Advances in Consumer Research Volume 17, 1990 Pages 237-243
THE INTERACTION OF PERIPHERAL CUES AND MESSAGE ARGUMENTS ON COGNITIVE RESPONSES TO AN ADVERTISEMENT
Judith E, Hennessey, California State University, Northridge
Shirley C. Anderson, California State University, Northridge
Predictions of the Elaboration Likelihood Model (ELM) as proposed by Petty and Cacioppo were tested in an experiment examining subjects' responses to some announcements. The "peripheral" cues of source context (commercial or noncommercial) and endorsement (present or absent) were systematically manipulated along with the "central" cue, argument strength (weak or strong). The effect of involvement with these variables was examined by assessing whether or not subjects had completed a university writing proficiency requirement, which served as the subject of announcements in the experiment. Results were consistent with the ELM prediction that relative use of central and peripheral cues is mediated by involvement. Both high and low involvement subjects use both central and peripheral cues in generating cognitive responses, but were affected by these cues in opposite ways. While the finding for involved subjects has been reported in earlier work and predicted by the ELM, the finding for low involvement subjects is new.
In the literature of communication processing, the Elaboration Likelihood Model of Petty and Cacioppo (1979, 1981, 1987) is a popular explanatory theory (see Stiff, 1986). It posits that communication about a product or service occurs by one of two divergent focuses of intent: (1) central cognitive processing focuses on analysis of product-relevant information such as message content (arguments); alternatively (2) peripheral cognition focuses on the persuasive peripheral cues, (such as message length, the presence or absence of an endorsement and the background context) rather than the message content. This paper presents the results of an empirical test of a recent clarification of the model. Specifically, by controlling peripheral information and argument quality features of a message, the nature of the interactions of these features for involved and uninvolved subjects are examined.
In central cognitive processing the Elaboration-Likelihood is said to be high, and factors favor the communication recipients' motivation and ability to engage in issue-relevant thinking. These thoughts are expected to be enduring and strongly related to the message object. The second route, low elaboration-likelihood (or peripheral cognition), occurs if either motivation or ability to expend cognitive resources is low (e.g., the recipients are distracted or assess the task as unimportant). If the elaboration likelihood is low, the message "recipients adopt a strategy wherein they are more likely to attempt to derive a reasonable' attitude based upon a superficial analysis of the recommendation . . .based upon various cues in the persuasion context (e.g., the N more arguments for a recommendation, the better it must be)"--Cacioppo and Petty, (1981? p.74).
Factors such as involvement have been viewed as affecting people's motivation and ability $ to think carefully about the merits of an argument (i.e., process centrally) (Petty, Kasmer, Haugvedt ; and Cacioppo, 1987). The ELM predicts that low involvement recipients should be more sensitive to peripheral cues than to argument features of the advertisement (Petty and.Cacioppo, 1981; Petty, Kasmer, Haugvedt and Cacioppo, 1987). Petty and Cacioppo (1984) found support for ELM predictions that the thoughts of the highly involved subject will be less influenced by peripheral cues (length of message) than quality of message. In contrast, the uninvolved subjects were found to be more influenced by the peripheral cues than by the message quality manipulations.
Petty, Kasmer, Haugvedt and Cacioppo (1987, p. 235) diagram various source effects as sample peripheral cues effecting peripheral route processors. i Advertisers frequently attempt to bolster the effect of , a communication by strategic control of background i features. Public relations efforts focused on news releases of product information the marketer posits as news or more focused efforts on the purchase of commercial slots during news as opposed to "less serious" programs can be viewed as attempts to control peripheral qualities of the communication. Advertisers also attempt to enhance the copy points (the message itself) through the use of familiar, attractive, sexy, likeable or credible endorsers or spokespeople. Communication models typically isolate such source effects as independent from message effects (such as appeal used, order, number of items, quality of information, number and schedule-of repetitions). The background or 0 peripheral feature enhancements can be particularly | costly. The relative cost to the benefit on the target audience is of strategic interest in marketing. The ELM suggests that when motivation to process is low (product has no personal relevance) the "peripheral" features should dominate persuasive influence. Alternatively, message features such as strength and quality of copy points should be the primary influence when motivation of process is high (product has high personal relevance).
In the present study, the test of the ELM involves manipulation of two peripheral cues (1) endorsement--present or absent, and (2) context-news or advertisement] and one argument feature (argument--weak or strong). From the ELM a two-way interaction between involvement and argument strength is predicted since high involvement subjects are expected to be more sensitive to variations in arguments. Likewise, two-way interactions of involvement with context and involvement with endorsement are also predicted by the ELM.
