The Effects of Viral Advertisements Containing a
Sensation Marketing Event
-
Do interpersonal sources and high levels of perceived ad creativity
have a positive effect on brand attitudes and forwarding intentions?
Graduate School of Communication Track: Persuasive Communication Master’s Thesis
Supervisor: Dr. Barbara Schouten Date of completion: 22.01.2015
Name: Leslie Gieraths Student Number: 10602216
Email: leslie.gieraths@student.uva.nl Address: Zuiderzeeweg 82L
2
Abstract
This study aims at giving a better insight into the driving factors of an advertising video’s viral success and help practitioners to break through the advertising clutter. More precisely, it assesses the influence of sensation marketing elements in viral ads on perceived creativity levels (divergence and relevance), brand attitudes and forwarding intentions. In addition, it investigates the impact of a video’s source on brand attitudes and forwarding intentions. Results of an experimental online survey among 125 participants show that sensational videos increase divergence perceptions but do not have a significant influence on brand attitudes or forwarding intentions. If an ad, however, is perceived as relevant it indeed increases brand attitudes and, further, forwarding intentions. Moreover, the source of a video had been found to not affect advertising effectiveness. This research adds to the literature by narrowing the scientific gap around viral videos containing sensational elements, by comparing results for both creativity components (divergence and relevance) and by shedding light on viral video forwarding behavior.
Keywords: Viral video advertisement, sensation marketing, perceived creativity (relevance, divergence), source (interpersonal, external), brand attitudes, forwarding intentions
3
Introduction
Consumers’ everyday lives are overloaded with advertisements. We face them when waking up to a radio broadcast, on our way to work at the subway station or when watching our favorite TV show. Advertisements are all around us. However, this increasing amount of advertising clutter consistently leads to ad avoidance behavior and resistance to persuasive attempts (Dahlén & Edenius, 2007). Traditional forms of advertising have become inadequate for todays’ informationally-empowered consumers (Rust & Oliver, 1994). Therefore,
marketers are constantly developing new strategies to reach and engage consumers.
One of those strategies is the use of viral campaigns, which shift advertising exposure from involuntary to voluntary (Rust & Oliver, 1994). This means that the viewer of a viral advertisement watches and shares the ad freely and without being paid to do so in his social media network. Viral advertising campaigns therefore take advantage of electronic word-of-mouth and are capable of reaching large proportions of the target market in a very short period of time while at the same time being comparably low in costs (Grigoraş & Cârjă, 2009). A prominent example of an ad that went “viral” is the Old Spice campaign: Within one day the ad already had 6 million views on YouTube which increased website traffic by 300% and sales by 107% as compared to the previous year (Effie Awards, 2011). This clearly illustrates the great impact viral marketing can have.
A common trend in viral campaigns is the use of sensation marketing tools as for instance real-life marketing stunts or flash mobs. Sensation marketing stunts surprise people by implementing real-life events in public spaces that are unexpected, extraordinary and spectacular (Hutter & Hoffmann, 2011). They differ from non-sensational videos in that they are containing real people in real-life situations being surprised by the stunt. No actors have been hired and no script has been written. Therefore, sensation marketing videos are a lot cheaper than other videos which contain professional actors. They aim at drawing attention towards the ad and increasing ad liking in order to cause viewers to share the video. However, little is known about the factors that make such viral ads successful and which effects viral sensations marketing videos have on brand attitudes and forwarding intentions as compared to non-sensational videos. Since viral ads motivate consumers to become endorsers of the brand by voluntarily sharing the video with their social network (Chiu et al., 2007), it is expected that they have the potential to highly engage consumers and create more favorable brand attitudes than traditional forms of advertising. This might be due to the fact that consumers who voluntarily expose themselves to a video are most likely be more involved with it and
4
will activate less persuasion knowledge. When consumers do not perceive the ad as a persuasive attempt but rather as a piece of entertainment they will subconsciously create higher attitudes towards the brand (Matthes et al., 2007). In order to provide proof for these suggestions and to gain more insight into consumers’ evaluations of sensational as compared to non-sensational viral ads this research thus aims at showing that the use of sensation marketing elements in viral ads will lead to more favorable attitudes and greater forwarding intentions than viral ads without sensational elements.
Up to now, most prior literature on viral advertising covers success stories of viral campaigns. There are far more campaigns though that never went viral because the main driving forces that make an ad successful in the online world are still mostly unacknowledged. However, some studies suggested that whether someone has received a piece of information from an interpersonal or external source has an influence on the intentions to forward the message. Especially interpersonal sources were found to have a positive effect on forwarding intentions and brand attitudes because they are perceived as more credible than external sources (Chiu et al., 2007; Phelps et al., 2004). However, this relationship has only been analyzed in the context of other media, such as emails. Nevertheless, this study argues that these findings can also be applied to viral videos because both - emails and viral videos - take advantage of ‘electronic word-of-mouth’ in order to spread through the web. Therefore, the source people receive a video from is argued to influence forwarding behavior and brand attitudes and will be regarded as an independent variable in this study.
Another factor that has been found to affect brand evaluations focuses on the ad’s message, namely creativity. The more creative consumers perceive an ad to be the more favorable was their evaluation of the brand (Dahlén, Rosengren & Törn, 2008). Therefore, I expect perceived creativity to mediate the effects on brand attitudes with higher levels of perceived creativity having the strongest positive effect. Overall, this study’s overarching research question is postulated as follows:
RQ: Do viral advertising videos that contain a sensation marketing element have a more positive effect on brand attitudes and forwarding intentions than viral advertising videos without a sensation marketing element, is this relationship mediated by perceived advertising creativity and does a video sent by an interpersonal source create greater brand attitudes and forwarding intentions than a video sent by an external source?
5
The findings of this research will give marketing practitioners a better understanding of the driving factors of viral successes in order to break through the advertising clutter. More precisely, the findings give insight into whether viral videos containing a sensation marketing event are more likely to be successful than those without, whether perceived creativity of the ad plays a mediating role in this and whether the source of the video has an influence on brand attitudes and forwarding intentions. Hence, results of this research will provide marketers with recommendations in terms of successful viral video content and targeting. Marketing practitioners are enabled to make suggestions on whether a video that is perceived as more creative (more divergent and relevant) will increase brand attitudes and forwarding intentions. Moreover, practitioners could base their decision of whether it is more useful to target certain influencers that share a video with their social network or to approach
consumers directly via their brand’s own channels. In addition, this study also closes a gap in empirical research by investigating the influence of sensation marketing events and perceived creativity on brand attitudes and forwarding intentions. To my knowledge, both factors have not been investigated in the context of viral advertising videos. Furthermore, this research aims at providing proof that the findings regarding the sender of an email can be applied to the sender of a viral video and that interpersonal sources are more likely to cause more favorable brand attitudes and forwarding intention than external sources.
