When Humorous Advertising Elicit Inferences of
Manipulative Intent
When Humorous Advertising Elicit Inferences of Manipulative Intent
Master Thesis, MSc Marketing, Specialization Marketing Management
University of Groningen, Faculty of Economics and Business
June 22, 2015
Monica Joana Graciela Kuijt Rumayor
Student number: 2589818
Bloemstraat 4a
9712 LD, Groningen
tel.: +32(6) 39574573
e-mail: m.j.g.kuijt.rumayor@student.rug.nl
Supervisor / university
Dr. Debra Trampe
Second assessor / university
When Humorous Advertising Elicit Inferences of
Manipulative Intent
Abstract. This study examines the effect of humorous advertising on consumer inferences of
manipulative intent compared to serious advertising. Such an effect of humor is based on existing theory regarding attention-getting techniques and their effect on judgments of manipulative intent. This study also explores the mediating role of information processing depth using the cognitive response model and the Elaboration Likelihood Model to explain the mechanisms of the relationships between these constructs. By means of an online experiment, this study found evidence that humorous advertising, compared to serious advertising, does increase consumers’ inference of manipulative intent. Moreover, empirical support for the mediating role of information processing depth was not found. Managerial implication regarding the use of humor in advertising is given as to help marketers to use humor for getting consumers’ attention and while doing so reduce the risk of generating inferences of manipulative intent.
Key words: inference of manipulative intent, humorous advertising, information processing depth,
TABLE OF CONTENTS
1. INTRODUCTION 6
2. THEORETICAL FRAMEWORK 8
2.1 Increased Attention and Judgments of Manipulative Intent 8
2.2 Humor in Advertising 10
2.3 Mediating Role of Information Processing Depth 12
2.3.1 Cognitive Response Model 12 2.3.2 Elaboration Likelihood Model 13 2.3.3 Processing of Humorous Advertising 15
3. METHOD 15
3.1 Selection of Stimuli and Pretest 15
3.1.1 Selected Stimuli 15
3.1.2 Pretest Design, Participants and Procedure 16
3.1.3 Pretest Results 17
3.2 Research Design and Participants 18
3.3 Procedure 19
3.4 Measures 20
3.4.1 Demographics 20
3.4.2 Depth of Information Processing measure 20 3.4.3 Extent of Inference of Manipulative Intent measure 21
3.4.4 Manipulation check 21
3.4.5 Other measures 21
3.5 Coding Procedures and Creation of Variables from Qualitative Data 22
3.5.1 Information Processing Depth Score 22 3.5.2 Processing Depth Category 22
3.5.4 Amount of Thoughts Listed 23
3.6 Data Analysis 23
4. RESULTS 23
4. Manipulation Check 23
4.2 Correlations 24
4.3 Hypotheses Testing – Mediation Analysis 25
4.4 Hypotheses Testing – Additional Analyses of Variance 27
5. DISCUSSION 32
5.1 Theoretical Implications 32
5.2 Managerial Implications 35
5.3 Study Limitations and Future Research 35
REFERENCES 37
1. INTRODUCTION
Thirty-second ad spots during the Super Bowl XLIX, which attracted 114.4 million viewers, were sold for an average of $4.5 million and traditionally the two most used advertising approaches have been comedy and spectacle (Super Bowl Broadcast News 2015). Unlike the way most people feel about commercial breaks when watching TV, the airing of the Super Bowl ads is a popular and well-accepted “event” in itself. The amount of money brands invest in them and the fact that the majority of the brands utilize humor as their approach, must be an indication that such an approach works. However, research in advertising has provided evidence of a positive impact of humor on attention but when it comes to the persuasive effectiveness of humor, there appears to be a lack of consensus in research evidence.
According to Eisend (2009), “TV shows on humorous advertisements and awards for such advertisements indicate that humor in advertising has even become an important part of everyday life” (p.191). However, despite its cleverness and vast resources, advertising faces an enormous challenge since consumers increasingly choose not to believe all product claims they encounter in advertising (Obermiller, Spangenberg, and MacLachlan 2005). Consumers have become skeptical about advertising, and this skepticism can make consumers discount the message from the ad, increase their counterarguments, like the ad less, and even lower their attitudes towards brands (La Tour and La Tour 2009).
Moreover, it is a true fact that at least some moderate level of attention is needed for persuasion to occur from exposure to an ad, if anything for the simple fact that brands need to fight through the clutter of advertising in the market. As already mentioned, humor is an effective approach for getting consumers’ attention, but contrary to popular belief that any increase in attention is desirable, there are also negative outcomes associated with increased processing attention (Campbell 1995). One of the negative effects, and the focus of the present study, is consumers’ elaboration of negative judgments about the intention of the advertiser, which can lead to resisting persuasion.
happen to a guy if he uses the deodorant, and consequently conclude that the advertiser is trying to persuade him unfairly by lying.
That is one of many examples of situations where it is possible that “if we think about a persuasive communication and scrutinize the arguments contained in the message, we will discover weaknesses and inconsistencies and will therefore be reluctant to accept the recommendation” (Fennis and Stroebe 2010, p.159). Thus, the central research question this study aims to answer is the following: What is the effect of a humorous ad on consumers’ inferences of manipulative intent
compared to a non-humorous ad?
Additionally, for inferences of manipulative intent to occur, the viewer of the ad must engage in deeper information processing (Campbell 1995; Greenwald and Leavitt 1984; Petty and Cacioppo 1984). Opinions regarding processing of information in humorous ads are mixed; on one hand, the increased attention as a result of humor allows people to process brand claims in the ad more actively (Smith 1993), but on the other hand, humorous content may also hinder the acquisition of relevant information directed to brand claims in the ad (Cline, Altsech, and Kellaris 2003).
In the present study we will examine the first possible effect of humor. Therefore, this study includes depth of information processing as a mediating variable of the relationship between type of advertisement and inferences of manipulative intent. Accordingly, this research seeks to address the following additional questions: 1) What is the effect of a humorous ad compared to a non-humorous
ad on consumers’ depth of information processing? 2) How does the depth of information processing generated by the type of ad mediate the effect of the advertisement on consumers’ inferences of manipulative intent?
advertising.
Furthermore, this study compares the effects of humorous as opposed to serious advertisements. From a practitioner's standpoint, “comparing the persuasive effects of humorous and serious messages is preferable since the practitioner often is faced with the choice between alternative versions of the same appeal, thus designs that allow for comparisons can aid decision making” (Sternthal and Craig 1973, p.14).