A predicted interaction of peripheral cue variables, was reported in a study similar to the present one (Hennessey, Anderson and Bessolo, 1986). In addition to main effects on buying intention for an advertisement versus a public announcement context and for a testimonial versus a non testimonial format, these two variables interacted. In a public announcement, a testimonial was perceived as favorable, lack of testimonial was neutral. In the advertisement lack of a testimonial was neutral and presence of testimonial was perceived as negative, which lowered both perceived credibility of the message as well as intention to buy the product. Presumably the logic in the interaction was that in public announcements, endorsers always tell the truth, but in advertisements endorsers may lie. In the present study, a test will be made for this interaction effect. A two-way interaction is predicted between the two peripheral cues such that when both cues are viewed favorably by the subjects (news source and endorsement present) or negatively (advertisement source and no endorsement) the effect on thought will be more positive than when the two cues are inconsistent with one another. Should this two-way interaction be evidenced, then a three-way interaction between involvement and the two peripheral cues is also predicted since, from the ELM the low involvement subjects should be more influenced than high-involvement subjects by variations in the peripheral cues. The ELM, however does not specifically address the nature of interactions between peripheral features and argument features.
Three way interactions between the peripheral cues and argument quality as a function of involvement are predicted. Subjects with a high need or motivation to think (involved) should show the peripheral cues working most when arguments are weak. Thus an endorsement should be more effective (although its effect may take on a central character, e.g. "he should know what he is talking about") when arguments are weak. The effect of both types of variables on high involvement subjects is discussed in light of other studies in Petty, Kasmer, Haugtvedt and Cacioppo (1987). The communication literature has also reported conditions in which strong arguments are not as successful as other approaches (Psychology Today, 1975; Hadley, 1963; Hovland and Mandell, 1952). Thus we propose that low involvement may mediate the effect of arguments, such that when arguments are of high quality (strong) they could take on a peripheral character and in doing so lose their positive quality. Strong arguments, for subjects who do not need or want to think, may be perceived as a "hard sell" tactic resulting in negative effects (particularly pronounced when peripheral cues such as a strong endorsement reinforce this perception).
Since the ELM was proposed in 1979, research findings have generally supported it (Petty, Kasmer, Haugtvedt and Cacioppo, 1987; Lutz 1985; Moore and Hutchinson, 1983; Moore, Hausknecht and Thamoderan, 1986). Lutz in 1985 extended the ELM by proposing that cognitive responses, the measure most frequently used in the ELM supportive literature, should be highly related to attitude variables. Cognitive responses are the thoughts subjects state they had as they viewed the experimental stimulus and the subsequent evaluations of such thoughts. Wright (1973) specified a method for classifying cognitive response into positive and negative categories. The frequency of the dominant cognitive response can be examined as a function of manipulated variables. More recently, Lammers (1985) investigated alternative methods of evaluating cognitive response and found this early method as appropriate as several alternative methods. Attitude, credibility of the message, cognitive response and buying intention were measured in the present study. Buying intention is expected to reflect the tendencies of attitude and cognitive response.
Subjects were 165 student volunteers from upper division undergraduate marketing courses. Subjects were not paid nor given course credit for participating.
Design and Stimuli
Stimuli were printed announcements of a writing proficiency course designed to help students meet a graduation requirement. The peripheral cues were (1) endorsement/no endorsement, and (2) editorial context: public announcement/advertisement. One half of the announcements had a message of four lines (four copy points) presented with an endorsement by the Dean of the School of Business. The remaining half contained the identical message, without the Dean's endorsement. Editorial context was manipulated by telling half of the subjects that the University administration was announcing a private remedial course to help students pass the Writing Proficiency Exam (WPE) required for graduation. The other half of the subjects were told that a fictitious company was advertising a course to effectively prepare students to meet the WPE requirement. The message content (argument strength) was manipulated by changing metric information in the copy (e.g. 90% or 10% of course evaluators are also University WPE evaluators) and emphasis adjectives to either strongly or weakly support the claim that the course would help in meeting the WPE requirement. This simple metric manipulation allowed the researchers to assure that there was no change in the difficulty level of the strong versus weak message. This message content manipulation was pretested and post-tested for perceived strength. Finally, students were classified into a high involvement and a low involvement group depending on whether or not they had yet to meet the WPE requirement. The experimental design was a 2 (expert endorsement versus no endorsement) by 2 (commercial source versus non commercial source) by 2 (weak versus strong argument) by 2 (high versus low involvement) factorial design with credibility, and other attitude dimensions (such as weak-strong; like-dislike) and intention to act measured on semantic differential scales (Osgood, Suci & Tannenbaum, 1957) and with cognitive responses collected as described in Wright (1973).
