Does Humor in Advertising Enhance Systematic Processing?

Stephen M. Smith, Ohio State University
[ to cite ]:
Stephen M. Smith (1993) ,"Does Humor in Advertising Enhance Systematic Processing?", in NA - Advances in Consumer Research Volume 20, eds. Leigh McAlister and Michael L. Rothschild, Provo, UT : Association for Consumer Research, Pages: 155-158.

Advances in Consumer Research Volume 20, 1993      Pages 155-158

DOES HUMOR IN ADVERTISING ENHANCE SYSTEMATIC PROCESSING?

Stephen M. Smith, Ohio State University

We attempted to assess the extent to which the humorousness of an advertisement influences systematic processing of the ad copy. While it is often assumed that humor enhances systematic processing, the present prediction, based on results from research on positive moods and message processing, was that it would disrupt systematic processing. Results supported the latter hypothesis, as subjects who perceived the ad as humorous evidenced less differentiation between strong and weak advertising claims than did subjects perceiving the ad as nonhumorous.

INTRODUCTION

An important question in the field of consumer research is whether humor enhances the persuasiveness of an advertisement relative to nonhumorous advertisements. Thus, it is not surprising that a number of studies have addressed this issue (e.g., Duncan and Nelson 1985; Gelb and Zinkhan 1986). Taken as a whole, this body of research has not been conclusive in establishing the effectiveness of humorous advertisementsCan effect that advertisers apparently assume to exist (cf. Madden and Weinberger 1984).

More recently, efforts have been devoted to understanding the conditions under which humorous ads might be more effective than nonhumorous ads, rather than assuming that the effect is universal. In one such study, Chattopadhyay and Basu (1990) provided evidence consistent with the idea that humorous ads are more effective than nonhumorous ads when subjects' prior brand attitudes are positive, but are less effective than nonhumorous ads when subjects' prior attitudes are negative. However, the authors' interpretation rests on the ostensibly uncontested notion that humorous ads enhance attention to the ad and hence the extent of processing, a notion that was indirectly addressed in a 1973 synthesis by Sternthal and Craig.

Sternthal and Craig (1973) forwarded the somewhat contradictory propositions that humorous messages may attract attention, but may also detrimentally affect comprehension. Subsequent research has indicated that humorous ads are indeed more attention-grabbing than nonhumorous ads (e.g., Madden and Weinberger 1982), but it is certainly conceivable that this enhanced attention would be associated with enhanced processing of advertising claims, leading to more rather than less comprehension. However, this notion has not been directly tested in an advertising paradigmCto our knowledge, argument quality has not been included as a manipulated variable in a humor and advertising experimentCand it seems at least equally plausible that humor could grab attention while disrupting processing. For example, attention could be focused on the humorous part of the ad, and deflected away from the rest of the ad. This would seem especially plausible in situations where exposure is held relatively constantCi.e., when consumers don't get to attend selectively to ads of their choosing, such as is frequently the case in laboratory studies.

Indeed, in studies concluding that humor enhances ad processing, the effects often are either evident only for subjects' amount of ad-related elaboration (Chattopadhyay and Basu 1990), or are reported for the total number of elaborations, with no distinction made between ad- and brand-related elaborations (Lammers et al. 1983). Thus, it may be that humor attracts attention to the ad, but not to the ad claims.

Another possibility is that humor in an ad acts to enhance the consumer's mood, and this mood state influences consumers' ad processing. A number of studies have examined the influence of positive moods on message processing (e.g., Batra and Stayman 1990; Kuykendall and Keating 1990; Mackie and Worth 1989). While disagreement remains as to why the effect occurs, the typical finding emerging from these studies is that positive mood states disrupt systematic message processing. Following the above authors by manipulating claim strength, we hypothesized that:

H1: Participants exposed to nonhumorous ads varying in the strength of ad claims will base their responses to the ad, at least in part, on claim strength. Thus, we anticipate more positive responses to strong versus weak ad claims on these participants':

(a) brand attitudes;

(b) ad attitudes;

(c) perceptions of claim strength;

(d) purchase intentions; and

(e) brand-related elaborations

H2: Participants exposed to humorous versions of the same ads will be less influenced by the strength of ad claims than participants exposed to the nonhumorous versions, on each of the measures (a-e) listed above.