The overall structure of the study takes the form of five chapters, including this introductory chapter. Chapter two begins by laying out the theoretical framework of the research. The third chapter is concerned with the employed methodology for this study followed by the fourth chapter, which presents the results obtained from statistical analyses. Finally, the fifth chapter ends with a discussion of the research findings, including theoretical and managerial implications as well as limitations and direction for future research.
2. THEORETICAL FRAMEWORK 2.1 Increased Attention and Judgments of Manipulative Intent
For an advertisement to be effective, it must first of all attract consumers’ attention. Attention can be defined as the general distribution of mental activity to a specific task and can be allocated in varying degrees to the ad under conscious control or due to attention-grabbing stimuli properties (MacInnis and Jaworski 1989). Moreover, increased attention is known to have positive effects on consumers’ reactions to advertising. MacInnis and Jaworski (1989) explained that for instance, when attention towards the brand increased, the viewer of the ad was better capable of understanding the ad and perceiving the benefits of the brand.
Benefits concerning increased processing attention on advertisement effectiveness are well established in literature (Cialdini, Petty, and Cacioppo 1981; Petty, Cacioppo and Schumann 1983). However, there is also evidence of a negative effect of increased attention on consumers’ reactions to advertising. Even though this downside effect has been researched to a lesser extent, some of the negative outcomes that have been found are an increased likelihood of the consumer generating counterarguments and reduced persuasion. Greenwald and Leavitt (1984) argue that it is possible that under elaborative processing the viewer will evaluate and oppose the content of the message and so a boomerang effect is possible, meaning that the change in the viewer’s attitude is the opposite of that intended by the persuasive message.
manipulative intent, and is defined as “consumer inferences that the advertiser is attempting to persuade by inappropriate, unfair, or manipulative means” (Campbell 1995, p.227). In her research, Campbell (1995) argues that an increase in processing attention could lead consumers to second-guess the advertiser, think why the ad was created in such a way, develop arguments that challenge those presented in the ad, discount the message, develop negative attitudes towards the ad and the brand, and even resist persuasion.
Moreover, advertisers use a variety of tactics to attract consumers’ attention. Campbell (1995) studied the impact of two specific attention-getting tactics used in television ads, namely delayed brand identification (i.e., delay showing the brand and product category until the end of the ad with the purpose of creating suspense) and borrowed interest appeal (i.e., including something in the ad that creates arousal and involvement, for example cute puppies or beautiful people) on consumers’ inferences of manipulative intent She argues that these types of attention tactics are usually applied by advertisers because they force viewers to increase their processing attention towards the ad. Consequently, “more active processing may sometimes lead consumers to think about what the advertiser is doing in the ad, increasing the probability of negative processing consequences, such as inferences of manipulative intent” (Campbell 1995, p.228).
The author tested this theory through an experiment in which she manipulated the type of advertisement participants were exposed to (i.e., brand identification in the ad was late or early and borrowed interest appeal was a high or low fit with the product being advertised). After seeing the ad, participants responded to statements about the ad to measure the extent to which they thought the ad/advertiser was trying to manipulate them; this scale was developed by the author prior to the study. During the experiment all participants saw the same news program and one of the four experimental ads (all ads were the same and varied only on the experimental manipulation). After exposure to the ads, participants were first asked to list their thoughts and feelings they had about the ad, next they answered the inference of manipulative intent (IMI) scale.
(1995) study is that an attention-getting tactic that creates arousal and involvement is more likely to lead to consumer inferences of manipulative intent compared to an attention-getting tactic that creates curiosity and suspense.
The following section introduces the choice of an attention-getting tactic for the present study and explores the effects it may have on consumers’ inferences of manipulative intent.
2.2 Humor in Advertising
The present study focuses on another widely used attention-getting technique, humor, to explore how its use may result in consumer inferences of manipulative intent. Figure 1 shows this study’s conceptual model, which is explained in detail in these following sections.
FIGURE 1 Conceptual Model
It is imperative to start by defining what humorous advertising is. It may seem obvious, however as Sternthal and Craig (1973) noted in their study of humor in advertising, there are three different approaches to define humor. One defines humor in advertising in terms of the stimuli properties, that is, whether jokes, satire, irony, etc. are used. A second one defines it in terms of the response to stimuli, thus whether a particular message creates heightened arousal, smiles and laughter in the audience. The third alternative defines it in terms of the perception of the audience, the extent to which the audience perceived the message to be humorous.
Arousal is part of the conceptual theory in this study; however, since it is not the focus of the research it is not tested. There is reason and evidence to believe that there is a difference between a humorous ad and a serious ad on their effects on viewers’ feelings of arousal. For one, it is embedded in the definition of humorous advertising I just described (Sternthal and Craig 1973). What’s more, humorous advertising evokes arousal (Gulas and Weinberger 2006) because of the novelty, complexity and incongruity that characterizes this type of ads (Berlyne 1972).
Furthermore, humor is consistently found to be an effective way to catch viewers’ attention (Eisend 2009; Weinberger and Gulas 1992). Eisend (2009) conducted a meta-analysis of effects of humor in advertising on a number of different dependent variables, attention being one of them. The study included 38 manuscripts covering 43 independent studies carried out from 1960 to late 2006. Additionally, only studies that compared humorous to non-humorous advertising were considered. The results of the meta-analytical correlation carried out in the study demonstrated that humor in advertising significantly enhances attention.
A reason why humor might be so effective at grabbing consumers’ attention could be because humor may be perceived as a benefit for paying extra attention and listening to the entire message; generally, to be able to understand a joke one should hear the entire story. Chan (2011) found the effect of humor on increased attention in television advertising to be also largely supported in his study; he claims that “people tend to be attracted by unexpected things and are more easily pulled in by things framed within a context which deviates from their perceived norm than by something which is under their expectations” (p.330).
So far I have established that humor increases consumers’ attention toward the ad and generates higher levels of arousal. In addition to that, the arousal experienced from the humorous ad also intensifies the involvement of the consumer when processing the ad (Andrews, Durvasula, and Akhter 1990). Furthermore, the element of surprise that accompanies humor “mobilizes cognitive resources that are used to resolve the inconsistency between what was expected and what is actually encountered, hence unexpected information results in extended causal reasoning, counterfactual thinking and cognitive elaboration” (Fennis and Stroebe 2010, p.55). Finally, the results of Campbell’s (1995) study suggest that the use of an attention-getting tactic, which utilizes stimuli that increase processing attention, arousal and involvement, is more likely to lead to consumers making inferences of manipulative intent.
