Influencers on YouTube
The effect of paid versus unpaid Influencers on
brand engagement of consumers
Anne-Laurien van Zoonen 11859474 Master Thesis Graduate School of Communication Master’s Programme Corporate Communication Dr. S. C. De Bakker October 26th, 2018
Page 1 of 58
Acknowledgments
I would first like to thank my thesis advisor Dr. Suzanne C. de Bakker from the University of Amsterdam. Her door was always open whenever I ran into trouble or had yet another
clarification question about my research or writing. She allowed me my own freedom in writing this paper but managed to steer me away from danger time and time again when it was necessary. Thank you.
I would also like to thank everyone that participated in taking the time to participate in this study. Without them, I would not have been able to run the experiment. I would also like to thank everyone that shared my thesis at work or with their friends, it led me to double the participants I needed, and for that I am forever grateful.
Then a few people I would like to thank for their extensive help in getting me through this thesis. First and foremost, my little sister, Liselotte van Zoonen, for being the influencer I needed in my thesis, showing her face to literally everyone who participated, and listening to my stressing about deadlines. I would also like to thank Irene Overtoom, without her SPSS would still be beating me and I would not have managed to finish this thesis.
Finally, I must thank my family, and my parents, for putting up with me throughout the years and through the last months of my education. You were always there for me when I needed it and took an interest in what I was doing. Even in difficult times, you encouraged me not to quit, and I thank you for your patience, effort, and love.
Thank you.
Page 2 of 58
Abstract
Influencers can be seen on a plethora of social media platforms, and this thesis explores the platform YouTube. Literature explains that the use of influencers can lead to an increase in brand engagement. Aside from this, it is explained that there are two types of influencers, those that are paid, and those that are not, however, little research has been done on whether or not the effect on brand engagement is different for unpaid influencers when compared to paid influencers, which is the gap this thesis aims to fill. Furthermore, this thesis uses the persuasion knowledge model to explain that the messages posted by influencers can either contain a one-sided, or a sided message. Literature has shown that influencers using two-sided messages have a larger effect on brand engagement than influencers using one-two-sided messages, however no distinction is made between paid or unpaid influencers. Therefore, this thesis additionally examined whether or not the effect of message sidedness would be
different for paid influencers when compared to unpaid influencers on brand engagement. Literature also explains influencers with higher credibility have a larger effect on brand engagement, but there is no consensus on whether unpaid influencers are more or less
credible than paid influencers. This lack of consensus is the third gap that this thesis explores. The hypotheses were tested through an online experiment and survey with 283 participants. The participants were shown a fictional product review and then asked to complete a survey. The results of this study showed that there was no significant difference between the paid and unpaid influencer on the brand engagement of the consumer. Furthermore, results showed that there was also no significant interaction effect between influencer type and message sidedness, and that credibility did not play a mediating role in this relationship. Areas for further research proposed in the discussion are clearer disclosures for both influencer types, as well as the placement of negative arguments in two-sided messages.
Page 3 of 58
Introduction
With the rise of social media, and consumer participation on these platforms, it has become easier and quicker for consumers to gather information about various products and services that can be purchased. According to Ang (2011), and Wei and Lu (2013), the importance with the rise in social media and available information on those platforms, is that consumers now have the ability to research and gather information in numerous different places. Aside from this, social media platforms such as YouTube, Twitter, and Facebook give the consumers the opportunity to review companies, products, and services, and allows for other consumers to find these reviews (Ang, 2011). This also means, according to Ang (2011), that companies can use these platforms to not only inform themselves about consumer wishes, but they can use them to try and encourage users to become advocates, or “influencers” for their company or brand. How this works is that consumers are purchasing products and reviewing them in an online setting, where other consumers are free to look up the information. The content of this information, however, is not always controlled by the company that the review is about (Wei & Lu, 2013). In these cases, companies are working together with influencers, previously known as opinion leaders, to increase brand engagement and awareness (Bernritter, Loermans, Verlegh, & Smit, 2015).
The consumers on YouTube that post videos, known as YouTubers, upload videos to their personal channels, and one of the biggest and fastest growing communities on YouTube is the self-proclaimed ‘beauty’ community (Berryman & Kavka, 2017). This community involves a number of YouTubers that post videos of themselves testing new products that have been produced by cosmetics companies, and are grouped into four subgroups: haul, first impression, monthly favourite, and tutorial videos (Berryman & Kavka, 2017). The
difference in these videos is that a haul is where the YouTuber goes out and purchases the products and then shows them in a reveal to their followers, but their own opinions about the
Page 4 of 58 individual products themselves remains unspoken, they do not go into further depth about the products themselves (Jeffries, 2011). During a first impression video, one or two products are chosen, and the YouTuber describes their first impressions of the product, what it looks like and smells like, whether it is comfortable to hold, and they then use the product for the first time and review the product in a seemingly objective way. Whether or not the first
impression video is indeed objective, relies on whether the company that created the product that is used in the review is sponsoring the video or not. The monthly favourite is where the YouTuber has decided for themselves which products, of all the ones that they have reviewed in the month, is their favourite and they explain why this is the case. This could be for any reason, simply from the colour of the packaging to the ease of application. The last case, the tutorial videos, are slightly more complex. In essence, the YouTuber creates a ‘How To’ step by step explanation of how something is done, which can be anything from creating the prefect eyebrow, to a full face of drag make up step by step.
These four subgroups fall under one of the two types of video that Berryman and Kavka (2017) describe as being content videos, seen as more professional and based on the talents of the YouTuber, which is applicable to all communities on YouTube and not just the beauty community. In the case of the beauty community, these videos would be about beauty-related topics such as product reviews. The other type of video that Berryman and Kavka (2017) describe that can be found in all communities of YouTube, and therefore also in the beauty community, is a vlog, which can be seen as a behind the scenes diary, where the YouTuber actively engages in their daily business whilst filming themselves and voicing their opinions.
The difference between these influencers working with or separately from the company, is explained by Hall (2016) as being either earned influencer marketing or paid influencer marketing. Earned influencer marketing is unpaid, and the organization is using
Page 5 of 58 pre-existing relationships with the influencers, where the initiative of the message does not come from the company but from the influencers themselves, which is where the term earned stems from. What this means is that there are influencers on social media platforms that promote and engage with a brand, even though the brand has not prompted these influencers to do so. With paid influencer marketing, the company is exchanging money or free samples for the influencers to express their opinions about their company or product. This is done because influencers are seen as a new form of human advertisement on social media, that can influence audiences through social media platforms as spokespersons for companies (Freberg K., Graham, McGaughey, & Freberg L., 2011).
