425094

Embed Size (px)

Citation preview

  • 8/12/2019 425094

    1/8

    Journal of Consumer Research Inc.

    Humor in Television Advertising: A MomenttoMoment AnalysisAuthor(s): Josephine L. C. M. Woltman Elpers, Ashesh Mukherjee, and Wayne D. HoyerSource: Journal of Consumer Research, Vol. 31, No. 3 (December 2004), pp. 592-598Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/10.1086/425094.

    Accessed: 24/02/2014 07:04

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at.http://www.jstor.org/page/info/about/policies/terms.jsp

    .JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

    content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

    of scholarship. For more information about JSTOR, please contact [email protected].

    .

    The University of Chicago PressandJournal of Consumer Research, Inc.are collaborating with JSTOR to

    digitize, preserve and extend access toJournal of Consumer Research.

    http://www.jstor.org

    http://www.jstor.org/action/showPublisher?publisherCode=ucpresshttp://www.jstor.org/stable/10.1086/425094?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/10.1086/425094?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=ucpress
  • 8/12/2019 425094

    2/8

    592

    2004 by JOURNAL OF CONSUMER RESEARCH, Inc. Vol. 31 December 2004

    All rights reserved. 0093-5301/2004/3103-0011$10.00

    Humor in Television Advertising: A Moment-to-Moment Analysis

    JOSEPHINE L. C. M. WOLTMAN ELPERSASHESH MUKHERJEEWAYNE D. HOYER*

    Although humor represents a critical advertising technique around the world, pre-vious research has investigated only single point, retrospectively measured an-tecedents of perceived humor. Drawing on recent research indicatingthat moment-to-moment (MTM) responses have a significant effect on ad evaluations, weperform a MTM analysis of humor in television advertising. Results indicate thata dynamic transformation of surprise into humor lies at the heart of humor andthat key features of the MTM surprise and MTM humor traces drive overall per-ceptions of humor. We discuss the theoretical implications of these findings and

    outline promising directions for future research.

    H umor is one of the most widely used techniques inadvertising around the world, with about one out ofevery five television ads containing humorous appeals ( Al-den, Hoyer, and Lee 1993; Weinberger and Spotts 1989).However, few studies on this topic have addressed the factthat ads attempting humor vary dramatically in the level ofhumor they actually evoke in the target audience (Speck1991). While some ads are spectacularly successful at rais-ing a laugh, others may fail to do so. Such variation in

    perceived humor is likely to have important consequencesfor downstream variables of interest to marketers such asmessage credibility, recall, and attitude toward the ad andbrand (Chattopadhyay and Basu 1990; Shimp 1997; Wein-berger and Gulas 1992). Further, past research on humor inadvertising has typically used simple print manipulations of

    *Josephine L. C. M. Woltman Elpers is marketing and research directorat Verify Deutschland GmbH, Venloerstrasse 2527, 5062 Koln, Germany([email protected]). This article is based on research performedas part of the Ph.D. project of the first author at the Department of Mar-keting and Marketing Research, University of Groningen, The Netherlands.Ashesh Mukherjee is an associate professor at the Faculty of Management,McGill University, 1001 Sherbrooke Street West, Montreal, Quebec H3A1G5, Canada ([email protected]) . Wayne D. Hoyer is the James

    L. Bayless/W. S. Farish Fund Chair for Free Enterprise, Department ofMarketing, McCombs School of Business, One University Station, Uni-versity of Texas, Austin, TX 78712 ([email protected]).Financial support for this research project was provided by the NetherlandsAmerica Commission for Educational Exchange (NACEE), the Councilfor International Exchange of Scholars (CIES), the Center for CustomerInsight (McCombs School), and the Social Sciences and Humanities Re-search Council of Canada. All authors contributed equally, and the orderof authorship was determined by a random draw. Correspondence con-cerning this article may be addressed to any of the authors.The authorsthank Sicco Huisman, Billy Zahn, and Marianna Demori for their help indata collection.

