When does humor enhance or inhibit ad responses? The moderating role of the need for humor.

Research shows that humor may effect desirable responses by attracting attention, making ads more likeable, and perhaps more memorable (Rossiter and Percy 1997). Nevertheless, research has also shown that humorous ads do not always elicit more positive brand attitudes and purchase intentions

(Weinberger and Gulas 1992). In fact, studies have shown humor to have positive effects (e.g., Duncan and Nelson 1985), mixed effects (e.g., Chattopadhyay and Basu 1990), and no effect (e.g., Wu, Crocker, and Rogers 1989) on ad responses at different levels of the hierarchy of communication effects. Sutherland and Sylvester (2000) assert that humor remains one of the least understood elements in advertising.

Few would dispute the fact that humor can be an effective advertising tool; however, the contingencies that determine when it is effective are just beginning to come into clearer relief. Recent work by Alden, Mukherjee, and Hoyer (2000) examined some of the generative mechanisms underlying humor's influence on attitudes. They find that surprise, as a generator of humor, may not be its own virtue. Specifically, humor elicitation is greater when surprise is accompanied by playfulness, warmth, and relatively easy humor resolution. The impact of humor in ads may also depend on the type of humor employed (e.g., Weinberger and Campbell 1991), the type of product being advertised (e.g., Alden and Hoyer 1993), and the relation of the humor to the product or message (e.g., Spotts, Weinberger, and Parsons 1997). It is important to note, as suggested by Lewin's 1951 "field theory," that the impact of humor is likely to depend on the interaction of ad characteristics with characteristics of the individual processing the ad. This approach suggests that an individual's understanding of the environment is within the totality of inner and outer forces, which give meaning and significance to stimuli such as advertising.

The purpose of the present study is to examine the role of two person traits--"need for humor" (NFH) and "need for cognition" (NFC)--as contingencies underlying the effects of humor on advertising outcomes, as well as the underlying process driving the contingencies. This work extends prior research by Zhang (1996) and Cline, Machleit, and Kellaris (1999) by examining the interplay of humor in print ads with potentially important person traits.

THEORETICAL BACKGROUND

There are several process mechanisms by which humor may operate to shape consumer responses to ads. Among these are attention, which is a necessary but not sufficient condition for cognitive information processing, and affect, which may have both direct effects and mediated effects through the conscious processing of information.

Humor's Effects on Processing and Outcome Variables

Joe Sedelmaier, a guru of humorous ads, has long touted the claim that the presence of humor effectively increases attention to ads (Dee 1993). Extensive research examining the use of humor in a variety of contexts--including radio, television, and print advertising--has provided support for that position (Stewart and Furse 1986; Weinberger and Campbell 1991; Weinberger and Gulas 1992). Using industry data, Madden and Weinberger (1982) found that humorous magazine ads outperformed nonhumorous ads in terms of generating and holding attention. In addition, Weinberger et al. (1995) found evidence that humor is directly linked to both attention and recognition. Sports, Weinberger, and Parsons (1997) used industry data to show that humor enhanced initial attention, aided brand recall, and held attention for both high-risk/expressive goods (e.g., snack foods and beer) and high-risk/functional goods (e.g., appliances). Duncan and Nelson (1985) also demonstrate humor's positive impact on attention.

In their review of the literature, Weinberger and Gulas (1992) cite Duncan (1979) and Madden (1982) and conclude that humor-relatedness differentially affects attention. In fact, they argue that controlling for the relatedness of humor makes the findings of the experimental studies in advertising unanimous in their support for humor's positive impact on attention. Because increased attention creates opportunities for more extensive information processing, humor in an ad may lead to an increase in the extent of message processing. On the other hand, research has shown that humor can have distracting effects on information processing (Duncan 1979; Sternthal and Craig 1973). It follows that if a consumer focuses on the humor in the ad, he or she may not attend to message arguments at all (Smith 1993). More specifically, Smith suggests that consumers' attitudes toward the ad and the brand can be enhanced by humor, but only in the case of ads characterized by weak (versus strong) arguments. Recently, Cline and Kellaris (1999) demonstrated that, when paired with ancillary humor, weak arguments for a low-involvement consumer good may work better than strong arguments in creating favorable attitudes. Related research (Altsech, Cline, and Kellaris 1999) shows that, given related (versus ancillary) humor, strong arguments generate more favorable attitudes and purchase intentions.

When humor is perceived as appropriate for advertising a particular product or service it can also make the ad more likeable (Rossiter and Percy 1997). According to the distraction hypothesis, a persuasive message that is discrepant with a position strongly held by consumers will be more successful in generating attitude change if the consumer is distracted during the message presentation (Festinger and Maccoby 1964). This suggests that distraction may enhance message persuasiveness by interfering with the audience's subvocal attempts to counterargue against the dissonant information. In an advertising context, affective responses elicited by humor may divert a consumer from counter-arguing against a discrepant message. Reduced counterarguing, in turn, may increase acceptance of a humorous message (e.g., Cline and Kellaris 1999; Duncan 1979; Sternthal and Craig 1973). Recent research by Arias-Bolzmann, Chakraborty, and Mowen (2000) find support for the distraction hypothesis. They show that absurd humor can result in proportionately more positive cognitive responses and hence more positive attitudes than no humor for those negatively predisposed toward the product class.

Humor may also operate by elevating consumers' moods. As Isen's work suggests, events that have even a slightly positive impact on an individual's mood can have a dramatic impact on the way that person acts or thinks (Isen 1984, 1987; Isen, Daubman, and Nowicki 1987; Kahn and Isen 1993). Humor is an ad feature likely to put people in a good mood. Wicker et al. (1981) found a number of emotion-related scales to be significantly correlated with perceived funniness. Olson and Roese (1995) found support for the contention that perceivers use their own reactions to humorous stimuli (e.g., mirth), together with information about the environment, to infer the emotion-eliciting qualities of humor. Thus, if subjects express joy, they are likely to infer funniness. Madden, Allen, and Twible (1988) used radio ads to demonstrate that, given disguised conditions, a humorous ad generates significantly more positive affect than its nonhumorous counterpart. Positive reactions to humor may also "transfer" to the advertised brand through an associative learning mechanism (Madden, Allen, and Twible 1988).

Individual Differences and Boundary Conditions

Irrespective of the mechanisms through which ad characteristics such as humor operate, reactions to ads are unlikely to be a function of the ad alone. It is likely that individual differences play a rote in the processing of humorous ads as well. According to Lewin's (1951) general "field theory," behavioral responses are shaped by the interaction of the stimulus environment and the person. Lewin suggests that behavior is determined by two pairs of factors: (1) the person (his or he: needs, beliefs, etc.) and the situation he or she is in, as well as (2) cognition and motivation, which respectively determine what a person will do and whether he or she will do it. Therefore, a complete understanding of the environment can only be reached by taking into account people's subjective interpretation of what surrounds them--and person traits play an important role in individuals' perceptions of environmental stimuli. By extension to the present context, responses to ads should be a joint function of ad characteristics, such as humor, and person characteristics. Individual difference variables may thus act as "boundary conditions" shaping the influence of humor on responses to ads.

Given the widespread evidence that attitude toward the ad, [A.sub.ad], has a mediating role on attitude toward the brand, [A.sub.b] (Lutz 1985; MacKenzie, Lutz, and Belch 1986; Shimp 1981), [A.sub.ad] might also be expected to mediate the effect of humor on [A.sub.b]. Indeed, Zhang (1996) recently showed [A.sub.ad] to be a mediator of humor's effects on [A.sub.b] for subjects low in NFC, although it is unclear whether humor can be an antecedent of attitudes for subjects high in NFC.

Following research by Thorson and Powell (1991; see also Thorson and Powell 1993a, 1993b) that examined the nature of humor, Cline, Machleit, and Kellaris (1999) identified a new person trait, the "need for levity" (NFL), which represents an individual's craving for humor and whimsy. NFL refers to an individual's tendency to seek out levity, which centers on humor (amusement, wit, nonsense) and whimsy (caprice, spontaneity, free-spiritedness). This trait is closely related to the idea of "sense of humor," but defines a construct more extensive than sense of humor alone, with each of its four dimensions (internal-humor, external-humor, internal-whimsy, and external-whimsy) playing a role in defining an individual's overall need for levity. Following Cline (1997), we focus on NFL's two humor dimensions, a subscale hereafter referred to as "need for humor" (NFH).

