Anda di halaman 1dari 14

THE INFLUENCE OF HUMOR STRENGTH AND HUMOR–MESSAGE

RELATEDNESS ON AD MEMORABILITY
A Dual Process Model
Thomas W. Cline and James J. Kellaris

ABSTRACT: This research examines contingencies that shape the effects of humorous appeals on consumers’ recall of
ads, as well as the processes underlying such effects. Results of experimentation show that ads are more memorable when
humor is both strong and related to the message, and this interaction is mediated by attention and mood. Stronger humor
appeals also induce higher recall among individuals with a high “need for humor” (NFH).

Following the early research on humor in advertising (e.g., that humorous ads outperform nonhumorous ads on sustained
Duncan 1979; Speck 1987; Sternthal and Craig 1973), numer- attention, even after controlling for initial attention.
ous studies have attempted to explain the relationship between
humor and ad outcomes (e.g., Alden and Hoyer 1993; Alden, HUMOR STRENGTH
Mukherjee, and Hoyer 2000; Chattopadhyay and Basu 1990;
Cline, Altsech, and Kellaris 2003; Krishnan and Chakravarti Although some prior studies examined effects of humor’s pres-
2003; Lee and Mason 1999; Spotts, Weinberger, and Parsons ence (versus absence) in an ad (e.g., Lee and Mason 1999), the
1997; Weinberger and Campbell 1991). Nevertheless, a lack of impact of humor should depend in part on its strength (e.g.,
systematic empirical results contrasts with humor’s widespread Alden, Mukherjee, and Hoyer 2000). Our use of the term humor
use and an intuitive sense among advertising practitioners strength refers to the extent of humor elicitation provoked by an
that humor enhances ad persuasion (Madden and Weinberger ad (Wyer and Collins 1992). Humor strength can be thought
1984). Accordingly, the goal of our research is to provide a of as “how funny” an ad is, rather than merely referring to
clearer picture of humor’s role in advertising by examining whether or not an ad employs humor (e.g., Elpers, Mukherjee,
important stimulus factors and personality traits that shape and Hoyer 2004). Hence, in the present research, we vary the
the effects of humorous appeals on consumers’ recall, as well strength of humor rather than its mere presence/absence.
as the processes underlying such effects.
HUMOR RELATEDNESS
HUMOR AND ATTENTION
Weinberger and Gulas (1992) argue that controlling for the
Although some of what is currently known about humor’s ef- relatedness of humor makes experimental findings unanimous
fects on advertising response may be equivocal, a strong case in their support of humor’s positive impact on attention. In
can be made for humor’s impact on attention. The majority fact, the relatedness of humor to the product or message may
of studies conducted in industry as well as in laboratory set- also be a strong predictor of the success of an ad. Madden
tings bear this out. Madden and Weinberger (1982) find that (1982) finds that a radio commercial with product-related
humorous magazine ads outperform nonhumorous ads on three humor is perceived as more interesting than one in which the
common surrogates for attention. Weinberger et al. (1995) humor is unrelated to the product. Weinberger and Campbell
find evidence that humor is directly linked to attention and (1991) define related humor as being linked to the product and
recognition. Spotts, Weinberger, and Parsons (1997) show the fabric of the commercial. Results of their study show that
that humor enhances initial attention, aids brand recall, and related humor offers recall advantages over unrelated humor
holds attention. In a laboratory setting, Speck (1987) finds for high-involvement/feeling goods. Weinberger et al. (1995)

Thomas W. Cline (Ph.D., University of Cincinnati) is an associate


The authors thank Karen Machleit and Steve Posavac for their helpful
professor of marketing, Alex G. McKenna School of Business, Eco-
comments and suggestions on earlier versions of this work. This research
nomics, and Government, St. Vincent College. was supported in part by a CoB Research Fellowship at the University of
James J. Kellaris (Ph.D., Georgia State University) is a professor of Cincinnati. A portion of the work was conducted while the second author
marketing, College of Business, University of Cincinnati. was visiting at Bond University, Queensland, Australia.

Journal of Advertising, vol. 36, no. 1 (Spring 2007), pp. 55–67.


© 2007 American Academy of Advertising. All rights reserved.
ISSN 0091-3367 / 2007 $9.50 + 0.00.
DOI 10.2753/JOA0091-3367360104
56 The Journal of Advertising