    humor (e.g., cartoon drawings) in which the humorous ads(although statistically different from low humor ads) areonly mildly amusing at best. These manipulations of humor,although valid from the point of view of experimental con-trol, do not throw much light on the processes underlyingmore complex audiovisual forms of humor used in televisionadvertising. They also do not provide much insight into theworkings of really funny ads, which are the ones likely tohave significant effects on consumer-level outcomes in the

    real world.Consequently, researchers have recently begun building

    theory to explain how different elements of advertising con-tent affect the level of perceived humor. In particular, Alden,Mukherjee, and Hoyer (1999, 2000) have proposed a two-stage model of humor in television advertising. In the firststage of this model, the incongruity of the situation depictedin the ad generates feelings of surprise, with higher levelsof surprise resulting if the situation is familiar to the au-dience. Here, incongruity refers to the extent to which adcontent differs from generally expected beliefs, attitudes,and/or behaviors. Feelings of surprise are subsequentlytransformed into humor in the second stage of the model,when the incongruity is easily resolved, and when the ad is

    high in playfulness and warmth. It is worth noting that pre-vious tests of the Alden et al. (2000) model have used static,retrospectively measured predictors of perceived humor. Forexample, in the two studies reported by Alden et al. (2000),study participants provided overall judgments of variablessuch as surprise and humor at the end of each ad. Thesemeasures were then subjected to correlation analysis to val-idate the proposed model of humor. However, a more com-plete test of the Alden et al. (2000) model, which posits thatsurprise is dynamically transformed into humor, would re-

    This content downloaded from 95.76.197.1 on Mon, 24 Feb 2014 07:04:28 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 425094

    3/8

    MOMENT-TO-MOMENT ANALYSIS OF HUMOR 593

    quire analysis of individuals moment-to-moment (hence-forth MTM) responses on variables such as surprise andhumor. Such an MTM analysis of humor would also add torecent research investigating other MTM responses to ad-vertising, such as general affective reactions, perceived en-tertainment value, and perceived information value (e.g.,

    Baumgartner, Sujan, and Padgett 1997; Woltman Elpers,Wedel, and Pieters 2003).

    Thus, the purpose of the current research is to conduct adynamic, MTM analysis of humor in television advertising.Building on the model proposed by Alden et al. (2000) andon research related to the aggregation of MTM affectiveresponses into overall evaluations (e.g., Baumgartner et al.1997), we develop and test hypotheses about the effects oftwo MTM variables, namely, MTM surprise and MTM hu-mor, on overall perceived humor. The results make a con-tribution to existing research by providing a deeper under-standing of the process underlying the generation of humorin advertising, especially in terms of the key dynamic var-iables of MTM surprise and MTM humor.

    THEORETICAL DEVELOPMENT

    The contrast resolution theory of humor suggests that amoment-to-moment transformation of surprise into humoris the central process underlying this phenomenon (Aldenet al. 2000; Raskin 1985). Humorous episodes are said tofirst generate surprise by the presence of incongruous sit-uations. Surprise is then converted into humor when incon-gruities are resolved in a playful context. Research in non-advertising contexts using verbal humor such as jokes, puns,and satire also supports the notion that both surprise andhumor are generated during the course of humorous episodes( Herzog and Larwin 1988; Wyer and Collins 1992). Given

    the importance of the dynamic transformation process fromsurprise to humor, the key proposition of the present researchis that surprise and humor are likely to vary in an MTMmanner during the course of humorous advertising.

    MTM Surprise, MTM Humor, and OverallHumor

    Although the contrast resolution theory suggests that aMTM transformation of surprise into humor is the core pro-cess underlying humor, previous static tests of this theoryhave not verified the dynamic transformation of surprise intohumor during the course of the ad (e.g., Alden et al. 2000).If the dynamic surprise-humor transformation proceeds fromsurprise to humor, then the peak of the MTM surprise traceshould precede the peak of the MTM humor trace in thehigh overall humor ads, where peak refers to the highestlevel of the MTM trace. Such precedence of the peak ofMTM surprise, compared to the peak of MTM humor, isunlikely to be observed in low humor ads since the dynamicsurprise-humor transformation is less likely to occur in lowhumor ads. These arguments are summarized as follows:

    H1: A characteristic of ads that generate high levels

    of overall humor is that the peak of MTM surpriseprecedes the peak of MTM humor.

