Resistance is futile.
The effect of creative media advertising (CMA) on attitudinal responses and behavioral intentions through perceived persuasive intent (PPI) and resistance
Master’s Thesis
Graduate School of Communication Master’s Program Communication Science Supervisor: Dr. Marieke Fransen
Anna Velsen
Student-ID: 10842128 29th January 2016
Abstract
This study contributes to the understanding of why creative media advertising (CMA) can be considered as an effective alternative to traditional media advertising. A model is proposed in which CMA decreases perceived persuasive intent (PPI), which leads to decreased resistance towards the advertisement. Ultimately, this positively affects several behavioral intentions as well as attitudinal responses. Theoretically, we argue that due to its implicit communication and the novel implication of the medium, CMA succeeds in
interfering with categorization processes so that consumers perceive the message to comprise less of a persuasive intent. Based on the framework of psychological reactance, we suggest that decreased PPI leads to less resistance, due to less threat that is put on the consumer. Conducting an online experiment (N = 158), we were able to confirm that PPI and resistance are important underlying mechanisms accounting for the effectiveness of CMA (vs.
traditional media advertising) on brand- and post-related eWOM intentions, purchase
intentions, as well as on brand attitudes. Three further analyses revealed these relationships to be explained even stronger through resistance as an only mediator, humor and resistance as well as surprise and resistance (in order of strength). Moreover, we found these mediators to additionally explain the relationship of medium type on attitude towards the ad. Thus, the findings suggest that CMA is a highly valuable alternative to traditional media advertising in reducing resistance. This reduction is explained by consumers perceiving CMA to be more humorous, more surprising and also to comprise less of a persuasive intent. Finally, we demonstrate that consumers have more positive ad and brand attitudes as well as they are more likely to engage in post- and brand related eWOM intentions and purchase intentions, which is why CMA should be highly valued by marketers who seek persuasion.
Resistance is futile.
The effect of creative media advertising (CMA) on attitudinal responses and behavioral intentions through resistance and perceived persuasive intent (PPI)
Introduction
In order to counterbalance the consequences emerging from advertising clutter, communications professionals are constantly looking for alternative ways to reach their customers. Additionally, the marketing environment is challenged to communicate with a more marketing literate consumer who has gained quite a profound understanding of
marketers’ persuasion tactics (Karo, 2002; Hogg, 2003). In order to address these problems, creative media advertising (CMA), where the medium itself implicitly communicates the message (Dahlén & Edenius, 2007) can be seen as an alternative advertising method that is capable to stand out from the crowd as well as to challenge consumers’ preconceived notions (Baltes & Leibing, 2008).
However, at this time previous studies were able to show positive effects of CMA on consumers (Dahlén, 2005; Eelen & Seiler, 2016; Wottrich & Voorveld, 2016) but especially as no research yet investigated the underlying effects of PPI and resistance, the current research contributes to the detailed understanding of CMA effectiveness. More precisely, we intend to contribute to the understanding of the extent to which media types influence
consumers’ attitude toward the ad (Aad), brand attitude, post- and brand-related electronic word-of-mouth (eWOM (P) and eWOM (B)) intention and purchase intention, and whether this relationship is mediated by PPI and resistance.
According to Dahlén and Edenius (2007), using CMA makes the commercial purpose less obvious. Consumers are less likely to zigzag their way through commercial messages without ever being affected by them, as they less frequently consider the CMA as comprising a persuasive intent. This even works on consumers whose myriad advertising exposures
trained them to reveal persuasive intents smartly. Perceiving a lower PPI may in turn be a crucial presupposition to decrease consumers’ resistance to persuasion, which again is as an initial condition in order for messages to be persuasive (Knowles & Linn, 2004).
In sum, the present research has three key contributions: First, we extend the work of Dahlén and Edenius (2007) regarding PPI as an underlying mechanism to explain the
effectiveness of CMA. Moreover, our research shows that, besides PPI as one component of this framework, resistance constitutes an additional fundamental feature in explaining
consumers’ responses to CMA. Thus, we show the capability of CMA to decrease PPI, which in turn decreases resistance, which again decreases consumers’ attitudinal responses and behavioral intentions.
Second, we empirically examine and identify resistance as a consumer-related driver of consumers’ attitudinal responses and behavioral intentions. Although some previous work has provided a first investigation of various outcomes within the CMA literature (Dahlén & Edenius, 2007; Dahlén, Granlund, & Grenros, 2009; Dahlén, Rosengren, & Törn, 2008; Eelen & Seiler, 2016), our paper is the first to empirically examine the direct effect of resistance on consumers’ responses. Especially as eWOM has been widely noted to be one of the most influential channels of communication in the marketplace (Fan, Miao, Fang, & Lin, 2013), no research yet empirically examined to what extend consumers’ resistance affects their intention to engage in eWOM.
Third, we extend the knowledge about CMA by answering the question to what extent CMA differs from traditional media advertising with regard to its inherent potential to
influence consumers’ attitudinal responses and behavioral intentions. This allows us to assess the feasibility of CMA as an alternative marketing strategy to traditional advertising formats.
In total, this study aims at investigating the underlying mechanisms accounting for consumers to respond more positively to CMA in contrast to traditional media advertising. By conducting an online experiment we intend to answer the following questions:
RQ: To what extent does media type (CMA vs. traditional media advertising) influence behavioral intentions (eWOM (P) intention, eWOM (B) intention, purchase intention) and attitudinal responses (Aad, brand attitude), and is this relation mediated by PPI and resistance?
Theoretical Background
Creative media advertising (CMA)
Dahlén (2005) was the first to clearly define CMA (also: creative media choice) and to differentiate it from traditional advertising in two respects: Creative media advertisements in contrast to traditional advertisements can be understood in terms of featuring a medium that implicitly communicates the message itself and serves as a distinct and relevant cue for the brand. The fact that the medium itself can play a substantial role in the persuasive
communication process has been intensively studied within the medium context literature. For instance, serving as a distinct cue, the medium context being congruent with the persuasive message of an ad has previously been found to positively affect certain outcomes (Moorman, Neijens, & Smit, 2002). Adapting this line of thought to the field of non-traditional media advertising, Dahlén (2009) found CMA to function as a rhetorical figure, where consumers have to draw conclusions from the connection of content and medium. The advertisers’ effort to provide this visual metaphor is ultimately rewarded by the consumer through enhanced brand and ad evaluations (Dahlén, 2009).
This first part of the definition blends in with the second considerable feature of CMA. In addition to the medium being an active part in the communication process, a novel or innovative implementation of a commercial message is used in order to advertise a certain brand or product (Dahlén, 2009). Recent research tested theses effects of CMA by creatively implementing eggs (Dahlén & Edenius, 2007), fire extinguishers (Dahlén, 2009; Wottrich & Voorveld, 2016) or straws (Eelen & Seiler, 2016) to display a commercial message.
Altogether, it has been shown that using a medium innovatively to implicitly communicate a commercial message exceeds consumers’ conventional experience with advertisements ultimately resulting in increased consumer responses such as Aad (Dahlén 2005; Dahlén & Edenius 2007), brand attitude (Dahlén 2005), eWOM (Eelen & Seiler 2016) and purchase intention (Wottrich & Voorveld 2016; Dahlén, Granlund, & Grenros, 2009).
