclaimed attribute was perceived as important by the subjects may have made this problem even greater. [As part of the dependent measures, subjects were asked to rate the importance of various toothpaste attributes on a seven-point scale ranging from very important to very unimportant. The mean importance rating for whitening ability was 5.55.] Most studies of message sidedness use an unimportant or trivial attribute as a disclaimer and thus the disparagement may not seem so salient or crucial to the message recipient. This suggests that admitting inferiority on an important product attribute may result in negative perceptions by the message recipient and might overwhelm any Positive credibilitY effects.

Hass and Linder's (1972) counterargument explanation for message sidedness affects is also relevant here. They suggest that a two-sided message is more persuasive than a one-sided message provided that the negative information cites only those unfavorable arguments that are already known to the receiver and is refuted in a compelling manner by the favorable information. The whitening ability disclaimer used in the two-sided messages may have provided the receiver with valid arguments to use against the advertiser rather than inhibit counterarguing. Also, since whitening ability was perceived as an important attribute, the favorable message arguments may not have compensated for the negative beliefs created by the use of the disclaimer.

The absence of a message sidedness by medium interaction, whereby--a two-sided message would be more effective in print versus television, suggests that the message reception environment did not influence the recipients' ability to process the message. The results of the manipulation check did show that the disclaimer was perceived in both the print and message conditions. Also, since the television commercial was not shown in the context of a program, the processing capabilities and response opportunity of the message recipients was probably very similar to that afforded by the print messages. Evidence of this comes from the results indicating that total ideation was higher for message recipients in the television versus print conditions. The lack of a modality effect is consistent with the results of the study of Chaiken and Eagly (1976) who found that media differences in message processing diminished when the message was easilY understood.

The level of cognitive response activity engaged in by recipients of the television messages in this study is quite high.' The average number of cognitive responses per subject was 3.72. This level of ideation is much higher than was found in the study by Belch (1981) where the commercials were embedded in an actual program. However, it should also be noted that the subjects in Belch's study were recruited from church groups while business students were used in the present study. The student group may be more likely to generate a greater number of responses to a commercial.

SUMMARY

A compelling hypothesis in the social psychology and advertising literature is that a two-sided message is more effective than a one-sided message. Presumably the presentation of information contrary to the communicator' s best interest leads to a dispositional attribution of truthfulness and enhances source credibility. This inhibits counterarguing against the communicator and stimulates the processing of favorable message arguments. However, the results of this study show no significant differences in the communication effectiveness of two-sided versus one-sided advertisements in either a print or television medium. While some advertising studies have shown favorable effects of two-sided versus one-sided messages, others have produced results that are consistent with this study.

Future research should be directed toward trying to determine the conditions under which a two-sided or varied product claim appeal may or may not be successful. Several factors have been mentioned which might be considered in future studies such as the mediating effect of enhanced source credibility, differences in beliefs resulting from one- and two-sided messages and the importance of the disclaimed attribute. Research on message sidedness must begin to consider these factors more closely in order to better understand when and why a two-sided advertisement will be more effective than a one-sided appeal.

REFERENCES

Beaber, R.J. (1975), "The General Characteristics of Covert Resistance Mechanisms and Their Relationship to Attitude Change and Speaker Perception," unpublished doctoral dissertation, The University of Southern California.

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. 333-349.

Brehm, J.W. ( 1966), A Theory of Psychological Reactance, New York: Academic Press.

Burnkrant, Robert E. ( 1974), "Attribution in Marketing Research: Problems and Prospects," in M.J. Schlinger (ed.), Advances in Consumer Research, Vol. II, Association for Consumer Research.

Chaiken , S. and Eagly, A.H. ( 1976), "Communication Modality as a Determinant of Message Persuasiveness and Message Comprehensibility," Journal of Personality and Social Psychology, 34, 605-614.

Chu, Goodman ( 1967), "Prior Familiarity, Perceived Bias and One-Sided Versus Two-sided Communications ," Journal of Experimental Social Psychology, 3, 243-254.

Eagly, A. H., Wood, W. and Chaiken, S. ( 1978), "Causal Inferences About Communicators and Their Effect on Opinion Change," Journal of Personality and Social Psychology, 36, 424-435.

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

Hansen, Robert A. and Scott, Carol A. (1976), "Comments on Attribution Theory and Advertiser Credibility," Journal of Marketing Research, 13. 193-197.

Hass, R.G. and Linder, D. (1972), "Counterargument Availability and the Effects of Message Structure on Persuasion," Journal of Personality and Social Psychology, 23, 219-233.

Hovland, C.I., Lumsdaine, A.A. and Sheffield, F.D. (1949), Experimentation on Mass Communication, Princeton, NJ: Princeton University Press.

Jacoby, Jacob J., 01 son, Jerry C. and Haddock, R.A. (1971), "Price, Brand Name and Product Composition Characteristics as Determinants of Perceived Quality," Journal of Applied Psychology, 55, 570-579.

Jones, Edward E. and Davis, Keith E. (1965), "From Acts to Dispositions," Advances in Experimental Psychology (Vol. 2), L. Berkowitz ( ed . ), New York: Academic Press, 219-266.

Jones, R.A. and Brehm, J.W. (1970), "Persuasiveness of One- and Two-Sided Communications as a Function of Awareness There are Two-Sides," Journal of Experimental Social Psychology, 6, 47-56.

Mazis, Michael B. (1976), "A Theoretical and Empirical Examination of Comparative Advertising," Faculty Working Paper, College of Business Administration, University of Florida.

Settle, Robert 8. and Golden, Linda L. (1974), "Attribution Theory and Advertiser Credibility," Journal of Marketing Research, 11 (May), 181-5.

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

Swinyard, William R. (1981), "The Interaction Between Comparative Advertising and Copy Claim Variation," Journal of Marketing, 18, 187-191.

Wright, Peter L. (1973), "The Cognitive Processes Mediating Acceptance of Advertising," Journal of Marketing Research, 10, 53-62.

Wright, Peter L. (1980), "Message-Evoked Thoughts: Persuasion Research Using Thought Verbalizations," Journal of Consumer Research, 7, 151-175.

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