    If surprise is dynamically transformed into humor duringthe course of humorous ads, then the later the peak of MTMsurprise in an ad, the higher should be the peak of MTM

    humor. The longer the time period between the start of thead and the point of resolution of the incongruity (i.e., thepeak of surprise), the longer the window of opportunity forsurprise to be transformed into humor. As a result, drawingout the time before the peak of MTM surprise should in-crease the amount of time during which surprise is trans-formed into humor, leading to a higher peak of MTM humor.This proposition is consistent with a large body of researchon event schemas and story structures (e.g., Brewer andLichtenstein 1981) suggesting that storylines that build upsuspense over an extended period of time are preferred tothose that have a shorter suspense span. Similarly, Loew-enstein (1987) and Loewenstein and Prelec (1993) havefound that people prefer delayed over instant gratification

    when faced with a bounded sequence of events. The pre-ceding arguments are summarized as follows:

    H2: The later the peak of MTM surprise in the ad, thehigher the peak of MTM humor.

    MTM Humor and Overall Humor

    The previous hypotheses examined the process underlyingthe transformation of MTM surprise into MTM humor,which in turn influences the level of overall perceived hu-mor. An implicit assumption in the earlier discussion wasthat MTM humor and overall humor have a one-to-one re-lationship, whereby all MTM humor responses during the

    course of the ad are mentally summed into an overall humorrating. However, research on extended hedonic sequencessuggests that certain salient features of the MTM trace, suchas the peak, final moment, and linear trend, may be moreimportant in shaping overall evaluations. Here, peak refersto the highest level, final moment refers to the ending level,and linear trend refers to the first order slope of the MTMtrace (Baumgartner et al. 1997; Loewenstein and Prelec1993; Ross and Simonson 1991; Varey and Kahneman1992). For example, Ross and Simonson (1991) showed thatevents that end on a happy note (i.e., with high peak andfinal moment) are evaluated more favorably than those thatdo not, even if the latter sequence generates more positivefeelings during its early stages. Further, individuals have

    been shown to prefer an increasing (over decreasing) trendin wages (Loewenstein and Sicherman 1991) and an in-creasing (over decreasing) trend in academic performanceover time (Hsee and Abelson 1991).

    More directly relevant to the current research, Baum-gartner and associates (1997) found that the peak, final mo-ment, and positive linear trend of MTM affective responses(i.e., momentary like/dislike reactions) had a significant ef-fect on overall ad liking. Further, they found that peaks andfinal moments had a greater impact on overall ad liking than

    This content downloaded from 95.76.197.1 on Mon, 24 Feb 2014 07:04:28 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 425094

    4/8

    594 JOURNAL OF CONSUMER RESEARCH

    the sum of momentary MTM affective responses to the ad.In the present case, humor is a more complex affectivereaction than the simple like/dislike affect examined byBaumgartner and associates (1997), since humor involves atwo-stage appraisal of stimuli. Hence, it is not clear if theeffects of peak, final moment, and linear trend obtained by

    Baumgartner and associates (1997) for general like/dislikeMTM affect will also emerge in the case of MTM humor.We investigate this issue by advancing the following hy-potheses, which are based on the results of Baumgartnerand associates (1997):

    H3: The higher the peak/final moment of MTMhumor,the higher the level of overall humor.

    H4: The peak/final moment of MTM humor predictsthe level of overall humor better than the sum ofindividual MTM humor responses.

    H5: The greater the magnitude of the positive linear

    trend in MTM humor, the higher the level of over-all humor.

    STUDY

    Stimuli

    A sample of 30 ads was used as stimuli in this study.Fifteen of these ads were humorous in intent, while the otherfifteen ads generated positive affect but were not intendedto be funny. Two major criteria were used to select thestimulus ads. First, to ensure variance in the focal variableof humor, it was considered important to select ads thatspanned the entire range of perceived humor from low to

    high. Second, to ensure that familiarity with the ad did notconfound humor ratings, it was also considered importantto select ads that were relatively unfamiliar to the audience.

    These criteria were met by selecting ads through a two-stage stratified sampling process. In the first stage, a numberof ads that were potentially high on humor were selectedfrom a commercially available collection of English-lan-guage ads from around the world (www.adfilms.com). Someof these ads were extremely funny, more so than regularads seen on network television. Further, none of these adshad been aired in commercial broadcasts in the UnitedStates, thus making it unlikely that participants in the presentstudy would be familiar with them. Next, a set of intended-humor ads that were potentially high or medium on humor

    was selected from a pool of ads recorded off the air inCanada, which made it unlikely that the American partici-pants in the study would have seen them. In a pretest, 15student participants rated the above ads on a single item,three-point scale of perceived humor (not at all/somewhat/very funny). Based on their responses, 15 ads were randomlychosen such that perceived humor varied from medium(somewhat funny) to high (very funny).