Decreased PPI in response to CMA
PPI can be understood as consumers’ awareness of a message being designed and constructed by advertising professionals in order to change their thoughts, emotions and behaviors (Friestad & Wright, 1999; Friestad et. al, 1994). Friestad and Wright (1994) introduced the idea that throughout their life, consumers develop a general knowledge about how, when and why persuasive messages are intended to influence them. This experience-based knowledge helps them to interpret, filter, assess and respond to those messages and thus, in coping with sales appeals and persuasive attempts (Wright, 1986). Hence, persuasion knowledge leads consumers in being aware of persuasive attempts (Wright, 1986; Dahlén & Edenius, 2007). Still, the question remains of how different media types used for advertising can influence the perception of persuasive intent.
Based on the categorization perspective it can be argued, that people categorize what they perceive and bring their perceptions in line with categories that already exist in their mindset on the basis of conspicuous cues (Fiske & Pavelchak, 1987; Mervis & Rosch, 1981). Regarding consumer psychology, so-called prototypes of a category, meaning representatives being most typical of a category, serve as guidance when it comes to categorizing new stimuli (Loken, Barsalou & Joiner, 2008). In line with this, Dahlén and Edenius (2007) argue that advertisements setting themselves apart from the surrounding make a categorization as advertising more feasible. Contrariwise, blurring this contrast makes it harder to categorize advertisements as advertising and thus, counteracts a possible transfer of category-specific
negative affect and beliefs. In other words, a stimulus incongruent from advertising expectations accounts for better processing (Goodstein, 1993).
However, recent although limited research has investigated the ability of CMA to break through these filters consumers use to detect advertisements as such, with persuasion knowledge having a categorizing function (Friestad & Wright, 1994). Dahlén and Edenius (2007) compared an elevator and a newspaper magazine in transferring the commercial message of an energy drink. In fact, the research proved consumers to be less likely to disclose the persuasive intent and categorize it as advertising, when the commercial message was placed in a creative media setting.
To recap this point, consumers use their persuasion knowledge to categorize
persuasive messages as advertising. Consequently, we expect CMA to interfere consumers’ categorization processes by making the persuasive intent less obvious. Thus, consumers may perceive CMA as having a lower persuasive intent than an ad placed on a traditional medium. Thus, we hypothesize:
H1: CMA generates less PPI than traditional media advertising.
Decreased resistance in response to decreased PPI
Where persuasive attempts put pressure on the consumer to change, resistance is particularly becoming apparent as a reactive response (Knowles & Linn, 2004). As a pioneer in the discussion among resistance and persuasion, McGuire (1964) defined resistance as being capable of withstanding a persuasive attack, which accounts for its consideration as the antithesis of persuasion (Knowles & Linn, 2004). Accordingly, the consumer perceiving a persuasive intent of the advertiser has been found to result in the likelihood of resisting a message in early research on the communicators’ role in persuasion (Eagly, Wood, & Chaiken, 1978; Laran, Dalton, & Andrade, 2011; Petty & Cacioppo, 1979; Wood & Eagly, 1981).
attempts in many situations (Fransen, Smit, & Verlegh, 2015). According to the theory of psychological reactance as one of the best-known frameworks for understanding why people resist persuasion (Brehm & Brehm, 1981), individuals have an inherent aspiration to preserve their autonomy and independence (Fransen, Smit, & Verlegh, 2015). Consequently, when people perceive their freedom to be constrained or threatened, motivation to reassert this independence is triggered (Brehm & Brehm, 1981). Brehm and Brehm (1981) suggest that with an increased level of perceived interference of the persuader, consumers feel a constraint to be free in their choice. And also the persuasion literature found especially appeals
including an apparent persuasive intent to facilitate an experience of threat to one’s freedom (McGrane, Toth, & Alley, 1990; Weinstein, Grubb, & Vautier, 1986). Hence, reactance can be considered to be the motive of a consumer whose freedom is threatened for resisting a persuasive attempt (Fransen, Smit, & Verlegh, 2015).
In response to perceived threat, there are a number of distinct mechanisms through which resistance can occur. In other words, people use different resistance strategies to cope with persuasion attempts. Based on forewarning and health message literature, resistance strategies can be divided into various clusters. Fransen, Smit, and Verlegh (2015) differentiate avoidance, contesting, biased processing, and empowerment strategies, of which contesting and empowerment strategies are relevant in the current study context. Contesting describes strategies by which consumers actively challenge the message, source or persuasion tactics used (Fransen, Smit, & Verlegh, 2015). Such studies include counter argumentation, where the consumer explicitly counters an argument posed by the source and source derogation, which describes the neutralization of threatening information through undermining the credibility of the message source. Prior research indicated PPI to significantly determine both counter argumentation (Brock, 1967; Zuwerink & Cameron, 2003) and source derogation (Festinger, 1962; Wright, 1974; Zuwerink & Cameron, 2003), with both strategies being reduced as a result of decreased PPI. Strategies used to assert one’s own, existing views rather
than challenging the persuasive communication are called empowerment strategies (Fransen, Smit, & Verlegh, 2015). As one of these strategies, attitude bolstering describes
argumentation in support of one’s original attitude and was previously found to be one of the most common strategies used by consumers to resist persuasion (Zuwerink & Cameron, 2003). Further empowerment strategies are self assertion, where consumers bring to mind that the perceived message can not change their attitudes or behavior, as they are confident about them. Additionally, the third-person effect implies consumers to resist the message by
reminding themselves of important others who share one’s original attitude. For this research context we also include negative affect as one additional strategy, which describes the
expression of negative emotions at the source or its arguments (Zuwerink & Cameron, 2003). Negative affect has been preciously categorized as rather affective than cognitive response, with the persuasive intent in a message being used as a simple negative cue (Giner-Sorolila & Chaiken, 1997).
To recap this point, perception of a persuasive intent can constitute a threat to
freedom, and thus create resistance in consumers. Furthermore, resistance can occur through one of the above-mentioned strategies, of which various studies have previously been found to be caused by PPI. By implication, the reviewed literature suggests that consumers
perceiving a message as comprising less of a persuasive intent are in turn less likely to react with resistance. As a result, taking into consideration all previous findings, the following hypothesis is formulated:
Attitudinal responses and behavioral intentions resulting from decreased resistance
Attitude towards the ad and attitude towards the brand
Resisting an advertising message can have several effects on the way receivers respond to a message and may change personal attitudes towards the ad and the brand. Even though no study tested the overall concept of resistance affecting attitudinal responses yet, various researchers investigated the persuasive effects of different resistance strategies. Research conducted on counter argumentation indicates individuals to maintain their attitudes after successfully counter arguing an attempt to change (Tormala & Petty, 2002). In line with this, Petty and Cacioppo (1977) found counter argumentation and attitude to be negatively correlated. Source derogation was found to have a similar effect on attitudes, with people derogating the source producing subsequent attitudes that conform less to message assertions (Knight Lapinski, & Boster, 2001).
EWOM intention and purchase intention
With emerging digital media, customer-to-customer communication has expanded into the online sphere, where one-to-one communication has developed into a one-to-many
communication (Hansen & Lee, 2013). EWOM describes “any positive or negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions via the Internet” (Hennig-Thurau, Gwinner, Walsh, & Gremler, 2004, p. 39). Engaging in eWOM behavior is of special importance for marketers, as consumers through (ultimately) passing along an advertisers’ message can potentially become a persuasion agent carrying the message to their peers (Hansen & Lee, 2013). Furthermore, in the Facebook environment, eWOM behavior can be post-related or brand-related. Former includes a consumer to like, comment on, or share the post including the advertisement. Consumers engaging in brand-related eWOM comprises actions of visiting or following the brand page of the sender. Prior research has shown that consumers more
likely engage in eWOM, if they agree with the provided content (Boyd, Golder, & Lotan, 2010). Thus, when consumers resist the message communicated in an advertisement, they may not form the intention to engage in any kind of eWOM. Conversely, we expect
consumers, who resist an advertising message to be less likely to form post- or brand-related eWOM intentions. In a similar vein, and based on prior research suggesting that message acceptance predicts purchase intentions (Belch, 1982), we argue that consumers are less likely to develop purchase intentions, whenever they resist a persuasive appeal. These findings indicate, that resistance might be negatively related to attitudinal responses and behavioral intentions, which is why we propose the following hypothesis:
H3: Resistance has a negative effect on (1) attitude towards the ad, (2) brand attitude, (3) eWOM (P) intention, (4) eWOM (B) intention, and (5) purchase intention.