focus specifically on thematic relatedness and find a positive ef- MEDIATING ROLE OF MOOD
fect for related humor. Spotts, Weinberger, and Parsons (1997)
use Speck’s (1991) typology to categorize humor relatedness. Research shows that a positive mood can facilitate recall (Isen
They suggest that the most effective type of relatedness to use 1987) and lead to greater receptiveness to persuasive commu-
in an ad depends on the product category. Speck (1987, 1991) nications (Galizio and Hendrick 1972). Bower and colleagues
outlines a broad typology of humor that incorporates the re- (Bower, Gilligan, and Monteiro 1981) provide insight into the
latedness of humor in an ad on three levels: (1) pragmatic, (2) processes through which mood states can bias cognitive activ-
thematic, and (3) structural. Pragmatic relatedness refers to ity. They demonstrate that people better remember material
the hierarchy of the humor–message relationship (i.e., humor- that is congruent with their mood states at the time of encod-
dominant or nonhumorous). Semantic (thematic) relatedness ing (“mood-congruent hypothesis”). These mood-congruent
describes the relationship between the humor and the product. memory effects are echoed in Isen (1978), who found that
Structural (syntactic) relatedness refers to the syntactic place- people in a positive mood during retrieval recalled more posi-
ment of the humor within the ad, that is, whether the humor tive traits, whereas those in a negative mood recalled more
is meaningful to the message. negative traits. Isen (1989) goes beyond the notion of mood-
In one of the few laboratory studies to address the issue congruent message elaboration, arguing that positive mood
of humor relatedness to brand claims, Krishnan and Chakra- states may broaden cognitive organization (i.e., they encourage
varti (2003) draw on Alba, Hutchison, and Lynch (1991) and people to categorize a wider range of stimuli together) and,
Meyers-Levy (1991) to define humor relevance as the degree of consequently, promote more integrated and flexible thinking.
pertinence of the execution to the brand claims. This definition This cognitive flexibility, in turn, may enhance the processing
dovetails with Heckler and Childers’s (1992) conceptualiza- of verbal content in an ad. Thus, it is possible that positive
tion of relevancy as material pertaining directly to the meaning moods increase elaboration and subsequent recall of the mes-
of the primary message (e.g., brand claims). Krishnan and sage. Isen (1989) also suggests that positive affect may promote
Chakravarti (2003) demonstrate that the relevancy of humor the use of heuristics or reliance on peripheral cues.
to the brand claims can positively influence memory for brand Humor is an ad feature that is likely to induce good moods.
claims. They argue that humorous executions strongly linked Littmann (1983) cites joy as one of the principal effects of
to some or all other ad components (e.g., brand claims) produce humor. Wicker et al. (1981) find a number of emotion-related
facilitation effects. scales to be significantly correlated with perceived funniness.
Kellaris, Cox, and Cox (1993) provide a conceptual frame- Olson and Roese (1995) find support for the contention that
work for understanding humor’s attention-gaining value perceivers use their own reactions to humorous stimuli (e.g.,
on advertising outcomes. They find that attention-gaining mirth), together with information about the environment, to
background music enhances message reception when the infer the emotion-eliciting qualities of humor. Thus, if par-
music evokes message-congruent (as opposed to incongruent) ticipants express joy, they are likely to infer funniness. O’quin
thoughts. By analogy, humor may operate similarly. Kellaris, and Aronoff (1981) find that humor lessens self-reported ten-
Cox, and Cox’s (1993) notion of congruity corresponds with sion and increases enjoyment of a task. Smith (1993) argues
Heckler and Childers’s (1992) concept of relevancy. Thus, it that humor in an ad can act to enhance the consumer’s mood,
seems likely that humor’s attention-gaining mechanisms (i.e., and that this mood influences how individuals process the ad.
humor strength) translate into positive effects for recall, so Madden, Allen, and Twible (1988) distinguish between affec-
long as the humor is linked to the brand claims. In contrast, tive reaction and cognitive evaluation in advertising response
incidental humor (i.e., humor that is unrelated to the brand data. They find that a humorous ad generates significantly
claims) could actually inhibit brand-claims recall. In the latter more nonevaluative positive affect than its nonhumorous
case, all that is recalled is the funny part of the ad. In sum- counterpart. Machleit and Wilson (1988) provide additional
mary of our expectations, we anticipate that humor–message support for this affective/evaluative dichotomy; they use the
relatedness will moderate the influence of humor strength term “emotional feelings” to describe nonevaluative, affective
such that: responses. In the present study, we follow Bruner and Hensel
(1996, p. 435) in using the label “mood” to measure the non-
H1: The impact of humor strength on message claims recall evaluative affect that has been induced by an object in a person
will be more positive when humor–message relatedness is high (see also Batra and Ray 1986). Our four-item, seven-point se-
than when it is low. mantic differential scale is borrowed from Yi (1990). We chose
the term “mood” to specifically avoid evaluative connotations
H2: The joint impact of humor strength and humor–message associated with “affect.”
relatedness on message claims recall will be mediated by To the extent that humor is related to the ad, it may be
attention to the ad. important to understanding the brand claims. Thus, related
Spring 2007 57

humor, which induces positive mood states, may facilitate brand mor mediate the impact of humor’s relatedness on recall of mes-
claims recall via systematic, or central route processing (Cline, sage claims, and (3) individual differences in NFH moderate the
Altsech, and Kellaris 2003). In summary, we anticipate that influence of humor strength on claims recall. Thus, an experi-
ment crosses humor strength (lower versus higher), humor–mes-
H3: The joint impact of humor strength and humor–message sage relatedness (lower versus higher), and NFH (higher versus
relatedness on message claims recall will be mediated by lower) in a 2 × 2 × 2 between-subjects factorial design with a
mood. no-humor control condition. We manipulate humor strength
and humor–message relatedness and measure NFH. The primary
INDIVIDUAL DIFFERENCES IN dependent variable is message claims recall.
NEED FOR HUMOR
Participants
Individual differences are likely to affect consumers’ receptiv-
ity to humorous stimuli (Zhang 1996). Failure to account for Two hundred fifty-three students enrolled in upper-level
such differences may help explain differences in past studies undergraduate marketing courses at a large, Midwestern uni-
(Duncan and Nelson 1985; Duncan, Nelson, and Frontczak versity participated in the study for course credit. Ages ranged
1984). Gelb and Zinkhan (1985) point out that because what from 18 to 48 (median = 21); 51% were female. Participants
is funny to one person may not be to another, individual dif- completed the experiment independently during a single ses-
ferences in perceptions of humor should be considered. More sion and received academic credit for participating.
recently, Elpers, Mukherjee, and Hoyer (2004) performed a
moment-to-moment analysis of humor in television advertis-
Stimulus Materials and Pretests
ing, demonstrating that moment-to-moment surprise and
humor drive individual perceptions of humor. Humor Strength
We propose that need for humor (NFH) should capture
important individual differences that shape responses to hu- To manipulate humor strength, we developed three versions
mor. NFH is a personality trait that refers to one’s tendency of a simulated print ad (higher humor strength, lower humor
to generate and seek out humor (Cline, Machleit, and Kellaris strength, and no humor control). A format was patterned
1998). We expect NFH to moderate the effects of humor after prior research using print ads (e.g., Arias-Bolzmann,
strength on ad memorability. Based on differential motiva- Chakraborty, and Mowen 2000; Cline, Altsech, and Kel-
tion to process humorous stimuli, individuals high in NFH laris 2003; Edell and Staelin 1983; Heckler and Childers
are expected to “seek out” and attend to humorous stimuli 1992; Houston, Childers, and Heckler 1987; Krishnan and
more readily than those who are low in NFH. Thus, high- Chakravarti 2003). Each ad contained a brand name, a head-
NFH individuals should be prone to recognize and respond line, a product picture, two cartoon figures, brand claims,
to different levels of humor in an ad. In contrast, people low and a tag line. Following Krishnan and Chakravarti (2003),
in NFH are less likely to acknowledge or respond to humor the headline was used as the primary manipulation of humor
of any strength (Dixon et al. 1989). Analogous to “need for strength. Following Lee and Mason (1999), cartoon illustra-
cognition,” NFH should distinguish individuals inclined to tions reinforced the humor manipulation. The size, position,
process ad information based on the humor in the ad from and fundamental meaning of the cartoon figures were held
those likely to process ads primarily on the basis of nonhumor- constant across conditions.
ous elements. Therefore, NFH should moderate the effects of An extensive program of pretesting was conducted to select
humor strength such that: the product category, brand name, and brand claims, and to
test the humor-strength manipulation. The first two pretests
H4: The impact of humor strength on message claims recall will (n = 54) asked participants to rate products on two separate di-
be significantly more positive among individuals with high levels mensions: personal relevancy and humor expectancy. Based on
of NFH than among those with low levels of NFH. midscale ratings for both humor-expectancy rating (M = 4.28
on a seven-point scale) and personal relevancy (M = 3.94 on a
seven-point scale), coffee was selected as the product category.
STUDY Coffee is representative of a low-risk, convenience product
Overview category in which the use of humor is both frequent and
believed to be effective. Weinberger and Campbell (1991)
The goals of this study are to test the hypotheses that (1) humor’s find the incidence of ad humor to be highest in the low in-
strength interacts with humor–message relatedness to influence volvement/feeling cell of the FCB (Foote, Cone, and Belding)
claims recall, (2) attention and positive mood generated by hu- Matrix, a cell made up of “personal pleasures,” including coffee
58 The Journal of Advertising