INVOLVEMENT BY SOURCE INTERACTION ON CREDIBILITY
The manipulation checks indicated that subjects rated the message with weaker metric information (number of course evaluators who are also WPE evaluators, number of weeks to complete course) as a significantly weaker message [p < .05]. Further, subjects who had not yet taken the WPE rated the message as more meaningful and significant than subjects who had already met the requirement [p = .05], thus supporting the successful operationalization of involvement. Credibility judgments were examined as a function of our design to assess the intended effect of our Source Context manipulation. We expected to see source impact Credibility and possibly interact with Involvement such that the credibility judgments of involved subjects would show less of a tendency to reflect the peripheral Source manipulations. Analysis of the variable Credibility revealed a significant interaction between Involvement and Source (F(1,155) = 8.15, p = .005). As can be seen from Figure 1, high involvement subjects judged the ad more credible than the news context while low involvement subjects judged news more credible than the ad context. No other main effects or interactions were significant in the analysis of credibility. Analysis of attitude components revealed that while there was a main effect for argument strength on the potency component (weak/strong), there was a main effect for endorsement on the evaluative component (like/dislike, good/bad, pleasant/unpleasant). Advertisements with the dean's endorsement were generally evaluated more favorably. No other main effects or interactions were evident in the analysis of the attitude components.
As described earlier in this paper, Wright (1973) indicates that the dominant type of cognitive response (negative or positive) is most sensitive to independent variable effects. Positive and negative cognitive responses were classified into categories by two independent judges as responses to cues or arguments. An independent third judge was used to resolve inconsistencies. Negative responses were dominant when thoughts were peripheral cue relevant while neither positive nor negative responses clearly dominated when thoughts were about message arguments. The number of positive and negative cognitive responses were tabulated for each cell in our design. Table 1 provides the results of our analysis on the number of negative cognitive responses to cues and the number of positive responses to arguments. Analysis of positive cognitive responses to message arguments revealed a significant interaction between Involvement and Argument Strength (F( ,147) = 5.21, p = .02) (see Figure 2(a)). Further, a significant interaction between Endorsement and Argument Strength was found in this analysis (F(1,147) = 4.32, p = .04) (see Figure 2(b)). Analysis of the number of negative cognitive responses to message arguments mirrored these effects, although only the Argument Strength by Endorsement interaction was significant in this analysis (F(1,14 7) = 4.61, p = .03). As can be seen in Figure 2, involved subjects were relatively more positive toward strong arguments than toward weak arguments while low involvement subjects were more positive toward weak arguments than strong arguments. Earlier work supports this pattern in high involvement subjects. The pattern shown for uninvolved subjects is new, although consistent with our hypotheses. Contrary to our predictions derived from the Hennessey et al. (1986) study, there was no 3-way interaction between involvement and the two peripheral cues. The data indicate that the Source manipulation was relatively weak compared to the endorsement manipulation. The 2-way interactions of these two peripheral cues were also not found.
ANALYSIS OF DOMINANT COGNITIVE RESPONSE TO MESSAGE ARGUMENTS AND TO PERIPHERAL CUES
Analysis of negative cognitive responses to peripheral cues revealed a marginally significant Involvement-by Endorsement by Argument Strength interaction showing a-differential advantage of no endorsement and no effect of argument strength for involved subjects. Strong arguments without an endorsement and weak argument with endorsement showed a differential advantage for low involvement subjects (F(1,147) = 3.51, p = .06). This finding is illustrated in Figure 3. No other main effects or interactions were evident in this analysis.
Analysis of attitude showed no significant effects for this design. However, analysis of buying intention revealed a significant main effect for Endorsement (F(1,155) = 5.47, p = .02) and a main effect for Involvement (F(1,155 = 11.54, p = .001). Both an endorsement and involvement directly stimulate buying intention. No significant interactions were found amongst the variables on buying intention.
