Advances in Consumer Research Volume 18, 1991 Pages 813-818
HUMOR IN TELEVISION ADVERTISING: THE EFFECTS OF REPETITION AND SOCIAL SETTING
Yong Zhang, University of Houston
George M. Zinkhan, University of Houston
This study investigates the effect of humor in advertising on three dependent measures of advertising effectiveness: perceived humor, attitude toward the brand, and ad recall. Also examined are the effects of multiple exposures and the effects of social setting (size of the audience). Humor is found to influence consumers' brand attitude and their brand information recall. Perceived humor appears to be affected by social setting, but unaffected by another mediating factor: frequency of exposure.
Humor has been used extensively in consumer product advertising on TV, radio, and in print media. Estimates of its usage range from 15% to over 40% (Kelly and Solomon 1975; Markiewicz 1974). Implicit in such practice is the rooted belief that humor produces desirable effects in persuading consumers to adopt products. Understandably, a considerable amount of effort has been spent on investigating the relationship between humor and a diverse array of response variables.
Among these variables are the humor perceived by the consumer when viewing or hearing the ads, brand attitude, and ad recall. It has been proposed that the effect of humorous ads passes beyond temporary amusement and influences message recall (Gelb and Zinkhan 1986). Eventually, as a result of this process, consumers in the target audience form positive brand attitudes toward the product. These hypothesized humor effects may be mitigated by a variety of other factors such as the number of exposures which the target audience experiences and the social context in which the humorous message is viewed or heard. In this study, perceived humor, recall and brand attitude constitute the three dependent variables of interest. These variables are expected to be related. The two key independent variables are repetition and social setting. The major research question is: to what extent do repetition and social setting influence perceived humor, memory, and attitude?
HUMOR AND REPETITION
Several authors have studied the effect of repetition and humor in advertising. Belch (1982) found that consumers' cognitive responses follow separate patterns depending on the number of repetitions. Positive responses do not decline over repetitions as has been predicted by the two-factor theoretical repetition effect model (Berlyne 1970). According to this model, the effect of repetition is determined by two opposing psychological factors: positive habituation and tedium. As the number of exposures increases, the domination of the first factor declines and that of the second factor increases, causing the deterioration of the ad's persuasive power. Further, Belch and Belch (1984) demonstrated in a later study that unaided recall and intention to use Federal Express, designed as a measure of persuasion, did not reveal any significant effect of humor, number of exposures, and the interactions between the two. These results however are contradicted by another study (Gelb and Zinkhan 1985) which found that humor can be used to improve commercial effectiveness and that perceived humor significantly declines as the number of repetitions increases.
THE INFLUENCE OF SOCIAL SETTING
A second mitigating factor of humor effect is social setting. --Humor and the intimately related result it elicits, laughter, have long been considered to be a social phenomenon. They are expected to occur within patterns of social interactions, and are regulated by society in much the same way as other social-physiological reactions such as yawning (Coser 1959). Humor and laughter are often shared, and they are defined as part of the interactive process of social life. In this regard, laughter is often thought to be contagious; but most humor-related consumer behavior studies have ignored this social dimension to humor.
Repetition causes wear-out of humor. The wear-out of humor refers to the phenomenon that humorous ads lose their humor after a number of reruns. It indeed may cause adverse responses, such as irritation, on the part of the viewer or listener. According to a proposed theory, this wear-out effect can be mitigated by social setting. Zinkhan and Gelb (1990) argue that humor has a social dimension and that if the audience listening to or watching a commercial consists of more than one person, then the social setting dimension increases the likelihood that a message will be perceived as humorous, even after repetition. Their proposition is based on the findings of several studies in this area (see Butcher and Whissell 1984; Aiello et al. 1983; Brown et al. 1982).
HUMOR AND ADVERTISING EFFECTIVENESS
While few researchers argue about the attention-grabbing quality of humorous ads, whether or not humorous ads are a more effective means of persuasion than serious ads remains the focal point of argument. Scott et al. (1990) argue for the view that, to be effective, commercials should: a) contribute to the main point of the message and b) pertain to an advertised product which is appropriate for the use of levity. They showed that such promotional messages can have a positive effect both on enjoyment and behavior (i.e., attendance of social events). Osterhouse and Brock (1970) demonstrated that humor distracts an audience during a persuasive communication. Such distraction leads to decreased levels of counterargumentation and, therefore, to an increase in message reception and persuasion.
Duncan and Nelson (1985) and Madden and Weinberger (1984), among others, concede that humor increases commercial effectiveness by drawing more attention from consumers. But they think it is less appropriate in gaining higher levels of comprehension and persuasion. Their results appear to support the view that despite the fact that humor improves attention, it may inhibit message comprehension and reception. Based on the same argument, it has been suggested that attitude as a measure of persuasion be included in experiments studying humor effects (Sternthal and Craig 1973).