As previous studies have found that brand evaluations can be enhanced by the presence of humor in an ad (e.g., Duncan and Nelson 1985) or by eliciting a positive mood state in the respondent (e.g., Batra and Stayman 1990), we also hypothesized that:

H3: Compared to subjects viewing the nonhumorous versions, subjects exposed to humorous versions of the ad will respond more favorably to the ad on each of the above measures (a-e).

METHOD

Subjects and Design

Sixty-seven undergraduate students enrolled in a marketing course participated for course credit. Subjects were randomly assigned to see either a humorous or nonhumorous advertisement. The copy of the ad consisted of either weak or strong claims, and subjects were randomly assigned on this factor as well. The original design was thus a 2 (Humorousness: High vs. Low) X 2 (Claim Strength: Strong vs. Weak) randomized factorial.

Procedure

Subjects participated in groups of five to twelve each. They were seated in a classroom, and were separated by at least 10' to minimize the likelihood that they would see another subject's materials. The experimenter instructed subjects that the study concerned the effects of personality factors on advertising effectiveness, and that they would be reading and evaluating two print ads, which were contained in a 6-page booklet presented to each subject. Subjects were allowed to proceed at their own pace, but no subject took longer than 20 minutes to complete all of the experimental materials.

The target ad was presented first, and was an advertisement for a fictitious life insurance company that was constructed to look quite similar to ads encountered in Fortune magazine. Humorousness of the ad was manipulated via a cartoon that appeared at the top of the ad. The cartoon depicted a man sitting on a cloud, presumably in heaven. For subjects in the humor condition, a caption indicated that the man was thinking, "Wish I'd brought a magazine." For subjects in the nonhumorous condition, this caption was deleted. The claim strength manipulation was carried out by varying the actual claims presented in the ad copy. Subjects in the weak claims condition read a set of five ad claims that pretest subjects had rated as predominantly weak (e.g., "we've consistently received solid B- ratings from insurance analyst A.M. Best"; "our fiscal management practices have never been questioned by the S.E.C."). [As one reviewer noted, the weak arguments could potentially be seen as additional attempts at humor. However, no effects on perceptions of humor were observed due to claim strength, F = 0.09.] Subjects in the strong claims condition read a set of five ad claims that were judged relatively strong in pretesting (e.g., "we've consistently received A+ ratings from insurance analyst A.M. Best"; "our fiscal management practices exceed industry standards"). [The argument quality pretest sample consisted of 14 undergraduate subjects who saw either the weak or strong arguments, but did not see the cartoon. They reported their attitudes toward the advertised brand on scales identical to those completed by the actual participants, and also completed a thought-listing task. Those reading the strong claims reported greater liking for the advertised brand (M = 6.95) than did subjects reading the weak claims (M = 5.52), F = 7.24, p =.02. The weak claims also elicited a marginally greater number of counterarguments on the thought-listing task (M = 1.14) than did the strong arguments (M = 0.28), F = 3.72, p < .08.]

The primary dependent measures were collected on the next two pages of the booklet. The first three items asked subjects to indicate their attitude toward the "Metro Group" insurance company described in the ad. Each item was presented as a 9-point scale, with the end anchors "bad-good," "not likeable-likeable," and "useless-useful." Subjects then were asked to list whatever thoughts they had in response to the ad, "including thoughts about the insurance company, thoughts about the ad itself, or even totally unrelated thoughts." Twelve lines of space were provided for the thought-listing task, which was not timed.

Next subjects were asked to indicate their attitudes toward the advertisement itself, again using 9-point scales with the end anchors "unpleasant-pleasant," "uninteresting-interesting," "not amusing-amusing" and "not funny-funny." The latter two items constituted our check on the humor manipulation.

The next page asked subjects to rate the strength of the reasons given in the ad for purchasing insurance from the Metro Group, using two 9-point scales anchored by "bad-good" and "weak-strong." Purchase intent was measured by asking subjects to respond to the statement, "If I were looking to buy life insurance right now, I would be interested in speaking to someone from the Metro Group," by completing two 9-point scales anchored by "not likely-very likely" and "impossible-very possible."

All subjects then saw a second print advertisement, for a fictitious automobile, and answered several similar questions regarding their reactions to this ad. The final page solicited demographic information about the subjects, and embedded in this last sheet was an item asking subjects to indicate their general interest in purchasing life insurance on a 9-point scale anchored by "none at all" and "very much." This data was collected with the specific intention of using it as a covariate. The last item in the booklet asked subjects to indicate what they thought was the purpose of the experiment. Inspection of responses to this item indicated that subjects were unaware of the study's purpose.