(Vaughn 1980, 1986). In addition, Weinberger et al. (1995) in the headline. Speck (1987, 1991) identifies three underlying
find that the category comprising coffee in their Product Color humor dimensions: (1) incongruity-resolution—level of respondent
Matrix had the highest incidence of humor for magazine ads surprise, (2) dispositional humor—identification with or detach-
and uniformly positive humor effects on three Starch scores: ment from the humor’s victim, and (3) arousal-safety—the degree
“Noted” (initial attention), “Seen-Associated” (aided-brand of effected relief. Speck describes five combinational humor
recall), and “Read Most” (held attention). types: (1) comic wit—incongruity-resolution humor; (2) senti-
To control for prior knowledge and familiarity, little-known mental humor—arousal-safety; (3) satire—incongruity-resolution
brand names for coffee were selected from a sample of 50 and dispositional humor; (4) sentimental comedy—incongruity-
actual coffee-producing firms. A third pretest (n = 42) was resolution arousal-safety humor; and (5) full comedy—incongru-
conducted to identify unfamiliar brand names that were neutral ity-resolution, dispositional, and arousal-safety humor. Speck
with respect to their “signaling” of humor. Participants rated uses semiotic theory to define three levels of humor–message
12 brand names for humor expectancy on seven-point scales. relatedness: (1) pragmatic—is the ad fundamentally humorous
Because students rated the Wachusett brand name relatively (humor-dominant) or generally nonhumorous? (2) semantic—the
low on humor expectancy (M = 2.91 on a seven-point scale), thematic relationship of the humor and the message, that is,
and because no participant recognized the name Wachusett, it is the humor related to the product, its use, its name, or its
was selected as the coffee brand name. Hundreds of real brand benefits? (3) syntactic—the structural relationship between the
claims were also sampled from coffee advertisements on the humor and the message, that is, is the message separate from or
Internet. A subset of these claims was chosen on the basis of a part of the joke work? Finally, Speck draws on the ELM (Elabora-
fourth pretest (n = 31), which asked students to rate the ap- tion Likelihood Model) to describe three fundamental processing
propriateness of each claim to the product category. strategies: (1) central route—issue-relevant thinking, (2) peripheral
For each condition of humor strength, a “one-liner” was route—nonissue-relevant thinking, and (3) humor-dominant—a
created to constitute the headline. One-liners are a form of special case of peripheral processing where the message is gen-
comic wit used frequently in humorous ads (Speck 1987, 1991). erally structured as a joke. Taken together, Speck’s taxonomy
One-liners were chosen over more involved jokes to keep the allows for 20 humor-dominant ad forms—5 (humor type) × 2
headlines simple. Also, one-liners (e.g., puns) involve an incon- (semantic relatedness) × 2 (syntactic relatedness).
gruity-resolution process (versus arousal-safety or disparagement In the present research, we deliberately selected semanti-
processes). Using this humor process is consistent with Wyer cally related comic wit for all humor conditions and manipu-
and Collins’s (1992) theory of humor elicitation. Finally, Spotts, lated the structural (thematic) relatedness of the humor for the
Weinberger, and Parsons (1997) find that more than 80% of higher (versus lower) humor–message relatedness conditions.
humorous magazine ads use incongruity (the humor process Specifically, the headline and message claims are semantically
comprising comic wit). Manipulating humor strength in the related, that is, the claims “delicate, earthy flavor” and “natural,
form of one-liners (similar to Cline, Altsech, and Kellaris 2003; distinct flavor” relate to the joke in the headline. However, only
Krishnan and Chakravarti 2003) affords the best opportunity to in the conditions with higher humor–message relatedness is
control for potential confounds such as humor process, length, the message syntactically (structurally) related to the humor.
complexity, and evoked emotions (Lammers et al. 1983). Here, the play on words in the headline is carried through to
A final pretest (n = 75) evaluated perceived humor for each the tag line, and hence the loop in the joke work is closed.
of the three humor-strength conditions using completed ver- The efficacy of the humor–message relatedness manipula-
sions of the mock-up ads in a between-subjects test. The results tion was evaluated in a pretest (n = 26). Participants rated the
indicate that the treatments were perceived as differentially relatedness of the headline to the claims and tag line on a four-
humorous, F(1, 74) = 8.53, p < .001. In addition, the lower item measure developed by Krishnan and Chakravarti (2003).
humor strength ad (M = 3.33 on a seven-point scale) was Results indicate that the humor–message relatedness manipu-
perceived as more humorous than the no-humor control ad lation was successful for both low and high levels of humor
(M = 2.62 on a seven-point scale), t = 1.85, df = 51, p < .036 strength. The advertising stimuli are provided in Figure 1.
(one-tailed), and the higher humor strength ad (M = 4.36 on
a seven-point scale) was perceived as more humorous than the Measures
lower humor strength ad (M = 3.33 on a seven-point scale),
t = 2.07, df = 43, p < .023 (one-tailed). Message claims recall was prompted by asking participants,
“What major claims did the ad make? Please list as many as
Humor–Message Relatedness you can remember. What other details about the ad do you
recall?” Participants’ recall of the message claims was com-
To manipulate humor–message relatedness, the tag line and puted on the basis of a set of a priori rules. Two independent
claims either refer specifically to or do not refer to the humor raters, blind to the purposes of the study, coded the claims
Spring 2007 59

FIGURE 1
Experimental Stimuli

Control Humor Strength Lower/Humor–Message Humor Strength Lower/Humor–Message


Relatedness Lower Relatedness Higher

Humor Strength Higher/Humor–Message Humor Strength Higher/Humor–Message


Relatedness Lower Relatedness Higher

recall measure; disparities were resolved via discussion. Because mediator, mood, was measured by a four-item, seven-point
the interrater agreement (90.9%) and the correlation between semantic differential scale developed to measure the feeling
the two rater’s scores (r = .935; n = 253, p < .001) were both that has been induced in a person by an object (Yi 1990). In
high, the two scores were averaged to form a claims recall addition, cognitive responses (e.g., humor comprehension and
index, which served as the dependent measure. The proposed elaboration) were assessed via a thought protocol task (e.g.,
mediator, attention to the ad, was measured by a three-item, Madden, Allen, and Twible 1988).
seven-point Likert-type scale intended to determine the NFH was measured using a scale developed by Cline,
amount of attention devoted to the written message in the Altsech, and Kellaris (2003). Construct validity for the NFH
ad (Muehling, Stoltman, and Grossbart 1990). The proposed scale was assessed following the procedures recommended by
60 The Journal of Advertising