    A second set of 15 nonintended humor but positive affectads was selected by further reviewing the pool of ads re-

    corded off the air in Canada. A subset of ads featuringpositive affective cues such as pleasant scenery, catchy mu-sic, and attractive models was first identified. Then, in apretest, 15 student participants rated this subset of ads on adichotomous scale of intended humor (intended/not intendedto be funny) and a single-item, three-point scale of positive

    affect toward the ad (dislike a lot/neither like nor dislike/like a lot). Fifteen ads that were classified as nonintendedhumor and high positive affect were randomly chosen foruse in the study. The total set of 30 stimuli ads representeda range of product categories including soft drinks, beer,music, credit cards, cars, sauces, candy bars, perfumes, ho-tels, films, cameras, sport shoes, condoms, balls, crackers,mobile phones, and clothes.

    Procedure

    Data was collected from undergraduate students at a majoruniversity in the southwestern United States. MTM surpriseand MTM humor were measured from two separate groups

    of 25 undergraduate student participants (see Larsen andFredrickson 1999). MTM measures were assessed using acomputer-based procedure validated by Baumgartner andassociates (1997; see also Woltman Elpers et al. 2003). Thisprocedure allowed participants to record their MTM re-sponses to the stimuli ads, using a mouse, on a continuousscale (0600), anchored by not surprising at all/0 to verysurprising/600 (for MTM surprise) and not funny at all/0 to very funny/600 (for MTM humor). Participantswatched the stimuli ads on a large-screen TV in two differentrandom orders to guard against order effects. In order tominimize fatigue, the ads were shown in sets of ten, witha 5-min. break in between sets. To promote fluency in re-sponding to the ads, participants were first given an oppor-

    tunity to practice with two starter ads before commencingtheir MTM responses to the stimuli ads. The MTM datacollection took less than an hour to complete, and academiccredit was given for participation.

    Two other groups of 25 student participants from the sameuniversity provided overall assessments of the 30 ads. Onegroup provided overall assessments of humor using a seven-point scale of overall humor (not at all/very funny, alpha p.94). The other group provided overall assessments of adattitude (dislike/like the ad a lot, alpha p .85), ad familiarity(very unfamiliar/familiar ad, alpha p .79), and brand fa-miliarity (very unfamiliar/familiar brand, alpha p .98). Theads were shown on the same large-screen TV in two randomorders, and questions were asked on the computer screen

    immediately after each commercial. This data collection alsotook less than an hour, and academic credit was given forparticipation.

    Results

    Consistent with previous research on MTM responses toadvertising (e.g., Baumgartner et al. 1997; VandenAbeeleand MacLachlan 1994), a net MTM trace for each ad wasconstructed by averaging MTM responses across respon-

    This content downloaded from 95.76.197.1 on Mon, 24 Feb 2014 07:04:28 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 425094

    5/8

    MOMENT-TO-MOMENT ANALYSIS OF HUMOR 595

    TABLE 1

    TIME POINTS OF THE PEAKS OF MTM SURPRISE AND MTM HUMOR

    Ad identificationOverall humor

    rating

    Time at whichpeak in MTM

    surprise occurs

    Time at whichpeak in MTMhumor occurs

    Time differencebetween peaks

    Humorous ads:1 4.68 40 44 42 5.24 61 68 73 4.8 28 28 04 5.32 43 59 167 5.16 39 39 08 4.52 43 41 211 5.84 35 35 016 6.56 31 34 318 4.8 35 36 119 4.56 16 17 120 5.08 57 56 122 5.36 51 57 624 5.36 41 41 027 5.16 56 59 330 5.12 31 31 0