That there is a difference between earned influencer marketing and paid influencer marketing (Hall, 2016) is supported by both Thomson (2017) as well as Zajc (2015), although they use different terminology. Thomson (2017) explains that there are two types of
influencer, namely those that monetize their channel (paid influencer marketing) and those that do not (earned influencer marketing). According to Zajc (2015), influencers can have either one of two motives, either they have a social motive (earned influencer marketing) or they have an economic motive (paid influencer marketing). What this literature (Hall, 2016; Thomson 2017; Zajc, 2015) comes down to is that there are two types of influencers, those that are paid, and those that are not paid by the organization or brand to promote the product.
In research done by Lu, Chang W. and Chang H. (2014) the importance of using paid influencers as an effective marketing tool is discussed, but the effects of unpaid influencers is not examined. In the papers by Lu et al. (2014) and Wiess (2014), it is explained that
influencers on social media sites such as Facebook and Instagram can increase sales as well as brand engagement and an overall positive attitude towards the brand, however, no
distinction is made between paid or unpaid. Other research (Kapitan & Silvera, 2016; Taylor, 2015; Wei & Lu, 2013) further supports the positive effect of influencers on consumer brand
Page 6 of 58 engagement, but also argue that not enough research is done on the difference of this effect between paid and unpaid influencers, which is a gap that this thesis aims to fill.
Weiss (2014) explains that research done on influencers on social media has often excluded YouTube for the reason that the platform remains more unknown to marketers (Barnes & Jacobsen, 2014; Berryman & Kavka, 2017; Geurin & Buch, 2017). This is because it is seen as the site where the content of the videos that is created is less controllable than other sites such as Twitter and Facebook (Wei & Lu, 2013; Smith, Fischer, & Yongjian, 2012). The limited research done on the specific platform YouTube (Bambauer-Sachse & Mangold, 2013; Barens & Jacobsen, 2014; Berryman & Kavka, 2017; Geurin & Buch, 2017; Weiss, 2014; Smith et al., 2012) as well as the little amount of research available for the effect of unpaid influencers on brand engagement (Ang, 2011; Hall, 2016; Lu et al., 2014; Thomson 2017; Yoshida, Gordon, Nakazawa, Shibuya, & Fujiwara, 2018; Zajc, 2015), provides a niche opportunity to examine it. This has led to the first research question of this paper:
Research Question 1: To what extent do unpaid influencers have a larger effect on brand engagement than paid influencers?
As mentioned above, there is a gap in the literature on the effect of unpaid influencers on brand engagement compared to paid influencers, which this thesis aims to fill by
answering the proposed research question. Research has shown that influencers have a positive effect on brand engagement on social media platforms such as Facebook, Instagram and Twitter, but research is lacking when it comes to the social media platform YouTube. It will therefore be interesting to see what the difference of effect of unpaid versus paid influencers is on this social media platform. With the results of the study, marketers could
Page 7 of 58 plan campaigns on how to best utilize influencers on YouTube, both paid and unpaid, as it will be beneficial for an increase in brand engagement.
When influencers create videos, especially reviews about products or services, they have the choice of putting either only positive or negative aspects into the video, or both. This is known in literature as message sidedness (Yao, Joshi, Chang, McDonalds, Tran, Wheeler, & Hou, 2018). Knowles and Linn (2004) explain that a one-sided message consists of only arguments that make the message believable, so for example only positive arguments. A two-sided message, on the other hand, contains these arguments, as well as arguments against the original message (Knowles & Linn, 2004). This is done because presenting consumers with both sides of the argument is a strategy to decrease resistance (Uribe, Buzeta, & Velasquez, 2016). Decreasing consumer resistance is important because, as literature has shown
(Fransen, Verlegh, Kirmani, & Smit, 2015; Pierro, Giacomantonio, Pica, Giannini, Kruglanski, & Higgs, 2013; Uribe et al., 2016), two-sided messages are more effective in increasing brand engagement than one-sided messages, as two-sided messages are more persuasive. How they become more persuasive is further discussed later on during the explanation of the persuasion knowledge model.
What this means is that a review of a product is considered to be more credible if it is not only either positive or negative, but has both aspects. This is based in inoculation theory as well, where the presentation of counter argumentation in the message already inhibits counter arguing of the consumer (Compton, 2017; Fransen et al., 2015; Uribe et al., 2016). According to Uribe et al. (2016) further research into the effects of message sidedness by influencers on social media is necessary. What has already been shown in literature
(Compton, 2017; Fransen et al., 2015; Knowles & Linn, 2004; Pierro et al., 2013; Uribe et al., 2016; Yao et al., 2018) is influencers that use two-sided messages have a larger effect on brand engagement than those using one-sided messages. The gap, however, is that research is
Page 8 of 58 lacking as to how far the effects of message sidedness on brand engagement differs between paid and unpaid influencers, which has led to the second research question of this thesis:
Research Question 2: How does the sidedness of the message affect the relationship between influencers and brand engagement?
As mentioned above, messages are viewed as more credible if the content of the message is two-sided, as opposed to one-sided only. In literature, the credibility associated with social media messages is often referred to as “source credibility” (Lim, Cheah, & Wong, 2017; Ohanian, 1990; Lafferty, Goldsmith, & Newell, 2002; Pornpitakpan, 2004; La Ferle & Choi, 2005). Source credibility is built on three main components, namely: trustworthiness – the trust a listener has in a speaker -, expertise – the belief that the information presented by the speaker is accurate -, and attractiveness of the speaker (La Ferle & Choi, 2005; Ohanian, 1990). Lim et al. (2017) explain that social media influencers are considered as more credible than celebrity endorsements and sponsorships, because of the influencers’ ability to build the relationships with their audiences. Furthermore, Wei and Lu (2013) and Kapitan and Silvera (2016) explain that the credibility of the influencer has a larger effect on the relationship between influencer and brand engagement, than whether or not the influencer is paid or unpaid. These two papers additionally explain that when an influencer is paid by a brand, they do not necessarily know the product they are endorsing very well, and this could show to an audience. On the other hand, an influencer that is unpaid but passionate and
knowledgeable about a product will be viewed as more credible, and therefore have a larger influence on brand engagement than a paid, less credible influencer (Wei & Lu, 2013).