Gerbing and Anderson (1988). A median split was used to attempted humor, F(1, 211) = 35.25, p < .001, ω2 = .139. No
separate the sample into low and high groups for each dimension other main or interactive effects were observed. In addition,
(e.g., Srull, Lichtenstein, and Rothbart 1985). one-way ANOVAs with post hoc multiple comparisons indicate
A manipulation check for humor strength asked participants that the three levels of humor (control, lower humor, and higher
to record their reactions to the ad on a six-item, seven-point humor) were perceived as differentially humorous and received
semantic differential scale (humorous/not humorous, funny/not significantly different ratings for attempted humor (p < .001).
funny, amusing/not amusing, playful/not playful, not dull/dull,
and not boring/boring) adapted from Alden, Mukherjee, and
Humor–Message Relatedness
Hoyer (2000) and Chattopadhyay and Basu (1990). Further-
more, because asking individuals whether they perceive a mes- A two-factor ANOVA was performed on the perceived hu-
sage to be funny or amusing might vary significantly between mor–message relatedness dependent variable with manipulated
individuals, a second measure was employed. Using a one-item, levels of humor and humor–message relatedness as the fac-
seven-point semantic differential scale, participants were asked tors. Results indicated a significant main effect for perceived
to what extent the advertiser was attempting to be humorous (i.e., humor–message relatedness, F(1, 205) = 63.62, p < .001,
a more objective measure of humor). This scale is suggested by ω2 = .231, and a nonsignificant main effect for humor, F(1,
Duncan and Nelson (1985). 205) = .236, p = .628.
A manipulation check for humor–message relatedness asked Measures assessing claim strength, product familiarity, and
participants to record their reactions to the ad on a four-item brand familiarity were included as confounding checks. The
scale (related well/related poorly, were consistent/were inconsis- data were analyzed in a series of ANOVAs with manipulations
tent, fit well/fit poorly, corresponded well/corresponded poorly) of humor strength and humor–message relatedness as the fac-
taken from Krishnan and Chakravarti (2003). Finally, questions tors. Claim strength was not confounded with humor strength,
assessing argument strength, product familiarity, and brand F(1, 205) = .096, p = .901, humor–message relatedness, F(1,
familiarity were included as confounding checks. 205) = .093, p = .760, or the interaction of humor strength
and humor–message relatedness, F(1, 205) = 2.35, p = .127.
Procedure Similarly, product familiarity was not confounded with the
manipulations or their interaction (all p values >.513). Brand
Participants were told that the study was concerned with their familiarity was not confounded with the manipulations or their
reactions to a magazine advertisement prototype. Each partici- interaction (all p values >.554). In summary, results suggest
pant received a questionnaire booklet containing a stimulus that the stimulus ads manipulated the humor and relatedness
ad. After about 10 minutes, participants were asked to place constructs successfully and independently, and the treatments
the questionnaire on the floor and continue with a filler task were not confounded with other traits likely to provide alter-
(e.g., Heckler and Childers 1992) to clear their short-term native explanations.
memories. Subsequently, message claims recall was assessed
via a second questionnaire, which also contained manipulation
and confounding checks, as well as a final question pertaining The Impact of Humor Strength and
to the purpose of the study. At the conclusion of the study, Humor–Message Relatedness on Recall
participants were debriefed.
H1 proposed that claims recall would be more positively
influenced by humor strength when humor–message relat-
RESULTS edness is high than when it is low. ANOVA shows a main
effect of humor–message relatedness, F(1, 211) = 6.25,
Manipulation and Confounding Checks p = .013, ω2 = .024, with higher humor–message related-
Humor Strength ness (M = 1.45) engendering greater claims recall than lower
humor–message relatedness (M = 1.08), t = 2.51, df = 190,
Both subjective and objective measures of humor were used to p = .031. An interaction between humor strength and hu-
verify the effectiveness of the humor manipulation. Two-factor mor–message relatedness was also observed, F(1, 211) = 4.34,
analyses of variance (ANOVAs) were performed on (1) a six- p = .038, ω2 = .02. When humor strength is higher (versus
item measure of perceived humor (Cronbach’s α = .971) and lower) people recall more ad claims under conditions of
(2) a one-item measure of attempted humor with the humor higher (M = 1.66) versus lower humor–message relatedness
strength and humor–message relatedness as the factors. Only (M = .01), t = 3.26, df = 101, p = .001 (one-tailed). Thus, H1
the anticipated main effect for humor was significant for both is supported. In addition, when humor–message relatedness
perceived humor, F(1, 211) = 19.38, p < .001, ω2 = .080, and is higher (versus lower) people recall more ad claims under
Spring 2007 61

FIGURE 2 TABLE 1
Humor Strength × Humor–Message Relevancy on Descriptive Statistics
Recall
Claims recall

Mean n

Low humor
Low humor–message relatedness 1.156 52
High humor–message relatedness 1.204 50
Total 1.179 102
High humor
Low humor–message relatedness 1.009 54
High humor–message relatedness 1.661 56
Total 1.341 110
Total
Low humor–message relatedness 1.082 107
High humor–message relatedness 1.448 105
Total 1.263 212
.9