Theoretical Framework
Viral Video Advertising
Viral advertising is an online marketing tool and defined as “[…] a widely used form of unpaid communication through persuasive messages created by identifiable sponsors and distributed among peers on interactive, digital platforms.” (Eckler & Rodgers, 2014, p. 186). This definition was chosen as the most applicable one in the context of this study because it includes all key characteristics of viral advertisement. Unlike the distribution of traditional advertisements, viral advertising is unpaid because it is distributed on free media channels such as social media platforms (Eckler & Rodgers, 2014). It takes advantage of “electronic word-of-mouth”, a digital equivalent of word-of-mouth which is defined as the “sharing of information about a product, promotion, et cetera, between a consumer and a friend, colleague, or other acquaintance.” (Kaplan & Haenlein, 2011, p. 254). In the context of electronic word-of-mouth and viral advertising the information is distributed among peers on interactive, digital platforms such as social media websites. Electronic word-of-mouth has
6
been found to be more effective than traditional distribution forms of advertising such as print or television ads, because the information about a brand is shared among close interpersonal sources. These sources exert a stronger influence on the receiver’s behavior since they are perceived as more credible because they have no personal or commercial intentions and have nothing to gain from recommending a product or brand (Chui et al., 2007).
However, viral advertising differs from electronic word-of-mouth in that its content is created by an identified sponsor and not by the consumer himself (Eckler & Rodgers, 2014). Thus, the content of viral advertising videos – as compared to other non-commercial viral content – is not only entertaining but also persuasive. It aims at increasing brand awareness, forwarding intentions and purchase behavior (Eckler & Rodgers, 2014).
Especially forwarding is essential for the success of a viral video advertisement. The more people that share the ad the greater its reach will be. Grigoraş and Cârjă (2009) even defined viral advertising as an “infection”: Once a consumer receives the viral advertising message he is infected and also infects others by forwarding the message (Grigoraş & Cârjă, 2009). In addition to its great reach, (successful) viral advertising also has other advantages over traditional advertising. Viral ads are comparably low in costs because no advertising space or broadcasting time needs to be bought, they spread rapidly throughout the World Wide Web and, thus, create greater brand awareness (Grigoraş & Cârjă, 2009).
Viral Advertising Videos containing Sensation Marketing Events
Viral advertising videos may take different forms. They can be scripted or unscripted, employ a guerilla marketing tool or contain real actors. This study will focus on viral advertising videos that feature a sensation marketing event because it is an unconventional method that can create an interactive and heightened consumer experience while drawing attention to the brand and engaging consumers by making a commercial with “real” people (Grant, 2014).
Sensation marketing is a guerilla marketing instrument which aims at gaining big results at low expenses (Hutter & Hoffmann, 2011). Guerilla marketing is generally defined as an unusual, provoking, flexible, dynamic, innovative and creative marketing strategy, which is driven by a surprise, diffusion and low cost effect. When recipients are surprised they are more likely to talk about their experience to other persons or share the information with their network on social media websites which increases its reach causing no costs (Hutter & Hoffmann, 2011). Similarly, sensation marketing “[…] aims at surprising pedestrians in public places by actions that go beyond the scope of familiarity.” (Hutter & Hoffmann, 2011, p. 3). It differs from other guerilla marketing strategies in that it is a one-time-only real-life
7
event containing real people in a public setting instead of actors (Hutter & Hoffmann, 2011; Nufer & Kern, 2012). In this way it aims at creating an advertising perception that is not disturbing or annoying but rather an experience people want to share with others. Consumers do not perceive the event as a persuasive attempt but rather as entertainment. The resulting “aha-effect” has been found to lead to more positive attitudes towards the brand as compared to traditional ads (Nufer & Kern, 2012).
If a sensation marketing event is filmed by the company and uploaded to a social networking site it has the potential to rapidly spread through the web due to the surprise effect mentioned above. It thus offers a cheap strategy to overcome consumers’ reactance, to elude the wear-out effect and to increase advertising reach (Hutter & Hoffmann, 2011). This is probably why an increasing number of companies and advertising agencies are making use of this tool. According to a survey among 40 executives at the top U.S. creative ad agencies and media buying firms 72.1% of all agency clients are “interested” or “highly interested” in viral videos and most of them have already produced a viral video (Feed Company, 2008).
However, not every online video goes viral. Former research has suggested two main driving factors of a successful viral video: The sender (or source) of the video and its content. In the following, those two factors will be examined in more detail.
The Impact of Sensational Videos on Perceived Creativity
The literature provides a variety of definitions for advertising creativity but most are similar to Leo Burnett’s definition stating that advertising creativity is “the art of establishing new and meaningful relationships between previously unrelated things in a manner that is relevant, believable, and in good taste, but which somehow presents the product in a fresh new light” (El-Murad & West, 2004, p. 190). This definition also captures the two characteristics of creative advertisements: divergence and relevance (Smith & Yang, 2004). Divergence is defined as the originality of the ad. An advertisement needs to be novel, different, imaginative and unusual in order to gain attention. In addition, the ad as well as the advertised brand needs to be relevant, meaningful, appropriate, useful or valuable to the consumer (Smith & Yang, 2004; Smith et al., 2007). However, it needs to be kept in mind that advertising creativity is not always perceived in the same way. Depending on consumers’ individual prior experiences advertising divergence may be rated differently. Similarly, individual consumer goals, needs and desires may determine the degree of perceived relevance. Thus it is the consumer’s individual perception of the ad that stimulates his perceived level of advertising creativity (Smith et al., 2007).
8
Viral videos that inherit a sensation marketing event are expected to be both divergent and relevant. They are divergent because a sensation marketing stunt usually is surprising, different and imaginative. The branded real-life installations or performances surprise
passersby in an unusual setting and trigger their and the video audience’s attention (Hutter & Hoffmann, 2011). In addition, they are perceived as more relevant as traditional online advertisements by consumers because their content is regarded as a piece of entertainment instead of a persuasive attempt (Matthes et al., 2007). Consumers are therefore more likely to voluntarily expose themselves to these kinds of online videos than to traditional advertising videos. Since sensational videos are expected to be more divergent and relevant than regular online ads and since those two elements are defined as the key features of advertising
creativity it is hypothesized that sensational ads are also perceived as more creative than non-sensational ads.
H1: Viral advertising videos containing a sensation marketing event are perceived as more creative than viral advertising videos without a sensation marketing event.
The Impact of a Video’s Source on Brand Attitudes
So far, only the content of the viral advertisement has been examined as a driving factor of viral success. However, also the source of the message has been found to have a crucial influence on consumers’ attitudes towards a brand.
Prior literature has identified two main sources of information: external sources, as for instance commercials or companies, and interpersonal sources, like family members, friends or personal acquaintances (Keaveney & Parthasarathy, 2001). Most authors have found that especially interpersonal sources have a positive effect on brand attitudes. Chiu and colleagues (2007) argue that the reason for this is that interpersonal sources such as friends or family members are perceived as not having anything to gain from promoting a certain brand or advertising message. They are rated as more credible and thus have a stronger influence on brand attitudes than external sources (Chiu et al., 2007). Hence, the perceived intention of the source is the crucial factor that builds source credibility (Chiu et al., 2007).
This effect can be explained by the Reactance Theory, which states that people have the desire to resist persuasive attempts and to maintain their freedom (Brehm, 1966). Thus, if people perceive a source as not credible, of external origin and as having a persuasive attempt they will consequently try to avoid this message, start counterarguing and form negative attitudes towards the brand (Brehm, 1966).