With regards to brand engagement, source credibility is often used as a measurement scale to the effectiveness of influencers on brand engagement (Lim et al., 2017). What this
Page 9 of 58 means is that source credibility can be seen as a mediator between influencer type and brand engagement, which has led to the third research question of this paper:
Research Question 3: What role does the source credibility play in the relationship between influencer type and brand engagement?
Brand engagement is important for organizations because, aside from the performance outcomes mentioned above, long-term effects can also be actualized (Van Doorn, Lemon, Mittal, Nass, Pick, Pirner, & Verhoef, 2010). According to Van Doorn et al. (2010) these long-term effects, such as customer retention over time, are important for organizations to keep its competitive advantage.
Theoretical Framework
Brand Engagement
According to Maslowska, Malthouse, and Collinger (2017a, p.231) engagement is described as “[a] behavioural manifestation toward the brand or firm that goes beyond transactions”. This can be explained as engagement having two main aspects, namely; behavioural and psychological. In other words, that brand engagement goes further than only the monetary transaction of purchasing from a brand, but involves the creation of a relationship with the brand. As a result of this, brand engagement can lead to actions, such as Word-Of-Mouth (WOM) or User Generated Content (UGC) (Maslowska et al., 2017a).
According to McEwen (2004), brand engagement is the result of the combination of rational and emotional bonds. Active engagement with a brand can create brand engagement, and is a result of different processes, such as involvement with the brand, trust in the brand, and the affective attitude towards the brand (Bowden, 2009). What this means is that brand engagement is a process that only happens over time and not straight off the bat. Bowden (2009) proposes a three-step model that explains how brand engagement can be built, where the first stage happens when the customer purchases a product or service based on a cognitive
Page 10 of 58 basis. The second step happens when there is an increase in involvement that is naturally supported by heightened trust in repeated purchases (Bowden, 2009). Consequently, in the final step, a more affective commitment is created towards the brand itself, based on an emotional level for purchasing which leads to lasting brand engagement (Bowden, 2009). This demonstrates that brand engagement is created through different steps, and moves from a cognitive to an emotional bond with the brand (Van Doorn et al., 2010).
This shift can be better explained through the paper by Bernritter et al. (2017) where they take the action of liking posts on Facebook, and try to distinguish the driving factors of this action. What they explain, together with Calder, Malthouse, and Schaedel (2009) as well as Sprott, Czellar, and Spangenberg (2009) is that this emotional bond can also be created to strengthen the self-identity of the consumer (Bernritter et al., 2017). This personal form of engagement leans more towards consumers feeling an attachment to a brand, making repeat purchases, and feeling that the brand is a part of their identity (Calder et al., 2009). Bernritter et al. (2017) also explain that this feeling of self-identity creation influences consumers to indulge in brand engagement.
As a consequence of the rise of social media, brands have lower levels of control over message content as consumers are now more able to make their ideas and opinions heard to a wider audience (Genseler, Volckner, Liu-Thompkins, & Wiertz, 2013). None the less, the rise of social media also leads to an increase in brand engagement, that consequently leads to improved attitudes as well as favourable behaviour of consumers towards the brand (Brodie, Ilic, Juric, & Hollebeek, 2013; Gummerus, Liljander, Weman, & Pihlstrom, 2012; Kumar, Aksoy, Donkers, Venkatesan, Wiesel, & Tillmanns, 2010; Seraj, 2012; Sridhar & Srinivasan, 2012).
Page 11 of 58 In other literature (Ferreira & Coelho, 2015), results have been found to show that brand engagement is positively related to product involvement. What this means is that when a consumer is highly attached to a service or product, repeat purchasing is likely to happen even across the same brand (Ferreira & Coelho, 2015). As seen in the three-step process by Bowden (2009), repeat purchasing and involvement with the brand leads to emotional bonds with the brand, which in turn leads to increased brand engagement. This is further supported by Hollebeek, Glynn, and Brodie (2014) who have developed the Consumer’s Brand
Engagement (CBE) scale. This scale measures three levels that coincide with the three dimensions that they use to define engagement. The first level is the cognitive level, the second level is the affective, or emotional level, and the final level is the activation, or behavioural level. Hollebeek et al. (2014) explain that better organizational performance outcomes, such as an increase in sales or brand referrals, are a result of consumer’s brand engagement.
This thesis uses the emotional and psychological bond that consumers create with a specific brand as the definition for brand engagement. Furthermore, it recognizes that, according to the literature above, brand engagement can stimulate behavioural responses. These behavioural responses can lead to positive behavioural outcomes such as repeat
purchasing, but also content creation through UGC, as well as negative behavioural outcomes such as organizing protests against an organization (Van Doorn et al., 2010). This is
furthermore supported by Vivek, Beatty, and Morgan (2012) in that brand engagement can also have the behavioural outcomes of trust, affective commitment, and brand community involvement.
Persuasion Knowledge
Friestad and Wright (1994) explain the persuasion knowledge model, which is a model that shows how an individual’s personal persuasion knowledge influences their reactions to
Page 12 of 58 persuasion attempts. The importance in this model is that an individual’s level of persuasion knowledge is learned and developed through time. Within this model two sides are looked at, namely the “target”, that is the person who the persuasion attempt is aimed at, and the
“agent” who is responsible for creating the persuasion attempt. Per side, there are three knowledge structures; persuasion knowledge, agent knowledge, and topic knowledge. The first, persuasion knowledge, encompasses the knowledge a target has about the persuasion tactics used, and if they can recognize them. Agent knowledge deals with the knowledge that a target has about the salesperson or company of the persuasion attempt. Topic knowledge means the knowledge that an individual has about the topic which the persuasion attempt is about. For further clarification on the model, Appendix 1 can be consulted as it represents the figure proposed by Friestad and Wright (1994, p.2). Consequently, when an individual’s persuasion knowledge is activated, they understand that they are being persuaded and they then have the choice to be open to the persuasion attempt or to resist the persuasion attempt (Van Reijmersdal, Fransen, van Noort, Opree, Vandeberg, Reusch, & Boerman, 2016).