.7
Mood

H3 predicted that the joint impact of humor strength and


conditions of higher (M = 1.66) versus lower humor strength humor–message relatedness on ad claims recall would be medi-
(M = 1.20), t = 1.95, df = 103, p = .028 (one-tailed). The ated by mood. Results from H2 indicate that the independent
higher humor strength/high humor–message relatedness variable, humor strength × humor–message relatedness, is
condition (M = 1.66) produces greater recall than the control positively correlated with the dependent variable, claims recall.
group (M = 1.24), t = 1.85, df = 95, p = .034 (one-tailed). Second, the independent variable is positively correlated with
Claims recall for the control group does not differ significantly the proposed mediator, mood, r = .548, p < .001. Third, mood
from any other treatment condition. Figure 2 illustrates the is positively correlated with claims recall, r = .216, p = .001.
interaction. Table 1 provides descriptive statistics, and Table Finally, when the dependent variable is regressed on both the
2 provides the ANOVA. independent and mediator variables, the relationship between
the independent variable and the dependent variable becomes
nonsignificant (β = .018, t = .244, p = .808), suggesting full
Attention and Mood as Mediators
mediation and support for H3.
Attention
Concurrent Tests of Mediation
To test the hypothesis that the joint impact of humor strength
and humor–message relatedness on claims recall is mediated To explain further the process by which mood mediates the
by attention to the ad (H2), we used a procedure similar to joint effects of humor strength and humor–message relatedness
that described by Sobel (1982). First, the multiplicative in- on claims recall, additional analyses were conducted. Inde-
dependent variable, humor strength × humor–message relat- pendent support for H2 and H3 suggests that both attention
edness, is positively correlated with the dependent variable, to the ad and mood may mediate the joint impact of humor
claims recall, r = .131, p = .038. Second, the independent strength and humor–message relatedness on claims recall.
variable is positively correlated with the proposed mediator, What is not clear from the separate analyses, however, is the
attention to the ad (r = .200, p = .001). Third, the mediator nature of the relationship between attention and mood. Thus,
is positively correlated with the brand claims recall dependent the dependent variable, claims recall, was regressed on both
variable, r = .147, p = .019. Finally, when the dependent mediators (attention and mood) and the independent factor
variable, claims recall, is regressed on both the mediator (humor strength × humor–message relatedness). Results of
(attention to the ad) and the independent variable (humor this model indicate that mood accounts for most of the vari-
strength × humor–message relatedness), the relationship be- ance in claims recall. In addition, the results imply that mood
tween the independent variable and the dependent variable jointly mediates the relationship between attention and recall,
attenuates (β = .108, t = 1.69, p = .092), thereby providing as well as the relationship between the independent variable
support for a partial mediation model and for H2. and recall. Specifically, the coefficient for mood is significant,
62 The Journal of Advertising

TABLE 2
Claims Recall: Humor Strength × Humor–Message Relatedness

Experimental method

Sum of Mean
squares df square F Significance

Main effects
Combined 8.31 2.00 4.15 3.75 .03
Humor strength 1.22 1.00 1.22 1.10 .30
Humor–message relatedness 6.92 1.00 6.92 6.25 .01
Two-way interactions
Humor strength × humor–message relatedness 4.81 1.00 4.81 4.34 .04
Model 13.12 3.00 4.37 3.95 .01
Residual 230.16 208.00 1.11
Total 243.28 211 1.15

Note: ANOVA = analysis of variance.

βmood = .186, t = 2.48, p = .014, whereas the coefficients for proposed by Ping (1996). The anticipated link between humor
humor strength × humor–message relatedness and attention strength × humor–message relatedness and mood is observed
are both nonsignificant, βhum×rel = .012, t = .161, p = .872, (γ21 = .054, t = 9.95, df = 3, p < .01), and a significant link
βattention = .084, t = 1.20, p = .194. between humor strength × humor–message relatedness and
Next, we examined the possibility that mood may operate attention is observed (γ11 = .026, t = 3.11, df = 3, p < .05).
through elaboration. The correlation between mood and total In addition, attention appears to influence mood (β21 = .151,
thoughts generated from the ad is nonsignificant, however, t = 3.33, df = 3, p < .05) and elaboration (β31 = .377, t = 4.54,
r = –.027, p = .672. Therefore, mood is unlikely to produce df = 3, p < .01). In contrast, mood is not significantly related
effects on message claims recall via elaboration. Moreover, in a to elaboration (β32 = –.195, t = 1.83, df = 3, n.s.). Finally,
separate model, when claims recall was regressed on both mood mood positively influences claims recall (β42 = .256, t = 3.62,
and elaboration, both predictors produced significant coeffi- df = 3, p < .05) and elaboration positively influences claims
cients and offered no indications of collinearity (βmood = .220, recall (β43 = .117, t = 2.47, df = 3, p < .05).
t = 3.62, p < .001; βelaboration = .166, t = 2.72, p = .007). This In addition to validating the proposed nomological net-
suggests that mood and elaboration may work independently work, the model fit the data exceptionally well. The χ2 value
to affect claims recall. was not significant (χ2 = .51, df = 3, p = .92). Furthermore,
Finally, elaboration was tested as a mediator to the relation- the goodness-of-fit index (GFI = .99), the adjusted goodness-
ship between attention to the ad and claims recall. Pairwise of-fit index (AGFI = .98), and the root mean square residual
correlations among elaboration, attention, and claims recall are (RMSR = .062) provide evidence that the model fit the data
significant, rattention, recall = .137 (p = .030), rattention, elaboration = .255 well. No standardized residuals fell outside the range of –1.0 to
(p < .001), relaboration, recall = .160 (p < .011). When claims recall 1.0. Thus, the path model appears to explain the relationships
is regressed on both attention and elaboration, however, the between the interaction of humor strength and humor–message
association between attention and claims recall becomes nonsig- relatedness, attention to the ad, mood, elaboration generated
nificant (β = .103, t = 1.60, p = .111). This suggests that elabo- from the ad, and message claims recall. Humor strength and
ration mediates the relationship between attention and claims humor–message relatedness influence claims recall via a dual
recall. In summary of these findings, it appears that whereas process—attention and mood. Moreover, mood mediates—both
mood and elaboration produce independent effects on claims directly and indirectly—the joint effects of humor strength
recall, attention works through both mood and elaboration to and humor–message relatedness on claims recall. In contrast,
influence claims recall. In both cases, the interaction of humor attention appears to work through mood and elaboration to
strength and humor–message relatedness is likely to stimulate influence claims recall.
the process. To test these relationships in a nomological net-
work, a path model with reliabilities (Figure 3) was developed. NFH as a Moderator
The humor strength × humor–message relatedness served as
the exogenous variable and attention, mood, elaboration, and H4 predicted that NFH would moderate the impact of humor
claims recall were specified as endogenous variables. strength such that its effects on claims recall would be more
To test the interaction hypothesis, we used the methodology positive for people with higher levels of NFH than for people
Spring 2007 63

FIGURE 3
Path Model with Reliabilities

* p < .05.
** p < .01.