9
Apart from Chiu and colleagues other authors have found evidence for this effect, too. Phelps, Lewis, Mobilio, Perry and Raman (2004), for instance, who also assessed consumer responses and motivations to pass along emails, found that a message is of greater value when it comes from a friend. Participants claimed that they did not consider the brand information in the email as “junk” because they assumed the interpersonal source passed it on for a good reason (Phelps et al., 2004). However, when receiving an unsolicited email from a company, respondents felt irritated and deleted the mail without opening it (Phelps et al., 2004).
Since both - emails and viral videos - make use of electronic word-of-mouth and have the potential to rapidly spread through the web while having a great reach, this study argues that the findings regarding email forwarding can also be applied to viral videos. Therefore, the following is hypothesized:
H2: People who receive a viral advertising video from an interpersonal source are more likely to form positive attitudes towards the brand than people who receive it from an external/commercial source.
The Impact of Perceived Creativity on Brand Attitudes
Prior literature has already discussed the influence of perceived creativity on advertising effectiveness. More specifically, the level of perceived ad creativity is expected to influence attitudes towards the advertised brand and consumer behavior. In their conceptual framework McInnis, Moorman and Jaworski (1991) argue that the use of divergent and relevant stimuli enhances consumers’ motivation to attend to the ad and to process brand information.
Especially advertisements that are unusual, different, surprising, novel or complex have been found to increase consumers’ motivation to actively watch the ad. If the ad is also relevant to the recipient and evokes his/her curiosity it leads to an increase in brand information
processing (McInnis, Moorman & Jaworski, 1991). As a result, consumers elaborate more on and link together brand information in the ad. These high processing levels were found to produce stronger, more stable, accessible and enduring brand attitudes (McInnis, Moorman & Jaworski, 1991). Another important factor which influences brand attitudes is the ability to interpret brand information in an ad. Only when consumers have the skills and knowledge to process an ad will they be able to form positive attitudes towards the brand (McInnis,
Moorman & Jaworski, 1991). This can be explained by consumers’ need for closure (Smith & Yang, 2004). Novel, unusual or surprising ads evoke consumers’ desire to fully understand
10
the information in the ads, which leads to a deeper processing and stimulates more cognitive thoughts and more favorable attitudes towards the message content (Smith & Yang, 2004).
However, most advertisements are watched with low levels of involvement (Krugman, 1965). Under these circumstances creative ads may serve as peripheral cues since the
Elaboration-Likelihood-Model holds that the peripheral route is preferred when the likelihood of elaboration is decreased (Petty, Gleicher & Baker, 1991). As a result, attitudes towards the brand are changed by relatively simple cues and associations such as creative elements in the advertisement (Petty, Gleicher & Baker, 1991).
Former studies have already found a link between advertising creativity and brand attitude. Sheinin, Varki and Ashley (2011), for instance, showed that especially novel advertisements positively influenced brand attitudes. Another study by Dahlén, Rosengren and Törn (2008) found that creative ads might change brand perceptions. More creative ads signaled greater effort on the advertisers’ behalf than less creative ads. As a result, the creative brands were perceived as smarter, as being able to solve problems and develop valuable products. Thus, consumers became more interested in the advertised brand and rated it more favorably (Dahlén, Rosengren & Törn, 2008). These findings suggest that more creative ads may lead to more positive brand attitudes. In addition, the authors found evidence that consumers’ perceptions of the ads’ creativity mediate the effects on the brand and
increase the impact of the manipulated creativity (Dahlén, Rosengren & Törn, 2008). This suggests that perceived creativity mediates the effects between creative videos and brand attitudes. Likewise, Smith and colleagues (2007) found that perceived creativity mediates the effects of ad exposure on brand attitudes. Thus, it is hypothesized:
H3: The higher the perceived creativity is, the higher the brand attitudes will be.
H4: Perceived creativity mediates the relationship between the type of video and brand attitudes.
The Impact of Brand Attitudes on Forwarding Intentions
A successful viral advertisement moves marketer-to-consumer communication to consumer-to-consumer communication by encouraging consumers to forward marketing messages to others (Chiu et al., 2007). In doing so, individual consumers may contribute to the exposure of thousands or even millions of people to the video (Chiu et al., 2007). This means that the success of a viral advertisement mainly depends on people that forward it. The more people forward a video, the greater its reach and the higher the number of views will be. But the
11
actual number of total views is only available after the video has been distributed. Thus, a measurement is needed that predicts this behavior and, thus, viral success. For that reason, this study will look at forwarding intentions. According to the Theory of Reasoned Action (TRA), which examines the determinants that guide human behavior, intentions are the immediate antecedent of actual behavior and therefore make a valid proxy measurement of behavior (Fishbein & Ajzen, 2010).
Additionally, brand attitudes have been found to influence behavioral intentions. The TRA claims that attitudes are one of the main influencing factors on behavioral intentions (Fishbein & Ajzen, 2010). Moreover, Petty, Gleicher and Baker (1991) stated that a person’s general attitude (e.g. brand attitude) is capable of guiding behavioral responses (e.g.
forwarding intentions). This would suggest that a positive attitude towards the brand after exposure to a viral video would increase the likelihood of forwarding it. Huang and
colleagues (2013) already found scientific support in the context of viral video advertisements for this. They used the Categorization Theory as an exploratory framework which states that when two objects are classified into one category by the consumer, the attitude towards one object will have a significant impact on the attitude towards the other (Huang et al., 2013). Applying this theory to the viral advertising context, it means that when a consumer allocates a brand to a certain advertising video, the brand attitudes will spill over towards the attitudes towards the content of the video and further influence sharing intentions (Huang et al., 2013). Their results support this and show that the higher an individual’s brand attitudes are, the higher his/her forwarding intentions. Based on these theoretical and statistical findings, it is hypothesized:
H5: The more positive the attitudes towards a brand, the higher the intention to forward the video.
In addition, prior research also investigated the influence of a message’s source on consumer’s brand attitudes and forwarding intentions. As already stated above, a message’s source has an important influence on consumer’s attitudes towards the advertised brand in that message. These attitudes, in turn, have an influence on forwarding behavior. Interpersonal sources are perceived to be more credible than external sources (Chiu et al, 2007). Recipients do not activate their persuasion knowledge because they do not experience the message as a persuasive attempt (Brehm, 1966). Consequently, consumers’ attitudes towards a message are more favorable when a message comes from an interpersonal source than from an external
12
source. According to the TRA, these positive attitudes further lead to an increase of
behavioral intentions, as for instance forwarding intentions (Fishbein & Ajzen, 2010). Chiu and colleagues (2007) examined these relationships in the context of email forwarding behavior. Their findings suggest that people who received an email from an unfamiliar interpersonal source evaluate the message more positively and are more willing to forward it than people who received an email from a commercial source (Chiu et al., 2007). External sources even caused irritations among their participants leading them to avoid the message by not opening the email (Chiu et al., 2007). Therefore, it can be concluded that interpersonal sources cause more favorable attitudes because they are perceived as unbiased sources of information. These positive attitudes further lead to an increase of sharing intentions. Based on this it is hypothesized that:
H6: Brand attitudes will mediate the relationship between the source of the video and the intentions to forward the video.