Being aware of a persuasion attempt, and having persuasion knowledge activated, can lead to one of two reactions, either cognitive resistance or affective resistance (Van
Reijmersdal et al., 2016). When cognitive resistance is used, it means that an individual uses cognition in order to refute the persuasion attempt, meaning that they are using pre-existing knowledge to counterargue the message (Fransen et al., 2015). When affective resistance is activated it means that an individual uses emotion to counter the persuasion attempt, which could be a reaction such as getting upset or angry (Van Reijmersdal et al., 2016). These two reactions to persuasion attempts can be further divided into one of the three strategies proposed by Fransen et al. (2015), namely; avoidance (affective resistance), contesting (cognitive resistance), and empowering (can be both).
Page 13 of 58 persuasion attempt by either ignoring or simply clicking away from the advertisement.
Contesting is done when individuals counter the persuasion attempt by either coming up with counterarguments as to why the persuasive message is not true, or by source derogation, meaning contesting the source itself. When a consumer uses an empowering strategy, they are focused more on their own beliefs and the focus is no longer on the message itself, but instead they are finding reasons and information that confirms their own beliefs (Fransen et al., 2015). The activation of persuasion knowledge and the resulting resistance strategy used is important because companies can use influencers, as well as two-sided messages to counter these resistance strategies, which is explained further on in this thesis.
Banner Blindness
One persuasion avoidance strategy that consumers use is called banner blindness. For this tactic, a consumers’ persuasion knowledge has been activated because they understand that the banner is a persuasion attempt (Hervet, Guérard, Tremblay, Chtourou, 2011; Resnick & Albert, 2014). According to Resnick and Albert (2014), this results in consumers no longer looking at advertisements and banners on online webpages. What happens is that a consumer is aware that a persuasion attempt will be made, and they use an avoidance strategy to
counter the attempt (Resnick & Albert, 2014).
A strategy to counter banner blindness is for companies to use influencers in order to reach their target audience (Fransen et al., 2015; Hervet et al., 2011; Knowles & Linn, 2004; Resnick & Albert, 2014). This results in an advertisement showing up as a banner on the upper or right-hand corner of a web page, the advertisement is done through direct contact with an influencer. The reason that this method counters banner blindness is because individuals have different persuasion knowledge about influencers as persuasive methods rather than banners online. What marketers are doing is hiding the persuasive nature of the
Page 14 of 58 message through the influencers and countering the avoidance strategy of the consumers (Van Rijmersdal et al., 2016).
Influencers as a Strategy
Originally, Katz and Lazarsfeld define opinion leaders as ‘[t]he individuals who were likely to influence other persons in their immediate environment’ (Watts & Dodds, 2007, p.442). Nowadays, the term opinion leaders has shifted to influencers, which is the term that will be used in this paper. Exposure to online reviews is shown to influence individuals more than exposure to the media and is explained through the “two-step flow’ model of communication by Katz and Lazarsfeld back in 1995 (Watts & Dodds, 2007). What this model shows is that there are opinion leaders who take content from the mass media and pass it on to their followers, often adapting the content of the message to include their own personal
interpretations. This way, the consumers get their information second hand through these influencers, as opposed to directly from the mass media, which makes platforms such as YouTube important in distributing new products or campaigns. With influencer marketing, a company strives to promote their products or services through an influencer, and through the two-step flow model they will reach their target audience, as opposed to traditional marketing through the mass media (Hall, 2016).
Paid Influencers, Unpaid Influencers, and Disclosures
Wei and Lu (2013) explain that using celebrity influencers can increase audience
attentiveness, add glamour to the brand, as well as increase memorability and credibility of the advertisement. The paper goes on to explain, however, that this is not only relevant for celebrities, but also for influencers that have a significant number of followers, and that as follower amount increases, credibility of the influencer increases (Wei & Lu, 2013). An example is a collaboration campaign by L’Oreal from 2016 where they created a five-member beauty squad using five high profile influencers (with a significant number of
Page 15 of 58 followers). This beauty squad, that posted reviews about Paris Fashion week, has helped L’Oreal to reach a new audience online (Digiday, 2017). This demonstrates that social media influencers embody a new form of independent endorsers that can influence audiences through their social media channels (Freberg et al., 2011).
Wei and Lu (2013) also explain that it does not matter whether or not the influencer is paid, in both cases influencer posts can distribute information to consumers. Kapitan and Silvera (2016), on the other hand, argue that there is not yet enough research done on the different effects of unpaid influencers when compared to paid influencers. They feel that a difference may not necessarily be present between paid and unpaid influencers, but that the difference lies between whether or not the influencer is credible, and that credibility of an influencer does not rely on them being paid or not (Kapitan & Silvera, 2016). Kapitan and Silvera (2016) as well as Wei and Lu (2013) explain this to mean that influencers that have a better link to the product they are promoting and seem more trustworthy, have a bigger effect as an influencer than an influencer that has little or no link to the product, regardless of whether or not the influencer is paid. They also explain that influencers increase brand
engagement when consumers feel that they are knowledgeable about the product that they are endorsing (Kapitan & Silvera, 2016). The one difference that is proposed by Kapitan and Silvera (2016) between paid and unpaid influencers, is that consumers feel paid influencers are less honest and believable, decreasing consumer brand engagement.
Aside from Wei and Lu (2013) and Kapitan and Silvera (2016), Taylor (2015) also expresses the importance of researching the effect of influencers on brand engagement of consumers. Consumers, particularly generation Z, are spending an increased amount of time online and are prone to be affected by influencers (Taylor, 2015). A driving factor in
distinguishing the difference in effect between paid and unpaid influencers is through the use of disclosures. A disclosure is used in order for the audience to clearly see whether or not the
Page 16 of 58 influencer is being paid by the company to promote their product or service, and it has
become the law in many countries to include disclosures (Van Reijmersdal et al., 2016). The development of this law is done because it is viewed as unethical when companies hide their persuasion attempts through influencers without the consumer being aware of this (Asbeek Brusse, Fransen, & Smit, 2015).
As a result of disclosures, persuasion knowledge in consumers can be activated, which can lead to resistance strategies (Van Reijmersdal et al., 2016). The study done by Van Reijmersdal et al. (2016) shows that on the one hand, disclosures can negatively influence brand engagement, because persuasion knowledge is activated, and consumers feel that they do not want to be persuaded and therefore counterargue the attempt. On the other hand, Kapitan and Silvera (2016) explain that even though a disclosure is used, and consumers are aware of the persuasion attempt, if the consumers view the influencer as credible and
believable, there is a positive effect. This is further supported by Van Reijmersdal et al. (2016) who explain that disclosures can also increase brand engagement. This is because even though the persuasion knowledge is activated, consumers view the disclosures as
informative of the relationship between the influencer and the brand, and it does therefore not affect the credibility of the influencer (Van Reijmersdal et al., 2016). In other words, the disclosure only means that the brand is paying the influencer, but it does not mean that what the influencer is saying about the brand is scripted (Hwang & Jeong, 2016). Hwang and Jeong (2016) further explain this as influencers telling their audience that the post is sponsored, but that the opinions posed are their own, honest, and unbiased.