The present research investigates the potential moderating role of NFH in shaping responses to humor. This emphasis is consistent with Punj and Stewart's suggestion (1983) that the interplay of situational and individual variables (rather than just main effects) is more diagnostic in delineating the processes affecting consumer behavior. Recent research by Cline, Machleit, and Kellaris (1999; Cline 1997) suggests that consumers' need for humor should influence the extent to which humor is appealing to them. It follows that ads with higher (versus lower) levels of humor will more positively affect attitudes among consumers who have a higher (versus lower) NFH.

On the basis of theory and prior findings, the following contingency is anticipated:

H1: The positive influence of higher versus lower ad humorousness on attitude toward the ad will be greater among individuals with higher (versus lower) needs for humor.

STUDY 1

Method

Study 1 tests the hypothesis that an individual's NFH should moderate the impact of humor on ad responses. Humor was experimentally manipulated (lower versus higher levels) using print ads, and NFH was measured. Lower versus higher NFH groups were created via median split. Hence, the experiment used a 2 x 2 between-subjects, full-factorial design. The dependent variable was attitude toward the ad. The procedure involved assigning subjects to treatment groups, exposing them individually to print ads presented in a questionnaire booklet, and capturing their reactions using a self-administered, multi-item scale.

Subjects

Subjects were 105 undergraduate students drawn from a subject pool at a large, Midwestern university. Fifty-five (52.4%) were male; 50 (47.6%) female. Ages ranged from 18 to 48 years (mean = 22.9, median = 21). Course credit was offered as an incentive to participate. Complete responses on all the measures were obtained from all but five of the subjects, for a final, usable sample size of 100 (n = 100).

Stimulus Materials

Stimulus materials consisted of two versions of a magazine ad created to represent lower versus higher levels of humor content. The product category (a premium coffee) represents a low-risk, convenience product for which the use of humor is both frequent and believed to be effective (cf. Weinberger and Campbell 1991). The format of the ads was patterned after prior research using print ads (e.g., Cline and Kellaris 1999; Heckler and Childers 1992; Krishnan and Chakravarti 1994). Each ad contained a brand name, a headline, a product picture, two cartoon figures, brand claims, and a tagline. Following Krishnan and Chakravarti (1994) and Cline (1997), the headline was used as the primary manipulation of humor strength. The nonhumorous headline stated: "Coffee Doesn't Have to Taste Like 'The Ground' Just Because It's Made in the Morning." The more humorous headline included the pun "Why Do Most Coffees Taste Like Mud? Because They're 'Ground' Too Early in the Morning." Cartoon figures were used to reinforce the level of humor conveyed by the headline. All other ad elements were held constant.

An extensive program of pretesting was conducted to select the product category, brand name and copy, and to verify the humor manipulation. A group of 26 subjects, taken from the same pool as those in the main studies, evaluated the face validity of the humor scale. A consensus indicated that the endpoint 1 on a seven-point humorous/funny/amusing scale is interpreted as "serious," that the midpoint of the scale, 4, is interpreted as "mildly humorous," and that the endpoint 7 is perceived as "laughing out loud." A final pretest (n = 75) evaluated perceived humor for two conditions of humor strength, using completed versions of the mock-up ads in a between-subjects test. Pretest results indicated the "higher-humor" ad (M = 4.36) was perceived as relatively more humorous than the "lower-humor" ad (M = 3.33), t = 2.07, df = 43, p < .023 (one-tailed). Results suggest that a 3.33 indicates some nonzero-level of humor, and 4.36 indicates slightly greater than "mild" humor.

Measures

Attitude toward the ad ([A.sub.ad]) was measured using a three-item seven-point semantic differential scale adapted from multiple sources (e.g., MacKenzie and Lutz 1989; Madden, Allen, and Twible 1988). The items were preceded by simple instructions to "Answer the following questions by circling an appropriate number on each of the scales: good(7)/bad(1), pleasant(7)/unpleasant(1), and favorable(7)/unfavorable(1)." Responses were summed and averaged. Alpha reliability for the scale is .90.

NFH was measured using ten items from the "need for levity" scale in Cline, Machleit, and Kellaris (1999). These items represent both internal- and external-humor dimensions, which are subscales of the NFL measure. The subscales of the NFL dimension labeled "whimsy" were omitted, as whimsy is less relevant to the present research (cf. Cline, Machleit, and Kellaris 2002). Construct validity for the NFH subscale (internal and external dimensions) was assessed following the procedures recommended by Gerbing and Anderson (1988). The analysis shows that the NFH scale exceeds the minimum requirements for convergent, discriminant, and nomological validity (Bagozzi 1984; Bagozzi and Yi 1988). In addition, the confirmatory factor analysis demonstrates excellent fit. The two-factor model exceeds the minimum threshold on a wide range of goodness-of-fit measures (e.g., GFI = .91, A GFI = .85). The correlation between the internal- and external-humor dimensions is positive and significant ([[PHI].sub.12] = .30, t = 2.78, p < .01). Hence, for the purposes of the experiment, the dimensions were summed and averaged into a single "global" measure of NFH with [alpha] reliability of .83. High and low NFH groups were formed via median split (median = 5.22) and found to differ statistically in terms of NFH scores (t = 14.1, p < .001).

To facilitate a posttest check on the success of our humor manipulation, the questionnaire included a three-item seven-point semantic differential measure of perceived humor. The items were preceded by instructions to "Please indicate the extent to which you agree with the following statements by circling the appropriate numbers" and by the prompt "The advertisement that I just saw was ..." The items were humorous(7)/not humorous(1), funny(7)/not funny(1), and amusing(7)/not amusing(1). Responses were summed and averaged. Alpha reliability of the three-item composite scale was .97. The questionnaire concluded with some standard demographic items to facilitate sample description and an open-ended question to probe for hypothesis guessing ("Indicate what you believe the purpose of this study might be."). No subjects guessed the key hypothesis.

Procedure

The NFH measure was administered in a separate, unrelated study, which preceded the main part of the study by more than a week's time to reduce potential demand effects. Experimental subjects were processed in a large auditorium. They were told that the study sought their reactions to a magazine advertisement prototype. Each subject received a questionnaire booklet containing the stimuli and measures. The first page contained brief instructions and a consent form. The second page contained a reminder to read the instructions before viewing the ad. The third page contained a treatment ad. The measures appeared on the pages following the ad. A proctor monitored subjects' compliance with written instructions to work independently. No violations of this instruction were observed.

Results

Manipulation Check

To verify the integrity of the humor manipulation, an analysis of variance (ANOVA) examined the impact of the humor treatment and NFH on perceived humor. The analysis shows a significant main effect of the humor treatment on the perceived humorousness of the ad, F(1, 97) = 15.27, p < .001; [[omega].sup.2] = .05. Mean scores on the humor scale were 3.3 for the lower humor treatment group, 4.5 for the higher humor treatment group (t = -3.96, p < .001, one-tailed). There were no significant main (F = 1.02) or interactive (F = .45) effects involving NFH. We interpret this as evidence of successful, independent manipulation of humor.

Effects on Attitudes

To test the hypothesis that NFH should moderate the impact of humor on ad responses, we performed analyses of variance (ANOVAs) with attitude toward the ad as the dependent variable. Results show a significant, two-way interaction of humor and NFH on attitude toward the ad, F(1, 96) = 6.56,p < .021, [[omega].sup.2] = .052, and a significant main effect of humor, F(1, 96) = 4.9, p = .03. No main effect for NFH was observed, F(1, 96) = .94 (n.s.). The shape of this interaction is depicted in Figure 1.

[FIGURE 1 OMITTED]

A series of post hoc comparisons shows that among subjects with a higher need for humor, attitudes toward the ad differed significantly across lower (M = 3.52) versus higher (M = 4.64) humor treatment groups (t = -2.80, p < .005, one-tailed). Among lower NFH subjects, lower (M = 4.34) versus higher (M = 4.26) humor groups did not differ statistically (p > . 10).

Among subjects exposed to the less humorous ad, differences were observed between those lower (M = 4.34) versus higher (M = 3.52) in NFH (t = 2.52, p < .015, two-tailed). Statistical differences between lower (M = 4.26) versus higher (M = 4.64) NFH groups were not observed among those exposed to the more humorous ads. Thus, H1 is supported.

Discussion

Results of our initial experimentation suggest that the influence of humor on attitude toward the ad may depend on the individual's orientation to humor as reflected in NFH scores. Responses appear to be shaped by the joint interplay of ad characteristics (level of humor) and person characteristics (NFH). Results from Study 1 suggest that whereas those individuals lower in NFH may be indifferent to the level of humor in the ad, those higher in NFH may be "turned off" by understated or weak humor. This observation may represent an important contingency underlying the effective or ineffective use of humor in advertisements. Awareness of the general characteristics of an audience may suggest whether or not to use humorous stimuli. Nevertheless, further evidence is required to assess the generality of the effect and to examine process mechanisms underlying the effect.

There may be multiple causes of the interactive effect observed in Study 1. Perhaps the simplest among these is the differential levels of motivation to attend to and process humorous ads concomitant to individual differences in NFH. As noted earlier, Zhang (1996) observed a mediating effect of [A.sub.ad] on humor's effects on [A.sub.b] for subjects low in NFC. People with higher (versus lower) NFC are more motivated to engage in issue-relevant thinking (Cacioppo and Petty 1982; Cacioppo, Petty, and Morris 1983). People with lower levels of need for cognition may treat message source variables as "simple acceptance or rejection" peripheral cues (Petty and Cacioppo 1984, p. 669; see also Petty and Cacioppo 1981, 1986). Following Cline and Kellaris (1999), the present research focuses on the case of incidental humor. Thus, the humor is likely to operate as a peripheral cue to persuasion and may play a greater persuasive role for people who are low in NFC and not highly motivated to process the advertising message critically. In contrast, people who are high in NFC are more likely to process the issue-relevant ad claims than the incidental humor. This is consistent with previous research that found low NFC individuals' attitudes to be influenced more by peripheral cues and less by message arguments (Haugtvedt et al. 1992; Haugrvedt and Petty 1992). Low NFC has also been linked to more heuristic processing (Batra and Stayman 1990). Results from Study 1 indicate that for lower and moderate levels of humor, the influence of NFH appears to accentuate the negative effect of weak humor. However, NFC may play an important moderating role, such that in a three-way interaction, the joint influence of NFH and humor strength on attitudes emerges for those lower (versus higher) in NFC. In summary of these expectations:

H2: The joint influence of humor strength and NFH on A d will be more pronounced among individuals characterized by lower (versus higher) needs for cognition.

A second study was conducted to investigate the anticipated three-way interaction among humor strength, need for humor, and need for cognition.

STUDY 2

Method

An experiment manipulated level of humor content (lower versus higher) in a between-subjects factorial design using print ads different from those used in Study 1. NFH and NFC were measured in an earlier session; cases were matched by identification numbers, and subjects were categorized into relatively lower versus higher groups on the basis of median splits. The focal dependent variable was attitude toward the ad. The procedure involved randomly assigning subjects to treatment conditions, providing a cover story, exposing subjects to a test ad contained in a questionnaire booklet, and capturing responses via self-administered, multi-item scales.

Subjects

Subjects were 119 students drawn from a subject pool at a large, Midwestern university. Complete responses were obtained from 62 males (60%) and 42 females (40%) for a final sample of n = 104. Subjects' ages ranged from 19 to 41 years (mean = 21.9; median = 21). All subjects were enrolled in an introductory marketing class at the time of recruitment, and course credit was offered as an incentive to participate in the study. None of the subjects had participated in the previous study.

Stimulus Materials

As in Study 1, a full-color print ad for a consumer product was created. The ad was designed to be similar with respect to ad and layout, claims, and presentational order of claims (see Chattopadhyay and Basu 1990). The product category (a caffeine and vitamin supplement) was selected for its relevance to college student subjects. The ad featured a fictitious over-the-counter stimulant tablet ("Alertamin"). The headline read "trouble staying alert in class? Try new ALERTAMIN!" (followed by bullet points).

To manipulate lower versus higher levels of humor, the ad contained either a humorous or nonhumorous illustration. The nonhumorous version featured an illustration of an "ALERTAMIN caffeine and vitamin supplement." The more humorous version featured an illustration of a cartoon figure of a college student, asleep, snoring at a desk. The ads were screened by experts and judged to have differential humor content; a posttest manipulation check verified the success of the manipulation.

Measures

Dependent Variable. Attitude toward the ad was measured using four seven-point semantic differential scales patterned after Osgood, Suci, and Tannenbaum (1957). The items bad/good, tasteless/tasteful, unpleasant/pleasant, and unlikable/likable were preceded by the prompt "The AD I saw was:" (coefficient [alpha] = .84).

Measured Independent Variables. The person trait need for humor (Cline, Machleit, and Kellaris 2002) was measured at an earlier session using published scales as reported in Study 1. Construct validity of the NFH scale was assessed using the Gerbing and Anderson (1988) paradigm. The analysis shows that the NFH scale exceeds the minimum requirements for convergent, discriminant, and nomological validity (Bagozzi 1984; Bagozzi and Yi 1988). In addition, the confirmatory factor analysis demonstrates excellent fit. The two-factor model (representing internal- and external-humor dimensions) exceeds the minimum threshold on a wide range of goodness-of-fit measures (e.g., GFI = .92, AGFI = .87). The correlation between the internal and external humor dimensions was positive and significant ([[PHI].sub.12] = .49, t = 4.88, p < .01). Hence, for the purposes of the experiment, the dimensions were summed and averaged into a single "global" measure of NFH with [alpha] reliability of .83. NFC was also measured at an earlier session using the short form of the scale developed by Cacioppo and Petty (1982). Alpha reliability for the 18-item scale (Cacioppo, Petty, and Kao 1984) is .87.

Other Measures. The last page of the questionnaire contained items to assess the success and integrity of the humor manipulation. Perceived humor was measured using four seven-point agreement items labeled "humorous, funny, amusing, serious" (reverse coded; coefficient [alpha] = .85) preceded by the prompt "The AD I saw was ..." The questionnaire concluded with standard demographic measures.

Procedure

Trait variables were measured in a mass testing setting early in the school term. Approximately six weeks later, the experiment was administered in a large auditorium classroom. Questionnaire booklets containing a cover sheet, test ad, and measures were distributed. Systematic randomization was achieved by collating questionnaires by treatment. Subjects were instructed verbally and in print (on the cover sheet) to "take a moment to look at the proposed advertisement on the next page. Then continue to the questions about the ad itself and about the specific product featured in the ad. Please provide your honest opinions, and answer the questions in the order presented without turning back to the ad or to previous questions." No violations of these instructions were observed during the ten-minute procedure.

Results

Manipulation and Confounding Checks

To verify the integrity of the humor manipulation, an ANOVA examined the impact of the humor treatment and NFH on perceived humor. The analysis shows a significant main effect of the humor treatment on the perceived humor of the ad, F(1,112) = 44.6, p < .001; [[omega].sup.2] = .28. As in Study 1, mean scores on the humor scale were significantly lower for the lower (M = 2.7) versus the higher (M = 4.1) humor treatment group. There were no significant main or interactive effects involving NFH. We interpret this as evidence of successful, independent manipulation of humor.

Humor and NFH

ANOVAs indicate a significant two-way interactive effect of humor with NFH on attitude toward the ad, F(1, 89) = 4.44, p < .038; [[omega].sup.2] = .049. Results indicate that subjects higher in NFH exhibit more favorable attitudes toward ads with higher humor content (M = 4.46) versus lower humor content (M = 3.36), t = 3.19, df= 48, p < .01 (one-tailed). This result closely resembles the pattern found in Study 1.