Moreover, past research has already found that consumers also make inferences about brands based on how a message is communicated instead of only focusing on what is being communicated. The authors Modig, Dahlén and Colliander (2014) used the Signaling Theory to examine whether high perceived advertising creativity signals positive properties about the brand to consumers. Signaling Theory holds that a company can enhance consumers’ brand attitudes by signaling greater effort and ability of the brand (Dahlén, Rosengren & Törn, 2008). This means that after watching an ad, consumers form impressions of the expenses and efforts that have been invested in it and evaluate whether these efforts were exceptionally high. These so called ‘signals’ influence consumers’ evaluations of the brand (Modig, Dahlén & Colliander, 2014). Modig and colleagues (2014) found that perceived creativity is such a signal. Their results show that highly creative advertising videos as compared to less creative advertising videos not only have a positive effect on brand attitudes but also on word-of-mouth intentions (Modig, Dahlén & Colliander, 2014). Logically, if consumers have positive attitudes towards a brand, they are more likely to talk about it and share information such as advertising videos with others. Again, this effect can be explained by the TRA, which states that individuals’ attitudes influence behavioral intentions (Fishbein & Ajzen, 2010).
Therefore, it is hypothesized that perceived advertising creativity influences brand attitudes which, in turn, influence sharing intentions. Thus, the following is suggested:
13
H7: Brand attitudes will mediate the relationship between the consumer’s perceived creativity and the intentions to forward the video.
Figure 1 displays the conceptual model of this study taking all the variables and the expected mediators into account.
Figure 1. Conceptual Model
Methods
Participants
The target population of this study consists of frequent internet users that are familiar with or already participated in sharing content online. Therefore, a convenience sample that is
stratified by the factor “having internet access” was used as a sampling method. Invitations to participate were send out via the social network site Facebook and via emails. They contained a link to the survey and short instructions.
A total of N = 125 participants completed the survey fully and correctly answered the manipulation check questions. A total of n = 36 of them were randomly assigned to the ‘friend-sensation element’-condition and watched the sensation marketing video sent by a friend, n = 22 participants watched a sensation marketing video sent by the brand KLM, n = 28 people received the sensation video by a friend and another n = 39 received the non-sensation video by the brand KLM.
14
The majority of participants was female (57%) and 43% were male. The average age of the sample was approximately 27 years (M = 26.6, SD = 7.88) with ages ranging from 21 to 65 years. When looking at the highest level of education participants had attained, most respondents stated to have completed a Bachelor’s degree from a University (60%). Approximately 21% completed a Master’s degree, 10% graduated at a University/College preparatory Secondary School, 6% finished higher vocational education, 2% graduated at a technical/vocational Secondary School, 1% completed lower vocational education and another 1% received a PhD.
Design
In order to assess the relations as outlined in the hypotheses quantitative research was needed. More specifically, a 2 (viral video containing a sensation marketing event versus without sensation marketing event) x 2 (interpersonal source versus external source) between-subjects classical experimental design was chosen because it increased the internal validity of the study and had the ability to exclude external factors (Bryman, 2008). This was particularly important to be able to establish possible causal relationships between the variables.
A self-completion online questionnaire was chosen as the most suitable research method since it enabled collecting data from people of a wide range of sociodemographic backgrounds and geographical areas. Additionally, this tool mainly addressed regular internet users which might be familiar with sharing content online and which were also the target population of this research. Furthermore, participants were randomly assigned to one of the four conditions. After the data collection the responses to the survey were compiled and statistically analyzed as explained in the following.
Pre-test
A pre-test among mostly university students (N = 16, 44% male) was conducted before the actual survey. The goal was to decide on two viral video advertisements that were going to be used in the main experiment. Overall, three pairs of viral videos were pre-tested with one of them containing a sensation marketing event and the other without. The pair of videos that was going to be used should be of the same brand in order to rule out that external factors such as brand liking and brand familiarity influence the results.
Thus, all participants of the pre-test had to watch a total of six advertising videos of three brands, which had only been distributed in online media such as YouTube. The length of the videos ranged from 0.41 to 3.00 minutes. The first pair of videos was by the Dutch airline
15
KLM. At first, participants had to watch a video which showed real people in two public settings in Amsterdam and New York City. A special live-stream video installation, equipped with a screen, a touch screen, speakers and a camera, was placed in each of these squares inviting pedestrians to try to achieve the perfect transatlantic high-five (see Appendix 4, Figure 2) (Sluis, 2014). If participants from both cities managed to get the perfect aim and timing they stood the chance of winning plane tickets to the other city. The second video introduced KLM’s new “KLM Trip Planner” – a platform to easily organize trips with friends (Dutch Daily News, 2012). At first, a group of four friends who struggle to find a date for their Barcelona getaway is shown. After a while, the sentence “Skip the hassle … Start the fun” appears and the same group is shown while they check-in for their flight and have a great time in Barcelona (see Appendix 4, Figure 3). At the end, the slogan re-appears and a link to the Trip Planner as well as a short description of how it works is presented.
The second pair of videos was by the Dutch beer brewery Heineken. In the first video the beer manufacturer searched for people with “legendary” talents and invited real
contestants to an intimate casting in an empty room with a camera (Feeley, 2014). However, Heineken wanted to surprise the contestants by letting the audition room walls collapse as soon as the contestant was ready to perform and revealed a live audience of hundreds of people (see Appendix 4, Figure 4). All contestants had to overcome their “Stage Fright” and reveal their hidden talents. Afterwards the sentence “Everyone is legendary at something” was shown. In the second video, a bartender carries a pile of Heineken bottles, trips and drops them in slow-motion (see Appendix 4, Figure 5). In the following scenes, several men pause doing what they were just doing. A boxer stops fighting during a match and gets knocked out, a men stops kissing his girlfriend, etc. as if they would be really sad all of a sudden. In the last scene the bursting bottles appear again, as well as the Heineken logo and slogan “It’s all about the beer”.
The last two videos were by the international cell phone provider T-Mobile. The first video featured a huge installation in Barcelona inviting passersby to play a live-version of the mobile game Angry Birds (see Appendix 4, Figure 6). Participants had to use their mobile phones to launch the large-scale bird cannons across a display of blocks in order to destroy the pigs that hid between them (Diaz, 2011). At the end, the sentence “With our biggest range of smartphones, you can join the fun wherever you are.” appeared, as well as the T-Mobile logo and slogan “Life is for sharing.”. The second advertising video advertised T-Mobiles “4G Network” by spelling out its advantages. The video describes the “new world of sharing” that people connect to by sharing music, ideas, recipes, cars, etc. and outlines that T-Mobile is
16
helping customers to connect “to a richer way of living” by offering them the best network (see Appendix 4, Figure 7).
After watching each online ad separately, participants had to answer the same
questions for each video. In order to be able to rule out ceiling effects of brand attitudes in the later analysis, brand attitudes were pre-tested by asking the participants the following
question: “How much do you like the brand featured in this video on the following scale?” and using the 7-point bipolar answering scale “I don’t like it at all – I like it a lot”.