SIGNIFICANT INTERACTIONS IN ANALYSES OF MESSAGE ARGUMENT RESPONSES: INVOLVEMENT
SIGNIFICANT INTERACTIONS IN ANALYSES OF MESSAGE ARGUMENT RESPONSES: ENDORSEMENT
THE INTERACTION OF INVOLVEMENT ENDORSEMENT AND ARGUMENT STRENGTH ON NEGATIVE COGNITIVE RESPONSES TO PERIPHERAL CUES
Consistent with recent clarification of ELM expectation Petty et al (1987), our analysis shows that involved subjects use central and peripheral elements in processing a message. These cues affect both involved and less involved subjects differently. A new finding is that cognitive responses to strong copy were more positive for involved subjects than for uninvolved subjects. Although Petty et al (1987) emphasize the mediating role of Involvement in determining the dominant path to persuasion, they clearly leave open the possibility that Arguments may effect peripheral route processors. They further provide a summary of empirical findings that show some effect of peripheral cues on central route processors. This study is the first, however, to report an effect for argument variations on peripheral route processors (i.e. on less involved subjects).
Further, credibility was enhanced by the advertiser cue in the involved subject but dampened for low involvement subjects, who found a public announcement more credible. By classifying cognitive responses into peripheral cue comments and argument content comments, as opposed to simply examining the relative favorableness of the thoughts generated, insight was provided on why central and peripheral cues seem to affect high and low involvement subjects in opposite ways. For example, while comments on the price of the course tended to address how "cheap" or "expensive" the course was perceived by the involved subject, price comments for less involved subjects were more like "Rip-off' or "Another attempt to get our money", indicating a more superficial use of information than was evident with involved subjects.
Buying intention was, not surprisingly, strongly affected by involvement, as well as by the presence of an endorsement. In light of findings in this study, further research might focus on differences in the way endorsement is perceived and used by both high and low involvement subjects. Further work on the circumstances likely to produce interactions in message features (both cues and arguments) is still needed.
Hadley, H. (1963) 'The Non-Directive Approach in Advertising Appeals," Journal of Applied Psychology 37, 496-498.
Hennessey, J., Anderson, S. and Bessolo, M. (1986) "A Structural Model of Context and Endorsement Effects: Credibility as a Mediator of Attitude and Buying Intention," Proceedings of the American Psychological Association, Div. 23.
Hovland, C. and Mandell, W.(1952) "An Experimental Comparison of Conclusion-Drawing by the Communicator and by the Audience," Journal of Abnormal and Social Psychology, 47, 581-588.
Lammers, H. B. (1985) "A Multitrait-Multimethod Analysis of the Validity of Cognitive Response Assessment Procedures" in Hirschman, E. C., and Holbrook, M. B., (eds.) Advances in Consumer Research XII. Provo, Utah: Association for Consumer Research), 164- 168.
Lutz, R. J.(1985) "Affective and Cognitive antecedents of Attitude toward the Ad: A Conceptual Framework," in Alwitt, L. F. and Mitchell. A. A., (eds.) Psychological Processes and Advertising Effects: Theory, Research and Application, Hillsdale, N.J.: Lawrence Erlbaum Associates, Publishers).
Moore, D. L., Hausknecht, D. and Thamodaran, K.(1986) "Time Compression, Response Opportunity and Persuasion," Journal of Consumer Research, 13 (June) 85-99.
Moore, D. L., and Hutchinson J. W. (1983) 'The Effects of Ad Affect on Advertising Effectiveness," in Bagozzi and Tybout, eds., Advances in Consumer Research 9, 526-530.
Osgood, C. I., Suci, G. J., and Tannenbaum, P. H. (1957) The Measurement of Meaning. (Urbana, IL: University of Illinois Press).
Petty, R. E., Kasmer, J. T., Haugvedt, C. P. and Cacioppo, J. T. (1987) "Source and Message Factors in Persuasion: A Reply to Stiff's Critique of the Elaboration Likelihood Model," Communication Monographs, 54 (September), 233-249.
Petty, R. E. and Cacioppo, J. T.(1984) 'The Effect of Involvement on Response to Argument Quantity and Quality: Central and Peripheral Routes to Persuasion," Journal of Personality and Social Psychology, 46, 69-81.
Petty, R. E., and Cacioppo, J. T. (1981) Attitudes and Persuasion: Classic and Contemporary Approaches. Dubuque: A. C. Brown.
Petty, R. E., and Cacioppo, J. T.(1979) 'The Elaboration Likelihood Model of Persuasion. Advances in Experimental Social Psychology, 19, 31-33.
Psychology Today (1975) "Ads without Answers Make the Brain Itch." 9 (November) 78.
Stiff, J. B.(1986) "Cognitive Processing of Persuasive Message Cues: A Meta-Analytic Review of the Effects of Supporting Information on Attitude," Communication Monographs 53 (December) 75-89.
Wright, P.L.(1973) 'The Cognitive Processes Mediating Acceptance of Advertising," Journal of Marketing Research, 10 (February) 53-62.