    Nonhumorous ads:5 3.8 39 42 36 1.84 26 7 199 4.24 51 52 110 4.04 23 28 512 2.88 47 42 513 4.24 31 40 914 4.12 26 30 415 4.12 30 29 117 3.44 27 25 221 4.32 24 29 523 4.32 27 28 125 2.6 22 21 126 3.48 12 9 328 2.52 40 11 2929 3.48 52 45 7

    dents and taking the ad as the unit of analysis (all alphas 10.90). Confirming the success of the stratified sampling pro-cedure, the humorous ads were considered funnier (M p5.2 vs. 3.6, t(28) p 6.66, p ! .001) and generated morepositive ad attitude (M p 5.1 vs. 4.2, t(28) p 5.19, p !.001) than the nonhumorous ads. The two groups of ads didnot differ on ad familiarity (Mp 2.6 vs. 2.1, t(28) p 1.22,p ! .23) or brand familiarity (M p 3.9 vs. 4.3, t(28) p0.49,p ! .63). The ads scored 2.4 on average on the seven-point ad-familiarity scale, which was significantly less thanthe midpoint (i.e., 4.0) of the scale (t(29) p 7.72,p !.001),indicating that the test ads were relatively unfamiliar to

    participants.According to hypothesis 1, the peak in MTM surprise

    precedes the peak in MTM humor in humorous ads but notin nonhumorous ads. Table 1 shows for each ad the pointsin time at which the peak in MTM surprise occurs and thepoints in time at which the peak in MTM humor occurs. Asshown in the table, the peak in the MTM surprise tracepreceded the peak in the MTM humor trace or occurredsimultaneously in 13 out of 15 humorous ads. In contrast,the peak in the MTM surprise trace occurred after the peak

    in MTM humor in 8 out of the 15 nonhumorous ads. Inaddition, the overall humor rating was significantly higherin ads where the peak of MTM surprise preceded peak ofMTM humor (or where the peaks occurred at the same time),compared to ads where this was not the case (Mp 4.9 vs.3.4, t(28) p 4.75, p ! .001).1 These results are supportiveof hypothesis 1.

    Hypothesis 2 posited that the later the peak of MTMsurprise, the higher the peak of MTM humor. Consistentwith this hypothesis, the correlation between the peak in theMTM humor trace and the time to peak in the MTM surprise

    trace was found to be significantly positive (rp

    .43, t(28)p 2.55,p ! .05). Consistent with hypothesis 3, it was foundthat the peak (rp .87, t(28) p 9.28, p ! .001) and finalmoment (rp .91,t(28) p 11.25,p !.001) both correlatedstrongly with the overall humor rating. In support of hy-

    1Since three of the commercials could be regarded as outliers in termsof time difference between the peak of MTM surprise and the peak ofMTM humor, the data were also analyzed without these three data points.The effect was found to be similar after deletion of these ads from theanalysis (Mp 4.8 vs. 3.7, t(25) p 3.66, p ! .01).

    This content downloaded from 95.76.197.1 on Mon, 24 Feb 2014 07:04:28 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 425094

    6/8

    596 JOURNAL OF CONSUMER RESEARCH

    pothesis 4, the correlation between the peak in MTM humorand overall humor was significantly greater than the cor-relation between the sum of the individual humor responsesand overall humor (rp .87 vs. .57;t(27) p 5.53,p ! .001;see Steiger 1980). Also consistent with hypothesis 4, thecorrelation between the final moment of MTM humor and

    overall humor was significantly greater than the correlationbetween the sum of individual humor responses and overallhumor (rp .91 versus .57;t(27) p 7.32,p ! .001). Finally,consistent with hypothesis 5, the linear trend of MTM humorwas positively correlated with the overall humor rating (rp.77, t(28) p 6.41, p ! .001).

    GENERAL DISCUSSION AND FUTURE

    RESEARCH

    Humor is one of the most popular advertising techniquesaround the world. However, previous research on humor inadvertising has focused mostly on downstream effects of

    humor such as message credibility, recall, and attitude to-ward the ad and brand (Chattopadhyay and Basu 1990;Weinberger and Gulas 1992). In contrast, we know relativelylittle about the antecedents of perceived humor, or whatmakes an ad funny. In this context, the results of the presentresearch make several important contributions to theliterature.

    First, our research provides a more detailed understandingof the mechanism underlying humor through a more com-plete test of the Alden and associates (1999, 2000) modelof humor in television advertising, using dynamic, MTMmeasures of the central variables of surprise and humor. Theresults of our study confirm that in order to generate a hu-morous response, a sequential dynamic transformation that

    proceeds from surprise to humor must occur. The peak ofMTM surprise preceded the peak of MTM humor in virtuallyall the humorous ads but in relatively few of the nonhu-morous ads. Further, the overall humor rating was signifi-cantly higher in ads where MTM surprise preceded MTMhumor than it was in ads where they did not. These resultsprovide the first empirical test of the core process underlyingthe generation of humor in television advertising. Notably,the stimuli ads used in our research were selected using astratified sampling approach that captured a full range ofhumorous ads, from plainly unfunny to hilarious. As a result,the MTM analysis reported in this article is likely to beuseful in understanding the mechanics underlying real-worldtelevision advertising, including the extremely funny ads