A similar effect is expected from the use of humor on consumers’ inferences of manipulative intent. Thus, I propose the following first hypothesis:
H1: Consumers will make more inferences of manipulative intent when exposed to a humorous ad
This effect is however, expected to be mediated by the depth of information processing of consumers while and after being exposed to the ad. This study explores the idea that inferences of manipulative intent are the result of consumers’ increased willingness and interest to process the information of the persuasive message more deeply, which is brought about by consumers’ increased attention to the humorous stimuli.
2.3 Mediating Role of Information Processing Depth
Information processing depth is conceptualized in this study as the amount of effort that consumers might devote to thinking while and after being exposed to a persuasive message (Petty and Wegener 1999). This mediating variable is explained in the following subsections using two well-know models of processing of persuasive messages, namely the Cognitive Response Model and the Elaboration Likelihood Model (ELM).
2.3.1 Cognitive Response Model
The mediating role of viewers’ cognitive responses to persuasive communication is not new. As Greenwald (1968) described it, “when a person receives a communication and is faced with the decision of accepting or rejecting the persuasion, he may be expected to attempt to relate the new information to his existing attitudes, knowledge, feelings, etc. In the course of doing this, he likely rehearses substantial cognitive content beyond that of the persuasive message itself” (p. 149). What the author suggests with this model is that it is not so much the communication content, but the cognitive response content (i.e., the thoughts generated by the content of the persuasive message), that leads to acceptance or rejection of the persuasion attempt.
Through a series of experiments, the author obtained evidence that shows that the effects of persuasive communications are strongly mediated by cognitive responses elicited, rehearsed and learned during exposure. Moreover, the author describes the effects of the persuasive communications as persuasion when the viewer generates thoughts supporting the message, or
boomerang when the viewer generates opposing thoughts, also referred to as counterarguments.
Love and Greenwald (1978) tested the cognitive response theory in an experiment where they recorded cognitive responses from exposure to a persuasive communication and assessed recall and opinion change both immediately after and one week after the experiment took place. Measures of recall of communication content failed to predict opinion change, while recall of cognitive response content was moderately successful; these results support the notion that cognitive response content is a stronger determinant of persuasion.
Moreover, Petty, Wells and Brock (1976) also argued that the valence of thoughts (i.e., favorable vs unfavorable thoughts) is an important aspect of cognitive responses and it depends on the quality of (the arguments in) the communication, so that cognitive responses to strong arguments are likely to be positive while cognitive responses to weak arguments are likely to be negative. The valence of thoughts is relevant in this study because I argue that due to the nature of humorous advertising compared to serious advertising, the former are more likely to be perceived as having more weaknesses and inconsistencies, and therefore as having weaker arguments. This leads to a higher likelihood of consumers’ thoughts towards humorous advertising being more negative. As a consequence of the negativity of these thoughts, increasing message relevant thinking (i.e. processing depth) should increase the likelihood of inferences of manipulative intent and reduce the persuasive power of humorous advertising.
In the previous sections I have established that viewers’ attention, arousal and involvement are increased with humorous advertisements leading to inferences of manipulative intent when exposed to this type of persuasive communication (Andrews, Durvasula, and Akhter 1990; Eisend 2009; Gulas and Weinberger 2006; Weinberger and Gulas 1992). Following the same logic, it seems possible to expect humor to have a similar effect on information processing depth. In other words, due to increased attention and involvement resulting from humorous advertising, the information processing depth is also likely to be higher in comparison to serious advertising. Therefore, I propose the following hypothesis:
H2: Depth of information processing will be higher when consumers are exposed to a humorous ad
than when exposed to a serious ad. 2.3.2 Elaboration Likelihood Model
The ELM postulates a continuum of message elaboration anchored at a high end called the central route to persuasion, and at a low end called the peripheral route to persuasion (Petty and Cacioppo 1984); in simple words, it determines how deeply an advertisement is processed.
Under the central route the consumer performs an effortful processing of information; meaning that he/she will develop attitudes toward the ad based on the quality of the arguments that are central to the product/brand and positive or negative cues, which are not central to the product/brand. Conversely, under the peripheral route the consumer processes the information more superficially and form attitudes mainly based on the positive or negative cues. Moreover, under the central route, the viewer focuses on relevant aspects of the communication (i.e., directly linked to the product or issue at hand) whereas under the peripheral route the focus is more on irrelevant aspects of the communication.
More importantly, based on the possible negative effects of elaborate processing (Greenwald and Leavitt 1984) and of increased processing attention (Campbell 1995), one could say that the central route is more likely to result in consumers counterarguing the information presented, making inferences of manipulative intent, and resistance instead of persuasion would be the likely result.
Furthermore, the amount of effort or elaboration processing is determined by the consumer’s motivation, opportunity and ability (MOA) to process the message (Hoyer, MacInnis, and Pieters 2013). Motivation to think can be influenced by the relevance and consequences of the persuasive appeal (Petty and Cacioppo 1984); people are more motivated to increase their efforts to deeply evaluate an issue or a product when their involvement is high rather than low (Petty, Cacioppo, and Schumann 1983).
Moreover, two of the most influential variables that determine ability and opportunity in the persuasion literature that are considered for the present study are cognitive resources and distraction (Campbell and Kirmani 2000; Cialdini, Petty, and Cacioppo 1981; Hoyer, MacInnis, and Pieters 2013; Petty and Cacioppo 1984). When consumers’ resources are not limited they are able to engage in elaborate processing of information and are more likely to perform more deep and thoughtful evaluations of a message. When distraction from the ad is low, the potential for elaboration is high and people are likely to process the message via the central route.
2.3.3 Processing of Humorous Advertising
Opinions regarding processing of information in humorous ads are mixed; on one hand, the increased attention as a result of humor allows people to process brand claims more actively (Smith 1993), but on the other hand, humor may also hinder people’s acquisition of relevant information from ads (Cline, Altsech, and Kellaris 2003). In the present study we will examine the first possible effect of humor.
The ELM states that when advertising is processed through the central route, it increases the likelihood of counterarguing and negative judgments. Furthermore, central route processing is likely when the consumer is motivated and able to think about the issue at hand. Accordingly, I propose that a consumer might process a humorous ad through the central route, leading to development of inferences of manipulative intent.
The reasoning behind this expected effect is the following: increased attention and arousal experienced from the humorous ad motivates consumers and increases involvement when processing the message; in addition to that, consumers who view an advertisement attentively are considered cognitively unbusy observers (see Campbell and Kirmani 2000) that have the cognitive resources needed to be able to process the ad centrally; moreover, the increased attention generated by humor is expected to reduce any possible undesired effects of distraction, so that consumers are more likely to turn away from a serious ad than from a humorous ad; and finally as mentioned before, the thoughtful evaluation of a humorous ad is more likely to result in weaknesses and inconsistencies that would then turn into couterarguments towards the advertisement.