Serial multiple mediation effect
Altogether, we hypothesize CMA (vs. traditional media advertising) to positively affect attitudinal responses and behavioral intentions, due to decreased PPI, which consequently decreases resistance to the persuasive message. As a result, we propose the following serial multiple mediation model, which is also depicted in Figure 1:
H4: CMA leads to decreased resistance through decreased perceived persuasive intent, which consequently leads to more positive (1) attitude towards the ad, (2) brand attitude, (3) eWOM (P) intention, (4) eWOM (B) intention, and (5) purchase intention.
Figure 1. Conceptual model of hypotheses.
Method Design and Participants
To test the proposed hypotheses, an online experiment was conducted with a one-factor between subjects design with 2 levels (media type: creative vs. traditional). For this experiment, 188 participants were recruited via email or a Facebook status transmitting the invitation letter. 158 finished the survey (59.1% female; Mage= 26.61, SDage= 8.08),
participant’s level of education was quite high (Bachelor’s degree = 46,5%) and most of them indicated to have fluent English skills (fluent = 52,8%). Participants were randomly assigned to one of two conditions. Participation in the online experiment took about 10 minutes.
Procedure
After entering the survey (Appendix C) by clicking the provided link, participants were presented with an explanatory introduction including basic instructions, informed consent and a cover story. Expression of agreement by the participants was followed by a randomized presentation of one of the two stimuli included in this study. A forced minimum of 30 seconds to look at the advertisements was chosen to ensure basic observation of the stimuli by each participant. The second part of the study included all measurements, starting
with a thought-listing task measuring resistance, as thoughts were intended to be as related to the stimuli and unbiased by follow-up questions as possible. The arrangement of all additional measurements was deliberately chosen, taking triggering effects into account (eWom (P) intention, eWom (B) intention, perceived persuasive intent, purchase intention, attitude towards the brand, Aad). In the final part of the study, a manipulation check and control measurements (social media use, brand commitment, perceived surprise, perceived humor) as well as questions on demographics such as age and gender were included. At the end, the cover story was addressed again and the actual purpose of the study was revealed. Participants could then leave their critical feedback, were reminded of the authors’ contact details and asked to leave their email address in case they were interested in the results of the current study.
Independent variable
The independent variable of this study was media type with two conditions: creative vs. traditional media advertising. The group exposed to CMA was exposed to several photos that were taken of an actual creative medium Oreo ad (Appendix B, Figure 1). With an elevator coming down, an Oreo cookie placed on the cabin landed in the milk glass that was placed on the outside of the elevator, creatively expressing the Oreo typical dunk in milk. An existing CMA was chosen so that ecological validity was met and artificiality was low. The Oreo logo as well as the “Milk’s favorite cookie” slogan were enlarged due to visibility reasons and a “Reduced Fat” label was added to provide consumers an additional argument (Hamilton, Knox, Hill, & Parr, 2000). Participants in the traditional medium condition were exposed to a photo of a traditional poster ad that was created by an advertising professional (Appendix B, Figure 2). Through holding as many aspects constant as possible, we were able to decrease a possible threat to internal validity and lower the risk of mere exposure effects. The movement expressed by the CMA was imitated by showing Oreo cookies falling into a glass of milk on the poster. Logo, slogan and label were added in identical execution and at
the same location on the glass, as to make both ads comparable. Finally, the poster was integrated outside the same elevator to exclude influences caused by the setting. However, in this condition, the movement of the elevator was neglected and it was made sure, the poster on the elevator was perceived as traditional advertising. Although photographs are not the same as the actual media, we refer to McQuarrie and Mick‘s (1996) argument that
photographs are iconic signs yielding the same reactions as the objects they depict.
Measures
Perceived Persuasive Intent. Participants’ perception of the persuasive intent was measured based on a 7-point Likert scale used by Dahlén and Edenius (2007): (1 = strongly disagree, 7 = strongly agree). Participants were asked to indicate to what extent they agreed with the statements “The ad wants me to buy Oreo cookies”, “The purpose of the ad is to sell more Oreo cookies” and “It is a commercial message”. We adapted two additional items more related to persuasion from Tutaj and van Reijmersdal (2012): “The aim of this ad is to influence my opinion” and “The aim of this ad is to make people like Oreo.” All items have been adjusted to our research context. A principal component analysis (PCA) showed, that these items form a single uni-dimensional scale: only one component had an eigenvalue above 1 (EV = 3.34) explaining 66,83% of the variance. As intended, it appears that the scale measures perceived persuasive intent: the higher the scale score, the higher the persuasive intent perceived by respondents. The scale proved highly reliable, Cronbachs α = .87, and thus, the items were averaged to form an index measure (M = 3.96, SD = 0.87).
Resistance. We used a though-listing task to measure cognitive responses, as this technique has been found to be most suitable to obtain written listings (Petty, Wells, & Brock, 1976) and to test hypotheses regarding cognitive responses (Cacioppo & Petty, 1981). In line with these findings, participants were asked to write into thought listing boxes all thoughts that were produced during exposure. To obtain only the most salient thoughts and to minimize the
possibility for subjects to reflect on their thoughts, the thought-listing task was executed directly after advertisement exposure, as suggested by Cacioppo and Petty (1981). Additionally, measuring reactive thoughts included an instruction to ignore spelling, punctuation, and grammar.
Based on a codebook (Appendix D), a panel of two judges coded the given answers and evaluated the thoughts based on (1) how many positive thoughts were listed (1 = one positive thought, 5 = five positive thoughts; M = 1.24, SD = 1.21), (2) how many negative thoughts were given in total (1 = one negative thought, 5 = five negative thoughts; M = 1.15, SD = 1.29), (3) what the total number of thoughts listed was (1 = one box used, 5 = five boxes used; M = 3.91, SD = 1.12), (4) which out of six resistance strategy was/were used (0 = strategy not used, 1 = strategy used; counter argumentation [M = 0.31, SD= 0.47], source derogation [M = 0.21, SD= 0.41], third-person effect [M= 0.05, SD= 0.21], attitude bolstering [M = 0.01, SD= 0.08], self assertion [M= 0.03, SD = 0.18], negative affect [M = 0.11, SD = 0.31]) and (5) what the total number of thoughts containing resistance was (1 = one thought including resistance, 5 = five thoughts including resistance; M = 0.92, SD = 1.15). This procedure was developed based on Cacioppo & Petty’s (1981). Following Wright (1974), correct definitions of all included strategies were given in the codebook and used in a training of the judges prior to the final rating. On the basis of these definitions, the judges then had to rate. Examples of resistance thoughts filled in by participants were “Products with reduced fat/reduced sugars have other substitutes that are bad for you.” (counter
argumentation), “Marketingmasche (marketing scam).” (source derogation) or “Deze reclame is niet voor mij (this advertisement is not aimed at me).” (third-person effect). Intercoder reliability was assessed to assure the agreement between coders’ judgement (Landis & Koch, 1977). Cohen’s k indicated an almost perfect agreement (κ = .88, p < .001). Still, intercoder differences were discussed and final judgement was agreed upon. Consequently we calculated an index of overall resistance, by subtracting the number of positive thoughts from the
number of thoughts including resistance and divided the difference by the total number of thoughts given by each participant (Petty, Briñol & Tormala, 2002). Thus, high numbers indicate the generation of much resistance and low numbers describing less resistance (M = -0.12, SD = 0.56).