As explained in the literature above, there are different opinions on the effects of paid versus unpaid influencers on the brand engagement of consumers. On the one hand
influencers that are paid are said to have a negative effect on brand engagement because they are being paid. On the other hand, whether or not an influencer is paid should not have an
Page 17 of 58 effect on the brand engagement of the consumers, because the message that they are
conveying is still their own. It is explained that influencers are used as a persuasion strategy to hide the persuasive nature of the message, which circumvents the activation of persuasion knowledge in the audience. However, new laws have forced influencers to include
disclosures in their sponsored posts and reviews, which re-activates persuasion knowledge, making the whole concept of using influencers as a strategy superfluous. This is where Van Reijmersdal et al. (2016) hypothesize and test that the use of disclosures, even though they activate persuasion knowledge, do not negatively impact the effect of paid influencers. However, as there is dissension in the literature, the following hypothesis has been formulated for this thesis:
Hypothesis 1: An unpaid influencer will have a larger positive effect on brand engagement than a paid influencer.
The basis for this hypothesis stems from the thought that paid influencers, especially with the use of disclosures, activate persuasion knowledge. With the activation of persuasion knowledge, resistance strategies will be used by the audience. The same resistance strategies that the use of influencers aimed to thwart in the first place. The use of disclosures was tested by Van Reijmersdal et al. (2016) through the use of blog posts, which is different than the testing in this thesis which is done through the use of videos. Through the use of disclosures in videos of influencers, the effects are expected to be different than when disclosures are used in written blog posts.
Sidedness as a Persuasion Strategy
Another strategy to counter persuasion tactics is to create two-sided messages, but where the use of influencers is to counter the avoidance strategy, two-sided messages is used as a strategy to decrease counterarguing. If a consumer is using the contesting resistance strategy where they are counterarguing the message, then an effective resistance neutralizing, or
Page 18 of 58 omega strategy, is to create two-sided messages (Fransen et al., 2015). What this means is that in the persuasive message, both sides of the argument are presented in order to decrease counterarguing done by the consumer and to increase the credibility of the message (Uribe et al., 2016). As explained by Knowles and Linn (2004), a one-sided message only contains arguments that support the message, whereas a two-sided message gives both sides of the argument, so arguments that support the message as well as arguments that counter the claims. The reason for this is that two-sided messages already give the consumers the
counterarguments that could be thought of to counter the persuasion tactic (Knowles & Linn, 2004). What happens as a result is that consumers spend less cognitive effort to counterargue the message if they have already been given the arguments (Yao et al., 2018).According to Knowles and Linn (2004) and Pierro et al. (2013), two-sided messages are more effective than one-sided messages in a situation where consumers would think of counter arguments themselves, such as higher-educated individuals or those individuals with a bias against the original message.
With regards to influencers using two-sided messages, Majmundar, Cornelis, and Heuvinck (2017) explain that the inclusion of counterarguments in the message can increase credibility, and therefore have a positive effect on attitudes and intentions of consumers. This is further supported in the paper by Eisend (2006) who explains that two-sided messages increase brand engagement, as they increase source credibility, as well as Lu et al. (2014) who explain that one-sided negative reviews can have negative effects on the consumer brand engagement of a company. What is interesting is that the placement of negative messages has a larger effect if they are at the end of the message as opposed to the beginning (Eisend, 2006). On the other hand, including negative arguments in the message could also cause for confusion, resulting in a decrease of credibility of the influencer, having the opposite effect (Majmundar et al., 2017). This is also supported by Kuster and Eisend (2016) who explain
Page 19 of 58 that putting both negative and positive arguments in one message can have a negative effect on brand engagement.
From the above literature it can be concluded that message sidedness is used as a tool to counter the cognitive resistance strategy. Where literature (Eisend, 2006; Kuster & Eisend, 2016; Majmundar et al., 2017) has proposed that two-sided messages can have a negative effect, literature on two-sided messages in the field of influencers (Berryman & Kavka, 2017; Fransen et al., 2015; Majmundar et al., 2017; Uribe et al., 2016) has shown that two-sided messages by influencers have a more positive effect on brand engagement than a one-sided message. However, whether or not this effect on brand engagement is different for unpaid influencers compared to paid influencers is underdeveloped. In line with the literature, the following hypothesis, stipulating that two-sided messages have a larger effect, is proposed:
Hypothesis 2: An unpaid influencer will have a more positive effect on brand engagement than a paid influencer but this effect will be more pronounced for two-sided messages than one-sided messages.
Where on the one hand literature explains that the use of two-sided messages can lead to confusion and result in a negative effect on brand engagement, this research was not specific to influencers. When looking at the literature whose primary focus is the effect of influencers, it is shown that two-sided messages have a larger effect on brand engagement than one-sided messages. Through the use of presenting the audience with both the positive and negative aspects of, in this case a cosmetics product, it is thought that the two-sided message will yield a larger positive change in brand engagement than the one-sided message. The reason that it is thought that this will overall have a larger effect with unpaid influencers is because, as explained above, it is expected that unpaid influencers will have a larger positive effect on brand engagement than paid influencers. This means that it is hypothesized
Page 20 of 58 that unpaid influencers using two-sided messages will have the largest positive influence on consumer brand engagement. Paid influencers using one-sided messages is the group that is, consequently, expected to yield the smallest positive effect on consumer brand engagement.
Credibility
One of the leading models in source credibility theory is the source-credibility model proposed by Ohanian (1990). This model stipulates that the concept of source credibility is built up of three main cornerstones: expertise, trustworthiness, and attractiveness (Ohanian, 1990). It was a model that was modified from an earlier model proposed by Hovland and his associates in 1953 (Ohanian, 1990) but only included the two factors of expertise and
trustworthiness, excluding attractiveness. The factor of expertise is explained as how
knowledgeable an influencer shows to be about the subject that they are endorsing (La Ferle & Choi, 2005). La Ferle and Choi (2005) go on to explain trustworthiness as whether or not an audience believes that the message being send by the influencer is unbiased and sincere. Attractiveness, which is simply whether or not the audience find the influencer attractive, was added to the older model because research showed that the attractiveness of the influencer could have a positive or negative impact on the opinions and consequently the behaviors of the audience (Lafferty et al., 2002).