Therefore, I propose the following final hypothesis to examine the mediating role of information processing depth in the relationship between type of advertisement and extent of consumers’ inferences of manipulative intent:
H3: Consumers’ inferences of manipulative intent will be higher when depth of information
processing is high compared to when it is low.
The following section contains a description of the methodology employed in this study.
3. METHOD 3.1 Selection of Stimuli and Pretest
3.1.1 Selected Stimuli
role of consumers’ depth of information processing in this relationship. In order to select two commercials that would properly represent the conditions, commercials broadcasted in English were considered. The chosen commercials had the same duration of one minute and were of the same brand (i.e., Audi) in order to avoid confounding effects that could arise from using different brands. The choices of stimuli for the present study where the following:
Serious ad. (https://www.youtube.com/watch?v=xZjykPITuq0) This commercial tells a story about how Audi’s first cars were made and how from then until now Audi has managed to do what was thought impossible and that couldn’t be done, excellent cars using true engineering.
Humorous ad. (https://www.youtube.com/watch?v=QeCW-M4hDYo) This commercial shows in a funny way what would happen to a little boy if he breaks the rules by exaggerating what would happen to him; then it ends by stating that with Audi you can be different in a world full of rules.
Decoy ad. (https://www.youtube.com/watch?v=OnNF-kJ5bI4) This ad was selected as the decoy ad because it is an informative ad of a very different product category, beauty products, hence significantly different from the experimental conditions. This ad was simply added as a filler ad in the experiment in order to simulate closer how consumers are exposed to advertisement in television, since consumers are exposed to more than one advertisement during commercial breaks.
3.1.2 Pretest Design, Participants and Procedure
Both target ads were pretested for their perceived seriousness and humorousness. The ads were tested in a short within-subjects design online survey, in which each participant was shown both the serious and the humorous Audi commercials. There were no missing values in the data and no participants were excluded from the analysis presented in the following sub-section.
Data were collected from 20 participants of whom the majority of respondents were female, as well as between 20 and 30 years old. For the majority of the sample the highest level of education reached was a bachelor’s degree. Finally, the majority of the respondents were European; table 1 presents the demographic statistics of the sample.
TABLE 1
Demographic Statistics of Pretest Sample
Male 35%
Gender
Female 65%
Between 20 and 30 85% Between 31 and 40 5% Over 40 5% High school 25% Bachelor’s degree 50% Master’s degree 20% Doctorate degree 0% Education Other degree 5% European 75% Nationality Non-European 25%
Moreover, participants were told they would be watching some videos and afterwards they would be answering some questions. They were first asked to answer the demographic questions, then they were shown the serious Audi commercial, and after watching it were asked to answer two 7-point semantic differential scales anchored by serious/funny and boring/entertaining as well as one attribution statement (i.e. The Audi car commercial I just saw made me laugh) varying from 1 (strongly disagree) to 7 (strongly agree). This 3-item scale was reliable (α=.70) and an overall
Serious_Index was created to measure participants perception of the serious Audi commercial.
After participants saw the first commercial and answered the scale, they were shown the humorous Audi commercial and after watching it were asked to answer the same scales. The 3-item scale in this case was also reliable (α=.76) and an overall Humorous_Index was created to measure participants perception of the humorous Audi commercial.
3.1.3 Pretest Results
A paired samples t-test was carried out to compare the means of the Serious_Index and the
Humorous_Index for the sample of pretest participants. Table 2 summarizes the results. On average,
TABLE 2
Participants’ mean scores of both serious and humorous Audi ads Mean Standard Deviation Range Serious_Index 3.80 1.16 4.00
Humorous_Index 5.68 .85 4.00
Note: Serious_Index indicates the rating for the serious Audi commercial and Humorous_Index indicates the rating of the humorous Audi commercial
3.2 Research Design and Participants
An experiment was conducted to test all of the hypotheses proposed in this study. It was conducted online and participants were reached by distributing a link via email and social media sites.
The experiment followed a 2-conditions between-subjects design, in which the independent variable type of advertisement was manipulated by randomly assigning participants to one of the two experimental conditions (serious advertising vs. humorous advertising); each participant viewed only either the humorous or serious ad. The mediating variable depth of information
processing and the dependent variable extent of inference of manipulative intent were measured,
this is further described in section 3.4.
Data were collected from 110 participants in total. However, 6 of the participants were deleted because they indicated to have skipped the commercials in the video, 2 were from the serious advertising condition and 4 were from the humorous advertising condition. Therefore, there were 104 participants left in the analyses. Moreover, the humorous ad was viewed by 57 participants while the serious ad was viewed by 47 participants. Both of these sub-samples were similar in terms of demographics, meaning that random assignment of participants to conditions was successful. No missing values were found in the final sample.
TABLE 3
Demographic Statistics of Overall Sample and Conditions Overall Sample n = 104 Serious Condition n = 47 Humorous Condition n = 57 Male 46% 45% 47% Gender Female 54% 55% 53% Under 20 6% 4% 7% Between 20 and 30 84% 92% 79% Between 31 and 40 6% 4% 7% Age Over 40 4% 0% 7% High school 14% 13% 16% Bachelor’s degree 51% 47% 54% Master’s degree 28% 36% 21% Doctorate degree 2% 2% 2% Education Other degree 5% 2% 7% European 64% 57% 70% Nationality Non-European 36% 43% 30% 3.3 Procedure
first, participants were told that since it has been found that advertisements sometimes affect people’s opinions and feelings towards television programs, they first needed to answer some questions about the target ad (i.e., the Audi commercial).
At the beginning of the experiment, participants were asked to fill in demographic information. When they were done, all participants began watching the same video. After 2 minutes of the video, all participants saw the same decoy ad but were randomly assigned to see one of the two target ads. Once the participants saw both commercials, they continued watching the rest of the video. Immediately after the video was finished, the participants were asked to write down any thoughts, ideas, and feelings they had about the target ad. They were given 3 minutes to write each thought in a separate sentence. When they were finished or ran out of time, they were asked to answer an inference of manipulative intent scale (Campbell 1995). Furthermore as a manipulation check, participants answered how they perceived the target ad with two 7-point semantic differential scales anchored by serious/funny and boring/entertaining as well as one attribution statement (i.e., The Audi car commercial I just saw made me laugh) varying from 1 (strongly
disagree) to 7 (strongly agree).