with lower levels of NFH.. An ANOVA was performed on bring into clearer relief the value of using strong, relevant
claims recall, with humor strength and NFH as the factors. humor, as opposed to stronger, incidental humor. It appears
The anticipated interaction between humor strength and NFH that strong humor is not its own virtue; it must be connected
was observed, F(1, 195) = 4.37, p = .038, ω2 = .02. The results to the brand claims to facilitate recall. In contrast, weaker
indicate that when humor strength is higher (versus lower), humor, whether related to the message or not, does not aid
those higher in NFH recalled more ad claims (M = 1.59) than brand claims recall.
those lower in NFH (M = 1.07), t = 2.48, df = 97, p = .008 The data also provide additional insight into person-by-
(one-tailed). Thus, H4 is supported. Figure 4 illustrates this situation influences on advertising responses. For example,
interaction. Table 3 provides descriptive statistics, and Table 4 an individual’s tendency to generate humor (NFH) appears
provides the ANOVA. to moderate humor strength effects such that the impact of
humor strength on claims recall is significantly more posi-
Discussion tive for people with high levels of NFH than for people with
low levels of NFH. When humor strength is higher (versus
As anticipated, the data demonstrate that humor’s attention- lower), individuals higher in NFH recall more ad claims than
gaining mechanisms (i.e., humor strength) may translate into those lower in NFH. In addition, those higher (versus lower)
positive effects for memory. Specifically, it appears that humor in NFH recall more ad claims when humor strength is higher
strength and humor–message relatedness jointly influence than when it is lower.
participants’ recall of advertising claims. The results show The data also support the notion that attention and mood
that when humor strength is higher (versus lower), partici- operate via dual processes to mediate the joint impact of hu-
pants recall more ad claims when the humor is relevant to the mor strength and humor–message relatedness on claims recall.
claims. The data also show that when the humor is related Having first established the importance of strong, structurally
to the message, participants recall more advertising claims related humor, this dual process model helps explain why the
when humor strength is higher than when it is lower. When strong, relevant humor leads to higher brand claims recall.
the humor is both stronger and related to the message, the These results are interesting because they suggest that humor
humorous ad outperforms the control ad with respect to mes- may trigger both cognitive (attention) and affective (mood)
sage claims recall. These results are important because they routes to memorability.
64 The Journal of Advertising

FIGURE 4 TABLE 3
Humor Strength × NFH (Need for Humor) on Recall Descriptive Statistics

Recall

Mean n

Low humor
Lower NFH 1.21 49
Higher NFH 1.10 48
Total 1.16 97
High humor
Lower NFH 1.07 48
Higher NFH 1.59 51
.9 Total 1.34 99
Total
Lower NFH 1.14 104
.7 Higher NFH 1.38 92
Total 1.25 196
.5 Note: NFH = need for humor.

GENERAL DISCUSSION our knowledge of the consumer’s sense of humor in general,


and under what conditions it influences responses to humor-
The goal of this research was to test the generative mechanisms ous communication.
by which humor–message relatedness influences brand claims The incremental contributions of the present research are
recall. In addition, we introduced the personality variable twofold. First, whereas prior research has shown a positive re-
“need for humor” (NFH) as a moderator of humor’s effects on lationship between humor and attention, and between related
advertising recall and investigated the mediating roles of at- humor and recall, previous studies have not explicitly investi-
tention, mood, and elaboration in a nomological network. gated the joint interplay of humor strength and humor–message
This research makes a number of theoretical contributions. relatedness. The present study specifically addresses this im-
First, by examining the joint effects of humor strength and portant interaction and finds that with respect to claims recall,
humor–message relatedness on advertising recall, this research neither humor strength nor humor–message relatedness is its
provides evidence that message-relevant stimuli may produce own virtue. Relatively stronger humor must be related to the
consistent effects across various attention-gaining contexts. For brand or message to engender higher claims recall. Second,
example, the findings suggest that humor–message relatedness previous research has not investigated the generative mecha-
interacts with humor’s attention-gaining properties in a man- nisms that follow from the interplay of humor strength and
ner that is analogous to the interaction between music–message humor–message relatedness. The present research specifically
congruity and music’s attention-gaining properties (Kellaris, evaluates the influence of attention and mood in a nomological
Cox, and Cox 1993). Thus, the present research provides a network of variables. We find that attention and mood oper-
foundation for suggesting that attention-gaining message ate in partnership, jointly mediating the interactive effects of
appeals, in general, may engender better recall when they are humor strength and humor–message relatedness on claims
structurally related to the message theme. Second, by exam- recall. The path model uncovers this dual process. Mood directly
ining the psychological mechanisms that underlie consumer mediates humor’s influence on claims recall. In contrast, atten-
responses to humorous stimuli (e.g., attention, mood, and tion operates indirectly through both mood and elaboration to
elaboration), as well as the theoretical principles that activate influence claims recall. Thus, both variables help explain why
these mechanisms (i.e., humor–message relatedness), this re- the interaction between humor strength and humor–message
search leads to better explanations for why humorous effects relatedness influences claims recall.
obtain or fail to obtain. Attention and mood, for example, are The present findings also have practical implications in
shown as dual mediators of the impact of humor strength and terms of guidelines for advertisers who use humorous ads
humor–message relatedness on recall for the message claims. and imply that the effects of message-congruent humor may
Third, NFH, like other individual difference variables, explains generalize to other attention-gaining appeals, such as music or
an additional source of variation in advertising outcomes sexual attraction. On the basis of humor’s prevalence and the
through its role as a moderator. The NFH construct broadens belief in its universal effectiveness, NFH may be useful as a
Spring 2007 65

TABLE 4
Recall: By Humor Strength by NFH

Experimental method

Sum of Mean
squares df square F Significance

Main effects
Combined 3.74 2 1.87 1.70 .19
Humor strength 1.14 1 1.14 1.04 .31
NFH 2.27 1 2.27 2.06 .15
Two-way interactions
Humor strength × NFH 4.82 1 4.82 4.37 .04
Model 8.57 3 2.86 2.59 .05
Residual 211.93 192 1.10
Total 220.50 195 1.13

Note: ANOVA = analysis of variance; NFH = need for humor.