Attitude towards the ad. Following previous research, we used a 7-point semantic
differential scale to measure participants’ attitude towards the ad. The scale consisted of five items on which participants had to indicate their overall impression of this advertisement: bad/good, dislike/like, unpleasant/pleasant, unfavorable/favorable and negative/positive (Ang & Low, 2000; Dahlén, 2005). PCA showed that included items formed a single
uni-dimensional scale (EV = 4.59) explaining 91,86% of the variance. Accordingly, an index was produced by averaging the responses to the items with high numbers indicating a more positive attitude towards the ad (M = 4,64, SD = 1,74) and reliability checks showed a high Cronbach’s alpha (α = .98) permitting that Aad be computed as the average of these items.
Attitude towards the brand. To assess participants’ attitude towards the brand, a seven-point semantic scale including six items based on Till and Baack (2005) was used. Participants were asked to rate their overall impression of the brand on the following items: bad/good, dislike/like, unpleasant/pleasant, unfavourable/favourable, negative/positive, bad quality/good quality. All items were found to form a single uni-dimensional scale (EV = 4.86) explaining 81,04% of the variance and also Cronbachs alpha was found to be highly reliable (α = .95). Thus, an index was produced with higher numbers indicating a more positive brand attitude (M = 5.13, SD = 1.35).
eWOM intention. The extend to which consumers are likely to engage in electronic word-of-mouth after exposure to the Oreo ad was assessed through a twofold measurement. Firstly,
participants were asked to rate their post-related behavioral intentions (eWOM (P)) by indicating how probable they would share the post, comment on the post and like the post (based on Huang, Lin, & Lin, 2009). In addition, to evaluate consumers’ eWOM intention regarding the brand (eWOM (B)), participants were asked whether they would visit and/or follow the Oreo brand page on Facebook after exposure (Lee & Hansen, 2013). We used a 5-point bipolar scale adapted by Bock, Zmud, Kim and Lee (2005), including three items: unlikely/likely, improbable/probable, impossible/possible.
For both eWOM (P) and eWOM (B) intention, separate PCAs was first conducted on respectively 9 and 6 items measuring the likelihood of liking, commenting and sharing the post and visiting and following the brand page. Findings show, that the included items formed two single uni-dimensional scales. For eWOM (P) intention, only one component had an eigenvalue above 1 (EV = 6.74) explaining 74,94% of the variance, while for eWOM (B) intention the PCA showed one component having an eigenvalue above 1 (EV = 4.85) as well, which accounts for 80,88% of the variance. Therefore, the scales indicate the respondents’ eWOM intention as follows: the higher the scale score, the higher a respondents’ eWOM intention. Scales for eWOM (P) intention (Cronbachs α = .96) as well as for eWOM (B) intention (Cronbachs α = .95) proved highly reliable, which is why two index measures were formed (Mpost = 2.00, SDpost = 1.15; Mbrand = 2.29, SDbrand = 1.24).
Purchase intention. Purchase intention was measured through a Juster-scale. This one-item measurement has been used in earlier studies to estimate consumers’ purchase intention (Day, Gan, Gendall, & Esslemont, 1991). Participants were asked to indicate on a 7-point
probability scale (1 = no chance, 7 = certain) their certainty on the following statement: “How likely would you be willing to buy the product that is advertised in the near future?” (M = 4.19, SD = 1.96).
Demographics. We assessed various demographic data such as gender, age, nationality and highest level of education. Due to the English questionnaire and the international scope of participant recruitment, we additionally enquired the participants’ proficiency in English. To exclude confounding effects of prior exposure, participants were asked, whether they had seen the ad before (yes = 12.6%, no = 87.4%) and whether they had been exposed to the ad in another study before (yes = 1.9%, no = 98.1%).
Manipulation Check. To see, whether the manipulation did work and whether there is a significant difference in perceived creativity of the independent variable, consumers' judgments of the creativity of both advertisements was measured. Based on Dahlén,
Rosengren and Törn (2008), the respondents were asked to rate the creativity of the ad: ‘To what extent do you think that the advertisement you just saw is creative?’ (1 = not at all creative, 7 = very creative; M = 4.35, SD = 2.09).
Control variables. Various control variables were measured to ensure that any effects found were not caused by differences between the experimental groups. Prior literature revealed a confounding effect of prior social media experience on intention to share content (Lee & Ma, 2012). Hence, we adapted one single item from the scale of prior social media experience to control for an unwanted influence: “How often do you use social media to share pictures and videos?” (1 = not at all; 5 = very often; M = 4.28, SD = 0.86).
Since brand loyalty can be defined as an emotional or psychological attachment to a brand within a product class (Lastovicka & Gardner, 1977), prior research has found this variable to be correlated with consumers’ eWOM intention (Tsao & Hsieh, 2012). Participants’ brand loyalty level was therefore measured by using one item adapted from Raju, Unnava and Montgomery (2009). Participants had to indicate their agreement on the
following statement using a seven-point scale: “I can see myself as being loyal to Oreo” (1 = strongly disagree; 7 = strongly agree; M = 4.75, SD = 1.67).
For explanatory reasons, we also included surprise as an additional measure. Prior research had found significant effects of surprise within the creative media advertising context (Dahlén, 2005; Hutter & Hoffmann, 2014). Based on this literature, we gave participants the instruction “Please indicate to what extent you agree with the statements. The advertisement is…” and asked them to indicate their agreement using multiple 7-point bi-dimensional semantic differential scales including three items: unsurprising/surprising, unusual/usual, unconventional/conventional (EV = 2.53; α = 0.91, M = 4.10, SD = 1.92), accounting for 84.32% of the variance.
A second factor that has shown to account for effects in creative media advertising is perceived humor (Rauwers & Van Noort, 2016). Thus, perceived humor of the ad was
assessed based on prior advertising research (Cline & Kellaris, 2007; Chattopadhyay & Basu, 1990). A 7-point bi-dimensional semantic differential scale including six items was used: humorous/humorous, not playful/playful, not funny/funny, not amusing/amusing, dull/not dull, boring/not boring. Two items were reverse coded (not amusing/amusing and boring/not boring). PCA revealed the item not amusing/amusing to load on an additional factor, why exclusion of this item was advocated for. The remaining items formed a highly reliable scale indicating perceived humor (EV = 3.43; α = 0.88, M = 4.23, SD = 1.62) explaining 68.66% of the variance.
Results
Manipulation check
Due to the main focus of this study, comparing creative and traditional media advertisements, it was essential to ensure that participants perceived the advertisements as varying in creativity. Therefore, an independent samples t-test was conducted in order to test,
whether the manipulation of the independent variable media type was successful. Results show, that participants in the CMA condition significantly perceived the advertisement to be more creative (M = 5.73, SD = 1.48) than participants in the traditional medium ad condition (M = 2.87, SD = 1.58), t (156) = 11.78, p < .001, 95% CI [2.38, 3.34]. Thus, it can be inferred that the intended manipulation was successful.