After all the relevant variables were measured, participants answered a few questions regarding the video documentary as well as some questions to control for how often participants watch that type of programs, how interested they are about the topic, and if they had seen it before. Finally, as a control measures for mood and distraction participants were asked to answer first a mood scale (Peterson and Sauber 1983) and then a one 7-point semantic differential scale anchored by not at all distracted and very distracted. At that point the experiment was over, the participants were debriefed about the true purpose of the study and thanked for participating.
3.4 Measures
Because of the interest in the process by which inferences of manipulative intent can arise, the measurements were administered in the following order to avoid contamination of processes.
3.4.1 Demographics
Participants were asked to answer a number of standard demographic questions regarding their age, gender, country of residence, and level of education.
3.4.2 Depth of Information Processing measure
Thought Listings. Collected thought listings by asking participants to write any thoughts,
ideas, and feelings they had while viewing the ad: It has been found that sometimes advertisements
writing down these thoughts, please separate them into individual ideas and write them down as separate sentences.
3.4.3 Extent of Inference of Manipulative Intent measure
Inference of Manipulative Intent Scale. Participants responded to a six-item scale
developed by Campbell (1995) containing five attribution statements with responses varying from 1 (strongly disagree) to 7 (strongly agree) (e.g., The way this ad tries to persuade people seems acceptable to me; The advertiser tried to manipulate the audience in ways that I don’t like) and one 7-point semantic differential scale anchored by unfair and fair (see Appendix 1 for full scale). The scale was highly reliable (α=.91).
3.4.4 Manipulation check
Participants responded to two 7-point semantic differential scales anchored by serious/funny and boring/entertaining as well as one attribution statement (i.e., The Audi car commercial I just saw made me laugh) varying from 1 (strongly disagree) to 7 (strongly agree). The scale was reliable (α=.79).
3.4.5 Other measures
Decoy Documentary Questions. Participants responded to a six-item scale containing
attribution statements with responses varying from 1 (strongly disagree) to 7 (strongly agree) (e.g., I found the documentary entertaining; I found the documentary useful).
Control Measures for Documentary. Participants answered how often they watch
television and how often they watch documentaries with responses varying from 1 (never) to 7 (all
the time). They also indicated how interested they are in sharks with responses varying from 1 (not at all interested) to 7 (very interested). Finally, they indicated whether they had seen the
documentary before or not.
Control Measure for Mood. Participants responded to the Mood Short Form (MSF), a
four-item scale developed by Peterson and Sauber (1983) composed of attribution statements with responses varying from 1 (strongly disagree) to 7 (strongly agree) (e.g., Currently, I am in a good mood; As I answer these questions I feel cheerful) (see Appendix 1 for full scale). The scale was highly reliable (α=.88).
Control Measure for Distraction. Participants responded to a one 7-point semantic
3.5 Coding Procedures and Creation of Variables from Qualitative Data
3.5.1 Information Processing Depth Score
Thoughts were coded by whether they indicated central processing (i.e. thoughts related to the intention of the ad = 1) or peripheral processing (i.e., thoughts related to the content of the ad = -1) and an information processing depth score per participant was computed as the number of central thoughts plus the number of peripheral thoughts. This newly computed scores were added as a new variable.
Examples of central thoughts given by participants were “I was annoyed that the commercial was trying to appeal to an emotional side of me” and “Initially couldn't make the connection between the car, the boy and the woman e.g. who's the target audience of this commercial?”. Moreover, examples of peripheral thoughts given by participants were “I found the part where the kid is greeting other tough prison inmates to be the best part” and “I quite liked the Audi commercial. It was a short documentary in itself about cars, so I was interested”.
3.5.2 Processing Depth Category
Based on their final information processing depth score, participants were assigned to a processing depth category by creating a new variable. If after the coding procedure described previously (i.e., central thought = 1, peripheral thought = -1) the final processing depth score of a participant was positive, then this participant was assigned to the high/central processing category. On the other hand, if the final processing depth score of a participant was zero or negative, then this participant was assigned to the low/peripheral processing category.
These last two coding procedures (i.e., processing depth score and processing depth category) were performed to analyze the depth of information processing. Similar procedures have been used by Chaiken (1980), Love and Greenwald (1978), Osterhouse and Brock (1970), and Greenwald (1968).
3.5.3 Thought Valence Score
To measure the overall favorability of thoughts generated by the ads, thought listings were coded as positive (1), neutral (0), or negative (-1) and a thought valence score per participant was computed as the number of negative reactions plus the number of positive reactions. This scores were then included as a new variable.
Audi and focused on their achievements through time” and “The Audi commercial is for both adults and children”. Examples of negative thoughts given by participants were “It was a nice car commercial, but I found it to be a bit boring, nothing special” and “I was not convinced to want an Audi or to break any rules just because of that commercial”.
3.5.4 Amount of Thoughts Listed
The amount of thoughts listed by each participant was counted and made into a new variable.
3.6 Data Analysis
In order to test the model and hypotheses of this study a mediation test was carried out using two different approaches, namely causal steps approach (Baron and Kenny 1986) and bootstrapping (Hayes 2009). Moreover, additional analyses of variance were performed to further test the hypotheses. The following section presents a detailed description of the analyses mentioned here together with their results.
4. RESULTS 4.1 Manipulation Check
An independent samples t-test was carried out to compare participants’ responses to how humorous or serious they perceived the target ad they were exposed to in their respective condition. The serious vs. humorous index means and standard deviations of the two experimental conditions are summarized in table 4. Higher numbers indicate that the ad was perceived as more humorous. On average, participants rated the humorous Audi commercial significantly more humorous (M = 4.82, SE = .20) than the serious Audi commercial (M = 3.36, SE = .16), t(102) = -5.70, p = .00. This indicated that the manipulation was successful.
TABLE 4
Serious vs. Humorous Index Means and SD per Condition
Condition: Type of Advertisement Mean Standard Deviation Serious 3.35 1.08 Serious vs.
4.2 Correlations
As a first step before testing the hypotheses of this study, correlations were tested with the purpose of observing whether other variables have an effect on the variables of interest in this study. Appendix 2 contains the correlations table of the independent variable type of advertisement, the mediating variable information processing depth score and the dependent variable inference of manipulative intent score, together with the additional variables processing depth category, thoughts valence score and amount of thoughts listed; as well as demographic variables and the two main control variables mood and distraction. The reason for this was to identify possible covariates affecting the mediating and dependent variables as well as any of the additional variables which are used to further test the hypotheses, in order to control for their effect in subsequent analyses.