    that marketers are always attempting to create.Second, the findings expand our understanding of how

    structural elements within the ad can increase the effective-ness of humorous ads. For example, examination of theMTM surprise-humor transformation process revealed thata late peak of MTM surprise was an important element ingenerating a humorous response. This is a novel insight sinceprevious research has been silent on desirable temporal char-acteristics of surprise generated within humorous ads. In oursample, we found that the average time to peak of MTM

    surprise in the top 10 humorous ads was about 35 sec. Giventhat the average total length of these ads was approximately40 sec., an emergent empirical guideline for humorous adswould be to engineer a surprise peak approximately 90%into the ad. Further, our results converge with Brewer andLichtenstein (1981), Loewenstein (1987), and Loewenstein

    and Prelec (1993), indicating that individuals may have ageneral tendency to prefer postponement (and savoring) ofexpected future rewards such as revelation of mystery in anovel or laughter at the end of an ad.

    Third, our findings replicate and extend previous researchon the integration of general MTM affective responses toadvertising (e.g., Baumgartner et al. 1997) in the specificcase of humor. Our results indicate that key features of MTMhumor, such as peak and final moment, predict overall ret-rospective judgments of humor better than the integratedsum of MTM humor responses. These results show that theinitial findings in the work of Baumgartner and associates(1997) for simple ad affect also holds in the case of a morecomplex affective reaction to advertising, namely, humor.Given this coincidence in the pattern of results, it is possibleto speculate that other affective reactions to advertising (e.g.,

    joy, fear, empathy) may also show similar relationships be-tween MTM reactions and overall evaluations. However,further research is needed to test this proposition. Morebroadly, our results at the subminute commercial level con-verge with previous research that has found similar relianceon key features, rather than on the sum of all experiences,in hedonic sequences extending over days (e.g., Loewenstein1987), months (e.g., Dube and Morgan 1996), and years(e.g., Loewenstein and Sicherman 1991). These accumu-lating findings indicate that individuals may have a generaltendency to rely on summary features of hedonic sequences

    rather than on the more normatively important sum of in-dividual experiences in the sequence (Fredrickson and Kah-neman 1993; Redelmeier and Kahneman 1996; Varey andKahneman 1992).

    The Alden and associates (1999, 2000) model impliesthat the transformation of MTM surprise to MTM humor,and therefore the level of overall perceived humor, is mod-erated by ease of resolution, playfulness, and warmth of thead. Hence, future research should experimentally investigatethe moderating effects of ease of resolution, warmth, andplayfulness at the MTM level. Ongoing research by thecurrent authors has taken a first step in this direction bymanipulating ease of resolution through the level of cog-nitive load, operationalized by a digit-memorization task

    (e.g., Shiv and Fedorikhin 1999; Swann, Hixon, and Stein-Seroussi 1990). Specifically, participants in a pilot studywere asked either to keep an eight-digit number in memorywhile viewing the test ads (i.e., low ease of resolution) orto view the test ads without any memorization task (i.e.,high ease of resolution). Consistent with the notion that easeof resolution facilitates the transformation of surprise intohumor, hypotheses 1 and 2, which are predicated on thesuccessful transformation of MTM surprise into MTM hu-mor, were supported when ease of resolution was high but

    This content downloaded from 95.76.197.1 on Mon, 24 Feb 2014 07:04:28 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 425094

    7/8

    MOMENT-TO-MOMENT ANALYSIS OF HUMOR 597

    not when ease of resolution was low. Similarly, it was foundthat an index of the surprise-humor transformation, namelythe correlation between peak of MTM surprise and peak ofMTM humor, was greater in the high ease of resolutioncondition than it was in the low ease of resolution condition.Future research should extend this line of investigation by

    examining the moderating roles of playfulness and warmththrough manipulations of these variables in the ad contextor ad content. In addition, future research can model theeffects of multiple antecedents of humor (e.g., surprise, easeof resolution, playfulness) using a regression approach,which would allow researchers to isolate the partial andinteractive effects of these antecedents on perceived humor.