Randomization
Additionally, we conducted independent-samples T-tests and Chi-square analyses to assure that there were no differences between the experimental groups with respect to brand loyalty, t (156) = -0.99, p = .320, 95% CI [-0.78, 0.26]; social media use, t (156) = 0.82, p = .412, 95% CI [-0.16, 0.38]; prior ad exposure, t (156) = 2.82, p = .406, 95% CI [-0.15, 0.06]; nationality, χ2 (19) = 19.38, p = .433; education, χ2 (3) = 2.56, p = .464; English proficiency, χ2 (4) = 6.29, p = .178; gender, χ2 (1) = 1.85, p = .174; and age, t (156) = -.065, p = .949, 95% CI [-2.64, 2.47]. Due to successful randomization, all participants who finished the online experiment were included in further analyses.
Medium impact on PPI and PPI on resistance
To test hypothesis H1, an independent samples t-test was conducted on PPI, with medium type as between subject variable (see Table 1 in Appendix A for all direct effects of medium type). The results of the t-test revealed a marginally significant effect of medium type on PPI, with participants in the creative medium condition reporting a lower PPI (M = 3.83, SD = 1.05) than those exposed to the traditional medium advertisement (M = 4.08, SD = 0.61), t (156) = -1.86, p = .065, 95% CI [-0.53, 0.02]. As we could confirm a marginally significant negative effect of medium type on PPI, we can infer that the creative medium advertisement accounts for less PPI. Thus, the first hypothesis is confirmed.
In addition, a linear regression analysis with resistance as dependent variable and PPI as independent variable proved significant, F (1, 157) = 10.30, p = .002. The regression model can therefore be used to predict resistance, but the strength of the prediction is weak: 6 per
cent of the variation in resistance can be predicted on the basis of PPI (R2 = .06). PPI has a significant, moderately strong association with resistance, b = 0.16, b* = 0.25, t = 3.31, p = .002, 95% CI [0.41, 0.25]. In order to find stronger effects, we conducted further analyses to test the effect of PPI on counter argumentation and source derogation as the most frequently used resistance strategies. Results of the first linear regression analysis with counter
argumentation as dependent variable and PPI as independent variable proved significant, F (1, 157) = 5.53, p = .020, R2 = .34. Nonetheless, the association of PPI with counter
argumentation was only weak, b = 0.10, b* = 0.18, t = 2.35, p = .020, 95% CI [0.02, 0.18]. Additionally, we did not find significant results for PPI to predict source derogation, F (1, 157) = 1.62, p = .205, R2 = .01. Due to these results, and the fact that the index of overall resistance provides a more comprehensive measurement than single strategies, we will only integrate the index of overall resistance in further analyses. In addition, we can conclude that PPI significantly predicts resistance, which is why H2 is confirmed.
Resistance on attitudinal responses and behavioral intentions
To test H3, we conducted separate linear regression analyses to assess the ability of resistance to predict attitudinal responses and behavioral intentions. Therefore, resistance was used as continuous predictor and Aad, brand attitude, brand-related eWOM intention, post-related eWOM intention, purchase intention as dependent variables respectively. Results showed, that resistance is significantly, strongly associated with Aad (b = -2.16, b* = -0.69, t = -12.03, p < .001, 95% CI [-2.52, -1.81]) with 48 per cent of the variation in Aad being explained on the basis of resistance (F (1, 157) = 144.08, p < .001, R2 = .48). Further, a moderately strong association of resistance with brand attitude (b = -1.16, b* = - 0.48, t = - 6.77, p < .001, 95% CI [-1.50, -0.82]) was found. Resistance also explains a significant proportion of variance in brand attitude(F (1, 157) = 45.78, p < .001, R2 = .23). In addition to these findings on attitudinal responses, we could also prove resistance to significantly predict behavioral intentions. Results showed that resistance explains a significant amount of the
variance in eWOM (P) intention (F (1, 157) = 57.03, p < .001, R2 = .23) with both variables being strongly associated (b = 1.07, b* = -0.52, t = -7.55, p < .001, 95% CI [-1.35, -0.79]). Resistance also predicts eWOM (B) intention significantly (F (1, 157) = 34.90, p < .001, R2 = 0.18), where a moderate association was indicated (b = -0.95, b* = -0.43, t = -5.91, p < .001, 95% CI [-1.27, -0.63]). In line with this, purchase intention was proven to be significantly predicted by resistance, (F (1, 157) = 33.33, p < .001), with resistance having a moderate association with purchase intention (b = -1.47, b* = -0.42, t = -5.77, p < .001, 95% CI [-1.98, -0.97]).
Serial multiple mediation through PPI and resistance
As prior analyses indicated, a serial multiple mediation effect of PPI and resistance could be expected. Before conducting serial multiple mediation analyses, we tested for direct effects of medium type on all response variables to assure an association without mediators. Separate independent-samples t-tests showed significant direct effects of medium type on attitude towards the ad (t (156) = 6.31, p < .001, 95% CI [1.07, 2.05]), brand attitude (t (156) = 3.00, p < .01, 95% CI [0.22, 1.05]), eWOM (P) intention (t (156) = 4.50, p < .001, 95% CI [0.44, 1.12]), eWOM (B) intention (t (156) = 3.07, p < .01, 95% CI [0.21, 0.97]), and
purchase intention (t (156) = 1.98, p < .05, 95% CI [0.39, 0.01]). Means across all dependent variables were higher for participants in the CMA condition indicating that people who were exposed to CMA had more positive brand and ad attitudes as well as they showed stronger behavioral intentions to engage in eWOM (P) and eWOM (P) and to buy the product (see Table 1 in Appendix A for means). The most positive response with the biggest mean difference between groups was found for Aad (Mcreative = 5.39, MD = 1.56).
Based on these results, further analyses tested the serial multiple mediation effect of medium type on all response variables through PPI and resistance. Preacher and Hayes’ (2008) procedure (Model 6) was used in order to estimate the path coefficients in the current
mediator model (figure 2). This method includes 95% bootstrap confidence intervals for calculating total and indirect effects of medium type on various outcomes through resistance. Bootstrapping with 1000 resamples was used to estimate the bias corrected and to indicate confidence intervals (BCACI). Figure 2 represents the tested serial multiple medial model with medium type as the independent variable, PPI as the first mediator, resistance as the second mediator, and one attitudinal response (Aad, Brand attitude) or one behavioral (eWOM (P) intention, eWOM (B) intention, purchase intention) measure as the dependent variable. Separate serial multiple mediation analyses were run for each dependent variable. The c-path in the model indicates the direct effect of medium type on the respective response variable, independent of all mediators (c’), and the total effect of medium type on the
respective response variable (c), composed of the sum of both direct effect and indirect effect through the mediators (Hayes, 2013).
Figure 2. Tested serial multiple mediation model: Effect of medium type on attitudinal responses and behavioral intentions via PPI and resistance.
In line with significant effects of medium type on all response variables indicated above, total effects (c) of the mediation analysis proved to be equally significant (bc(Aad) = -1.56, SE = 0.25, p < 0.001; bc(BrandAttitude) = - 0.63, SE = 0.21, p < 0.01; bc(eWOM(P) Intention) = -0.78, SE = 0.17, p < 0.001; bc(eWOM(B) Intention) = -0.59, SE = 0.19, p < 0.05; bc(Purchase Intention)= -0.61, SE = 0.31, p < 0.05;). This confirms that without mediation, there is a direct effect of
medium type on all outcomes (with the strongest effect on Aad). Still, direct effects indicated that there was no direct effect (c’) of medium type on response variables anymore, when taking resistance into account, with the exception of one significant direct effect on Aad (bc’ = 0.68, SE = 0.21, p < 0.01) and one marginally significant direct effect on eWOM (P) intention (bc’ = -0.30, SE = 0.16, p < 0.10).