segmentation tool. Clearly, advertisers cannot administer NFH Basu 1990) or that other humor dimensions (e.g., disposi-
scales to members of their target audiences; however, market tional humor; Speck 1987, 1991) or humor structures (e.g.,
research can identify media that draw groups of people with message-dominant/image-oriented; Spotts, Weinberger, and
high levels of NFH (e.g., Mad Magazine readers and Late Show Parsons 1997) may be more effective in a print context. The
with David Letterman watchers). In addition, this information use of incongruous humor, however, is justified on the grounds
can be used to determine which product categories or brands that it appears to be the most pervasive humor in magazine ads
tend to be popular with specific media users (e.g., Mad Maga- and for nondurable items such as coffee (Spotts, Weinberger,
zine readers may tend to be heavy users of B-movies or video and Parsons 1997).
games). Thus, NFH may be helpful both in media selection The results of the experimentation are intuitively appealing.
and in targeting audiences for specific products. Each year, advertisers spend millions of dollars on humorous
ads. Some contain strong, relevant messages that engender
LIMITATIONS AND FUTURE RESEARCH recall. Recently, several brands have generated remarkable
brand awareness with strong, related humor. Geico’s gecko and
The results of this research are necessarily qualified by the Aflac’s duck appear to be deliberate attempts at linking the
limitations of the studies’ designs, each of which suggests an source of the humor with the brand name, and hence increas-
opportunity for additional research. First, the experimenta- ing recall. Orbit gum links its “dirty mouth” humor with its
tion used mock advertisements in a controlled setting. Thus, primary attribute, clean-tasting gum. Classic successes like the
the extent to which the findings are limited by the artificial Eveready’s Energizer Bunny were designed to link the humor
context and the representativeness of the ads and product is with the claims. The message is indirect but clear—like the
unknown. Second, both studies use comic wit, manipulated in bunny, the batteries keep going and going. In contrast, it ap-
the headline of a print ad, to convey humor that is independent pears that millions of dollars may be wasted on advertisements
of the brand name, thus limiting generality to other types of that make no connection between the humor and the claim,
humor. This choice, however, allows for control over the for- subsequently creating brand confusion or inhibiting recall.
mat of the ads across conditions and thus avoids confounding
other ad characteristics with humor (Krishnan and Chakravarti CONCLUSION
2003). Furthermore, it enables clear identification of the locus
of effects to specific ad components (e.g., brand claims). Third, The pervasive use of humor in advertising attests to the wide-
whereas all ads were designed such that the humor was related spread belief that humor enhances the effectiveness of ads. The
to the brand claims (i.e., thematically related), the structural research reported here examines contingencies that shape the
relatedness of the humor was manipulated via the tag line. effects of humorous appeals on an important outcome of ad-
Thus, future research should examine the structural location vertising—consumers’ recall of ad claims—as well as processes
of the humor in an ad. Fourth, despite extensive pretesting, it through which and boundary conditions within which such
could be argued that humor is more appropriate in a broadcast effects operate. In so doing, this research seeks to elucidate
medium (e.g., Alden and Hoyer 1993; Chattopadhyay and when and how humor contributes to remembering ads.
66 The Journal of Advertising

REFERENCES Elpers, Josephine, Ashesh Mukherjee, and Wayne Hoyer (2004),


“Humor in Television Advertising: A Moment-to-Moment
Alba, Joseph W., J. Wesley Hutchison, and John G. Lynch (1991), Analysis,” Journal of Consumer Research, 31 (3), 592–598.
“Memory and Decision Making,” in Handbook of Consumer Galizio, Mark, and Clyde Hendrick (1972), “Effect of Musical Ac-
Research, Thomas S. Robertson and Harold H. Kassarjian, companiment on Attitude: The Guitar as a Prop for Persua-
eds., Englewood Cliffs, NJ: Prentice-Hall, 1–49. sion,” Journal of Applied Social Psychology, 2 (4), 350–359.
Alden, Dana L., and Wayne D. Hoyer (1993), “An Examination Gelb, Betsy, and George Zinkhan (1985), “The Effect of Repeti-
of Cognitive Factors Related to Humorousness in Television tion on Humor in a Radio Advertising Study,” Journal of
Advertising,” Journal of Advertising, 22 (2), 29–37. Advertising, 14 (4), 13–20, 68.
———, Ashesh Mukherjee, and Wayne D. Hoyer (2000), “The Gerbing, David W., and James C. Anderson (1988), “An Updated
Effects of Incongruity, Surprise, and Positive Moderators Paradigm for Scale Development Incorporating Unidimen-
on Perceived Humor in Television Advertising,” Journal of sionality and Its Assessment,” Journal of Marketing Research,
Advertising, 29 (2), 1–15. 25 (May), 186–192.
Arias-Bolzmann, Leopoldo, Goutam Chakraborty, and John C. Heckler, Susan E., and Terry L. Childers (1992), “The Role of
Mowen (2000), “Effects of Absurdity in Advertising: The Expectancy and Relevancy in Memory for Verbal and Visual
Moderating Role of Product Category Attitude and the Information: What Is Incongruency?” Journal of Consumer
Mediating Role of Cognitive Responses,” Journal of Adver- Research, 18 (4), 475–492.
tising, 29 (1), 35–49. Houston, Michael J., Terry L. Childers, and Susan E. Heckler
Batra, Rajeev, and Michael L. Ray (1986), “Affective Responses (1987), “Picture-Word Consistency and the Elaborative
Mediating Acceptance of Advertising,” Journal of Consumer Processing of Advertisements,” Journal of Marketing Research,
Research, 13 (September), 234–249. 24 (December), 359–369.
Bower, Geoffrey H., Stephen G. Gilligan, and Kenneth P. Mon- Isen, Alice M. (1987), “Positive Affect, Cognitive Processes and
teiro (1981), “Selectivity of Learning Caused by Affective Social Behavior,” in Advances in Experimental Social Psychol-
States,” in Affect and Social Cognition, Susan Fiske and ogy, Lawrence Berkowitz, ed., New York: Academic Press,
Margaret Clark, eds., Hillsdale, NJ: Lawrence Erlbaum, 203–253.
291–331. ———. (1989), “Some Ways in Which Affect Influences Cogni-
Bruner, Gordon C., II, and Paul J. Hensel (1996), Marketing Scales tive Processes: Implications for Advertising and Consumer
Handbook, vol. 2, Chicago: American Marketing Association. Behavior,” in Advertising and Consumer Psychology, Alice M.
Chattopadhyay, Amitava, and Kunal Basu (1990), “Humor in Ad- Tybout and Patricia Cafferata, eds., New York: Lexington
vertising: The Moderating Role of Prior Brand Evaluation,” Books, 91–117.
Journal of Marketing Research, 27 (November), 466–476. Kellaris, James J., Anthony D. Cox, and Dena Cox (1993), “The
Cline, Thomas W., Moses B. Altsech, and James J. Kellaris Effect of Background Music on Ad Processing: A Contin-
(2003), “When Does Humor Enhance or Inhibit Ad Re- gency Explanation,” Journal of Marketing, 57 (October),
sponses? The Moderating Role of Need for Levity,” Journal 114–125.
of Advertising, 32 (3), 31–46. Krishnan, H. Shanker, and Dipankar Chakravarti (2003), “A
———, Karen A. Machleit, and James J. Kellaris (1998), “Is Process Analysis of the Effects of Humorous Advertising
There a Need for Levity?” in Proceedings of the Society for Executions on Brand Claims Memory,” Journal of Consumer
Consumer Psychology 1998 Winter Conference, Margaret C. Psychology, 13 (3), 230–245.
Campbell and Karen A. Machleit, eds., Washington, DC: Lammers, H. Bruce, Laura Liebowitz, George E. Seymour,
American Psychological Association, 155–157. and Judith E. Hennessey (1983), “Humor and Cognitive
Dixon, Paul N., William Willingham, Donald A. Strano, and Responses to Advertising Stimuli: A Trace Consolidation
Cynthia K. Chandler (1989), “Sense of Humor as a Mediator Approach,” Journal of Business Research, 11 (2), 173–185.
During Incidental Learning of Humor-Related Material,” Lee, Yih L., and Charlotte Mason (1999), “Responses to Informa-
Psychological Reports, 64, 851–855. tion Incongruency in Advertising: The Role of Expectancy,
Duncan, Calvin P. (1979), “Humor in Advertising: A Behavioral Relevancy, and Humor,” Journal of Consumer Research, 26
Perspective,” Journal of the Academy of Marketing Science, 7 (2), 156–169.
(4), 285–306. Littmann, Jane R. (1983), “A New Formulation of Humor,”
———, and James E. Nelson (1985), “Effects of Humor in a Advances in Descriptive Psychology, 3, 183–207.
Radio Advertising Experiment,” Journal of Advertising, 14 Machleit, Karen A., and R. Dale Wilson (1988), “Emotional
(2), 33–40. Feelings and Attitude Toward the Advertisement: The Roles
———, ———, and Nancy T. Frontczak (1984), “The Effect of Brand Familiarity and Repetition,” Journal of Advertising,
of Humor on Advertising Comprehension,” in Advances in 17 (Fall), 27–35.
Consumer Research, Thomas C. Kinnear, ed., Provo, UT: As- Madden, Thomas J. (1982), “Humor in Advertising: Applications
sociation for Consumer Research, 432–437. of a Hierarchy-of-Effects Paradigm,” Ph.D. diss., University
Edell, Julia A., and Richard Staelin (1983), “The Information of Massachusetts–Amherst.
Processing of Pictures in Print Advertisements,” Journal of ———, Chris T. Allen, and Jacquelyn L. Twible (1988), “At-
Consumer Research, 10 (1), 45–61. titude Toward the Ad: An Assessment of Diverse Measure-
Spring 2007 67