Relationships between main variables and additional variables. The results show that
there is a marginally significant positive relationship between type of advertisement and information processing depth score, r = .18, p = .07. This means that with humorous advertising the information processing depth score increases. There is also a marginally significant positive relationship between type of advertisement and inference of manipulative intent score, r = .18, p = .07. This means that with humorous advertising the inference of manipulative intent score also increases. Moreover, there is a significant positive relationship between type of advertisement and processing depth category, r = .23, p = .02, meaning again that with humorous advertising processing depth increases. Information processing depth score is also obviously positively related to processing depth category, r = .72, p = .00, because a positive (negative) score is related to the high (low) category.
Furthermore, information processing depth score has a significant negative relationship with thoughts valence score, r = -.26, p = .01. This means that as information processing depth increases the valence of thoughts decreases. Similarly, inference of manipulative intent score is negatively related to thoughts valence score, r = -.61, p = .00, meaning that as the valence of thoughts increases the extent of inference of manipulative intent decreases. Moreover, there is a significant positive relationship between type of advertisement and amount of thoughts listed, r = .19, p = .05, indicating that with humorous advertising the amount of thoughts listed increases. Finally, processing depth category is also positively related to amount of thoughts listed, r = .24, p = .01, meaning that high processing depth also increases the amount of thoughts listed.
Effects of demographics variables and control variables. There are no significant
score and mood, r = -.28, p =.00. This indicates that as mood increases the extent of inference of manipulative intent decreases. Similarly, there is a significant positive relationship between thoughts valence score and mood, r = .30, p = .00, indicating that as mood increases the valence of thoughts also increases. For these reasons, the variable mood was included in further analyses as covariate for the variables inference of manipulative intent and valence of thoughts.
Moreover, there is a significant negative relationship between distraction and processing depth category, r = -.22, p = .02. This indicates that as distraction increases processing depth decreases. Similarly, there is a marginally significant negative relationship between distraction and amount of thoughts listed, r = -.22, p = .06, indicating that as distraction increases the amount of thoughts listed decreases. Finally, there is a highly significant negative relationship between the two covariates mood and distraction, r = -.39, p =.00, meaning that as one of them increases the other decreases.
4.3 Hypotheses Testing – Mediation Analysis
In order to test the full model of this study, a mediation test was performed. Mediation can
be tested by estimating a series of regression models as recommended by Baron and Kenny (1986), this approach is called the causal steps approach. However, this method has been found to have substantial drawbacks and thus another method, bootstrapping, is recommended as it is a more valid and powerful approach for testing intervening variable effects (Hayes 2009).
Using a special PROCESS developed by Hayes (2013) it is possible and convenient to test mediation using this method. The objective was to test whether the effect of type of advertisement on the extent of inference of manipulative intent is mediated by depth of information processing. At the same time, the effect of the covariate mood on inference of manipulative intent was controlled for.
A full mediation effect was expected; in other words, I expected the indirect effect of type of advertisement on the extent of inference of manipulative intent to be much more significant and larger than its direct effect.
Causal steps approach. The results of this first approach consisting of a series of
TABLE 5
Mediation Results – Causal Steps Approach Dependent Variable IMI°
Score Processing Depth Score (mediator) IMI° Score Model 1 Model 2 Model 3 Independent Variables
Type of Advertisement .23** .40* .21*
Information Processing
Depth Score (mediator) .04
Mood (covariate) -.29*** -.29***
R Square .12 .03 .12
° Inference of Manipulative Intent ***p < .01 / **p < .05 / *p < .10
Model 1 of the mediation model was overall significant, F(2,101) = 6.62, p = .00. It shows that the regression of type of advertisement on inference of manipulative intent score, ignoring the mediator, was significant, b = .23, t = 2.02, p = .05. Thus, this result established that there is an effect that may be mediated. Additionally, the direction of the effect is as expected; hypothesis one of this study suggested that consumers will make more inferences of manipulative intent when exposed to a humorous ad than when exposed a serious ad. The results indicate that humorous type of advertisement increases the inference of manipulative intent score, thus hypothesis one is supported.
Model 2, also called “path a” of the mediation model, was overall marginally significant, F(1,102) = 3.39, p = .07. It shows that the regression of type of advertisement on information processing depth score was marginally significant, b = .40, t = 1.84, p = .07. Thus, this result established that there is a marginally significant effect of the independent variable type of advertisement on the mediator depth of information processing. Additionally, the direction of the effect is as expected; hypothesis two of this study suggested that depth of information processing will be higher when consumers are exposed to a humorous ad than when exposed to a serious ad. The results indicate that hypothesis two is supported.
information processing is high compared to when it is low. However, according to these results there is no effect of information processing depth and therefore hypothesis three is not supported.
Bootstrapping approach. This approach served to test the significance of the indirect effect
of type of advertisement on the extent of inference of manipulative intent (i.e. the effect via information processing depth). The results showed that there is not a significant indirect effect, b = .01 with a 95% confidence interval which did include zero, CI [-.02, .08]; that is to say the effect was not significantly greater that zero at α = .05.
Therefore, this approach confirmed the previous results that indicated that information processing depth does not mediate the relationship between type of advertisement and extent of inference of manipulative intent.
4.4 Hypotheses Testing – Additional Analyses of Variance
In this section, the results of additional analyses are presented. The purpose of these tests was to further explore the hypotheses proposed in this study.
Inference of Manipulative Intent. To further test hypotheses one and three, which suggest
that higher inferences of manipulative intent occur when exposed to a humorous ad compared to a serious ad and that inferences of manipulative intent are also higher when depth of information processing is high compared to when it is low, a two-way ANCOVA was conducted.
This test was conducted in order to compare the inference of manipulative intent mean scores of humorous vs. serious types of advertisement and central/high vs. peripheral/low processing depth categories, while accounting for the effect of the covariate mood on inference of manipulative intent. Results are displayed in table 6 and figure 2, higher scores indicate that participants made more inferences of manipulative intent.
The results of this two-way ANCOVA showed that the covariate, mood, was significantly related to inference of manipulative intent scores, F(1,103)= 10.02, p= .00, ηp2= .09. Moreover,
there was a significant main effect of type of advertisement on inference of manipulative intent, F(1,103)= 5.41, p= .02, ηp2= .05, but no significant main effect of processing depth category on
inference of manipulative intent, F(1,103) = 2.28, p= .13, ηp2= .02, after controlling for the effect of
TABLE 6
Inference of Manipulative Intent Score Means and (SD) Processing Depth Category
Type of Advertisement High/Central Low/Peripheral Total Serious 2.33 (.17) 2.73 (.99) 2.71 (.96)
Humorous 2.79 (1.20) 3.24 (1.36) 3.14 (1.33)
Total 2.71 (1.09) 2.99 (1.21)
FIGURE 2
Mean Inference of Manipulative Intent Scores
These results support hypothesis one. According to hypothesis one, it was expected to observe a higher average inference of manipulative intent score from the humorous type of advertisement than in the serious type. The total average inference of manipulative intent score for participants in the humorous
condition (M= 3.14) was higher than the total average score of participants in the serious condition (M= 2.71), thus the direction of the effect was as hypothesized.