Notwithstanding, bootstrapping revealed significant indirect effects through the two mediators PPI and resistance for brand attitude (indirect effect −0.03, SE = 0.03, 95% BCBCI [−0.11, −0.01]), eWOM (P) intention (indirect effect −0.02, SE = 0.02, 95% BCBCI [−0.08, −0.01]), eWOM (B) intention (indirect effect −0.02, SE = 0.01, 95% BCBCI [−0.06, −0.01]) and purchase intention (indirect effect −0.04, SE = 0.04, 95% BCBCI [−0.14, −0.01]). For Aad, bootstrap CI straddles zero (from -0.18 to 0.00), which is why evidence is not
sufficiently strong to claim an indirect effect. It should be noted, that all indirect effects are weak. Still, we can partly confirm H4 with the exception of Aad. This means that participants being exposed to the creative medium advertisement, perceived a lower persuasive intent than participants in the traditional medium condition. In turn, those participants generated less resistance, which consequently led to a more positive brand attitude and more favorable behavioral intentions. The strongest mediation effect of PPI and resistance was found on purchase intention.
To make sure the order of mediators in this serial multiple mediation model could not be switched, all models were tested with the mediators reversed (resistance predicting PPI). This did only lead to significant models for Aad, post-related and brand-related eWOM with even weaker effects, providing evidence that the data are in line with the expected order.
Additional analyses
Due to the marginally significant effect of medium type on PPI and the rather weak results from the serial multiple mediation analysis, we conducted an additional independent-samples t-test on medium type and resistance. The difference between groups proved to be highly significant, t (156) = - 5.85, p < .001, 95% CI [-0.63, -0.31], with participants in the CMA condition generating less resistance (M = -0.35, SD = 0.48) than participants in the traditional medium advertising condition (M = 0.12, SD = 0.52; see Table 1 in Appendix A). Consequently, we decided to test an additional simple mediation model (Figure 3) using Preacher and Hayes’ (2008) procedure (Model 4) with medium type as independent variable, resistance as the only mediator and attitude towards the ad, brand attitude, eWOM (P)
intention, eWOM (B) intention and purchase intention, as dependent variables. Simple mediation analyses were run for each dependent variable. Consistent with previous tests and following Hayes (2012), the c-path in the model indicates the direct effect (c’) and the total effect(c). Associated results for each path are shown in Table 3 in Appendix B.
Figure 3. Tested mediation model: Effect of medium type on attitudinal responses and behavioral intentions via resistance.
Results from simple mediation analyses proved significant total effects (c) to be consistent with prior mediation tests. Significant direct effects (c’) were found for Aad (bc’ = -0.66, SE = 0.21, p < 0.05) and eWOM (P) intention (bc’ = -0.33, SE = 0.17, p < 0.05).
attitudinal responses with an indirect effect of medium type on Aad through resistance (indirect effect -0.90, 95% BCa CI [-1.29, -0.61]) as well as on brand attitude through resistance (indirect effect -0.53, 95% BCa CI [-0.84, -0.30]). For effect sizes we used kappa-squared, expressing the indirect effect as a ratio to the maximum possible indirect effect (Field, 2009). Numbers revealed a large-sized effect for the mediation including Aad (κ2 = 0.27, 95% BCa CI [0.19, 0.37]), while the mediation including brand attitude represented a medium-sized effect (κ2 = 0.19, 95% BCa CI [0.12, 0.28]). Still, a full mediation could be found for the model including brand attitude, while resistance only partially mediated the relationship between medium type and Aad. Additionally, bootstrapping indicated significant indirect effects of medium type on all behavioral intentions with an effect of medium type on eWOM (P) intention (indirect effect -0.45, BCa CI [-0.71, -0.26]), eWOM (B) intention (indirect effect -0.42, BCa CI [-0.69, -0.21]) and on purchase intention (indirect effect -0.71, BCa CI [-1.07, -0.39]) through resistance. Effect sizes revealed medium-sized effects for all outcomes with purchase intention being the strongest intention (κ2eWOM(P) = 0.19, 95% BCa CI [0.11, 0.28], κ2eWOM(B)= 0.16, 95% BCa CI [0.09, 0.25], κ2purchase = 0.17, 95% BCa CI [0.10, 0.25]). Based on the non-significant direct effects proposed above, we can provide maximum evidence for mediation for eWOM (B) intention and purchase intention. Resistance on the other hand only partially mediated the relationship of medium type and eWOM (P) intention.
Therefore, we can propose a negative indirect effect of medium type on all responses, with CMA generating less resistance, which in turn results in more positive attitudinal and behavioral outcomes. The strongest effect could be found for Aad. Still, resistance was not able to explain all of the relationship between medium type and Aad. The strongest full mediation was found for the effect of medium type on brand attitude through resistance.
Humor and surprise
Due to prior literature on creative media advertising, we integrated humor and surprise as additional variables in this study. Expectedly, two separate independent samples t-test
indicated participants in the creative medium condition to perceive the ad as more humorous (M = 5.34, SD = 0.97) and as more surprising (M = 5.40, SD = 1.22) than in the traditional medium condition (Mhumor = 3.04, SDhumor = 1.29; Msurprise = 2.71, SDsurprise = 1.51). This difference proved significant for both humor (t (156) = 12.67, p < .001, 95% CI [1.94, 2.66]) and surprise (t (156) = 12.31, p < .001, 95% CI [2.25, 3.11]). Additional separate linear regression analyses indicated both variables to significantly predict resistance. Results show that humor predicts resistance significantly (F (1, 157) = 74.95, p < .001, R2 = 0.32). A significant, strong association of humor with resistance was shown (b = 0.20, b* = 0.57, t = -8.66, p < .001, 95% CI [-0.24, -0.15]). Also, surprise was found to predict resistance significantly (F (1, 157) = 37.95, p < .001, R2 = 0.20). Nonetheless, this association of surprise with resistance indicated to be only moderate (b = 0.13, b* = -0.44, t = -6.16, p < .001, 95% CI [-0.17, -0.09]).
Based on these results, we tested two additional serial multiple mediation models with medium type as the independent variable, humor (Figure 4) or surprise (Figure 5) as the first mediator respectively, resistance as the second mediator, and one attitudinal (Aad, brand attitude) or one behavioral response (eWOM (P) intention, eWOM (B) intention, purchase intention) as the dependent variable. Again, separate serial multiple mediation analyses were run for each mediator and each dependent variable, while direct (c’), total (c) and indirect effects were indicated similarly to prior tests (see Table 4 in Appendix A for humor; Table 5 in Appendix A for surprise).
Serial multiple mediation through humor and resistance
Table 4 in Appendix A shows the results of the serial multiple mediation model testing the mediation effect of humor and resistance. While total effects (c) were found to be the same as above, only the direct effect (c’) of medium type on Aad (bc’ = 0.63, SE = 0.22, p < 0.01) was significant. Direct effects for all other tested independent variables were
medium type through the two mediators humor and resistance for Aad (indirect effect −0.54, SE = 0.12, 95% BCBCI [-0.78, -0.34]), brand attitude (indirect effect −0.44, SE = 0.11, 95% BCBCI .721, -.276]), eWOM (P) intention (indirect effect −0.34, SE = 0.11, 95% BCBCI [0.58, 0.16]), eWOM (B) intention (indirect effect −0.33, SE = 0.11, 95% BCBCI [0.60, -0.15]) and purchase intention (indirect effect −0.60, SE = 0.15, 95% BCBCI [-0.95, -0.35]). From these results we can conclude that participants being exposed to the creative medium advertisement perceived the ad to be more humorous than participants in the traditional medium condition. In turn, those participants generated less resistance, which consequently led to more favorable attitudinal and behavioral responses. The strongest full mediation could be found for the effect of medium type on purchase intention mediated by humor and
resistance.