    The goal of the present research was to systematicallyinvestigate the process underlying the generation of humorin advertising. Hence, participants in our study were in-structed to watch the stimuli ads in their entirety while eitherproviding their MTM responses during the ads or their over-all evaluations at the end of the ads. However, when con-

    sumers watch advertising in the real world, they have theoption of switching the channel during the course of the adif the ad does not appeal to them. Thus, to be effective, real-world humorous ads may need to introduce a playfulnesssignal relatively early in the ad, so that viewers are inducedto watch the entire ad. Recent research (Woltman Elpers etal. 2003) has used MTM measures of ad entertainment valueand ad informativeness to predict when viewers are likelyto discontinue watching television advertising in general.Building on this research, future studies should investigatedeterminants of viewer switch-out in the case of humorousads, including factors such as the point of introduction ofthe playfulness signal and the viewers level of involvementwith the brand.

    In the present research, we investigated the effects ofMTM responses on overall humor and showed that higheroverall humor scores were related to a more positive attitudetoward the ad. However, ads that have a high score on overallperceived humor or attitude toward the ad are not necessarilyeffective in terms of recall, recognition, claim credibility,brand attitude, or choice (Weinberger and Campbell 1991).Thus, future research should investigate the effect of MTMhumor on key downstream variables, such as the credibilityof brand claims, brand memory, and brand attitudes. Futureresearch should also investigate moderators of MTM effectson downstream variables such as the relevance of humor tobrand benefits, expectations of humor for the brand/product

    category, and the time point of introduction of brand in-formation in the ad (see Chattopadhyay and Basu 1990;Howard and Barry 1994). Finally, in the current research,we do not provide any execution guidelines for the natureof scenes that can generate high levels of humor. Followingthe procedure used by VandenAbeele and MacLachlan(1994), future research could content analyze scenes in adswhere MTM humor reaches its peak/final moment. Suchanalysis would complement the MTM pattern results re-ported in this article and provide advertisers with concrete

    ad design strategies to increase the level of overall perceivedhumor.

    [Dawn Iacobucci served as editor and DurairajMaheswaran served as associate editor for this article.]

    REFERENCES

    Alden, Dana L., Wayne D. Hoyer, and Chol Lee (1993), Identi-fying Global and Culture-Specific Dimensions of Humor inAdvertising: A Multinational Analysis, Journal of Market-ing, 57 (April), 6475.

    Alden, Dana L., Ashesh Mukherjee, and Wayne D. Hoyer (1999),Extending a Contrast Resolution Model of Humor in Tele-vision Advertising: The Role of Surprise, Humor: Interna-tional Journal of Humor Research, 12 (1), 1522.

    (2000), The Effects of Incongruity, Surprise and PositiveModerators on Perceived Humor in Television Advertising,

    Journal of Advertising, 29 (2), 116.Baumgartner, Hans, Mita Sujan, and Dan Padgett (1997), Patterns

    of Affective Reactions to Advertisements: The Integration ofMoment-to-Moment Responses into Overall Judgments,Journal of Marketing Research, 34 (May), 21932.

    Brewer, William F. and Edward H. Lichtenstein (1981), EventSchemas, Story Schemas and Story Grammers, in Attentionand Performance,Vol. 9, ed. John Long and Alan Baddeley,Hillsdale, NJ: Lawrence Erlbaum, 36379.

    Chattopadhyay, Amitava and Kunal Basu (1990), Humor in Ad-vertising: The Moderating Role of Prior Brand Evaluation,

    Journal of Marketing Research, 27 (November), 46676.Dube, Laurette and Michael S. Morgan (1996), Trend Effects and

    Gender Differences in Retrospective Judgments of Consump-tion Emotions,Journal of Consumer Research, 23 (Septem-ber), 15662.

    Fredrickson, Barbara L. and Daniel Kahneman (1993), DurationNeglect in Retrospective Evaluations of Affective Episodes,

    Journal of Personality and Social Psychology, 65 (July),4555.

    Herzog, Thomas R. and David A. Larwin (1988), The Appreci-ation of Humor in Captioned Cartoons, Journal of Psy-chology, 122 (November), 597607.

    Howard, Daniel J. and Thomas E. Barry (1994), The Role ofThematic Congruence between a Mood-Inducing Event andan Advertised Product in Determining the Effects of Moodon Brand Attitudes,Journal of Consumer Psychology, 3 (1),127.