Humor, NFH, and NFC

The ANOVA also reveals the anticipated three-way interaction among humor, NFH, and NFC on [A.sub.ad], F(1, 89) = 5.35, p < .023; [[omega].sup.2]= .058. Subjects with lower NFC and higher NFH exhibit more positive attitudes toward ads with higher (M = 4.82) versus lower (M = 3.05) humor content, t = 5.01, df = 21,p < .001 (one-tailed). In addition, subjects with lower NFC and higher NFH respond more positively to the ad with higher humor content (M = 4.82) than do those with lower needs for cognition and humor (M = 4.13, t = 1.71, df = 21, p < .05, one-tailed). Furthermore, subjects with lower needs for cognition and humor report more favorable attitudes toward the ad with lower humor (M = 4.53) than do those with lower NFC and higher NFH (M = 3.05, t = 3.11, df = 20, p < .01, one-tailed). Finally, subjects with both lower needs for cognition and humor do not report differential attitudes toward ads with higher (versus lower) levels of humor. Thus, H2 is supported. The three-way interaction on [A.sub.ad] for the tangible product is depicted in Figure 2.

[FIGURE 2 OMITTED]

Discussion

Results from Study 2 provide additional support for the notion that NFH, like other person traits, may explain a source of variation in attitudes through its role as a moderator. The results indicate that consumers high in NFH may form more favorable attitudes toward ads with humorous content. That humorous ads are well received among individuals characterized by high needs for humor seems straightforward. People who crave humor may enjoy numerous forms of humor (e.g., jokes, comedy, wit), including humorous print ads.

Study 2 also examined the joint interplay of NFC and NFH on the effects of humorous advertising. The findings provide preliminary evidence that the moderating role of NFH on the effect of humor on [A.sub.ad] may be more pronounced among subjects with lower needs for cognition. How might NFC moderate the influence of NFH on attitudes such that the positive effects of humor are more pronounced among people who are low (versus high) in NFC?

First, high NFC may motivate systematic processing. To the extent that high NFC individuals tend to focus on the fundamental arguments of an ad, their need for humor may be "blocked." Hence, they do not process the peripheral humor. For this group, NFH should not moderate the influence of humor content on attitudes toward the ad. By comparison, individuals lower in NFC may use heuristic processing, with humor as a peripheral cue. It follows that, if these individuals are also higher in NFH, they should respond more positively to humorous ads.

A second possibility is that individuals who are higher in NFC process both the central arguments and the humor, but any positive affect generated by the humor is lost in the process. For this group, processing humor may be like dissecting a frog--in the words of E. B. White, "it dies during the analysis." Specifically, individuals higher in NFC (and higher in NFH) may dilute the value of the humor via rational processing, whereas those higher in NFC (and lower in NFH) may not enjoy humor in an ad--no matter how strong. In both cases, humor is unlikely to generate positive affect. In contrast, those individuals lower in NFC and higher in NFH may find humor to be its own virtue and, consequently, transfer positive affect to the ad, whereas those lower in NFC and NFH may not find it amusing. Thus, when NFC is low, it seems reasonable to expect NFH to moderate the influence of humor strength on [A.sub.ad]--particularly if the humor is compelling.

A third possibility centers on the distracting properties of humor. The distraction hypothesis (Festinger and Maccoby 1964) suggests that distraction may enhance message persuasiveness by interfering with the audience's subvocal attempts to counterargue against the dissonant information. In an advertising context, affective responses elicited by humor may divert a consumer from counterarguing. Reduced counter-arguing, in turn, may increase acceptance of a humorous message (e.g., Duncan 1979). Thus, it is possible that individuals who are lower in NFC (and higher in NFH) will generate fewer counterarguments and respond more favorably to humorous advertising than those who are higher in NFC or lower in NFH. Whereas Studies 1 and 2 manipulated low and moderate levels of humor, we anticipate the significant interaction effect for humor strength and NFH on attitudes toward the ad to emerge at higher levels of humor strength:

H3a: The positive influence of high versus low ad humorousness on attitude toward the ad will be greater for individuals characterized by a higher (versus lower) NFH.

H3b: The interactive effect of humor (three levels) and NFH on attitudes toward the ad will be more pronounced for those lower (versus higher) in NFC.

Previous research supports an "affect transfer" explanation for humor's influence on attitudes. Madden, Allen, and Twible (1988) demonstrate that a humorous ad can generate significantly more positive affect than a nonhumorous version. We anticipate need for humor (cognition) to augment (attenuate) this effect. There is also support for the notion that attitudes elicited by humor may divert consumers from counterarguing against a discrepant message. Recent research by Arias-Bolzmann, Chakraborty, and Mowen (2000) shows that humor can result in proportionately more positive cognitive responses, and hence more favorable attitudes, than no humor. We anticipate perceived humor to reduce counterarguments and NFC to increase counterarguments. Finally, recent research by Alden and Hoyer (2000) finds a direct correlation between perceived humor and [A.sub.ad] (also cf. Brown and Stayman 1992). We expect to replicate this direct link between perceived humor and A d. In summary of these expectations:

H4: The influence of perceived humor, NFH, and NFC on attitudes toward the ad will be mediated by attention to the ad. affect, and counterarguments.

H5: Perceived humor will have a direct influence on attitude toward the ad.

A third study was conducted to investigate whether the two-way interaction of humor strength and NFH emerges at higher levels of humor (H3a), to test the robustness of the three-way interaction (H3b) and to explore process explanations underlying the interaction (H4, H5).

STUDY 3

Method

This study was designed to extend Study 2 by manipulating three (versus two), distinct levels of humor and to collect further evidence concerning the processes underlying the effects observed in the previous studies. Level of humor content (low, moderate, and high) was manipulated in a between-subjects factorial design using stimuli adapted from an unfamiliar magazine advertisement. NFH and NFC were both categorized into relatively lower versus higher groups on the basis of low/ high splits. The dependent variables included attitude toward the ad, attention paid to the ad, ad-induced affect, and counterarguments. The procedure followed Studies 1 and 2.

Subjects

Subjects were 277 students drawn from subject pools at two mid-Atlantic universities. Complete responses were obtained from 149 males (54%) and 124 females (46%) for a final sample of n = 273. Subjects' ages ranged from 18 to 38 years (mean = 21.7; median = 22). None of the respondents had participated in the previous two experiments.

Stimulus Materials

To manipulate humor, three versions of a full-color print ad were adapted from an ad for "Revelstoke Canadian Spiced Whisky" found in VIBE magazine. The original ad was scanned and manipulated via graphical software. The brand name was changed to the fictitious name "Preston's." A pretest (n = 34) verified that no subjects were familiar with the brand. The three ads were created to be identical regarding ad and layout, claims, and tag-line, and the headline was used as the primary manipulation of humor strength (Chattopadhyay and Basu 1990).

The product category (liquor) was chosen for several reasons. First, because a majority of the subjects were legally old enough to drink, this product category was relevant. Second, liquor is one of the few product categories for which an extensive set of humorous print ads is available for pretesting (cf. Geuens and De Pelsmacker 1998). Third, Weinberger et al. (1995) found that incidence of humor in magazines is highest in the low-risk/expressive product category of their "Product Color Matrix." This category, labeled "Yellow Goods" or "little treats," includes items such as tobacco and alcohol products. Furthermore, Madden and Weinberger (1984) surveyed 150 U.S. advertising agencies about their views on the use of humor. The categories receiving the largest support for using humor include alcohol products. Fourth, the product category, liquor, was thought to minimize the impact of potential confounding factors and provide a similar variance for both males and females (cf. Arias-Bolzmann, Chakraborty, and Mowen 2000).

An extensive program of pretesting was conducted to verify the humor strength manipulation and to create unfamiliar ads. A sample of 30 subjects, taken from the same pool as those in Study 3, evaluated the humorousness of five versions of the ad in a within-subjects pretest. Three of the ads emerged as providing distinct and equal intervals of humor strength; the means for low, moderate, and high humor were 2.9, 4.4, and 5.9, respectively (on a seven-point scale). In addition, none of the subjects indicated that they had ever seen the ads before. A second pretest (n = 69) validated the high level of humor (M = 5.7). A final pretest (n = 76) evaluated perceived humor for all three conditions, using completed versions of the mock-up ads in a between-subjects test. The results indicated that the treatments were perceived as differentially humorous: The "high-humor" ad (M = 5.29) was perceived as relatively more humorous than the "moderate-humor" ad (M = 3.71), t = 8.83, df= 74, p < .001 (two-tailed), and the "moderate-humor" ad (M = 3.71) was perceived as relatively more humorous than the "low-humor" ad (M = 2.4), t = 3.89, df = 74, p < .001. In summary, pretests suggest that three distinct levels of humor were achieved.

Measures and Procedure

Attitude toward the ad ([A.sub.ad]) was measured using the same scale used in Study 2. Attitude toward the brand ([A.sub.b]) was measured using five seven-point semantic differential scales based on Osgood, Suci, and Tannenbaum (1957). The items good/bad, appealing/unappealing, interesting/uninteresting, desirable/undesirable, acceptable/unacceptable were preceded by the prompt "The PRODUCT/SERVICE featured in the ad is ..." Responses were summed and averaged. Attention to the ad was measured by a three-item seven-point Likert-type scale intended to determine the amount of attention devoted to the written message in the ad (Muehling, Stoltman, and Grossbart 1990). Ad-induced affect was measured by a four-item seven-point semantic differential scale developed to measure the feeling that has been induced by an object in a person (Yi 1990). Alpha reliabiliries for the three-item scales exceeded .90. In addition, counterarguments were assessed via a thought protocol task (cf. Madden, Allen, and Twible 1988).

NFH was measured using the same 12 NFH items used in Study 2, representing both internal- and external-humor dimensions of NFH. Similar to Studies 1 and 2, construct validity of the NFH scale was assessed following the procedures recommended by Gerbing and Anderson (1988). The analysis shows that the NFH scale exceeds the minimum requirements for convergent, discriminant, and nomological validity (Bagozzi 1984; Bagozzi and Yi 1988). In addition, the confirmatory factor analysis demonstrates excellent fit. The two-factor model (representing internal- and external-humor dimensions) exceeds the minimum threshold on a wide range of goodness-of-fit measures (e.g., GFI = .93, AGFI = .89). The correlation between the internal- and external-humor subscales was positive and significant (p < .001). Hence, they were summed and averaged into a single "global" measure with a reliability of .85. High and low NFH groups were formed via median split (median = 5.0) and found to differ statistically in terms of NFH scores (t = -23.7, p < .001).

As in Study 2, NFC was measured using the short form of the scale developed by Cacioppo and Petty (1982). Alpha reliability for the 18-item scale (Cacioppo, Petty, and Chuan 1984) is .89. Finally, ad and brand familiarity were measured on one-item semantic differential scales with the endpoints familiar(7)/unfamiliar(1) (reliabilities for both scales exceeded .90).

Results

Manipulation and Confound Checks

To verify the integrity of the humor manipulation, an ANOVA examined the impact of the humor treatment and NFH on perceived humor. The analysis shows a significant main effect of the humor treatment on perceived humor of the ad, F(1,225) = 60.74, p < .001; [[omega].sup.2] = .35. Mean scores on the humor scale were 2.7 for the low humor treatment group, 3.8 for the moderate humor group, and 5.4 for the high humor group. Although there was a main effect of NFH, F(1,225) = 8.80, p < .004; [[omega].sup.2] = .04, there were no interactive effects involving NFH. We interpret this as evidence of successfully manipulating three distinct levels of humor.

To verify that the ad and brand were unfamiliar to subjects, frequency counts of the seven-point one-item familiarity measures were examined. The median responses were 1.0 (unfamiliar) for all three treatment groups.

Humor and NFH

The anticipated two-way interaction of humor strength with NFH on [A.sub.ad] emerged, F(1,168) = 3.70, p < .05; [[omega].sup.2] = .03. Subjects higher in NFH exhibit more favorable attitudes toward ads with strong humor (M = 4.432) versus weak humor (M = 3.21), t = 3.84, df = 89, p < .001 (one-tailed). Moreover, the highest level of humor engenders more favorable attitudes for those higher (M = 4.32) versus lower (M = 3.48) in NFH, t = 2.94, df = 87, p < .004. Thus, H3a is supported.

Humor, NFH, and NFC

The ANOVA also reveals a three-way interaction between humor (three levels), NFH, and NFC on [A.sub.ad], F(2, 115) = 3.26, p < .042, [[omega].sup.2] = .054. Similar to Study 2 results, subjects with lower needs for cognition and higher needs for humor exhibit more positive attitudes toward ads with moderate (M = 4.47) versus low (M = 2.88) humor content, t = -2.58, df = 16, p < .01 (one-tailed). By comparison, subjects with both lower NFC and NFH reveal more positive attitudes toward the low (M = 3.79) versus moderate (M = 3.03) humor content, t = 1.68, df = 20, p < .05 (one-tailed).

Furthermore, subjects with lower NFC and higher NFH exhibit more positive attitudes toward ads with moderate humor content (M = 4.47) than subjects with both lower needs for cognition and humor (M = 3.03), t = -2.54, df = 19, p < .01 (one-tailed). Similarly, subjects with lower NFC and higher NFH exhibit more positive attitudes toward ads with high (M = 4.96) humor content than do subjects with both lower needs for cognition and humor (M = 3.39), t = -2.48, df = 19, p < .01 (one-tailed). In contrast, subjects with both lower needs for cognition and humor reported more positive attitudes toward ads with low humor content (M = 3.79) than those with lower NFC and higher NFH (M = 2.87), t = 1.98, df = 17, p < .04 (one-tailed). The three-way interaction is depicted in Figure 3.

[FIGURE 3 OMITTED]

In summary, H3b is supported. Study 3 extends the results of Study 2 by providing additional evidence of an interactive effect of humor (with three distinct levels of humor strength) and NFH on ad attitudes toward the ad. For high NFC subjects, humor strength and NFH appear to have little effect on attitudes. For low NFC subjects, however, the influence of the interaction between humor strength and NFH is dramatic and in the expected direction.

Process Explanations

As previously elaborated, there are at least three plausible explanations for why NFC further qualifies the interactive influence of humor content and NFH on attitudes toward advertisements (i.e., systematic versus heuristic processing, affect, and distraction from counterargumentation). Data from Study 3 were analyzed to test these process explanations, both individually and in a nomological network.

The first process explanation suggests that those higher in NFC will focus attention on central arguments rather than on the humor. An ANOVA examined the interaction of humor content level, NFH, and NFC on attention paid to the ad. Although NFC failed to produce a main effect, the ANOVA shows a significant three-way interaction, F(1,158) = 4.61, p < .033; [[omega].sup.2] = .04. Subjects higher in NFC and higher in NFH paid more attention to the humorous ad (M = 5.3) than those higher in NFC and lower in NFH (M = 4.47), t = -2.66, df = 40, p < .01 (two-tailed). On this basis, it seems reasonable to conjecture that those subjects higher (versus lower) in NFC may be able to quickly identify that an ad contains humor, yet only those individuals who are also higher (versus lower) in NFH maintain their attention to the humorous elements in the ad.

To test the notion that individuals higher (versus lower) in NFC find less humor in an ad, a two-way ANOVA examined the interaction of humor strength and NFC on perceptions of humor. The ANOVA reveals a two-way interaction between humor (three levels), and NFC on perceptions of humor content, F(2, 128) = 3.31, p < .039; [[omega].sup.2] = .049. Contrary to expectations, individuals higher in NFC find the ad with high humor content more amusing (M = 5.92) than those lower in NFC (M = 5.10), t = 2.20, df = 38, p < .04 (two-tailed). Thus, the first process explanation is not supported. It appears that, as a group, those individuals higher (versus lower) in NFC do not pay more attention to the ad and, indeed, they find the humorous ad amusing.

A two-way ANOVA was performed to test the second process explanation, which postulated that given lower NFC, humor generates differential affect for those who are higher (versus lower) in NFH. The results reveal a significant two-way interaction of humor and NFH on affect, F(1, 81) = 5.17, p < .03; [[omega].sup.2] = .04. Subjects higher in NFH generated more affect from the ad with high humor content (M = 4.6) than from the ad with low humor content (M = 3.7), t = -2.63, df = 37, p < .02 (two-tailed). In addition, subjects higher in NFH generated more affect for the ad with high humor content (M = 4.6) than did those lower in NFH (M = 3.6), t = -2.97, df = 36, p < .01 (two-tailed). Additional analysis brings into clearer relief the source of the affect for those lower in NFC and higher in NFH. Results from an ANOVA reveal a significant two-way interaction of humor and NFH on perceived humor, F(1, 81) = 4.61, p < .04; [[omega].sup.2] = .03. Subjects lower in NFC and higher in NFH found the ad with high humor content significantly more amusing (M = 5.9) than did those lower in needs for cognition and humor (M = 4.5), t = -3.68, df 21.5, p < .01 (two-tailed). Thus, the second process explanation receives preliminary support. For those lower in cognition, need for humor appears to influence perceptions of strong humor as well as the affect that it induces.