Figure 4. Tested serial multiple mediation model: Effect of medium type on attitudinal responses and behavioral intentions via humor and resistance.
Serial multiple mediation through surprise and resistance
Testing a third serial mediation model, we now tested the mediating effect of surprise and resistance on the relationship of medium type on all outcomes. While total effects (c) were expectedly the same, direct effects (c’) were all insignificant for this model. As can be seen in Table 5 in Appendix A, bootstrapping revealed significant indirect effects for all response variables. Media type had an indirect effect thought surprise and resistance on Aad
(indirect effect −0.42, SE = 0.12, 95% BCBCI [-0.80, -0.13]), brand attitude (indirect effect −0.25, SE = 0.11, 95% BCBCI [-0.53, -0.08]), eWOM (P) intention (indirect effect −0.21, SE = 0.09, 95% BCBCI [-0.54, -0.05]), eWOM (B) intention (indirect effect −0.19, SE = 0.09, 95% BCBCI [-0.43, -0.05]) and purchase intention (indirect effect −0.34, SE = 0.15, 95% BCBCI [-0.66, -0.11]). Due to the fact that all direct effects were insignificant and all indirect effects significant, surprise and resistance fully mediate all five models, with medium type showing the strongest association with Aad through both mediators. Summarizing, compared to the traditional advertising medium, the creative advertising medium causes more surprise in the recipient, which elicits less resistance, which in turn leads to more positive behavioral and attitudinal responses.
Figure 5. Tested serial multiple mediation model: Effect of medium type on attitudinal responses and behavioral intentions via surprise and resistance.
As with previous tests, we tested this model with reversed mediators (resistance predicting humor as well as resistance predicting surprise) in order to ensure the correct order of mediators. For humor, we could find two significant, but very weak indirect effects on Aad and eWOM (P) intention, when entering the mediators in reversed order. When testing a serial multiple mediation with resistance and surprise in reversed order, one single and very weak mediation effect was found for Aad. From these results, we can infer, that the data has been assessed in the correct order.
General conclusion and discussion
This study aimed not only at understanding the different effects of advertisements placed on different medium types on behavioral and attitudinal outcomes, but also at
clarifying the underlying mechanisms of PPI and resistance, accounting for this effect. Thus, we proposed a model, where CMA in contrast to traditional medium advertisements was expected to cause less PPI in the consumer, which in turn causes less resistance, which ultimately affects behavioral and attitudinal outcomes more positively. We were able to find support for this model on all response variables, except for Aad. Notwithstanding, due to weak effects, we tested three additional models on all behavioral and attitudinal responses. The first alternative single mediation model including only resistance, proved fully significant for all variables. Surprisingly, a second alternative serial multiple mediation model with humor and resistance was fully supported as well. The same applies to a third model testing mediating effects of surprise and resistance. The strongest overall mediation effect was found for resistance mediating the relationship of medium type and Aad. The effect of medium type on purchase intention via humor and resistance constituted the strongest serial mediation effect.
The results of this study support the notion that decreasing resistance strongly accounts for persuasion. As a basic principle in achieving persuasion, the reduction of resistance is required: ‘Resistance hounds persuasion the way friction frustrates motion. To accomplish the latter, you have to expect and, preferably, manage the former.’ (Knowles & Linn, 2004, p. 3)We succeeded to confirm this common notion in a rarely investigated field of research, by providing significant effects of CMA on all responses through resistance. A more positive Aad and brand attitude after exposure to CMA (vs. traditional medium advertisement) were explained by resistance. Consumers additionally were more likely to intend to buy Oreo cookies after exposure and to have stronger intentions to engage with the post (like, share, comment) or the brand (visit, follow) on Facebook, due to decreased
resistance.
Being the primary aim of this study, we further succeeded in detangling these underlying mechanisms on a multiple mediation level. As proposed, we found evidence for the underlying mechanism of PPI: When exposed to the CMA (vs. traditional medium advertisement), consumers were less likely to perceive the advertisement as comprising a persuasive intent, which resulted in less resistance and consequently in more positive
behavioral intentions and brand attitude. We could not find this effects for Aad and it should be mentioned, that all other serial multiple mediation effects including PPI were very weak. Larger indirect effects for all responses were found, when humor and resistance were investigated in order to explain underlying effects of CMA. While humor has been widely studied in traditional advertising research, the concept has not been subject to many studies dealing with CMA. In their study Rauwers and Van Noort (2016) found evidence for an effect of creative media ads on ad attitude, brand attitude and purchase intention, mediated by perceived humor. Studies from other fields found humor to significantly counter negative responses to persuasive messages such as counter argumentation (Nabi, Moyer-Gusé, & Byrne, 2007) or negative affect (Eisend, 2009). Notwithstanding, we were the first to confirm this effect within the creative media context and even to extend these findings across
attitudinal as well as behavioral responses, by testing a serial multiple mediation effect through humor and resistance. Derived from the literature on humor effects in traditional advertising, this study contributes to the theory that humor may serve as a distraction from counter-argumentation (as one strategy of resistance) and thus reduces negative cognitions related to the ad (Eisend, 2011; Petty, Wells & Brock, 1976). We found consumers to perceive CMA as more humorous, which led to less resistance and in turn they developed higher intentions to buy Oreo cookies, to engage in eWOM regarding post and brand as well as they showed more positive ad and brand attitudes.
(2007) were able to give a hint on a possible effect of surprise within the creative media setting. Also, Dobele (2007) proved surprise to have a significant influence within viral campaigns and Eelen and Seiler (2016) even integrated surprise as a mediator in their study. Still, we were the first to prove CMA to increase surprise, which in turn reduced resistance, which consequently accounts for stronger intentions and more positive attitudes. Overall, the indirect effects of this model testing two mediators were less strong than those for humor and resistance. Still, all mediation effects turned out to be full mediation effects. Thus, perceived surprise can be understood as a central driving force in order to reduce resistance and
consequently cause more positive responses in the consumer.
In general, our study contributes to the understanding of the complex conditions responsible for the effectiveness of CMA (vs. traditional medium advertisement). Increase in consumers’ intentions to buy a product, to become active in social media as well as positive attitudes after being exposed to CMA can be explained by a reduction of resistance.
Explaining these effects on an even more profound level, we could hold humor, surprise as well as PPI (in order of strength) responsible.
Managerial Implications
Besides its important contribution to the poorly studied field of CMA, various
practical implications can be drawn from this current study. Refuting the common notion that CMA would not be an alternative to traditional media advertising (Wanner, 2011), especially the strong direct effects of medium type on all outcomes in this study show, that CMA can indeed cause more positive responses in consumers than traditional media advertisements. Particularly as attentions have been found to be proximal predictors of intention and intentions significantly account for considerable variance in actual behavior (Ajzen & Fishbein, 1980; Ajzen, 1991), the current findings indicate that marketers can increase effectiveness of their commercial messages by placing them in CMA. In this way, they can
expect consumers to be more positive about the ad, the brand and also to eventually buy the advertised product.
Weaker but still highly significant results were found for participants being exposed to CMA appearing to be more likely to develop post-and brand-related eWOM intentions. Not only could we prove CMA of being able to “get legs and walk away on its own” (Dru, 2002), this study even confutes an often-postulated disadvantage of CMA of being inferior in
reaching target audiences. Each interaction with brands such as likes, shares or follows figures into the Facebook algorithm (Carah, 2014). This in turn increases the probability of brand content to appear in the news feeds of users’ friends, generates more buzz and possibly makes the ad ‘go viral’ (Karo, 2002). Thus, marketers should read our indications as a
possibility to generate organic reach through advertising their brand choosing creative media.