ment Indices Under Different Processing ‘Sets,’” Journal of Spotts, Harlan E., Marc G. Weinberger, and Amy L. Parsons
Marketing Research, 25 (3), 242–252. (1997), “Assessing the Use and Impact of Humor on Ad-
———, and Marc G. Weinberger (1982), “The Effects of Humor vertising Effectiveness: A Contingency Approach,” Journal
on Attention in Magazine Advertising,” Journal of Advertis- of Advertising, 26 (3), 17–32.
ing, 11 (3), 8–14. Srull, Thomas K., Meryl Lichtenstein, and Myron Rothbart (1985),
———, and ——— (1984), “Humor in Advertising: A Practi- “Associative Storage and Retrieval Processes in Person Mem-
tioner View,” Journal of Advertising Research, 24 (4), 23–29. ory,” Journal of Experimental Psychology, 11 (2), 316–345.
Meyers-Levy, Joan (1991), “Elaborating on Elaboration: The Dis- Sternthal, Brian, and C. Samuel Craig (1973), “Humor in Ad-
tinction Between Relational and Item-Specific Elaboration,” vertising,” Journal of Marketing, 37 (4), 12–18.
Journal of Consumer Research, 18 (3), 358–367. Vaughn, Richard (1980), “How Advertising Works: A Planning
Muehling, Darrel D., Jeffrey J. Stoltman, and Sanford Gross- Model,” Journal of Advertising Research, 20 (5), 27–33.
bart (1990), “The Impact of Comparative Advertising on ——— (1986), “How Advertising Works: A Planning Model
Levels of Message Involvement,” Journal of Advertising, 16 Revisited,” Journal of Advertising Research, 26 (1), 57–66.
(3), 41–50. Weinberger, Marc G., and Leland Campbell (1991), “The Use
Olson, James M., and Neal J. Roese (1995), “The Perceived Fun- and Impact of Humor in Radio Advertising,” Journal of
niness of Humorous Stimuli,” Personality and Social Psychology Advertising Research, 31 (December), 44–52.
Bulletin, 21 (9), 908–913. ———, and Charles S. Gulas (1992), “The Impact of Humor
O’quin, Karen, and James Aronoff (1981), “Humor as a Tech- in Advertising: A Review,” Journal of Advertising, 21 (4),
nique of Social Influence,” Social Psychology Quarterly, 44 36–59.
(4), 349–357. ———, Harlan E. Spotts, Leland Campbell, and Amy L. Par-
Ping, Robert A., Jr. (1996), “Latent Variable Interaction and sons (1995), “The Use of Humor in Different Advertising
Quadratic Effect Estimation: A Two-Step Technique Using Media,” Journal of Advertising Research, 35 (May/June),
Structural Equation Analysis,” Psychological Bulletin, 119 44–56.
(1), 166–175. Wicker, Frank W., Irene M. Thorelli, William L. Barron III, and
Smith, Stephen M. (1993), “Does Humor in Advertising Enhance Marguerite R. Ponder (1981), “Relationships Among Af-
Systematic Processing?” in Advances in Consumer Research, vol. fective and Cognitive Factors in Humor,” Journal of Research
20, Leigh McAlister and Michael L. Rothschild, eds., Provo, in Personality, 15, 350–370.
UT: Association for Consumer Research, 155–158. Wyer, Robert S., and James E. Collins (1992), “A Theory of Hu-
Sobel, Mark E. (1982), “Asymptotic Intervals for Indirect Effects mor Elicitation,” Psychological Review, 99 (4), 663–688.
in Structural Equation Models,” in Sociological Methodology, Yi, Youjae (1990), “Cognitive and Affective Priming Effects of
S. Leinhart, ed., San Francisco: Jossey-Bass, 290–312. the Context for Print Advertisements,” Journal of Advertis-
Speck, Paul S. (1987), “On Humor and Humor in Advertising,” ing, 19 (2), 40–48.
Ph.D. diss., Texas Tech University. Zhang, Yong (1996), “Responses to Humorous Advertising:
——— (1991), “The Humorous Ads,” in Current Issues and Re- The Moderating Effect of Need for Cognition,” Journal of
search in Advertising, James H. Leigh and Claude R. Martin, Advertising, 15 (1), 15–32.
Jr., eds., Ann Arbor: University of Michigan Press, 1–44.

Anda mungkin juga menyukai