Furthermore, past research has shown that the emotional tone of a video affects forwarding behavior and brand attitudes. Especially videos that evoke high-arousal emotions (e.g. awe, anger, anxiety) were found to be more likely to go viral than those that contained deactivating emotional content (e.g. sadness) (Berger & Milkman, 2012). Also, a rather pleasant emotional tone elicits more favorable brand attitudes and sharing intentions (Ecker & Bolls, 2011). For these reasons, the emotional tone of the two videos used in the research had to be kept constant in order to avoid it from affecting results. Thus, the emotional tone of the viral ads was assessed during the pre-test by using an adapted scale by Biener and colleagues (2004), asking “Overall, how would you rate the video on the following scales?”. Participants had to rate the two scales ‘positive emotions’ (“funny”, “happy”, “entertaining”) and
‘negative emotions’ (“frightening”, “sad”, “disturbing”) on a 7-point answer scales with [1] being the lowest and [7] the highest possible score. These items have been found to be reliable by prior researchers with standardized alphas for the multi-item scales being αpos = .72 and
αneg = .83 (Biener et al., 2004).
Lastly, participants had to rate the relevance of the brand’s products or services. The aim of this was to find a brand that was regarded as “useful”, “important” and “relevant” by the participants. Therefore, a three-item scale with a seven-point bipolar answer scale by Miyazaki, Grewal and Goodstein (2005) was used to measure the brand’s relevance. This scale has been found reliable with Cronbach’s alpha being α = .90 (Miyazaki, Grewal & Goodstein, 2005). The full questionnaire of the pre-test can also be found in appendix 5.
The results of the pre-test revealed that KLM’s overall brand attitudes were rated the highest (M= 4.66, SD = .83), followed by overall brand attitudes towards Heineken (M= 4.28, SD = 1.66). T-Mobile was found to have the least favorable overall brand attitudes (M= 4.00, SD = .98). However, in terms of relevance of the brand for the consumer, T-Mobile scored the highest (M = 4.29, SD = 1.19), followed by KLM (M= 3.95, SD = 1.06) and lastly, Heineken (M= 2.97, SD = 1.08) (table 1). Two repeated-measures ANOVAs have been conducted to compare all brand means for brand attitudes and brand relevance. For brand attitudes,
17
Mauchly’s test indicated that sphericity can be assumed, χ2
(2) = 2.49, p = .288. The results show that brand attitudes did not significantly differ between the three brands, F(2, 30) = 1.22, p = .308, ω2 = .08. Similarly, for brand relevance Mauchly’s test also indicated that sphericity can be assumed, χ2(2) = .99, p = .608. Furthermore, a significant, moderate effect among the three brands on respondents’ brand relevance evaluations has been found, F(2, 30) = 10.10, p < .001, ω2 = .40. An overview of the post hoc results of the two repeated-measures ANOVAs for the three brands is provided in table 2, appendix 1.
In addition, a paired-samples t-test was conducted to compare the emotional tone of the videos. KLM’s videos were found to evoke the most similar emotions as compared to the other brands’ videos. There was a significant difference in the evaluation of positive emotions between KLM’s sensation video (M = 5.77, SD = .92) and KLM’s non-sensation video
conditions (M = 3.92, SD = 1.43); t(15) = 4.98, p < .001). Furthermore, there was a significant difference in the evaluation of negative emotions between KLM’s sensation (M = 1.02, SD = .08) and non-sensation video (M = 1.13, SD = .24); t(15) = -2.08, p < 0.1). The mean
differences of the two videos’ positive emotional tone (M = 1.85, SD = 1.49) and negative emotional tone (M = -.10, SD = .20) were found to be the smallest as compared to the video pairs by T-Mobile (Mpos = 1.63, SDpos = 1.33; Mneg = .44, SDneg = 1.19) and Heineken (Mpos =
1.96, SDpos = 2.10; Mneg = -.65, SDneg = 1.51). A detailed reporting of T-Mobile’s and
Heineken’s descriptives can be found in table 3 through 5 (appendix 1).
Overall, the two videos by KLM scored the highest on brand attitudes, were found to be perceived as relevant by the respondents and had the highest similarity in terms of
emotional tone. Therefore, KLM’s videos were selected for the main study, which is described in more detail in the following.
Procedure
At first, participants were informed about the nature of this research, the guaranteed anonymity of responses and the usage of the data for research purposes only. It was highlighted that the survey should be filled out voluntarily and individually on the basis of respondents’ personal opinion. Afterwards, it was claimed that by participating in the survey respondents gave the researcher the permission to use the data for her Master thesis.
Additionally, an email address was given in case respondents had any questions or
complaints. After continuing to the survey, participants were randomly assigned to one of the two source (interpersonal source n = 64, external source n = 61) and to one of the two video conditions (sensational video n = 58, non-sensational video n = 67). After the manipulations,
18
all participants had to answer the same questions regarding the mediator, manipulation check, socio-demographics, dependent and control variables. These will be described in more detail below. The complete Qualtrics survey can also be found in appendix 6.
Variables
Independent variables
The type of video participants were exposed to and the source they received it from were regarded as the independent variables. The experimental groups were told that they received the video from a friend, whereas the control groups were told that received it from the brand itself. Previously, this source manipulation has proven sufficient to create interpersonal and external source status (Chiu et al., 2007). The two groups were then further split up between the two video conditions and received either a viral video without a sensation marketing event or with a sensation marketing event. The two videos were from the Dutch airline KLM. Manipulation Check
In order to assess whether the manipulation of the independent variables was successful, a manipulation check was included in the survey. Participants were asked who the sender of the video was they just saw in order to test whether the source manipulation succeeded. Answer options included “One of your best friends” or “KLM”.
Additionally, respondents had to indicate who they think the people in the video that they have just seen were with answer options being “Real people” and “Actors”. In doing so, it was possible to analyze whether the respondents also perceived the videos as sensation marking events or not.
Mediator
Perceived creativity is regarded to be the mediating variable in this study. It was measured using ten items assessing elements of divergence, relevance and overall creativity, which were adapted from Smith et al. (2007). The question
“
How much do you agree or disagree with the following statements?” was asked first, followed by three items that measured divergence: “The ad was different.”, “The ad was uncommon.” and “The ad was unusual.”. Next, another three items assessed the ad’s relevance: “The viewing experience was relevant to me.”, “The viewing experience was useful to me.” and “Overall, the ad and the brand were NOT really applicable to me.”. Afterwards, four items examined participants overall levels of perceived advertising creativity: “In general, the ad was very creative.”, “The ad should win an award19
for creativity.” and “The ad was not very inventive and displayed little creativity in its design.”. All items were measured on a 7-point Likert scale ranging from [1] = ‘strongly disagree’ to [7] = ‘strongly agree’. The remaining overall creativity item asked participant to “Please rate the ad’s overall creativity on a scale from 1 to 7 where 1 means that you think the ad was “not very creative” and 7 means that you think it was “very creative”!”. Former
research has shown that overall creativity, overall divergence and overall relevance were all reliable measures (Cronbach’s alphas were α = .78, α = .88, and α = .83, respectively) (Smith et al., 2007). Again, this variable was measured on an interval level of measurement and the items of this scale were later combined in order to get an overall score for perceived ad creativity.
Dependent variables
Brand attitudes and forwarding intentions are the dependent variables of this study. In order to measure participants’ brand attitudes after exposure to the viral video Machleit and Wilson’s (1988) scale was adapted. Participants had to indicate their attitude towards the brand on seven different items: “dislike it very much / like it very much”, “useless / useful”, “worthless / valuable”, “unimportant / important”, “not beneficial / beneficial”, “not fond of / fond of” and “unenjoyably / enjoyable” using a 7-point differential scale. Thus, brand attitudes were measured on an interval level of measurement. This scale has been found to be reliable with Cronbach’s alpha ranging between α = .93 and α = .95 (Machleit & Wilson, 1988).