    Hsee, Christopher K. and Robert P. Abelson (1991), VelocityRelation: Satisfaction as a Function of the First Derivative ofOutcome over Time,Journal of Personality and Social Psy-chology, 60 (March), 34147.

    Larsen, Randy J. and Barbara L. Fredrickson (1999), Measure-

    ment Issues in Emotion Research, in Well-Being: The Foun-dations of Hedonic Psychology, ed. Daniel Kahneman, EdDiener, and Norbert Schwarz, New York: Russell Sage Foun-dation, 4160.

    Loewenstein, George (1987), Anticipation and the Valuation ofDelayed Consumption, Economic Journal, 97 (September),66684.

    Loewenstein, George and Drazen Prelec (1993), Preferences forSequences of Outcomes, Psychological Review, 100 (Janu-ary), 91108.

    Loewenstein, George and Nachum Sicherman (1991), Do Workers

    This content downloaded from 95.76.197.1 on Mon, 24 Feb 2014 07:04:28 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 425094

    8/8

    598 JOURNAL OF CONSUMER RESEARCH

    Prefer Increasing Wage Profiles?Journal of Labor Econom-ics, 9 (1), 6784.

    Raskin, Victor (1985), Semantic Mechanisms of Humor, Boston:D. Reidel.

    Redelmeier, Donald A. and Daniel Kahneman (1996), PatientsMemories of Painful Medical Treatments: Real-Time and Ret-rospective Evaluations of Two Minimally Invasive Procedu-res, Pain, 66 (July), 38.

    Ross, William T. and Itamar Simonson (1991), Evaluations ofPairs of Experiences: A Preference for Happy Endings,Jour-nal of Behavioral Decision Making, 4, 27382.

    Shimp, Terence A. (1997), Advertising, Promotion and Supple-mental Aspects of Integrated Marketing Communications,Fort Worth, TX: Dryden.

    Shiv, Baba and Alexander Fedorikhin (1999), Heart and Mind inConflict: The Interplay of Affect and Cognition in ConsumerDecision-Making,Journal of Consumer Research, 26 (De-cember), 27892.

    Speck, Paul S. (1991), The Humorous Message Taxonomy: AFramework for the Study of Humorous Ads, in Current Is-sues and Research in Advertising, Vol. 13, ed. James Leighand Claude R. Martin, Jr., Ann Arbor: Michigan Business

    School.Steiger, James H. (1980), Tests for Comparing Elements of a

    Correlation Matrix, Psychological Bulletin, 87(2), 24551.Swann, William B., Jr., Gregory J. Hixon, and Alan Stein-Seroussi

    (1990), The Fleeting Gleam of Praise: Cognitive Processes

    Underlying Behavioral Reactions to Self-Relevant Feedback,Journal of Personality and Social Psychology, 59 (July),1726.

    VandenAbeele, Piet and Douglas L. MacLachlan (1994), ProcessTracing of Emotional Responses to TV Ads: Revisiting theWarmth Monitor, Journal of Consumer Research, 20

    (March), 586600.Varey, Carol and Daniel Kahneman (1992), Experiences Extendacross Time: Evaluation of Moments and Episodes, Journalof Behavioral Decision Making, 5 (JulySeptember), 16985.

    Weinberger, Marc G. and Leland Campbell (1991), The Use andImpact of Humor in Radio Advertising, Journal of Adver-tising Research, 31 (December/January), 4452.

    Weinberger, Marc G. and Charles S. Gulas (1992), The Impactof Humor in Advertising: A Review, Journal of Advertising

    Research, 31 (December/January), 4452.Weinberger, Marc G. and Harlan E. Spotts (1989), Humor in U.S.

    versus U.K. TV Advertising,Journal of Advertising, 18 (2),3944.

    Woltman Elpers, Josephine L. C. M., Michel Wedel, and Rik G.M. Pieters (2003), Why Do Consumers Stop Watching TV

    Commercials? Two Experiments on the Influence of Moment-to-Moment Entertainment and Information Value,Journal of

    Marketing Research, 40 (November), 43753.Wyer, Robert S. and James E. Collins (1992), A Theory of Humor

    Elicitation,Psychological Review, 99 (October), 66388.

    This content downloaded from 95.76.197.1 on Mon, 24 Feb 2014 07:04:28 AMAll bj JSTOR T d C di i

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp