Running head: HUMOR AND ADVERTISING 1 Humor in Advertising: An Associative Processing Model

This article reviews a research program on the effects of humor in advertising on positive and negative brand associations and brand choice, and integrates the findings into a single, overarching model. Based on the Associative and Propositional Processes Model of Evaluation (Gawronski & Bodenhausen, 2006, 2007, 2011), we propose that repeated pairings of a novel brand with brand-unrelated humor forms positive brand associations, which mediate spontaneous brand choice. This associative process was found to be independent from the level of distraction posed by humor and from awareness of the stimulus pairings. In fact, the distraction posed by humor benefits persuasion by preventing negative brand associations. Previous marketing research, which mainly viewed humor as a cue in peripheral processing, was rather pessimistic about the persuasive impact of humor. In contrast, this research program suggests that a repeated pairing of a brand with humor affects the brand’s underlying associative structure, which may lead to stable attitude changes that guide overt spontaneous brand choice. Theoretical and practical implications are discussed. Running head: HUMOR AND ADVERTISING 3 Humor in Advertising: An Associative Processing Model Persuasion is among the central issues of social behavior, and persuasion attempts can be found almost everywhere. Just walking down the street or surfing the Internet we are already exposed to countless advertising messages. When we read a newspaper or turn on the TV, we see political candidates trying to sway people to vote for them. Even at home, our friends and spouses try to lure us into doing things like join their birthday parties, babysit their children, or take out the trash. The literature on persuasion has addressed a multitude of persuasion variables. Some variables influence deliberative persuasion processes such as the impact of argument quality, biased information processing, and cognitive dissonance, whereas others influence intuitive persuasion processes such as the impact of source credibility, source attractiveness, and the sheer number of arguments. Somewhat surprisingly, it has left out systematic research on one very prevalent message element: the use of humor. It is estimated that between 30 and 42% of ads are intended to be humorous (Markiewicz, 1974; Weinberger, Spotts, Campbell, & Parsons, 1995). In the present article, we aim to address this gap in the social psychological literature by reviewing a research program on the effects of humor in advertising and by introducing an associative model of humorous advertising. Prior Research on Humor in Advertising Although social psychologists have not paid much research attention to the topic, humor is one of the most frequently used and studied message strategies in the advertising and marketing literature. This research has revealed several interesting and sometimes paradoxical effects of humor on various marketing outcomes. There is broad agreement among advertising researchers and practitioners that humor enhances the amount of attention paid to ads (e.g., Madden & Weinberger, 1982). Several Running head: HUMOR AND ADVERTISING 4 experiments demonstrated that humor directly increases positive attitudes towards ads and brands (Chung & Zhao, 2003; Chattopadhyay & Basu, 1990; Cline & Kellaris, 1999; Gelb & Zinkhan, 1986; Lee & Mason, 1999). Humor can also enhance persuasion indirectly by positively biasing ad elaboration (e.g., Allen & Madden, 1985), or by increasing motivation to process ads (e.g., Zhang & Zinkhan, 2006). Other researchers have noted that humor may disrupt critical processing of advertising claims (Cline & Kellaris, 1999), and may reduce negative responses to advertisements like counterarguing (Nabi, Moyer-Guse, Byrne, 2007) or reports of reactance (Skalski, Tamborini, Glazer, & Smith, 2009, for reviews see Eisend, 2009, 2011). On the downside, there is also evidence that humor can harm the memory for products and brand claims (e.g., Gelb & Zinkhan, 1986; Krishnan & Chakravarti, 2003), which suggests that humor distracts attention from products. Even more importantly, marketing studies generally found that humor has little or no effect on behavioral persuasion variables such as behavioral compliance, purchase intentions, or brand choice (Chattopadhyay & Basu, 1990; Scott, Klein, & Bryant, 1990; Zhang & Zinkhan, 2006). In fact, a comprehensive review about the effect of humor in advertising concluded “the current conclusion from the overall literature concurs with the view that humor does not offer significant advantages over non-humor when persuasion is the goal” (Weinberger & Gulas, 1992, pp. 56-57). This pessimistic view about the persuasive impact of humor among marketing researchers could be explained by the fact their research was mainly guided by the Elaboration Likelihood Model (ELM, Petty & Cacioppo, 1986). Humor was assumed to affect consumer behavior only through the peripheral route to persuasion. Persuasion through the peripheral route occurs when the consumer is unable or unwilling to engage in much message-relevant thought, and instead uses simple Running head: HUMOR AND ADVERTISING 5 mental shortcuts to process a persuasive message (e.g., the sheer number of arguments presented or the attractiveness of the source). Peripheral processing only generates weak and short-lived attitude changes that are hardly persuasive on a behavioral level (Petty & Cacioppo, 1986). Our Research Approach In the present article we deviate from the ELM approach, and instead propose an associative processing model of humor in advertising. We start by the notion that humor is a source of positive affect, and humorous ad campaigns usually present viewers with multiple brand-humor pairings. We propose that this repeated humorbrand pairing forms positive brand associations that guide spontaneous brand choice. In contrast to peripheral processing, associative processing is thought to always influence brand attitudes; when consumers process the ad only superficially, but also when they process it more thoroughly (although this is a domain of debate). Brand associations formed on the basis of multiple pairings with affective stimuli are robust (Sweldens, Osselaer, & Janiszewski, 2010) and are considered fundamental for persuasion (Greenwald & Leavitt, 1984; Osselaer & Janiszewski, 2001). Compared to the general conclusions of marketing research, our model proposes that humorous ad campaigns can promote overt brand choice, which may help to explain why so many advertisers choose to use humor in their ads. Importantly, our model also speaks to the particular psychological mechanisms involved in the processing of humorous ads. Although marketing research has made important progress in identifying the kinds of messages, market segments, and consumer types in which humor can impact persuasion, it spoke less to the particular psychological mechanisms involved. Besides using a different theoretical approach, we also used a different experimental approach than most previous marketing studies. We chose to employ a Running head: HUMOR AND ADVERTISING 6 less obtrusive, and in some respects more ecologically valid way to investigate the effect of humorous advertising. In the majority of marketing studies participants were presented with a single humorous or non-humorous ad, and afterwards reported their thoughts about the ad and brand, and/or rated their attitudes and purchase decisions on explicit Likert-type rating scales (e.g., Chattopadhyay & Basu, 1990; Cline & Kellaris, 1999; Lammers, Leibowitz, Seymour, & Hennessey, 1983; Lee & Mason, 1999; Zhang & Zinkhan, 2006). An important limitation was that the research relied exclusively on self-report measures to assess attitudes and purchase intentions. Moreover, the research participants were usually aware of the experimenter’s goal to investigate advertisements. It is unlikely that the explicit way in which research participants were consulted about their opinions about ads in marketing studies resembles the way consumers deal with ads in the real world. In real advertising situations consumers are frequently exposed to multiple humorous ads promoting the same brand, and this happens mostly involuntarily and without being asked for an opinion afterwards. We also took into account the fact that automatic processes may govern the impact of ads on brand choice. The average consumer is exposed to thousands of ads per day (Jhally, 1998). This abundance of advertising information exceeds the attentional resources of most consumers, which means that a lot of product information passes by unnoticed and will not be deliberately evaluated at all (see Heath, 2001). Although consumers are usually inattentive of what is said about brands in ads, they certainly form impressions of them. They unintentionally pick up brand information throughout the day, from tasting, using or wearing a particular product to seeing its particular packaging or logo in the shop or in a marketing campaign. When making a spontaneous purchase decision (e.g., in the supermarket), consumers often Running head: HUMOR AND ADVERTISING 7 rely on this automatically retrieved brand information (see Bargh, 2002; Dijksterhuis, Smith, Van Baaren, & Wigboldus, 2005). Furthermore, we took into account that real advertising situations potentially evoke resistance and ad avoidance. Ads typically force themselves upon consumers, sometimes in unwanted places (e.g., in e-mail and SMS text messages), and/or at inconvenient times (e.g., interrupting a popular TV show). More generally, every ad is an overt attempt to change the opinion and behavior of its perceiver, which automatically triggers resistance (Sherman, Crawford, & McConnell, 2004; for a review, see Knowles & Linn, 2004). Thus, compared to experimental participants on whom previous research conclusions were based, actual consumers may experience more resistance when confronted with commercial advertising, and may react with more avoidance or irritation. To simulate a situation of general low involvement and information overload, we presented brands in an information-rich environment including other brands, various humorous and non-humorous stimuli and lots of filler material. Care was taken to disguise the hypotheses under investigation, the particular brands and ad variables we were interested in, and the particular pairings between brands and humorous stimuli. We also minimized the obtrusiveness of the dependent measurements by using implicit measures of memory and attitudes, and behavioral observations. In several experiments we examined the impact of resistance by experimentally manipulating resistance, measuring individual differences in resistance, or employing field settings in which we encountered natural resistance. Before presenting the model it is useful to define the term humor as it is used here. The present work subscribes to an incongruity-resolution theory of humor, which defines humor as a cognitive-linguistic problem-solving task that elicits Running head: HUMOR AND ADVERTISING 8 positive affect (Goel & Dolan, 2001; Raskin, 1985; Suls, 1972). A typical joke contains a setup that causes perceivers to make a prediction about the likely outcome. The punch line violates this expectation, and perceivers will look for a cognitive rule that will make the punch line follow from the setup. When this cognitive rule is found, the incongruity is resolved and the joke is perceived as funny. Thus, the processing of humor comprises two characteristic stages: (1) resolution of a schema-incongruity in order to “get the joke” (cognitive stage), and (2) experience of positive affect (affective stage). Content analyses show that indeed about 69 to 82% of humorous ads throughout the world are based on incongruity-resolution (Alden, Hoyer, & Lee, 1993; Spotts, Weinberger, & Parsons, 1997). It is important to note that our experiments and model only concern low involvement consumer decisions, which can be defined as spontaneous choices between brands or products that consumers make without much deliberation of the choice alternatives (e.g., choosing between two brands of ballpoints or soft drinks). Indeed, the consumer market segments that most frequently employ humor in advertisements (ads) are non-durable, low involvement products such as snack foods, soft drinks, and beer (Madden & Weinberger, 1984; Weinberger & Campbell, 1991). As mundane as these spontaneous consumer decisions may seem, the sales of low involvement consumer goods amount to billions of dollars. The soft drink market is worth $64 billion in the United States alone (Oakland Tribune, 2004). Moreover, these kinds of small decisions that consumers make in daily life have a significant impact on medical, social, and economic outcomes on both individual consumers and society as a whole (Verplanken & Wood, 2006). Thus, we think it is important to examine how advertising influences low involvement consumer decisions. Running head: HUMOR AND ADVERTISING 9 We also wish to point out that our theory and experiments are only relevant to advertising in which the humor is unrelated to the brand. We chose to study only brand-unrelated humor because it is more common in advertising than brand-related humor. According to content analyses of Spotts and colleagues (1997), 71% of humor in ads worldwide is brand-unrelated. Thus, although we cannot draw conclusions about the effects of brand-related humor, we can still generalize our findings to most humorous ads. Although our research does not pertain to deliberate consumer decisions (e.g., deciding on houses, political voting) or to brand-related humor, we do speculate on the effect of humor in these domains in the discussion. Finally, we chose to study only novel instead of mature brands because this gives a clear picture of the effects of humor on brand perceptions and brand choice per se, without having to incorporate possible confounding factors like prior knowledge and experience with the brand. Thus, our findings can only be generalized to novel, non-mature brands, as previous research has indicated that humor affects novel and mature brands in qualitatively different ways (Chattopadhyay & Basu, 1990; see also Gibson, 2008). Foundations of the Model Our model derives in essential ways from three theoretical approaches. From the research on schema-incongruity (e.g., Heckler & Childers, 1992; Houston, Childers, & Heckler, 1987) we derived our hypotheses regarding the effects of humor on attention and memory processes. Research on schema-incongruity distinguishes two types of incongruity in ads: unexpectedness and irrelevance. Unexpectedness refers to advertising information that is somehow incongruent with prior expectations or schemas. Consider, for example, a TV ad for a British food manufacturer showing an Englishman ordering a shepherd's pie in a restaurant. After taking the order, the Running head: HUMOR AND ADVERTISING 10 waiter suddenly jumps on the table and starts singing “O sole mio”. Research shows that unexpected information (i.e., the waiter jumping on the table and starting to sing) receives enhanced attention and elaborative processing. Similarly, our model predicts that the inherent schema-incongruity in humor is likely to attract attention and enhance recall of the humorous parts of the ad. Irrelevance, on the other hand, concerns the extent to which a piece of information pertains directly to, and helps the identification of, the theme or message of a commercial. Research shows that irrelevant information receives little attentional processing and, if encoded, is poorly linked within the associative network. Because the waiter singing “O sole mio” is irrelevant for shepherd's pie (it would be more relevant, however, if the guest ordered lasagna), this product will be poorly processed. Similarly, our model predicts that humor does not support the processing of advertising information that is irrelevant to the humor, such as a relatively meaningless brand name (see also Krishnan & Chakravarti, 2003). In fact, given that humor requires solving incongruities and therefore poses cognitive demands, it is likely that humor withdraws cognitive resources from the processing of humorunrelated brand names. Thus, we hypothesized that the solving of incongruities within humorous ads impairs memory for humor-unrelated brands. A second theoretical approach we relied heavily on is the Associative and Propositional Processes Model of Evaluation (APE-model) of Gawronski and Bodenhausen (2006, 2007, 2011). We used this model to derive our hypotheses regarding the effects of humor on implicit attitudes and explicit attitudes. This model differentiates two qualitatively distinct mental processes related to attitude change, namely associative or propositional processes. Associative processes are defined as mental processes based on the activation of mental associations in memory, which are Running head: HUMOR AND ADVERTISING 11 formed by feature similarity and contiguity between stimuli in space and time. In contrast, propositional processes are defined as the validation of the information that is implied by the activated associations, which is guided by the principles of logical reasoning. The most important difference between associative and propositional processes is their (in)dependency of subjective truth. Whereas associations can be activated in memory regardless of whether the person considers the information implied by these associations accurate or inaccurate, propositional processes are inherently concerned with a subjective assessment of the validity of activated information. For example, the propositional implication of a positive automatic reaction to a brand on a supermarket shelf (e.g., “I like Brand X cookies”) may be rejected when it is inconsistent with other propositions that are currently considered (e.g., “Brand X cookies are very heavily advertised” and “I shouldn’t buy cookies just because they are advertised a lot”). The behavioral outcomes of propositional processes are typically measured with self-report scales (also termed explicit attitude measures), while the behavioral outcomes of associative processes are typically measured with response latency measures (also termed implicit attitude measures). Implicit attitude measures are better measures of the outcome of associative processes because they tap directly into the evaluations that are automatically activated upon exposure of the attitude object. One particularly thorny issue is the use of the term implicit attitude. Recently, researchers have noted that although the measure is implicit, the attitude is not, since there is no evidence that the individual is unaware of the attitude (Fazio & Olson, 2003; Gawronski, LeBel, & Peters, 2007). We completely agree with this position. Running head: HUMOR AND ADVERTISING 12 However, we use the term implicit attitude in this article because it is common in the