This study intends to further investigate the plausible relationships between humor and measures of advertising effectiveness operationalized here as message recall and brand attitude. The above discussion suggests the following three hypotheses:
H1: Humorous message has a positive effect on message recall, attitude towards the advertised brand, and therefore, the effectiveness of the ad.
H2: Multiple exposures of TV audiences to the same humorous message diminishes the perceived humor of ads, and therefore, advertising effectiveness.
H3: Social setting influences the degree of perceived humor. That is, as the number of people in the audience increases perceived humor increases.
An experiment was conducted using 216 student subjects, all of whom were undergraduate business majors in a large urban university. Just under 40% of the subjects were females. The stimulus TV ads were for soft drinks and had been previously ailed on network television. As a guise, subjects were told that they would be watching some music videos and would be later asked to indicate their musical preferences.
Three variables were manipulated in the experiment: humor in the ads (Message), number of ad exposures (Repeat), and the size of the audience exposed to the ads (Size). Humor in the ads had two levels: an ad that contained humor and an ad that had no humor. The ads were pretested to ensure the effectiveness of the humor manipulation.
Half of the subjects were exposed to the humorous ad, the remaining subjects were exposed to the non-humorous ad. Both versions of the ad contained similar information about the product, but differed concerning the inclusion of the humor stimulus. The commercials were imbedded in 30 minutes of prerecorded music videos. The stimulus video programs contained three different numbers of exposures of the ads. The numbers of ad exposures varied from one to three to five repetitions. The placement of the ads were approximately at equal intervals at breaks between music excerpts in an effort to minimize potential confounding due to unequal spacing. The size of the audience had three levels: one-person audience, three-person audience, and six-person audience. The experimental surroundings where the subjects received the stimuli for the different conditions were made identical.
Thus, this study had a factorial design with all the factors completely crossed. With this design, it was possible to investigate the main effect of the factors and the interaction effects among them. The study had a balanced design with twelve subjects in each condition. All subjects were assigned to the conditions in a randomized fashion.
After the subjects were exposed to the commercials, they were requested to complete a questionnaire containing the three dependent measures: brand attitude (Attitude), perceived humor (Humor), and ad recall (Recall). Both perceived humor and brand attitude were measured using six-point Likert-type scales anchored by the words "strongly disagree" and "strongly agree". The third scale (Recall) was a composite score which was calculated by summing the--number of correct answers to ten questions about information contained in the ad. The minimum possible score was zero while the maximum score was ten on this true-false recall test.
The three dependent variables represent three components in a much investigated theoretical framework, which specifies that high levels of perceived humor lead to better recall, which in turn contributes to the formation of positive attitude towards the advertised brand (Gelb and Zinkhan 1986). This kind of hierarchical model is amenable to a multivariate approach for the analysis of the experimental data since the dependent variables are expected to be related in the sense that they may be regarded as different facets of the overall measure of the "effectiveness" of the ads. Therefore, MANOVA technique was used to assess the humor effect in this analysis.
MANOVA has advantages over ANOVA with a series of individual F-tests in this situation (Hair et al. 1987). The individual F-tests in ANOVA result in an inflated Type One Error which may cause false positive, and render significant tests spurious and irreplicatable (Haase and Ellis 1987). MANOVA provides a single overall test of group differences at a specified alpha level, thus avoiding the multiple F-tests and the resulting Type One Error inflation problem. MANOVA also enables us to test the linear combination of the dependent variables that provides the strongest evidence of overall group differences. Such evidence is not provided by the univariate tests.
The MANOVA results are summarized in Table 1. All the interaction effects are insignificant at 0.05 significance level. This result helps to reduce the uncertainty associated with the interpretation of the main effect results. The first hypothesis concerns the effect of humor on brand attitude and recall. This hypothesis is supported by the Message main effect results (F = 19.78, p < 0.0001).
MANOVA RESULTS OF THE MAIN EFFECTS AND THE INTERACTION EFFECTS
MEAN DIFFERENCE ON PERCEIVED HUMOR FOR EACH LEVEL OF REPETITION
The second hypothesis was not supported by the MANOVA main effect results (F = 1.63 for Repeat, n.s.). This finding is echoed by a plot of the mean differences in perceived humor over the treatment. As illustrated in Figure 1, the mean perceived humor scores (3.82, 3.49, 3.74) do not vary much across the repetition levels.
Another MANOVA was performed with respect to the social and repetition factors on subjects' responses to the humorous treatment. The results are presented in Table 2. Audience size had a significant influence on the dependent variables. Univariate follow-up indicated that the source of the variation resulted from the mean differences of perceived humor over the audience size levels (see Figure 2). When the audience consisted of more than one person, mean scores of perceived humor considerably increased. Therefore, the third hypothesis that social setting positively influences perceived humor was supported by the results of this study. As is shown in Table 2, the repetition effect was again not significant.
The significant omnibus MANOVA result on the Message factor (see Table 1) shows that there is an overall group difference caused by the two levels of the Message factor across the three dependent measures. It would be of interest to find out which measures represent the difference. Therefore, three post hoc tests were subsequently carried out using univariate factorial ANOVA. Attitude, Humor and Recall were used separately as the dependent variable in the ANOVA procedures, and the results are presented in Table 3. Again, significant results are obtained for Attitude and Recall. There are significant mean differences over the levels Of Message. Thus, the hypothesis that humorous ads have a positive effect on message recall and brand attitude is fully supported by this study.
TWO-WAY MANOVA RESULTS ON HUMOROUR AD
PERCEIVED HUMOR SCORES FOR THE HUMOROUS AD ACROSS EACH LEVEL OF AUDIENCE SIZE
POST HOC 3-WAY FACTORIAL ANOVA TEST RESULTS FOLLOWING A SIGNIFICANT MANOVA ON THE FACTOR MESSAGE
The univariate results on the dependent measure Humor is also significant with regard to Message, further confirming the existence of the main effect of the humor manipulation. One interesting finding is that the interaction effect of Message and Repeat on the dependent variable Humor is significant at p < 0.05. This suggests that the effect of ad type (Message) is not the same across the levels of repetition (Repeat). Given the fact that the repetition main effect is non-significant over all dependent variables, as is shown by the MANOVA results, one might suspect that this interaction is caused by the strong effect of Message factor. Another speculation is that the omnibus test by MANOVA did not provide enough power to discern the interaction in the multivariate analysis. MANOVA has been shown to be extremely conservative (Hummel and Sligo 1971). Although the MANOVA tests failed to support the hypothesized effect of repetition, this significant interaction points the way for a follow-up study of the effect of ad type across levels of repetition --one of the enduring themes of research in this area.
This study is supportive of a theoretical model which specifies that humorous ads tend to produce higher levels of perceived humor, positive brand attitude, and brand information recall. The results of this study differ from findings by Belch and Belch (1984) who found humor did not affect recall levels. But similar results are obtained regarding the repetition effect on perceived humor and brand attitude. Based on this sample, humorous ads have more persuasive power than serious ads possibly through a third mediating process such as the reduction of counterargumentation. And humor is perceived to be funnier when someone else is present. The findings here also indicate that repetition does not influence perceived humor and the overall effectiveness measure of advertising. This is inconsistent with the theoretical propositions and empirical results of several studies. As is suggested by the significant Message and Repeat interaction in the univariate procedure, the pattern of wear-out of humor as a function of repeated exposure clearly deserves further investigation.
Several cautions should be heeded when interpreting the results of this study. First, the contrived experimental settings may not have fully induced the social effect of humor. For example, a more realistic and interactive audience composed of friends or family members may produce more marked results. This may be an important area for future research, as social effects on humor may vary depending upon the composition of the audience watching or hearing the advertising message. When all of the members of an audience are friends, or family members, messages may be perceived as more humorous than in instances where a group of relative strangers are exposed to the same message.
Second, the strong Message effect which we observed might be dampened if a less homogeneous sample were tested. Education level of the subjects is a factor which may influence perceived humor, and it has been shown that more highly educated people are more sensitive to humor (Brooker 1981).
Third, repetition typically refers to the frequency of viewing a message over a period of time (weeks, months). Here, we completed all repetitions within a 30 minute period. It is also quite common for audience members to be exposed to the same commercial several times during a short period of time (e.g. one hour viewing session) in the natural environment. However, some of the theoretical work on repetition has specified that effects over time (e.g., forgetting) may play an important role in determining commercial effectiveness particularly over a longer period of time. Repetition effect of humor over longer periods of time should also be studied.
Beyond these three caveats, there may be other reasons for concerns. For example, there may be other (unmeasured) variables which could play important moderating roles. Involvement may be one such variable. Under conditions of high involvement, the presence of humor may lead to reduced comprehension; and this effect may be reversed under low involvement conditions. Alternatively, there may be important, non-linear effects over both social and repetition conditions which should be investigated in later studies.
Given these cautions, this study offers another piece of empirical evidence in the quest for a better understanding of humor in advertising. The results suggest that humor increases ad information recall and positively affects brand attitude. This process may be positively moderated by social setting. Due to its potential usefulness, humor and its role in commercial advertising will continue to interest researchers and practitioners alike. The social aspect of humor appreciation deserves further investigation. Those conducting future research may wish to focus on alternative social factors, beyond size of the audience. Despite the fact that social effects have been widely studied concerning other aspects of buyer behavior (e.g., attitude formation), it is only rarely that we have considered laughter and humor to be inextricably linked to patterns of social exchange. Without a social context, there is little grounds for judging a potentially humorous message as funny or not.
Aiello, John R., Donna E. Thompson, and D. M. Brodzinsky (1983), "How Funny is Crowding Anyway? Effects of Room Size, Group Size, and the Introduction of Humor," Basic and Applied Social Psychology, 4 (No. 2), 193-207.
Appel, Valentine (1971), "On Advertising Wear Out," Journal of Advertising Research, 11, 11-13.
Belch, E. (1982), "The Effects of Television Commercial Repetition on Cognitive Response and Message Acceptance," Journal of Consumer Research, 9 (June), 56-65.
Belch, George E. and M. A. Belch (1984), "An Investigation of the Effects of Repetition on Cognitive and Affective Reactions to Humorous and Serious TV Commercials," Advances in Consumer Research, 11, 4-10.
Berlyne, D. E. (1970), "Novelty, Complexity, and Hedonic Value," Perception and Psychophysics, 8, 279-286.
Brooker, George (1981), "A Comparison of the Persuasive Effects of Mild Humor and Mild Fear Appeals," Journal of Advertising, 10, 2940.
Brown, Gary E., Paul A. Dixon, and D. Hudson (1982), "Effect of Peer Pressure on Imitation of Humor Response in College Students," Psychological Reports, 51, 1111-1117.
Butcher, Jennifer and Cynthia Whissell (1984), "Laughter as a Function of Audience Size, Sex of the Audience and Segments of the Short Film 'Duck Soup'," Perceptual and Motor Skills, 59, 949-950.
Calder, Bobby J. and Brian Sternthal (1980), "Television Commercial Wearout: An Information Processing View," Journal of Marketing Research, 17 (May), 173-186.
Chattopadhyay, Amitava and Junal Basu (1989), "Prior Brand Evaluation as a Moderator of the Effects of Humor in Advertising," Paper under reviewing.
Coser, Rose L. (1959), "Some Social Functions of Laughter," Human Relations, 12, 171-182.
Craig, C. Samuel, B. Sternthal and C. Leavitt (1976), "Advertising Wearout: An Experimental Analysis," Journal of Marketing Research, 13 (November), 365-372.
Duncan, Calvin P., and Janes E. Nelson (19853, "Effects of Humor in a Radio Advertising Experiment," Journal of Advertising, 14, 3340, 64.
Gelb, Betsy D. and George M. Zinkhan (1986), "Humor and Advertising Effectiveness After Repeated Exposures to a Radio Commercial," Journal of Advertising, 15 (No. 2), 15-20, 34.
Hair, Jr., J. F., Rolph E. Anderson and Ronald L. Tatham (1987). Multivariate Data Analysis, 2nd ed., Macmillan Publishing Company, New York.
Haase, Richard F. and Michael V. Ellis (1987), "Multivariate Analysis of Variance," Journal of Counseling Psychology, 34 (No. 4), 404413.
Hummel, Thomas and Joseph Sligo (1971), "Empirical Comparison of Univariate and Multivariate Analysis of Variance Procedures," Psychological Bulletin, 76 (No. 1), 49-57.
Kelly, J. P. and P. J. Solomon (1975), "Humor in Television Advertising," Journal of Advertising, 4, 33-35.
Madden, Thomas J.and Marc G. Weinberger, (1984), "Humor in Adverting: A Practitioner View," Journal of Advertising Research, 24, 23-29.
Osterhouse, Robert and Timothy Brock (1970), "Distraction Increases Yielding to Propaganda by Inhibiting Counterarguing," Journal of Personality and Social Psychology, 15 (August), 344-358.
Scott, Cliff, David M. Klein, and Jennings Bryant (1990), "Consumer Response to Humor in Advertising: A Series of Field Studies Using Behavioral Observation," Journal of Consumer Research, 16, 498-501.
Sternthal, Brian and Sanmuel Craig (1973),"Humor in Advertising," Journal of Marketing, 37, 12-18.
Zinkhan, George M. and Betsy D. Gelb (1985), "The Effect of Repetition on Humor in a Radio Advertising Study," Journal of Advertising, 14, 13-20, 68.
Zinkhan, George M. and Betsy D. Gelb (1990), "Humor, Repetition, and Advertising Effectiveness," Advances in Consumers Research Vol. 17, 438-441.