Unless otherwise noted, the results below are from ANCOVAs using subjects' ratings on the general-interest-in-purchasing-life-insurance measure as the covariate. These ratings were at least marginally related to all of the dependent measures, with p-values ranging from .01 to .18.

RESULTS

Manipulation Checks

Ratings on the 2 claim strength measures were highly correlated (r = .87) and hence were averaged. A 2 (Ad Humorousness) X 2 (Claim Strength) ANOVA on this index yielded a main effect for claim strength, F = 14.14, p < .001, indicating that the manipulation was successful. Subjects reading the set of strong claims thought the claims were stronger (M = 6.98) than did subjects reading the set of weak claims (M = 4.52).

Ratings on the 2 perceived humorousness measures were also highly correlated (r = .89) and hence averaged. An ANOVA on this index, however, indicated no significant effects. Thus, our manipulation of humorousness was unsuccessful. Subjects exposed to the humorous ad did not rate the ad as significantly more humorous (M = 4.31) than did subjects exposed to the nonhumorous ad (M = 3.67), F = 1.18, p = .281.

Since results using the manipulated levels of humorousness as an independent variable were rendered virtually meaningless, they are not reported here. We elected to perform a median split on subjects' ratings of the humorousness of the advertisement. While an analysis of manipulated levels of humorousness would certainly have been preferable, we believed further analysis using the self-report data might shed light on how perceptions of humor in an ad affect persuasion. Indeed, the subjective nature of humor has led some scholars to call explicitly for analyses of self-reported humor in lieu of manipulations of humor (Duncan and Nelson 1985, p.33).

The median rating of humorousness in our sample was 4.0; as this fell below the midpoint, we placed subjects with scores falling on the median in the lower ("nonhumorous") half of the split. The results reported below are based on this median split.

Brand Attitudes

Subjects' responses to the three brand attitude measures were highly intercorrelated (alpha = .93) and were thus averaged to form a single index. A 2 (Rated Ad Humorousness) X 2 (Claim Strength) ANCOVA performed on these scores indicated a main effect for perceived humorousness, F = 13.14, p = .001. Supporting H3a, subjects who perceived the ad as humorous liked the brand better (M = 7.25) than subjects who did not perceive the ad as humorous (M = 5.90). A significant main effect was also obtained for claim strength, F = 12.99, p = .001, as strong claims elicited more positive brand attitudes (M = 7.17) than did weak claims (M = 5.77).

More importantly, a marginally significant interaction between perceived ad humorousness and claim strength, F = 3.68, p = .060, qualified the main effects. Planned comparisons of the means within conditions (see Table 1) indicated that, as predicted in H1a and H2a, subjects low in perceived ad humorousness liked the brand better when the claims were strong than when they were weak, F = 16.59, p < .001, while subjects who perceived the ad as humorous did not differentiate between weak and strong claims, F < 1.

Ad Attitudes

Subjects' responses to the two ad attitude items were highly correlated (r = .77) and hence were averaged. A 2 X 2 ANCOVA on this index indicated a pattern of results similar to that found for brand attitudes. Subjects high in perceived ad humorousness rated the ad more positively (M = 6.84) than did subjects low in perceived ad humorousness (M = 5.09), F = 18.70, p < .001, supporting H3b, and subjects reading the strong claims rated the ad marginally more positively (M = 6.35) than subjects who read the weak claims (M = 5.32), F = 3.70, p = .059.

TABLE 1

CELL MEANS FOR PRIMARY DEPENDENT MEASURES

The interaction between perceived ad humorousness and claim strength was significant, F = 5.89, p =.018, and was again analyzed via planned comparisons. As hypothesized in H1b and H2b, subjects low in perceived ad humorousness rated the ad more positively when the claims were strong than when they were weak, F = 10.87, p < .01, while subjects high in perceived ad humorousness were not influenced by claim strength, F < 1.

Perceptions of Claim Strength

The average of subjects' responses to the 2 claim strength items were also submitted to a 2 X 2 ANCOVA. Results again indicated two main effects, as subjects high in perceived ad humorousness thought the claims were stronger (M = 6.48) than did subjects low in perceived ad humorousness (M = 5.26), F = 8.03, p = .006, providing support for H3c. Also, subjects reading the strong claims rated them as stronger (M = 6.98) than subjects reading the weak claims (M = 4.52), F = 34.65, p < .001.

Again, these effects were qualified by an interaction between perceived ad humorousness and claim strength, F = 10.43, p =.002 (see Table 1 for cell means). Planned comparisons indicated that subjects low in perceived ad humorousness were affected more by the claim strength manipulation, F = 45.52, p < .001, than were subjects high in perceived ad humorousness, F = 1.39, p > .25. These results provided support for H1c and H2c.

Purchase Intent

Responses on the two purchase intent measures were highly correlated (r = .95) and hence were averaged to form a single index. Scores on this index were analyzed in a 2 X 2 ANCOVA; results indicated significant main effects both for perceived humorousness, F = 12.97, p = .001, and for claim strength, F = 11.46, p = .001. Subjects rated their likelihood of purchase as greater when perceived ad humorousness was high (M = 6.65) than when it was low (M = 4.64), supporting H3d. Purchase likelihood was also seen as greater by subjects in the strong claims condition (M = 6.48) than those in the weak claims condition (M = 4.48).

The interaction between perceived ad humorousness and claim strength was marginal at best, F = 2.41, p = .126, but simple effects tests indicated that, as before, subjects low in perceived ad humorousness were influenced by the strength of the ad claims, F = 12.23, p =.001, while subjects high in perceived ad humorousness were not, F < 1. These results provided tentative support for H1d and H2d.

Number of Brand and Ad Elaborations

Subjects' responses on the thought-listing measure were classified as either brand-related, ad-related, or unrelated. Brand-related and ad-related thoughts were then disaggregated according to their valence: either positive (pro-brand or pro-ad), negative (counter-brand or counter-ad), or neutral. [These classifications were performed by two independent judges. The judges agreed on over 80% of the classifications, with disagreements resolved by discussion.] An index of total number of brand elaborations was created by summing each subject's number of positive, negative, and neutral brand-related thoughts. A 2 X 2 ANCOVA on this measure yielded two marginal main effects: subjects low in perceived ad humorousness generated slightly more brand elaborations (M = 2.46) than subjects high in perceived ad humorousness (M = 1.85), F = 3.64, p = .061, and subjects reading weak ad claims generated marginally more brand elaborations (M = 2.53) than subjects reading strong ad claims (M = 1.87), F = 3.63, p = .062.

An index of the total number of ad elaborations was created by summing the positive, negative, and neutral ad-related thoughts generated by each subject. Analysis of this index indicated only a marginal main effect for perceived ad humorousness, F = 2.55, p = .116, with subjects high in perceived ad humorousness generating somewhat more ad elaborations (M = 1.00) than subjects low in perceived ad humorousness (M = 0.57).

Note that the effects of perceived ad humorousness had opposite effects on brand and ad elaborations: subjects high in percieved ad humorousness appeared to think less about the brand, and more about the ad, than did subjects low in perceived ad humorousness.

Positivity of Brand and Ad Elaborations

An index of the net positivity of subjects' brand elaborations was created by subtracting the number of negative brand elaborations from the number of positive brand elaborations. A 2 X 2 ANCOVA on this index indicated a main effect for claim strength, F = 20.82, p < .001, as strong ad claims elicited more positively-toned brand elaborations (M = +.875) than did weak ad claims (M = -.967). Hypothesis 3e was not supported, as the main effect of perceived humorousness was not significant, p > .25.

The main effect for claim strength was qualified by a marginal perceived ad humorousness by claim strength interaction, F = 3.71, p = .059. As can be seen in Table 1, the net positivity of the brand elaborations of subjects low in perceived ad humorousness was strongly influenced by claim strength, simple effects F = 23.27, p < .001. The net brand elaborations of subjects high in perceived ad humorousness were, by contrast, not influenced by claim strength, F = 1.79, p > .15. These results supported H1e and H2e.

An index of the net positivity of subjects' ad elaborations was created in similar fashion, and also analyzed via 2 X 2 ANCOVA. Results indicated only a significant interaction effect, F = 5.44, p = .023 (see Table 1 for means). As was the case for net brand elaborations, subjects low in perceived ad humorousness were significantly influenced by claim strength, F = 6.04, p < .025, while the net ad elaborations of those high in perceived ad humorousness were not influenced by claim strength, F < 1.

DISCUSSION

The primary purpose of the present study was to determine the effects of humorous advertisements on the extent to which ad copy is processed by consumers. Contrary to the popular notion that humorous ads enhance ad processing via their impact on attention, we hypothesized that they would undermine ad processing, with ad exposure held fairly constant. Although our manipulation of ad humorousness was not effective, the results of our internal analyses provided support for our hypotheses.

Specifically, subjects scoring in the upper half on a median split on their self-reports of perceived ad humorousness (i.e., those who perceived the ad as more humorous) were less sensitive to the strength of the ad claims, as gauged by assessments of their brand attitudes, ad attitudes, perceptions of claim strength, purchase intentions, and cognitive responses to the ad, than were subjects who perceived the ad as less humorous. Indeed, the results of our simple effects tests indicated that the perception of humor in the advertisements led subjects to be relatively uninfluenced by the strength of ad claims. And while perceptions of humorousness tended to enhance both ad and brand evaluations, this enhancement only appeared in the case of weak ad claims (i.e., higher perceived ad humorousness did not enhance subjects' evaluations when claim strength was high).

While the results were consistent in supporting our experimental hypotheses, caution is advised in interpreting the present results. The present design included ads for only one product, and hence the results may be limited in their generalizability (e.g., Gardner and Scott 1990). In addition, the advertised service, life insurance, presents a context in which humor is likely to be seen as irrelevant to the claims (e.g., Weinberger and Campbell 1991). For product or service classes where humor is more relevant, it seems plausible that perceived humor may serve as a persuasive argument rather than serving to predict the extent of message processing (see Petty, Gleicher and Baker 1991 for a related discussion of the different functions mood can serve in persuasion contexts).

The life insurance ad is also likely to be low in personal relevance for the majority of our undergraduate sample, few of whom have any dependents. This implies that our effects may be limited to contexts in which the baseline level of message processing is relatively low. Put another way, our effects may be limited to situations characterized by low to moderate elaboration likelihood (see Petty and Cacioppo 1986). Given a highly involving product or service, we would expect all subjects to base their judgments primarily on claim strength.

Subsequent research should clarify the present results by replicating them with an effective manipulation of ad humorousness, and by indicating their generalizability to different product classes and different advertising media.

REFERENCES

Batra, Rajeev and Douglas M. Stayman (1990). "The Role of Mood in Advertising Effectiveness," Journal of Consumer Research, 17, 203-214.

Chattopadhyay, Amitava and Kunal Basu (1990). "Humor in Advertising: The Moderating Role of Prior Brand Evaluation," Journal of Marketing Research, 27, 466-76.

Duncan, Calvin P. and James E. Nelson (1985). "Effects of Humor in a Radio Advertising Experiment," Journal of Advertising, 14(2), 33-40.

Gardner, Meryl P. and John Scott (1990). "Product Type: A Neglected Moderator of the Effects of Mood," Advances in Consumer Research, 17, 585-89.

Gelb, Betsy D. and George M. Zinkhan (1986). "Humor and Advertising Effectiveness After Repeated Exposures to a Radio Commercial," Journal of Advertising, 15(2), 15-20.

Kuykendall, David and John P. Keating (1990). "Mood and Persuasion: Evidence for the Differential Impact of Positive and Negative States," Psychology and Marketing, 7(1), 1-9.

Lammers, H. Bruce, Laura Leibowitz, George E. Seymour, and Judith E. Hennessey (1983). "Humor and Cognitive Responses to Advertising Stimuli: A Trace Consolidation Approach," Journal of Business Research, 11, 173-85.

Mackie, Diane M. and Leila T. Worth (1989). "Processing Deficits and the Mediation of Positive Affect in Persuasion," Journal of Personality and Social Psychology, 57(1), 27-40.

Madden, Thomas J. and Marc C. Weinberger (1982). "The Effects of Humor on Attention in Magazine Advertising," Journal of Advertising, 11(3), 8-14.

Madden, Thomas J. and Marc C. Weinberger (1984). "Humor in Advertising: A Practitioner's View," Journal of Advertising Research, 24(4), 23-29.

Petty, Richard E. and John T. Cacioppo (1986). Communication and Persuasion. New York: Springer-Verlag.

Petty, Richard E., Faith Gleicher, and Sara M. Baker (1991). "Multiple Roles for Affect in Persuasion," in Emotion and Social Judgments, ed. Joseph P. Forgas, Elmsford, NY: Pergamon, 181-200.

Sternthal, Brian and C. Samuel Craig (1973). "Humor in Advertising," Journal of Marketing, 37, 12-18.

Weinberger, Marc G. and Leland Campbell (1991). "The Use and Impact of Humor in Radio Advertising," Journal of Advertising Research, 30(6), 44-52.

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