The third process explanation was tested by examining the relationships among the continuous variables: [A.sub.ad], affect, perceived humor, NFH, NFC, and counteraging. Because regression analysis is more effective than ANOVA for capturing information provided by continuous variables (Hays 1996, cited in Alden and Hoyer 2000), a stepwise regression equation was specified. First, counterarguing was regressed on the following independent variables: perceived humor, affect, NFH, and NFC. The coefficient for NFC was not significant (p > . 10); it was therefore excluded from further analysis. The results of the regression equation suggest that the model explains the data (F = 14.27, p < .001, [r.sup.2] = .10). The coefficient for perceived humor is significant ([[beta].sub.perceived humor] = -.118, t = -4.01, p < .001), and the coefficient for affect approaches significance ([[beta].sub.affect] = -.09, t = 1.78, p < .09). The results indicate that perceived humor is negatively related to counterarguing. Next, affect was regressed on the following independent variables: perceived humor, NFH, and NFC. The results show that each of the three predictors is significant ([[beta].sub.humor] = .186, t = 6.02, p < .001;[[beta].sub.NFH] = .341, t = 4.71, p < .001; [[beta].sub.NFC] = -.15, t = -2.14, p < .04). Perceived humor and NFH are positively related to affect, whereas NFC is negatively related to affect. Finally, Am was regressed on the following independent variables: counterarguments, perceived humor, and affect. The results suggest that the model explains a significant amount of the variation in [A.sub.ad] (F = 100.64, p < .001, [r.sup.2] = .726). The coefficient for counterarguments ([[beta].sub.counterargume] = -.211, t = -3.15, p < .002), perceived humor ([[beta].sub.perceived humor] = .180, t = 5.38, p < .001), and affect ([[beta].sub.affect] = .705, t = 11.89, p < .001) are significant. Whereas counterarguing is negatively linked to [A.sub.ad], perceived humor and affect are positively linked to [A.sub.ad]. Thus, the third process explanation receives partial support.

In sum, we find preliminary support for the notion that perceived humor, NFH, and NFC operate through affect and counterargumentation. To test this explanation in a nomological network, a structural equation model was estimated.

Path Model

A path model with reliabilities (see Figure 4) was specified with the exogenous variables--NFH, perceived humor, and NFC--and the four endogenous variables--counter-arguments, affect, [A.sub.ad], and [A.sub.b]. To estimate the model, summed scores rather than individual items served as indicators of the latent variables. In addition, the covariance (versus the correlation) matrix was used as the input matrix, which aids in achieving reliable maximum likelihood estimates (Bentler 1989) and helps overcome difficulties associated with fitting LISREL models to small samples (cf. Bentler and Chou 1987; Williams and Hazer 1986). Furthermore, the reliabilities (Cronbach's as) were used to compute the factor loadings and measurement errors for the latent variables. Each factor loading (path from the latent variable to its manifest indicator) was set equal to the square root of the reliability of the measured variable. The error variance of each manifest indicator was set equal to the product of the variance of the summed scale and the quantity one minus the reliability of the scale. The results reveal the anticipated links between [A.sub.ad] and [A.sub.b] ([[beta].sub.12] = .53, t = 9.37, df = 21, p < .01) and between affect and [A.sub.ad] ([[beta].sub.23] = .83, t = 11.34, df = 21, p < .01). In addition, the exogenous variables appear to operate via affect and counterarguments, which act as mediators. NFC appears to negatively influence affect ([[gamma].sub.33] = -.14, t = - .99, df = 21, p < .05), whereas NFH positively influences affect ([[gamma].sub.31], = .59, t = 2.90, df = 21, p < .01). Perceived humor appears to reduce counterarguments ([[gamma].sub.42] = -.14, t = -4.95, df = 21, p < .01) and positively influences affect ([[gamma].sub.32] = .18, t = 5.50, df = 21, p < .01). In turn, counterarguments appear to negatively influence [A.sub.ad] ([[beta].sub.24] = -23, t = -3.32, df = 21, p < .01), and [A.sub.b] ([[gamma].sub.22] = .15, t = 4.14, df = 21, p < .01). Finally, the model shows the anticipated direct link between perceived humor and [A.sub.ad] ([[gamma].sub.22] = . 15, t = 4.14, df = 21, p < .01).

[FIGURE 4 OMITTED]

In addition to validating the proposed nomological network, the data fit the model extremely well. The goodness-of-fit index (GFI = .98), adjusted goodness-of-fit index (AGFI = .93), and the root mean square residual (RMSR = .054) provide evidence for the efficacy of the specified model (Bagozzi 1984). Finally, only one standardized residual fell outside the range of -2.0 to 2.0. Thus, the path model explains the relationships between the interaction variables (perceived humor, NFH, and NFC), the mediating variables (counterarguments and affect), and the response variables (attitudes toward the ad and brand).

In sum, we find support for H4 and H5. In a nomological context, NFH, NFC, and perceived humor appear to influence attitudes by elevating affect and reducing counterarguments (H4). In addition, perceived humor is directly linked to [A.sub.ad] (H5).

GENERAL DISCUSSION

Need for Humor

In the present research, our interest lies in two subscales of NFH: internal-humor (the need to experience humor internally--humor generation) and external-humor (the need to experience humor from external sources--humor connoisseurship). The domain of the need-for-humor construct centers on an individual's tendency to enjoy or engage in (to seek out) amusement, wit, and nonsense. The present research offers some evidence that an individual's need for humor may play a significant role in moderating attitudes. The trait appears to operate as a motivational factor whereby individuals with higher levels of NFH respond more favorably to humorous ads. Hence, NFH may hold some potential for unlocking deeper insights into consumer information processing and provide important boundary conditions governing the operation of humorous stimuli on communication outcomes.

Evidence from three experiments suggests that NFH, like other person traits, may explain an additional source of variation in Aa through its role as a moderator. In addition, NFH may expand our knowledge of sense of humor in general and how it influences consumers' responses to humorous advertising. It seems that people who crave humor should enjoy numerous forms of humor (e.g., jokes, comedy, wit). The present research shows this to be true for humorous print ads. Results indicate that consumers high in NFH not only form more favorable attitudes toward humorous ads, but also may exhibit less favorable attitudes to ads with lower levels of humorous content.

Another explanation for need for humor's influence on A d may be that it serves as an intensifier variable with respect to attitude formation. Analogously to NFC, NFH may differentiate those inclined to form attitudes based on the humor in the ad from those who form attitudes primarily on the basis of nonhumorous elements. Prior research supports this proposition. Two independent studies, Cline 1997 and Cline, Machleit, and Kellaris 1999, provide evidence that an individual's NFH can significantly influence the extent to which the humor in an ad is processed.

On the basis of humor's prevalence in advertising and the belief in its universal effectiveness (Madden and Weinberger 1984), NFH may be useful as a segmentation tool. For example, it may be helpful in identifying audiences who are more likely, under certain conditions, to respond favorably to humorous ads. The present findings indicate that NFH can differentiate subjects' attitudes under varying conditions of humor content. Thus, for advertisers whose objectives include attitude change, humor targeted at audiences high in NFH may produce favorable results. Clearly, it would not be practical to administer NFH scales to members of target audiences; however, market research may identify media vehicles that draw audiences characterized by high NFH (e.g., Mad Magazine readers, Comedy Central watchers, or The Onion for Web surfers). In addition, advertisers can use this information to determine which product categories or brands tend to be popular with specific media users (e.g., National Lampoon readers may tend to be heavy users of B-films or video games). Thus, NFH may be helpful both in media selection and in targeting audiences for specific products. The present research suggests that need for humor--until recently an ignored person trait--may be highly relevant, influential, and worthy of additional investigation.

Need for Cognition and Need for Humor

The role of NFC on the effects of humorous advertising was examined for the first time in conjunction with NFH. The findings from Studies 2 and 3 provide evidence that the moderating role of NFH on the effect of humor on [A.sub.ad] may be more pronounced among subjects with lower NFC. This, in fact, is consistent with the Lewinian perspective, according to which "the valence of environmental objects and the needs of the individual are correlative" (Lewin 1938, p. 78). An individual's NFC affects his or her motivation to process an ad, which in turn should influence the moderating role of other person traits, such as NFH. Moreover, in a nomological context, we find support for the notion that need for humor, need for cognition, and perceived humor influence attitudes by elevating affect and reducing counterarguments.

Limitations and Future Research

There are several features of this study that impose limitations on the generality of the findings. First, the joint impact of NFH and NFC should be further examined to determine the extent to which the two traits may be in conflict and the conditions under which the influence of one may overshadow the other. Does a high NFC force NFH into dormancy, or do these traits interact to enhance processing? A manipulation of involvement may shed some light on this question. It is possible, for example, that in high-involvement situations, a consumer's need for humor becomes less pertinent and, consequently, less influential than it would be under low involvement. Furthermore, an involvement manipulation would overcome the limitation that, in the present studies, forced exposure may have triggered artificially high involvement.

Additional research is also needed to explore the possibility that individuals higher in NFC may be more demanding of humor, that is, humor must bring forth additional elaboration to be successful (cf. Wyer and Collins 1992). In the present research, data from Study 3 indicate that those higher in NFC perceived significantly more (less) humor in the high (moderate) humor ad than those lower in NFC. This suggests that those higher in NFC may react negatively to relatively weak humor. In contrast, those low in NFC and high in NFH (and less demanding of the humor) may have found the humor sufficiently amusing and subsequently transferred positive affect to the ad.

Additional research is also needed to evaluate the possibility that individuals high in NFC require humor-claims relatedness for the humor to be successful. Manipulations of argument strength, humor relevancy, and measures of perceived humor would bring into clearer relief how people high in NFC process ads. To the extent that humor is integral to understanding the advertiser's proposition, those higher in NFC (regardless of their need for humor) may treat the humor on rational terms--like any other argument in the ad. It follows that if the humor enhances the ad claims, it may positively influence attitudes for those higher in NFC. In the present study, all three experiments used humor that was not integral to understanding the claims. Thus, individuals higher in NFC may have rejected its value, and hence the positive effects of humor on ad outcomes were attenuated. In contrast, individuals who are lower in NFC are less likely to evaluate the diagnosticity of the humor. And, if they are also higher in NFH, humor may be its own virtue. Thus, in the present study, humor's effects may have transferred to the ad.

Future research may help clarify the relationship between NFH and perceived humor. In Studies 1 and 2, which included only low and moderate levels of humor, the relationship between NFH and perceived humor was statistically independent (r = .08 n.s., and r = .02 n.s., respectively). Furthermore, perceived humor was tested as mediating the joint impact of humor and NFH. The underlying logic is that if the two-way interaction of humor (low/moderate) and NFH becomes nonsignificant (or "significantly" less significant) after variation due to perceived humor is removed, there is evidence that the joint impact of humor and NFH may be mediated by perceived humor. In both studies, however, the two-way interaction of humor and NFH remains significant in the presence of the covariate perceived humor. Thus, on the basis of these analyses, the mediation hypothesis can be ruled out. In contrast, Study 3 (which included low, moderate, and high levels of humor) reveals a significant correlation between NFH and perceived humor (r =.20, p < .01). (This is also shown in Figure 4.) Interestingly, the relationship becomes nonsignificant when the analysis is limited to low and moderate levels of humor only. This pattern is corroborated by the results of an ANOVA, which examined the joint interplay of humor strength and NFH on perceived humor. With three levels of humor (low, moderate, and high), the interaction is significant. In comparison, the same two-way ANOVA with only low and moderate humor treatments reveals a nonsignificant interaction.

Thus, there is preliminary support for the notion that NFH may operate through perceived humor--if the humorous stimuli are well received. Following Alden and Hoyer (2000), future research might examine the possibility that successful (versus failed) humor explains an additional source of variation in perceived humor as a mediator to NFH. Additional testing of the NFH scale also should be undertaken to assure that the scale adequately captures critical components of an individual's desire or craving for humor. For example, Alden and Hoyer (2000) find that ad content "playfulness" is an important moderator in the relationship between surprise and perceived humor. This suggests that one's appreciation of playfulness may indeed be an important part of one's need for humor. Finally, subjects in the present study received only one exposure to an ad. During an advertising campaign, consumers are likely to receive multiple exposures. Hence, a longitudinal study would facilitate testing a "wearout" effect of humor. Also, more research is needed to clarify the impact of humor and its moderators across various product types, buying situations, and humor types, dramatically, over time.

REFERENCES

Alden, Dana, Ashesh Mukherjee, and Wayne Hoyer (1993), "An Examination of Cognitive Factors Related to Humorousness in Television Advertising," Journal of Advertising, 22 (2), 29-38.

--, and -- (2000), "The Effects of Incongruity, Surprise, and Positive Moderators on Perceived Humor in Television Advertising," Journal of Advertising, 29 (2), 1-15.

Alrsech, Moses B., Thomas Cline, and James Kellaris (1999), "Can Humor Enhance the Impact of Strong Message Arguments?" in Proceedings of the Society for Consumer Psychology 2000 Winter Conference, J. Inman, ed., San Antonio, TX: American Psychological Association.

Arias-Bolzmann, Leopoldo, Goutam Chakraborry, and John C. Mowen (2000), "Effects of Absurdity in Advertising: The Moderating Role of Product Category Attitude and the Mediating Role of Cognitive Responses," Journal of Advertising, 29 (1), 35-49.

Bagozzi, Richard P. (1984), "A Prospectus for Theory Construction in Marketing," Journal of Marketing, 48 (Winter), 11-29.

--, and Youjae Yi (1988), "On the Evaluation of Structural Equation Models," Journal of the Academy of Marketing Science, 16 (1), 74-94.

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

Bender, Peter M. (1989), Structural Equation Program Manual, Los Angeles: BMPD Statistical Software.

--, and Chih-Pi Chou (1987), "Practical Issues in Structural Modeling," Sociological Methods and Research, 16, 78-117.

Brown, Steve P., and Douglas M. Stayman (1992), "Antecedents and Consequences of Attitude Towards Ad: A Meta-Analysis," Journal of Consumer Research, 19 (June), 34-51.

Cacioppo, John, and Richard Petty (1982), "The Need for Cognition," Journal of Personality and Social Psychology, 42 (1), 116-131.

--, --, and C. Kao (1984), "The Efficient Assessment of Need for Cognition," Journal of Personality Assessment, 48 (3), 306-307.

--, --, and K. Morris (1983), "Effects of Need for Cognition on Message Evaluation, Recall, and Persuasion," Journal of Personality and Social Psychology, 45, 805-818.

Chattopadhyay, Amitava, and Kunal Basu (1990), "Humor in Advertising: The Moderating Role of Prior Brand Evaluation," Journal of Marketing Research, 27, 466-476.

Cline, T. (1997), "The Role of Expectancy and Relevancy in Humorous Ad Executions: An Individual Difference Perspective," Ph.D. diss., University of Cincinnati.

--, and J. Kellaris (1999), "The Joint Impact of Humor and Argument Strength in a Print Advertising Context: A Case for Weaker Arguments," Psychology and Marketing, 16 (1), 69-86.

--, Karen Machleit, and J. Kellaris (1999), "Is There a Need for Levity?" in Proceedings of the Society for Consumer Psychology 1998 Winter Conference, K. A. Machleit and M. Campbell, eds., Austin: American Psychological Association.

--, --, and -- (2002), "The Effects of Need for Levity on Advertising Outcomes," working paper, St. Vincent College, Latrobe, PA.

Dee, Juliet (1993), "Joe Sedelmaier and the Creative Process: Here's the Beef," Journal of Popular Culture, 26 (4), 173-185.

Duncan, Calvin (1979), "Humor in Advertising: A Behavioral Perspective," Journal of the Academy of Marketing Science, 7, 285-306.

--, and J. Nelson (1985), "Effects of Humor in a Radio Advertising Experiment," Journal of Advertising, 14 (2), 33-40.

Festinger, Leon, and N. Maccoby (1964), "On Resistance to Persuasive Communication," Journal of Abnormal and Social Psychology, 68, 359-366.

Gerbing, David W., and John C. Anderson (1988), "An Updated Paradigm for Scale Development Incorporating Unidimensionality and Its Assessment," Journal of Marketing Research, 25 (May), 186-192.

Geuens, Maggie, and Patrick de Pelsmacker (1998), "Feelings Evoked by Warm, Erotic, Humorous or Non-Emotional Print Advertisements for Alcoholic Beverages," Academy of Marketing Science Review, 98 (1).

Haugtvedt, Curt, and Richard Petty (1992), "Personality and Persuasion: Need for Cognition Moderates the Persistence and Resistance of Attitude Changes," Journal of Personality and Social Psychology, 63, 308-319.

--, --, and J. Cacioppo (1992), "Need for Cognition and Attitude Change: Understanding the Role of Personality Variables in Advertising," Journal of Consumer Psychology, 1 (3), 239-260.

Heckler, Susan E., and Terry L. Childers (1992), "The Role of Expectancy and Relevancy in Memory for Verbal and Visual Information: What Is Incongruence? Journal of Consumer Research, 18, 475-492.

Isen, Alice (1984), "Toward Understanding the Role of Affect in Cognition," in Handbook of Social Cognition, vol. 3, S. Wyer and T. Srull, eds., Hillsdale, NJ: Lawrence Erlbaum, 179-236.

-- (1987), "Positive Affect, Cognitive Processes and Social Behavior," in Advances in Experimental Social Psychology, vol. 20, L. Berkowitz, ed., New York: Academic Press, 203-253.

--, Kimberly Daubman, and G. Nowicki (1987), "Positive Affect Facilitates Creative Problem Solving," Journal of Personality and Social Psychology, 52, 1122-1131.

Kahn, Barbara, and Alice Isen (1993), "The Influence of Positive Affect on Variety Seeking Among Safe, Enjoyable Products," Journal of Consumer Research, 20, 257-270.

Krishnan, Shanker, and Dipankar Chakravarti (1994), "Humorous Executions in Advertising: Facilitation and Interference Effects on Brand Claims Memory," working paper, Indiana University, Bloomington.

Lewin, Kurt (1938), A Dynamic Theory of Personality, New York: McGraw-Hill.

--(1951), Field Theory in Social Science, New York: Harper. Lutz, Richard (1985), "Affective and Cognitive Antecedents of Attitude Toward the Ad: A Conceptual Framework," in Psychological Processes and Advertising Effects: Theory, Research and Applications, L. Alwitt and A. Mitchell, eds., Hillsdale, NJ: Lawrence Erlbaum, 45-56.

MacKenzie, Scott, and Richard Lutz (1989), "An Empirical Examination of Structural Antecedents of Attitude Toward the Ad in an Advertising Pre-test Context," Journal of Marketing, 53, 48-65.

--, --, and George Belch (1986), "The Role of Attitude Toward the Ad as a Mediator of Advertising Effectiveness: Determinants and Consequences," Advances in Consumer Research, 23, 130-143.

Madden, Thomas J., Chris Allen, and Jacquelyn Twible (1988), "Attitude Toward the Ad: An Assessment of Diverse Measurement Indices Under Different Processing 'Sets,'" Journal of Marketing Research, 25, 242-252.

--, and Marc G. Weinberger (1982), "The Effects of Humor on Attention in Magazine Advertising," Journal of Advertising, 11, 8-14.

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

Muehling, Darrel D., Jeffrey Stoltman, and Sanford Grossbart (1990), "The Impact of Comparative Advertising on Levels of Message Involvement," Journal of Advertising, 16 (3), 41-50.

Olson, Judy M., and Neal J. Roese (1995), "The Perceived Funniness of Humorous Stimuli," Personality and Social Psychology Bulletin, 21 (9), 908-913.

Osgood, Charles, George Suci, and Percy Tannenbaum (1957), The Measurement of Meaning, Urbana: University of Illinois Press.

Petty, Richard, and John Cacioppo (1981), Attitudes and Persuasion: Classic and Contemporary Approaches, Dubuque, IA: Brown.

--, and -- (1984), "Source Factors and the Elaboration Likelihood Model of Persuasion," Advances in Consumer Research, 11, 668-672.

--, and -- (1986), Communication and Persuasion: Central and Peripheral Routes to Attitude Change, New York: Springer-Verlag.

Punj, Girish, and David Stewart (1983), "An Interaction Frame work of Consumer Decision Making," Journal of Consumer Research, 10, 181-196.

Rossirer, John, and Larry Percy (1997), Advertising Communications and Promotion Management, 2nd ed., New York: McGraw Hill, 229, 241-242.

Shimp, Terrance (1981), "Attitude Toward the Ad as a Mediator of Consumer Brand Choice," Journal of Advertising, 10 (2), 915.

Smith, Stephen M. (1993), "Does Humor in Advertising Enhance Systematic Processing?" in Advances in Consumer Research, vol. 20, L. McAlister and M. L. Rothschild, eds., Provo, UT: Association for Consumer Research, 155-158.

Sports, Harlen, Marc Weinberger, and Amy Parsons (1997), "Assessing the Use and Impact of Humor on Advertising Effectiveness: A Contingency Approach," Journal of Advertising, 26 (3), 17-33.

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

Stewart, D., and David Furse (1986), Effective Television Advertising, Lexington, MA: D.C. Heath.

Sutherland, Max, and Alice K. Sylvester (2000), Advertising and the Mind of the Consumer, Sydney, NSW, Australia: Allen and Unwin.

Thorson, James, and F. C. Powell (1991), "Measurement of Sense of Humor," Psychological Reports, 69, 691-702.

--, and -- (1993a), "Development and Validation of a Multidimensional Sense of Humor Scale," Journal of Clinical Psychology, 48, 13-23.

--, and -- (1993b), "Sense of Humor and Dimensions of Personality," Journal of Clinical Psychology, 49, 799-809.

Weinberger, Marc G., and Leland Campbell (1991), "The Use and Impact of Humor in Radio Advertising," Journal of Advertising Research, 31, 44-52.

--, and Charles Gulas (1992), "The Impact of Humor in Advertising: A Review," Journal of Advertising, 21, 36-59.

--, Harlen Spotts, Leland Campbell, and Amy L. Parsons (1995), "The Use of Humor in Different Advertising Media," Journal of Advertising Research, 35, 44-56.

Wicker, E W., I. M. Thorelli, W. L. Barron, III, and M. R. Ponder (1981), "Relationships Among Affective and Cognitive Factors in Humor," Journal of Research in Personality, 15, 350-370.

Williams, Larry J., and J. T. Hazar (1986), "Antecedents and Consequences of Satisfaction and Commitment in Turnover Models: A Reanalysis Using Latent Variable Structural Equation Methods," Journal of Applied Psychology, 71, 219-231.

Wu, Bob, Kenneth Crocker, and Martha Rogers (1989), "Humor and Comparatives in Ads for High and Low Involvement Products," Journalism Quarterly, 66, 653-661.

Wyer, Robert S., and James E. Collins (1992), "A Theory of Humor Elicitation," Psychological Review, 99, 663-688.

Yi, Y. (1990), "Cognitive and Affecrive Priming Effects of the Context for Print Advertisements," Journal of Advertising, 19 (2), 40-48.

Zhang, Yong (1996), "Responses to Humorous Advertising: The Moderating Effect of Need for Cognition," Journal of Advertising, 25, 15-32.

Thomas W. Cline (Ph.D., University of Cincinnati) is an associate professor of marketing, St. Vincent College, Alex G. McKenna School of Business, Economics, and Government.

Moses B. Altsech (Ph.D., Pennsylvania State University) is an assistant professor of marketing, Edgewood College, College of Business Administration.

James J. Kellaris (Ph.D., Georgia State University) is professor of marketing, University of Cincinnati, College of Business Administration.