Wei and Lu (2013) explain that the use of influencers by organizations can be seen as a strategic tool to increase credibility of the message, and ultimately the company. This is further supported by La Ferle and Choi (2005) who explain that highly credible sources have the capacity to increase brand engagement more than less credible sources. Kapitan and Silvera (2016) agree and build on this, explaining that paid influencers are seen as less credible because they are viewed as less honest and believable, which in turn, decreases brand engagement. This is then again opposed by what Van Reijmersdal et al. (2016) state in their paper, which is that paid influencers, often with the use of disclosures, are not
Page 21 of 58 necessarily seen as less credible. This is because the disclosure is only seen as informative, it shows the audience the nature of the relationship between the influencer and the brand, but does not mean that the message being sent is not honest (Van Reijmersdal et al., 2016). Aside from this, Ohanian (1990) also explains that the credibility of the influencer no longer matters if the audience is already positively biased towards the message itself. This means that if the audience already likes the product from the review, the credibility of the influencer will no longer play a role on the brand engagement of the consumer. These two conflicting opinions found in the literature, whether or not credibility plays a role and whether unpaid influencers are seen as more credible, has led to the third hypothesis of this thesis, namely:
Hypothesis 3: Unpaid influencers will be viewed as being more credible, which results in a larger increase in brand engagement as opposed to paid influencers who will be viewed as less credible and therefore have a smaller effect on brand engagement.
The focus of hypothesis three is on the different effect that credibility has on the relationship between influencer type and brand engagement. In literature, it has been argued that two-sided messages increase credibility because both sides of the argument are given. However, there is a gap as to what effect credibility has dependent on whether or not the influencer is paid or unpaid. On the one hand literature explains that because an influencer is paid, they are seen as less honest and therefore less credible. On the other hand, it is argued that even though the influencer is paid, this does not have an effect on credibility because being paid does not influence the honesty of the message. What is believed to happen in this thesis, which is where the third hypothesis comes from, is that unpaid influencers will be seen as more credible, and therefore have a larger effect on brand engagement because they are seen as unbiased. Aside from this, hypothesis three is expected because of the motivational factors for an unpaid influencer to produce a review, which is because they are passionate and knowledgeable about the product they are reviewing, without being paid for it. This is what
Page 22 of 58 has led to the belief that they will be seen as sincerer, and more credible, resulting in a larger effect on brand engagement.
Below is Figure 1 of the paper, that shows the conceptual framework together with which hypothesis corresponds to which relationship.
Figure 1: Conceptual Framework
Methodology
Setting of the Study
The setting of the study comprised of showing participants one of four videos and then asking them to fill out a questionnaire. This was done through an experimental setting with a 2x2 between-subjects factorial design: Influencer (2 levels: paid vs unpaid) * Sidedness (2 levels: One-sided messages vs two-sided messages). For this experiment, the independent variables to be manipulated were influencer type and message sidedness, and the dependent variable tested for was brand engagement. The participants were found through social media sites
Page 23 of 58 Facebook and Reddit in order to collect the data. The videos were created to illustrate a YouTube beauty influencer giving a product review of the product Naked Skin by the cosmetics company Urban Decay. A different video was made per condition, as further explained below.
Table 1. Factorial Design
Influencer type
Paid Unpaid
Message Sidedness
One-Sided Message Condition 1 Condition 2 Two-Sided Message Condition 3 Condition 4
The Video Material
To create the materials used during the experiment, so the videos that were to be shown to the participants, new material was recorded. This was done in order to keep the internal validity between the groups equal, and rule out any factors that could have occurred if four different videos were taken from the internet. It also had to do with the concept that videos self-created are personal property, whereas videos taken from the internet would require permission due to copyright infringement. An influencer that makes YouTube content herself was asked to create a product review for this experiment. Only one video was made that corresponded to the video used in condition three, because it included both a positive as well as a negative review, and it was clearly stated that it was sponsored. This video was then edited and used to create videos for the other three conditions. This was done to ensure that the content seen in the video, as well as intonation, lighting, and non-verbal communication was the same across all conditions.
Page 24 of 58 The product that was chosen for this experiment was a skin concealer called Naked Skin created by the company Urban Decay. This specific product was chosen because a concealer is a basic layer for make-up. The video for condition one showed the influencer giving a positive review only, but she mentioned in the beginning that the video was sponsored by Urban Decay and a disclosure at the bottom of the video was used to make it clear to participants that they were watching a sponsored video. The video for condition two included only the positive review, and there was no mention in the video of the video being sponsored and there was no disclosure present, as these were both edited out of the video. The video for condition three contained both positive and negative aspects of the product in the review, as well as the influencer clearly stating that the video was sponsored with the disclosure clearly visible below the video. The video for condition four contained both the positive and negative reviews, but the influencer mentioning that the video was sponsored was edited out and no disclosure was present. Screenshots from the experiment are presented below (Figure 2) to clearly illustrate the disclosure. The videos were edited using the
programme HitFilm which the influencer used to also edit her own videos for her YouTube channel.
Page 25 of 58 Figure 2: Showing the difference between the video’s that include the disclosure (Right)
Manipulation Checks
Manipulation checks for this experiment were conducted to ensure that the participants perceived both experimental factors influencer type and message sidedness as they were manipulated. To see whether or not the participants perceived influencers as either paid or unpaid, two questions were asked, with a 5-point answer scale (1) definitely true to (5) definitely false. The two questions that were asked were: “The influencer and the brand Urban Decay are working together to produce this video” and “Urban Decay paid for the influencer to make this video”. This scale had been adapted from Van Reijmersdal et al. (2016) who measured persuasion knowledge through questions such as “the blog post is advertising”. A second paper that was used to create the scale for this thesis was from the paper by Evans, Phua, Lim, and Jun (2017) that asked participants to “indicate the extent in which you thought the Instagram post was advertising” on a 7-point Likert scale (1) strongly disagree to (7) strongly agree.
In order to test the reliability of this scale, Cronbach’s alpha was run and found to be .79 (M=7.53, SD=1.09), which means it was found to be reliable. From this, the two
Page 26 of 58 questions were computed into a new variable for influencer type and used for the
manipulation check, which was done using chi-square. A chi-square manipulation check showed that participants understood that the video they were watching was indeed paid for by the company. From the results of the test chi square = 95.39, p= .00 which means that the manipulation check was a success. The results of the test answers can be seen in Table 2 below.
Table 2: Which video they were watching compared to whether or not they understood that the video was sponsored or not.
Video Watched Urban decay paid for
the influencer to make this video
The influencer and the brand Urban Decay are working together to produce this video
Unsponsored Definitely true 7 10
Probably true 44 52
Neither true nor false 25 28
Probably false 59 44
Definitely false 10 12
Sponsored Definitely true 55 52
Probably true 56 46
Neither true nor false 15 16
Probably false 9 19
Definitely false 2 4
To create the variable of message sidedness, only one question from the experiment was used, namely: “What elements were present in the video”. There was a 4-point answer scale; (1) a positive review only, (2) a negative review only, (3) a positive and negative review, and (4) none of the above. This scale was extrapolated from earlier research
Page 27 of 58 (Cornelis, Cauberghe, & De Pelsmacker, 2014; Eisend, 2006) that explained that for a
message to be one-sided it can only have arguments that support one side of the message, whereas two-sided messages also have to present the counter arguments. From Table 3 below it can be seen that with a few exceptions all participants understood whether they were watching a one-sided or a two-sided review.
Table 3. Which video they watched compared with what they thought they were watching (Sidedness)
Which video they watched compared with what they thought they were watching What elements respondents thought were present in the video
Positive Review only Negative Review only A positive and negative review None of the above Total Which video they watched Unsponsored Positive 65 0 9 74 Unsponsored two-sided 10 1 60 71 Sponsored Positive 62 0 5 67 Sponsored two-sided 7 2 59 68 Total 144 3 133 3 283
To double check these results for the manipulation check, and to ensure that participants understood whether the video they were watching was indeed only one or two-sided, a chi-square manipulation test was run. From the results chi square = 178.49, p = .00 it can be seen that this manipulation check was also a success.
Page 28 of 58 The experiment was created (the video as well as the questionnaire) through the programme Qualtrics, which also randomly assigned participants to one of the four conditions. The experiment was distributed using the social media platforms of Facebook and Reddit, as well as personal distribution through Facebook Messenger and WhatsApp Messenger. The target sample was female participants, and to ensure that this was the case the sample was later filtered by gender as in the experiment itself the option was given of (1) male, (2) female, and (3) other. The reason that the target sample was female is because the experiment was about brand engagement of a cosmetics company. A total of 476 individuals started the experiment, out of which 333 completed the experiment. After filtering the participants on gender, 50 individuals were removed because they had either filled in their gender as (1) male, or (3) other. This means that the total sample size used in this experiment was 283 participants.
As seen in Table 4 Distribution of the four conditions, there was an equal distribution of the participants per condition. On average, around a quarter of participants was randomly placed into each of the four conditions.
Table 4: Distribution table of the four conditions
Frequency Percent %
Unsponsored Positive 75 26.5
Unsponsored Two-sided 71 25.1
Sponsored Positive 68 24.0
Sponsored Two-sided 69 24.4
Furthermore, the sample included participants from 37 countries, the largest with 166 participants (58.7%) being Dutch, followed by German with 16 participants (5.7%) and the United States with 11 participants (3.9%). Participants were also asked what their highest obtained level of education is, of which the largest group was a University degree with 198 (70%) of the participants. From the remaining participants, 33 (11.7%) had obtained a high school degree, 39 (13.8%) had obtained a HBO or University of applied science degree, and
Page 29 of 58 13 (4.6%) participants filled out the option other. The age range of the participants was from 17 years old to 70 years old, with the largest sample being between 17 and 32 (cumulative frequency at 32 being 91.2%), and a total of 26 participants had answered the question “do you know the influencer in the video” with (1) yes.
Measurement instrument I – Engagement
Aside from the creation of the variables for influencer type and message sidedness, the
variable Brand Engagement was also computed in order to test the hypotheses. Research done by Sprott et al. (2009) resulted in the creation and testing of the brand engagement in self-concept, or BESC scale. This scale incorporated that individuals link aspects of the brand, or the brand themselves, to their personal identity and a relationship was created. Sprott et al. (2009) explain that brand engagement comes from the linking of people’s self-concept to the brand. What this means is that consumers are using brands, or what they believe they stand for, and incorporating them in their identity, or self-concept. After Sprott et al. (2009) tested and finalised their scale, it resulted in an 8-item scale, measured on a 7-point scale from (1) strongly disagree to (7) strongly agree. An example of one of the items is “I have a special bond with the brand that I like” (Sprott et al., 2009). The full list of scale items can be found in Appendix 3.
Hollebeek et al. (2014) developed a scale with which they could measure Consumer’s Brand Engagement (CBE). This scale was comprised of three different dimensions that corresponded with the three levels in which they defined engagement, namely: cognitive, emotional (affective), and behavioural (activation). They based their scale on a literature review, where they explained that CBE was expected to be positively related to better organizational performance outcomes. These outcomes could be; increased sales, brand referrals, but also include the engagement that consumers have with the process of product development and co-creation activities. Hollebeek et al. (2014) then explained how and why
Page 30 of 58 their model was different from a model proposed by Calder et al. (2009) because that model would only reflect a consumer’s Online Engagement (OE). They explain that their model differs because the model proposed by Hollebeek et al. (2014) included consumers’ brand engagement and was not limited to consumers’ online engagement only. They also argued that their model incorporated the notion of a mutual relationship between consumers and brands, as opposed to them being independent factors, which previous research had failed to do. It is for these reasons that the model proposed by Hollebeek et al. (2014) was adapted and used in this research paper. An example of a scale item from Hollebeek et al. (2014) is “using a [brand] gets me to think about [brand]”. The full 10-item scale list can be found in
Appendix 4.
For this research paper, the questions used to measure Brand Engagement was created based on the three dimensions proposed by Hollebeek et al. (2014): cognitive, affective, and activation. Aside from this, a 7-point scale as used by Sprott et al. (2009) was used, (1) strongly disagree to (7) strongly agree to measure an individual’s brand engagement. From these two papers, own statements were created that incorporated elements from both scales, an example being “viewing this video has made me think about the brand Urban Decay”. A list of all six statements can be found in Appendix 2. This 6-scale item proved to be reliable after running Chronbach’s alpha with the result of .85 (M=4.45, SD=1.20) and these items were combined to create the variable of Brand Engagement.