Afterwards, the items were combined into one variable which indicated the overall brand attitude after exposure to one of the videos that was forwarded by either a friend or a commercial source.
Forwarding intentions were measured using two items from Eckler and Bolls’ scale (2011). The following two statements had to be rated on a 7-point Likert scale ranging from [1] = ‘strongly disagree’ to [7] = ‘strongly agree’: “This ad is worth sharing with others” and “I will recommend this ad to others”. Therefore, forwarding intentions were measured at an interval measurement level. Previous research has demonstrated a Cronbach’s alpha of α = .89 for these items (Eckler & Bolls, 2011; Chiu et al., 2007). Afterwards, the items were also combined into one variable measuring the overall intention to forward the video.
Control Variables
Last, participants were asked to answer some questions regarding the control variables ‘ad attitude’ and ‘brand familiarity’ as well as their socio-demographic characteristics.
20
Furthermore, they were asked whether they had seen the video before and whether they could guess the aim of the study.
In order to measure ad attitude a scale by Smith et al. (2007), which has been found to be reliable (α = .91), was used. Participants were asked to answer the following question: “What is your overall evaluation of the advertisement?” using the four bipolar items “bad / good”, “unpleasant / pleasant”, “unfavorable / favorable”, “not likable / likable”. All four items were measured on a 7-point bipolar scale and later combined to one overall ad attitude item. Thus, ad attitude was measured on an interval level of measurement.
Next, brand familiarity was measured using an adapted scale from Dawar and Lei (2009) in which participants had to rate their overall familiarity on a 7-point Likert scale with answer options ranging from [1] = ‘not at all familiar’ to [7] = ‘very familiar’. The scale has been found to be reliable by the authors with Cronbach’s alpha being α = .94. Again, this variable used an interval level of measurement.
Lastly, some information about participants’ socio-demographics were assessed. They included questions about participants’ age, gender and level of education. In addition,
participants were asked whether they had already seen the advertisement before and whether they could guess what the research was about. Afterwards, participants were thanked for their time to participate in the study and debriefed.
Data Analysis Plan
After the data collection has been completed the data set needed to be prepared for further analyses. First of all, the data needed to be checked for any errors. This was done by running frequencies for each variable in order to search for missing or erroneous data. Missing data was then excluded from further analyses. Additionally, respondents who correctly guessed the aim of this study were excluded from the analyses. As a next step, a manipulation check was conducted in order to test whether the manipulation of the independent variables was
successful. Only respondents who correctly answered who the source of the video was they just saw and whether the video contained actors or real people were included in the following analyses.
Some of the endpoints of the scales used were randomized in order to rule out
response effects. Therefore, they needed to be recoded so that all numbers on the scale had the same meaning. Moreover, most of the scales contained several items which then needed to be computed in order to create an overarching variable.
21
Afterwards, several principal component analyses were run to control whether all items of a scale load on one factor. If the items did not load on one factor the rotated
component matrix was checked to see which item did not belong to the factor. The scale for creativity, however, consisted of several dimensions. Therefore, the principal component analysis should also show multiple factors. Also, forwarding intentions only consisted of two items. A principal component analyses would therefore not have been feasible and a bivariate correlation analysis was run. As a next step, a reliability analysis was done for each scale in order to assess whether the scale was reliable and whether the reliability could be increased by removing items.
Before starting the main analyses a randomization check was done to test whether the randomization created equal experimental groups. A one-way ANOVA was run in order to test whether the four groups differed significantly in means of age, gender and education. Furthermore, the hypotheses needed to be tested for alternative explanations. This was done via several bivariate correlation analyses. Variables that needed to be controlled for were ‘ad attitude’, ‘brand familiarity’, ‘gender’, ‘age’, ‘educational level’ and ‘recognition of the video’. Only if there were significant correlations between the control and dependent variables did the control variables have be considered in future analyses.
The first hypothesis stated that viral advertising videos containing a sensation marketing event are perceived as more creative than viral advertising videos without a sensation marketing event. In order to test this and because some covariates had to be considered, an analyses of covariance (ANCOVA) was run.
The second hypothesis focused on the source of the video and claimed that people who received a viral advertising video from an interpersonal source are more likely to form
positive attitudes towards the brand than people who received it from an external/commercial source. Again, this relationship was tested by running an ANCOVA.
The third hypothesis stated that the higher the perceived levels of creativity are, the higher will the brand attitudes be. This hypothesis was tested by running a multiple regression analysis.
The fourth hypothesis claimed that perceived creativity mediates the relationship between the type of video and brand attitudes. According to Baron and Kenny (1986) three steps are necessary to test mediation. At first, a multiple regression analysis needs to test whether the type of video has a significant effect on brand attitudes. In order to do this, the type of video variable needed to be dummy coded. If the type of video significantly predicts brand attitudes, one can continue to the second step. The second step consists of a regression
22
predicting the mediator from the predictor variable. Again, the dummy coded type of video variable was used to predict creativity levels. Should the effect be significant one can continue to the third step, which consists of a regression predicting the outcome from both the predictor variable and the mediator. Should the creativity significantly predict brand attitudes and should the type of video now less strongly predict brand attitudes, a mediation effect has been found (Baron & Kenny, 1986; Field, 2013).
The fifth hypothesis assessed whether higher brand attitudes lead to higher forwarding intentions. Again, this relationship was tested by running a multiple regression analysis.
The sixth hypothesis predicted that brand attitudes will mediate the relationship between the source of the video and the intentions to forward it. In order to test for this relationship several regression analyses had to be run. Following the guidelines by Baron and Kenny (1986) the effect of the video’s source on the intention to forward it was tested first by conducting a multiple regression analysis. The source of the video had to be dummy coded to enable the analysis. Secondly, a multiple regression predicting brand attitudes from the video’s source needed to be run. Again, the dummy coded source variable had to be used. Last, a multiple regression predicting forwarding intentions from both brand attitudes and source had to be run. Only if (1) the source of the video significantly predicted forwarding intentions; (2) the source significantly predicted brand attitudes; (3) the source significantly predicted forwarding intentions when taking brand attitudes into account and (4) the source predicted forwarding intentions less strongly than before, can we speak of a mediating effect.
Finally, the seventh hypothesis claimed that brand attitudes will mediate the relationship between perceived creativity levels and the intentions to forward the video. Again, several regression analyses had to be run following the guidelines by Baron and Kenny (1986). At first, a multiple regression tested the effect of perceived creativity on forwarding intentions. If this relationship was found to be significant one could continue to the next step. Another multiple regression predicting brand attitudes from creativity needed to be run next. If, again, this relationship was found to be significant one could continue with the last step. A final multiple regression predicting forwarding intentions from both brand attitudes and perceived creativity had to be run. If creativity was found to significantly predict forwarding intentions when taking brand attitudes into account and when it did so less strongly than originally, a mediating effect has been found.
23
Results
Preliminary Analyses
Data preparation
Before starting the main analyses, all incomplete responses were excluded from further analyses, which narrowed the sample size of 331 down to 195 (response rate of 59%). In addition, the responses by people who correctly guessed the aim of the study were deleted. Two people came very close by guessing that the study is about “How social settings
portrayed in videos, shared by friends you trust, affects your perception of brands.” and “[…] whether seeing a video with such "human-touch" and engagement impacts the brand
perception.". The answers of both participants were excluded from further analyses since their knowledge of the research’s aim might have influenced their responses. Secondly, a
manipulation check found that only 125 out of 193 respondents correctly answered who the source of the video was they just saw and whether the video contained actors or real people. The ones who did not correctly answer the manipulation check were then excluded from further analyses.
Scale testing: Principal component- and reliability analyses
The scales described above had to be tested for whether all items load on one factor and for their reliability. In order to do this, several principal component- and reliability analyses were run. The brand attitudes scale was tested first. The Kaiser-Meyer-Olkin value for that scale had a meritorious value of .83 and therefore verified the sampling adequacy. Additionally, Bartlett’s test of sphericity was found to be significant, χ2
(21) = 503.51, p < .001. A principal component analysis with Varimax rotation was run to obtain eigenvalues for each factor in the data. Results showed that the seven items formed two components with eigenvalues above Kaiser’s criterion of 1 (component 1: EV = 4.22, R2
= .60; component 2: EV = 1.02, R2 = .15). Thus, these findings would suggest that brand attitudes consist of two subscales (see
Appendix 2, Table 6). However, a reliability analysis found that the scale had a strong
reliability when keeping all items in it, Cronbach’s α = .89; M = 5.03, SD = .82. Moreover, the reliability would not increase if any of the items would be deleted. Therefore, it was decided to keep all items in the original scale as suggest by the authors Machleit and Wilson (1988) and combine them in a new scale called ‘brand attitudes’.
Next, the scale for forwarding intentions was tested. The Kaiser-Meyer-Olkin value for that scale had a barely acceptable value of .50, which verified the sampling adequacy. Additionally, Bartlett’s test of sphericity was found to be significant, χ2
24
A principal component analysis was run to obtain eigenvalues. Results show that both items loaded on the same factor and had an eigenvalue above Kaiser’s criterion of 1, EV = 1.84, R2
= .92 (see table 7). Since the scale only consisted of two items, a bivariate correlation analysis was run instead of a reliability analysis to test whether both items measured the same concept. Both items had a significant, very strong correlation and thus confirmed the reliability of the scale, r = .84, p < .001. Thus, the two items were in a next step combined to create a new ‘forwarding intentions’ scale.
For perceived creativity, the Kaiser-Meyer-Olkin value was .85 which verified the sampling adequacy. Furthermore, Bartlett’s test of sphericity was found to be significant, χ2
(45) = 794.84, p < .001. The principal component analysis with Varimax rotation obtained two factors with eigenvalues above Kaiser’s criterion of 1 (component 1: EV = 5.26, R2
= .53; component 2: EV = 1.52, R2 = .15). Table 8 (Appendix 2) illustrates the rotated factor loading of each of the 10 components and demonstrates that the items of the suggested theoretical dimensions ‘Divergence’ and ‘Overall Creativity’ load on the same factor and that the dimension ‘Relevance’ forms an own factor. Thus, the findings of the principal component analyses suggest that perceived creativity consists of two dimensions (divergence and relevance). In a next step, reliability analyses were run for all items together and also separately for the two factors. The reliability of all creativity items was very high with Cronbach’s α = .89; M = 4.46, SD = 1.09. When looking at the two factors separately results for reliability clearly differ. Reliability for the divergence scale was very high, Cronbach’s α = .91; M = 4.62, SD = 1.24, while the reliability for relevance scale was moderately high,
Cronbach’s α = .74; M = 4.11, SD = 1.28. Additionally, reliability for the divergence scale would not increase if any of the items would be deleted. Reliability of the relevance scale, however, would slightly increase to Cronbach’s α = .79 if the item ‘Overall, the ad and the brand were applicable for me’ would be deleted. Nevertheless, it was decided to keep the item in the scale since the increase would be marginal and prior literature suggested that it belongs in the scale (cf. Smith et al., 2007). Similarly, the reliability of the ‘Creativity’ scale, which combined all items could also be slightly increased by deleting the item mentioned above (α = .898) or by deleting the item ‘The viewing experience was useful to me’ (α = .894). However, due to the minor increases that would result and due to prior literature suggestions it was decided to keep the items in the scale. Overall, for future analyses, all three scales were computed in order to look for differences in results. All ‘Overall creativity’ and ‘Divergence’ items were combined to the new ‘Divergence’ scale, all ‘Relevance’ items were combined to
25
the new identically named scale and an ‘Overall Creativity’ scale was created which combined all items.
Lastly, the ad attitude scale was tested. The Kaiser-Meyer-Olkin value for that scale had a meritorious value of .86, which verified the sampling adequacy. Additionally, Bartlett’s test of sphericity was found to be significant, χ2(6) = 387.93, p < .001. A principal component analysis was run to obtain eigenvalues. Results show that all four items loaded on the same factor which had an eigenvalue above Kaiser’s criterion of 1, EV = 3.30, R2 = .82 (see Appendix 2, Table 9). Furthermore, reliability for the scale was very high, Cronbach’s α = .93; M = 5.18, SD = 1.11. Moreover, this score would not increase if any of the items would be deleted. Thus, the four items were in a next step combined to create a new ‘ad attitude’ scale.
Randomization check
Before starting the main analyses a randomization check was done to test whether the randomization created equal experimental groups in terms of respondents’ age, gender and education. A one-way ANOVA indicated that there were no significant differences between the experimental conditions with respect to age, F(3, 121) = .19, p = .902. Additionally, crosstabulation analyses showed that there were no significant differences between the four experimental groups with respect to gender, χ2(3) = 6.88, p = .076, or education, χ2(18) = 14.36, p = .705. Hence, the randomization has been successful.
Correlation analyses
Furthermore, the hypotheses needed to be tested for alternative explanations. This was done via several bivariate correlation analyses. Variables that needed to be controlled for were ‘ad attitude’, ‘brand familiarity’, ‘gender’, ‘age’, ‘educational level’ and ‘recognition of the video’. Ad attitude strongly and significantly correlated with all three creativity scales, rD = .61, p < .001; rR = .52, p < .001; rAll = .66, p < .001, as well as moderately with brand attitudes, r = .37, p < .001 and strongly with forwarding intentions, r = .61, p < .001. This suggests that respondents’ attitude towards the ad positively correlated with all dependent variables. Ad attitude will therefore be used as a control variable in all further analyses. Moreover, there was a weak, significant association between brand familiarity and the creativity scale
‘Relevance’, r = .22, p = .013 and a moderate, significant association with brand attitudes, r = .44, p < .001. When looking at the bivariate correlation analysis results for age, only the ‘Relevance’ scale weakly correlated with this control variable, r = .21, p = .021. Furthermore,
26
there were no significant correlations between respondents’ gender as well as ad recognition and the dependent variables. Lastly, education had a slightly significant and weak correlation with the ‘Divergence’ scale of creativity, rs = .18, p = .049. However, since the effect was so
small and nearly crossed the significance threshold it was decided to not consider this variable in future analyses. All of the other significant correlations between the control and dependent variables had to be considered and controlled for in future analyses.
Hypothesis Testing
The impact of the viral video type on creativity perceptions
The first hypothesis stated that viral advertising videos containing a sensation marketing event are perceived as more creative than viral advertising videos without a sensation marketing event. An independent samples t-test using the overall creativity scaleshowed that the sensation video was indeed perceived as more creative (M = 5.11, SD = .82) than the non-sensation video (M = 3.90, SD = .98). This difference, -1.21, BCa 95% CI [-1.51, -.89], was significant t(123) = -7.40, p < .001. When using the two subscales of creativity, ‘divergence’ and ‘relevance’, the sensation video was also perceived as more creative (MD = 5.38, SDD =
.88; MR = 4.48, SDR = 1.11) than the non-sensational video (MD = 3.95, SDD = 1.11; MR =
3.79, SDR = 1.33). The mean differences for the divergence scale, 1.43, BCa 95% CI [1.79,
-1.09], as well as the relevance scale, -0.69, BCa 95% CI [-1.10, -0.25], were both significant tD(121.99) = -8.04, p < .001; tR(123) = -3.10, p = .002.
However, since the bivariate correlation analyses of the preparatory analyses showed that the overall creativity scale correlates with ad attitude an analysis of covariance
(ANCOVA) was run to take this covariate into account. Results show that ad attitude had a significant, moderate effect on overall creativity perceptions, F(1, 122) = 62.24, p < .001, partial η2 = .34. There was also a significant, weak effect of the type of viral video on creativity perceptions after controlling for the effect of ad attitudes, F(1, 122) = 26.80, p < .001, partial η2 = .18. Additionally, two ANCOVAs were run using either the divergence or relevance scale. Since divergence correlated with ad attitude and relevance correlated with age, ad attitude and brand familiarity those variables were included as covariates in the analyses. Results show that the covariate ad attitude was significantly and weakly related to divergence perceptions, F(1, 122) = 32.06, p < .001, partial η2 = .26. There was also a
significant, weak effect of the type of viral video on creativity perceptions after controlling for the effect of ad attitudes, F(1, 122) = 25.89, p < .001, partial η2 = .22. Moreover, results for the relevance scale show that the covariates age, F(1, 120) = 7.17, p = .013, partial η2 = .05,
27
ad attitude, F(1, 120) = 37.17, p < .001, partial η2 = .22, and brand familiarity, F(1, 120) = 5.44, p = .029, partial η2 = .04, were significantly and weakly or very weakly related to perceptions of the video’s relevance perceptions. However, the effect of the type of viral video on relevance perceptions was no longer significant when controlling for these three covariates, F(1, 120) = 0.64, p = .450. Thus the first hypothesis only holds for ‘overall creativity’ and ‘divergence’ perceptions but not for ‘relevance’ perceptions. Therefore, the first hypothesis was only partially accepted.
The impact of the viral video’s source on brand attitudes
The second hypothesis focused on the source of the video and claimed that people who received a viral advertising video from an interpersonal source are more likely to form positive attitudes towards the brand than people who received it from an external/commercial source. Results of an independent samples t-test showed that, on average, people who
received a viral video from a friend had slightly higher attitudes towards the brand KLM (M = 5.11, SD = .76) than people who received a viral video from an external source (the brand itself) (M = 4.95, SD = .87). However, this mean difference, -.16, BCa 95% CI [-0.48, 0.13], was not significant, t(123) = -1.12, p = .265.
Additionally, results of an ANCOVA controlling for ad attitudes and brand familiarity came to a similar result. The covariates ad attitude, F(1, 121) = 8.92, p < .001, partial η2 = .14, and brand familiarity, F(1, 121) = 13.43, p < .001, partial η2 = .19, were both significantly and weakly related to brand attitudes. However, no significant main effect was found for video source on brand attitudes when controlling for those covariates, F(1, 121) = 0.02, p = .828. Therefore, the second hypothesis was rejected.
Impact of perceived creativity on brand attitudes
The third hypothesis stated that the higher the perceived levels of creativity are, the higher the brand attitudes will be. The regression model with brand attitudes as a dependent variable and overall creativity, ad attitude and brand familiarity as independent variables is significant, F(3, 121) = 20.29, p < .001. The regression model can therefore be used to predict attitudes towards the brand. Approximately 34% of the variation in brand attitudes can be explained by overall creativity, ad attitudes and brand familiarity (R² = .34). Overall creativity had a
significant, weak association with brand attitudes, b* = 0.23, t = 2.33, p = .022, 95% CI [0.03, 0.32], while ad attitudes were found to have an insignificant, weak association, b* = 0.18, t = 1.81, p = .073, 95% CI [-0.01, 0.28]. Last, brand familiarity was found to have a significant,
28
moderately strong association with attitudes towards the brand, b* = 0.39, t = 5.17, p < .001, 95% CI [0.10, 0.23]. As overall creativity increases by one unit, brand attitudes increase by .17 units. Similarly, for each additional point on the ad attitudes and brand familiarity scales, brand attitudes increase by .13 and .16 units respectively. For all these effects the other independent variables are assumed to be held constant.
Furthermore, a second regression model which also used brand attitudes as dependent variable but divergence, ad attitudes and brand familiarity as independent variables was found to be significant and can therefore be used to predict brand attitudes, F(3, 121) = 18.20, p < .001. Approximately 31% of the variation in brand attitudes can be explained by divergence, ad attitudes and brand familiarity (R² = .31). Divergence had an insignificant, very weak association with brand attitudes, b* = 0.10, t = 1.03, p = .303, 95% CI [-0.06, 0.19]. Ad attitudes had a significant, weak association, b* = 0.27, t = 2.87, p = .005, 95% CI [0.06, 0.34, and brand familiarity a significant, moderately strong association with brand attitudes, b* = 0.41, t = 5.36, p < .001, 95% CI [0.11, 0.23]. One point extra on the divergence scale would increase brand attitudes by .07 units. Likewise, each additional point on the ad attitudes and brand familiarity scales is associated with an increase of .20 and .17 on the brand attitudes scale. Again, for all these effects the other independent variables are assumed to be held constant.
Finally, a third multiple regression analysis was run. Brand attitudes were used as dependent variables and relevance, ad attitudes and brand familiarity as independent variables. The regression model was significant and can thus also be used to predict brand attitudes, F(3, 121) = 24.41, p < .001. Furthermore, approximately 38% of the variation in brand attitudes can be explained by the three independent variables (R² = .38). Relevance, b* = 0.32, t = 3.74, p < .001, 95% CI [0.10, 0.31], and brand familiarity, b* = 0.36, t = 4.82, p < .001, 95% CI [0.09, 0.21], both had a significant, moderately strong association with brand attitudes, while ad attitudes had a significant and weak association with brand attitudes, b* = 0.17, t = 2.04, p = .044, 95% CI [0.00, 0.25]. For each additional point on the relevance scale brand attitudes increase by .21 units. Moreover, one point extra on the ad attitudes and brand familiarity scale would increase brand attitudes by .13 and .15 units respectively. All these interpretations are true only if the effects of the other independent variables are held constant.
A summary of the findings of all three models can be found in table 10 (Appendix 3). According to the results, the third hypothesis can only partially be accepted since only overall creativity and relevance significantly predicted brand attitudes.