Limitations & recommendations for future studies
A possible explanation for the fact, that we found only very weak mediation effects when testing PPI, could be that this measurement heavily relies on the requirement that consumers are not commonly aware of creatively used media being commercially intended. However, as several brands already made use of CMA to date, consumers might have become aware of this type of advertising. Moreover, we used a very familiar brand. In line with this, prior research on CMA found a significant negative effect on PPI only when testing an unknown brand (Dahlén & Edenius, 2007). Being exposed to a familiar brand name could have triggered the awareness of being exposed to commercial content in both conditions. A third considerable explanation for those weak effects can be found in the measurement of PPI. We included statements such as “The ad wants me to buy Oreo cookies”, which could have functioned as an additional commercial prime. Notwithstanding, the fact that we could find weak but significant mediation effects of PPI indicates that the limitation rather lies in the measurements. We suggest future research to avoid specific brand naming and rather
content as opposed to talking about advertising. Additionally, future research should also bear in mind the possibilty, that people perceive CMA as comprising a persuasive intent, but that they care less because they like the ad.
Also, whereas direct effects of medium type on attitudes and purchase intention showed very substantial differences, means for both measured eWOM intentions were quite fractional. This might be due to stimuli presentation, where participants were exposed to raw pictures. Integrating these pictures in a social media layout might account for a better
imagination to like, comment on or share the presented content on Facebook. Consequently, future research should provide participants with a more appealing design of an online environment.
Finally, since our study only focused on the social media platform Facebook, we suggest future research to examine as to what extend our findings apply to other social media platforms, such as Twitter, YouTube or Instagram. The unique architecture, culture and norms of each of these platforms might also account for diverse effects of CMA in these contexts (Smith, Fischer, & Yongjian, 2012).
Besides these limitations, this study adds to the literature, as it is the first to introduce resistance as an underlying mechanism accounting for the superior effect of CMA (vs. traditional media advertising) in generating positive attitudes as well as in prompting consumers to develop intentions to buy a product and communicate the advertisement online. Additionally, we proved serial multiple mediation effects of resistance in combination with humor, surprise and PPI respectively. As such, we can not only conclude CMA to be a more feasible marketing strategy than traditional medium advertising in changing consumers' attitudes and form their intentions but also form a significant basis for future research by introducing resistance as a new field to the CMA literature.
References
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behaviour. Prentice–Hall Englewood Cliffs.
Ajzen, I. (1991). The theory of planned behavior. Organizational behavior and human decision processes, 50(2), 179-211.
Ang, S. H., & Low, S. Y. (2000). Exploring the dimensions of ad creativity. Psychology &
Marketing, 17(10), 835-854.
Baltes, G., & Leibing, I. (2008). Guerrilla marketing for information services?. New Library
World, 109(1), 46-55.
Belch, G. E. (1982). The effects of television commercial repetition on cognitive response and message acceptance. Journal of Consumer Research, 9, 56-65.
Bock, G. W., Zmud, R. W., Kim, Y. G., & Lee, J. N. (2005). Behavioral intention formation in knowledge sharing: Examining the roles of extrinsic motivators,
social-psychological forces, and organizational climate. MIS quarterly, 29(1) 87-111. Boyd, D., Golder, S., & Lotan, G. (2010, January). Tweet, tweet, retweet: Conversational
aspects of retweeting on twitter. In System Sciences (HICSS), 2010 43rd Hawaii International Conference (pp. 1-10). IEEE.
Brehm, S. S., & Brehm, J. W. (1981). Psychological reactance: A theory of freedom and
control. San Diego, CA: Academic Press.
Brock, T. C. (1967). Communication discrepancy and intent to persuade as determinants of counterargument production. Journal of Experimental Social Psychology, 3(3), 296-309.
Cacioppo, J. T., & Petty, R. E. (1981). Social psychological procedures for cognitive
response assessment: The thought-listing technique. Cognitive assessment, 47(8), 309-342.
Carah, N. (2014). Like, comment, share: alcohol brand activity on Facebook. Foundation for Alcohol Research & Education.
Chattopadhyay, A., & Basu, K. (1990). Humor in advertising: The moderating role of prior brand evaluation. Journal of Marketing Research, 27(4), 466-476.
Cline, T. W., & Kellaris, J. J. (2007). The influence of humor strength and humor—message relatedness on ad memorability: A dual process model. Journal of Advertising, 36(1), 55-67.
Dahlén, M. (2005). The medium as a contextual cue: Effects of creative media choice.
Journal of advertising, 34(3), 89-98.
Dahlén, M. (2009). A Rhetorical Question: What is the Impact of Non-traditional Media for Low-and High-Reputation Brands?. Journal of Current Issues & Research in Advertising, 31(2), 13-23.
Dahlén, M., & Edenius, M. (2007). When is advertising advertising? Comparing responses to non-traditional and traditional advertising media. Journal of Current Issues &
Research in Advertising, 29(1), 33-42.
Dahlén, M., Granlund, A., & Grenros, M. (2009). The consumer-perceived value of non-traditional media: effects of brand reputation, appropriateness and expense. Journal of
Consumer Marketing, 26(3), 155-163.
Dahlén, M., Rosengren, S., & Törn, F. (2008). Advertising creativity matters. Journal Of Advertising Research, 48(3), 392-403.
Day, D., Gan, B., Gendall, P., & Esslemont, D. (1991). Predicting purchase behaviour. Marketing Bulletin, 2(3), 18-30.
Dobele, A., Lindgreen, A., Beverland, M., Vanhamme, J., & Van Wijk, R. (2007). Why pass on viral messages? Because they connect emotionally. Business Horizons, 50(4), 291-304.
Dru, J. M. (2002). Beyond disruption: changing the rules in the marketplace. John Wiley & Sons Inc.
Eagly, A. H., Wood, W., & Chaiken, S. (1978). Causal inferences about communicators and their effect on opinion change. Journal of Personality and Social Psychology, 36(4), 424.
Eelen, J., & Seiler, M. R. (2016). Creative Media Use Increases Online Sharing of Your Ad (but Seems Less Effective for Your Brand). Advances in Advertising Research, 6, 291-308).
Eisend, M. (2009). A meta-analysis of humor in advertising. Journal of the Academy of
Marketing Science, 37(2), 191-203.
Eisend, M. (2011). How humor in advertising works: A meta-analytic test of alternative models. Marketing Letters, 22(2), 115-132.
Fan, Y. W., Miao, Y. F., Fang, Y. H., & Lin, R. Y. (2013). Establishing the adoption of electronic word-of-mouth through consumers’ perceived credibility. International
Business Research, 6(3), p58-65.
Festinger, L. (1962). A theory of cognitive dissonance. Stanford university press.
Fiske, S. T., & Pavelchak, M. A. (1987). Category-based and attribute-based reactions to others: Some informational conditions of stereotyping and individuating processes.
Journal of Experimental Social Psychology, 23(5), 399-427.
Fransen, M. L., Smit, E. G., & Verlegh, P. W. (2015). Strategies and motives for resistance to persuasion: an integrative framework. Frontiers in psychology, 6, 1201.
Friestad, M., & Wright, P. (1994). The persuasion knowledge model: How people cope with persuasion attempts. Journal of Consumer Research, 21(1), 1–31.
Friestad, M., & Wright, P. (1999). Everyday persuasion knowledge. Psychology & Marketing,