Moreover, these results do not support hypthesis three. According to hypothesis three, a higher average inference of manipulative intent score was expected in the central/high processing depth category than the peripheral/low processing depth category. Despite the differences between the groups’ mean inference of manipulative intent scores not being statistically significant from eachother, some remarks regarding the direction of the effects are important. The total average inference of manipulative intent score for participants in the low/peripheral category was higher (M= 2.99) than the total average score of participants in the high/central category (M= 2.71), this results show the opposite direction than it was hypothesized.
Information Processing Depth. To further test hypothesis two, which suggests that depth of
This test was conducted in order to compare the information processing depth scores of humorous vs. serious types of advertisement. Results are displayed in table 7, higher scores (i.e., less negative scores) indicate that participants processed the advertisement deeper.
TABLE 7
Information Processing Depth Scores per Ad Type Type of Advertisement Mean (SD)
Serious -2.02 (1.87)
Humorous -1.23 (2.41)
The results showed that there is a marginally significant difference in the scores for serious advertisement (M = -2.02) and humorous advertisement (M = -1.23); t(102) = -1.84, p = .07. This suggests that type of advertisement has marginally significant effect on information processing depht.
Moreover, I expected higher average information processing depth score (i.e. higher amount of central thoughts) from participants in the humorous condition than in the serious condition. Despite the fact that the negative mean scores indicate that the majority of the thoughts that participants listed where peripheral, the results show that there is a difference between the two groups; where the average processing depth score of participants in the humorous ad condition is higher, or in this case less negative (M= -1.23) than the average processing depth score of participants in the serious ad condition (M= -2.02) indicating a higher amount of central thoughts in the humorous condition. This result is in line with the proposed effect and concludes that hypothesis two is supported.
Thought Valence. In this test I explored one of the additional variables, thought valence.
No specific hypothesis was formulated; however, this variable can be used to generate additional information of the effects of type of advertisement and depth of information processing, therefore the results could bring interesting insights to this study.
TABLE 8
Thought Valence Score Means and (SD) Processing Depth Category
Type of Advertisement High/Central Low/Peripheral Total Serious 1.00 (1.00) 1.30 (2.18) 1.28 (2.12)
Humorous .31 (2.93) .66 (2.42) .58 (2.52)
Total .44 (2.66) .98 (2.31)
The results of the two-way ANCOVA showed that the covariate, mood, was significantly related to thought valence score, F(1,103)= 10.40, p= .00, ηp2= .09. Moreover, there was no
significant main effect of type of advertisment on thought valence score, F(1,103) = .73, p= .39, ηp2= .00, as well as no significant main effect of processing depth category on thought valence
score, F(1,103) = .15, p= .70, ηp2= .00, after controlling for the effect of mood. There was also not a
significant interaction effect, F(1,103) = .00, p= .94, ηp2= .00.
Despite the differences between the groups’ mean thought valence scores not being statistically significant from each other, its important to mention some remarks regarding the
direction of the effects. I expected a lower average thought valence score (i.e. higher amount of
negative reactions) from participants in the humorous condition than in the serious condition. The resutls show that the total average thought valence score for participants in the humorous condition (M=.58) was lower than the total average score of participants in the serious condition (M= 1.28); thus, the direction of the effect was as hypothesized.
Moreover, I also expected a lower average thought valence score from participants in the high processing depth category than in the low processing depth category. The results confirmed such expectations, the total average thought valence score for participants in the low/peripheral category was higher (M= .98) than the total average score of participants in the high/central category (M= .44).
Amount of Thoughts Listed. In this test I explored another one of the additional variables,
amount of thoughts listed. No specific hypothesis was formulated; however, this variable can also be used to generate additional information of the effects of type of advertisement and depth of information processing. A two-way ANOVA was conducted in order to compare the average number of thoughts listed by participants exposed to humorous vs. serious types of advertisement and central/high vs. peripheral/low processing depth categories. Results are displayed in table 9, higher numbers indicate a higher amount of thoughts listed by participants.
TABLE 9
Amount of Thoughts Listed per Ad Type Processing Depth Category
Type of Advertisement High/Central Low/Peripheral Total Serious 2.67 (1.53) 2.61 (1.48) 2.62 (1.47)
Humorous 4.31 (3.09) 3.00 (1.35) 3.30 (1.94)
Total 4.00 (2.90) 2.81 (1.42)
The results of the two-way ANOVA showed that there is a marginally significant main effect of type of advertisement on the amount of thoughts listed by participants, F(1,103)= 3.09, p= .08, ηp2=
.03. There was no significant main effect of processing depth category on the amount of thoughts listed by participants, F(1,103) = 1.39, p= .24, ηp2= .01. There was also not a significant interaction
effect, F(1,103) = 1.18, p = .28, ηp2 = .01.
For these tests I expected a higher average number of thoughts listed from participants in the humorous condition than in the serious condition. The results indicate that there is a difference between the two groups, where the average number of thoughts listed by participants in the humorous condition was indeed higher (M= 3.30) than the average number of thoughts listed by participants in the serious condition (M= 2.62). Thus, the results confirmed the expectations of this analysis. In addition to that, I also expected the average number of thoughts listed by participants to be higher when processing depth was high compared to low. The results suggest that such an effect is not significant, however the direction of the effect is as expected, with the average number of thoughts in the high/central group (M= 4.00) being higher than the average in the low/peripheral group (M= 2.81).
inference of manipulative intent score was not significant, b = -.08, t = -1.14, p = .25. Therefore I conclude that the amount of thoughts listed by participants is not a good predictor of the extent of inferences of manipulative intent that they make.
5. DISCUSSION
This section describes the theoretical and managerial implications of the results presented in the previous section, as well as the limitations of this study and direction for future research.
5.1 Theoretical Implications
The purpose of this research was twofold, to understand the effect of humorous advertising on consumers’ inference of manipulative intent and whether the depth with which the information from advertising is processed mediates such an effect. Table 10 gives a summary of the hypotheses and whether they were supported by the results of the analyses described in the previous section.
TABLE 10
Summary of Hypotheses
Hypotheses Support
1 Consumers will make more inferences of manipulative intent when
exposed to a humorous ad than when exposed a serious ad. Yes
2 Information processing depth will be higher when consumers are
exposed to a humorous ad than when exposed to a serious ad. Yes
3 Consumers’ inferences of manipulative intent will be higher when
information processing depth is high compared to when it is low. No
Hypothesis one relates to the first question of interest in this study, namely what is the effect of a humorous ad on consumers’ inferences of manipulative intent compared to a non-humorous ad? According to the literature presented in sections 2.1 and 2.2, I predicted that a humorous ad would have a higher impact on consumers’ inferences of manipulative intent. This reasoning followed from the fact that humor increases the level of attention paid as well as the level of arousal, therefore consumers can become more involved, find inconsistencies and develop counterarguments toward the advertisement.
humor as well as negative effects of attention-getting tactics in advertising. Previous studies have explored the effects of humor on a numerous of different outcome variables (for an overview see Eisend 2009); however, no studies were found that explored its effect on inferences of manipulative intent. In addition to that, inferences of manipulative intent caused by attention-getting tactics used in advertising have been studied (Campbell 1995); however again, humor was not considered as one of those tactics. Therefore, the present study and this particular outcome are an important contribution to the growing literature concerning the negative side of humor in advertising.
Hypothesis two and three relate to the remaining questions of interest in this study, namely what is the effect of a humorous ad on consumers’ information processing depth compared to a non-humorous ad? and how does this information processing depth mediates the effect of type of advertisement on inference of manipulative intent? According to the literature presented in section 2.3, I suggested that humorous advertising would have a higher impact on information processing depth compared to non-humorous advertising; I also proposed that the resulting higher processing depth from humorous advertising would in turn increase consumers’ inferences of manipulative intent.
This reasoning was based on the fact that, as explained by the ELM and cognitive response model, it is the thoughts generated by consumers while exposed to advertising that determines how deeply an ad is processed and whether it leads to the acceptance or rejection of the persuasive message. In this study, thoughts that participants expressed about the advertisements during the experiment determined the depth of information processing (central vs. peripheral) while the corresponding level of inference of manipulative intent would indicate the extent to which the advertisement was successful at changing consumers’ opinion towards the product or brand.
Moreover, it was expected for humorous advertising to have a similar effect on information processing depth as it has on inference of manipulative intent, meaning that the increased attention and involvement created by humorous advertising compared to non-humorous advertising would lead to deeper processing of information. The results of this study confirm the expected outcome of the effect of type of advertisement on depth of information processing demonstrating that attention and arousal do increase consumers’ effort devoted to processing persuasive communication.
elaborate processing resulting from increased attention can lead to consumers counterarguing the information and making inferences of manipulative intent (Campbell 1995; Chaiken 1980).
However, a difference between Chaiken’s (1980) study and the present study is that in that study motivation to process the information and involvement were manipulated by instructing participants to process an advertisement in an elaborate way or not, while in this study participants were only presented with a humorous or serious advertisement expecting that the humorous ad would increase participants motivation and involvement through attention and arousal. Therefore, the level of information processing as well as its effect on inference of manipulative intent was treated in a more realistic way by presenting the advertisements the same way that consumers would normally encounter them on television and allowing them to decide whether they felt motivated to process the advertisements deeply.
Moreover, according to the literature, there are other ways to assess the depth of processing of information and the success of a persuasive message, namely amount of thoughts and valence of thoughts (Osterhouse and Brock 1970). This was also explored in this study as to provide additional information. Thought valence is an indication of the acceptance of an advertisement and therefore should be related to the extent of inference of manipulative intent that a consumer could make. In comparison to previous studies, in this study I did not find a large difference between types of advertising nor processing depth categories in thought valence. However, I did find that in line with expectations, thought valence strongly predicts the level of inference of manipulative intent. Since inference of manipulative intent is likely to arrive from inconsistencies and counterarguments found in advertisements, makes sense that negative thoughts will lead to higher inferences of manipulative intent, while the opposite is for positive thoughts.
The amount of thoughts on the other hand, is an indication of the level of processing that participants engaged in. One could argue that a higher amount of thoughts listed would result in deeper information processing; however, this is not necessarily the case. Even with a large number of thoughts, it could be that the majority of these thoughts reflect peripheral processing and as a result information processing depth would still be low rather than high. This reasoning is backed up by the results obtained in this study that show no difference of processing depth categories on the amount of thoughts listed by participants. Finally, compared to thought valence, amount of thoughts listed is not directly related to the extent of inference of manipulative intent.
on manipulative intent. The valence of thoughts however, does have an impact.
5.2 Managerial Implications
Based on the insights from this study, there are some practical implications that are important to mention. When making a choice between humorous or serious advertising, marketers should be aware of the fact that humor in advertising can be in most cases an effective tactic when it comes to grabbing consumers’ attention and increasing effortful processing of the message and information, but its effect should not be taken for granted. The reason I point this out is because at the same time humorous advertising seems to increase consumer inferences of manipulative intent. One of the reasons this happens is because it tends to generate more negative thoughts than positive thoughts, something that normally one would not have thought of about humor.
A way to prevent or lessen this effect of humor on inferences of manipulative intent would be to pay close attention to the strength of the arguments presented in the advertisement. Why? Because valence of thoughts depends on the quality of arguments, so that responses to strong arguments tend to be positive. Moreover, an outcome of this research points to the fact that generating positive thoughts in consumers when using humorous advertising is more important than encouraging consumers to thing about a product or brand deeply. Thus, when wanting to use humor in advertising as to attract consumers’ attention, marketers need to make sure that the core of their message contains strong arguments so that it will increase the likelihood of generating positive thoughts and reduce the likelihood of generating negative ones that would lead to inferences of manipulative intent.
5.3 Study Limitations and Future Research
One of the limitations of this study is the characteristics of the sample obtained. Since the majority of the sample where European young adults of ages between 20 and 30 years old and with a bachelor’s or master’s degree, it is not representative of an entire community of consumers. Therefore, the results of the study cannot be generalized to that extent.
Another limitation of this study was the online experimental method; when participants were taking part in this study they might have been distracted (even if they indicated not being distracted in the survey), which in this specific study could have an effect specially when it comes to listing their thoughts about the advertisements. It is possible that the same experiment if conducted in a laboratory setting with a more heterogeneous sample, would yield more and interesting insights. Thus, this is a possibility and recommendation for future research.