Measurement Instrument II – Credibility
Through literature research and experimental testing, Lock and Seele (2017) developed and tested a scale with which to measure credibility. The focus of their paper was to examine the perceived credibility of CSR reports specifically. They did, however, build on earlier research that examined credibility in a broader range, such as Jackob (2008), Hellweg and Andersen (1989) and Sias and Wyers (2001). In essence, credibility has been conceptualized in three
Page 31 of 58 main ways, namely source, recipient, and interaction models (Lock & Seele, 2017). From these conceptualizations, Lock and Seele (2017) developed a 19-point scale that was divided into four subsections, namely; trust, sincerity, appropriateness, and understandability. The results were measured on a 5-point scale, (1) being not at all adequate, and (5) being very adequate. An example of one of the items of Truth is “I am confident that the statements are true” and an example of one of the items of Sincerity being “I think the text is not
misleading”. A full list of scale items can be found in Appendix 5.
Based on the scale developed by Lock and Seele (2017), two questions were formed for this thesis to measure credibility, namely “The influencer is trustworthy” and “The influencer is well informed about the Cosmetics Industry”. The reason that only two questions were used to test the credibility of the influencer is because the first question incorporated the scale items used to test the trustworthiness of the influencer. The second question posed in the survey encompassed the scale items used to test expertise, as explained by Lock and Seele (2017). Attractiveness of the influencer was not included in the research because the participants knowing who the influencer is could have caused for an uneven bias in answering a question about perceived attractiveness of the influencer. Both of the
questions posed were measured on a 7-point scale from (1) strongly agree to (7) strongly disagree to stay in line with the other answer options in the experiment, which were also on a seven-point scale. These two statements proved to be reliable after running Chronbach’s alpha with the result of .71 (M=3.86, SD=1.14) and were therefore computed into the variable Credibility.
Experiment
The experiment was, as mentioned before, run through the Qualtrics Survey programme provided by the University of Amsterdam. The participants were found online through social media platforms Reddit and YouTube and the experiment was live online for one week. This
Page 32 of 58 was because after that point in time thrice as many respondents as necessary had been
collected. The experiment was broken down into different topics to create a clearer overview of the separate variables, and a copy of the experiment can be found in Appendix 6, which includes screenshots from each of the four videos. The experiment started with a thank you for participating and the terms and conditions of the experiment as proposed by the
University of Amsterdam. Only if the participants agreed to these terms and conditions could they continue with the experiment, the terms including statements such as “your anonymity will be safeguarded” as well as contact information to the Ethics Committee. Then
demographics were asked such as age, gender, education, and nationality. Afterwards, a baseline was created for brand engagement, with statements such as “I can identify with important brands in my life” (1) strongly agree to (7) strongly disagree, as well as questions based on their use of cosmetics, how often they purchase and use it. Then participants were randomly selected by Qualtrics into one of the four conditions, and based on their condition shown one of the four videos. Afterwards they were asked questions about whether or not they knew the brand and the influencer before watching the video. All four groups were asked to fill in the same questionnaire after watching the YouTube review. Consequently, for the manipulation check, participants were asked to select which elements were present in the video, (1) a positive review only, (2) a negative review only, (3) a positive and negative review, and (4) none of the above. Next, six statements were made about brand engagement, and participants had to answer based on the 7-point scale (1) strongly agree to (7) strongly disagree.
Results
The first hypothesis proposed was: “An unpaid influencer will have a larger positive effect on brand engagement than a paid influencer”. In order to test for this hypothesis, an independent sample t-test was conducted with influencer type (paid versus unpaid) as the independent
Page 33 of 58 variable and brand engagement as the dependent variable. From the results it was conclusive that there is no significant difference in the unpaid influencer condition (M=4.64, SD=1.11), and the paid influencer condition (M=4.56, SD=1.14), t(281) = -.52, p= .60. This means that the first hypothesis was rejected.
The second hypothesis proposed was: “An unpaid influencer will have a more positive effect on brand engagement than a paid influencer but this effect will be more pronounced for two-sided messages than one-sided messages”. In order to test for this hypothesis, a two-way analysis of variance (univariate ANOVA) was run with influencer type (paid versus unpaid) and message sidedness (positive versus two-sided) as the
independent variables, and brand engagement as the dependent variable. The results showed that there was no significant effect of influencer type on brand engagement, F(1, 283) = .20, p = .66, 𝜂2< .01, as well as no significant effect of message sidedness on brand engagement, F(1,283) = 2.78, p = .10, 𝜂2= .01. Furthermore, no significant interaction effect was found between influencer type and message sidedness, F(1, 283) = .72, p = .40, 𝜂2< .01. Therefore, hypothesis two was also rejected, showing that unpaid influencers do not have a larger effect on brand engagement than paid influencers, and this effect is not more pronounced for two-sided messages than one-two-sided messages.
The third hypothesis proposed was: “Unpaid influencers will be viewed as being more credible which results in larger increase in brand engagement as opposed to paid influencers who will be viewed as less credible and therefore have a smaller positive effect on brand engagement”. In order to test for this hypothesis, multiple regression analysis was used with engagement as the dependent variable, and credibility and influencer type as the independent variables. The results showed that 12.3% of the variance R2= .12, F(2,283)= 19.72, p= .00 was explained by credibility and influencer type, however, influencer type alone β= .01, p= .96 was not significant whereas credibility β= .37, p= .00 was significant on its own. This
Page 34 of 58 means that hypothesis three was also rejected, and that credibility does not mediate the
relationship between influencer type and brand engagement. What is interesting to see, however, is that credibility on its own does have an effect on brand engagement, which leaves room for future research.
As all three hypotheses were rejected, an extra independent sample t-test was conducted with message sidedness as the independent variable and credibility as the
dependent variable. From these results with the one-sided condition (M= 3.70, SD= 1.42) and the sided condition (M= 4.03, SD= 1.11), t(283) = 2.53, p= 0.01 it can be seen that two-sided messages are seen to be more credible than one two-sided messages.
Discussion and Conclusion
The results of this study show that all three of the proposed hypotheses proposed are rejected, but this does not mean that a relationship between influencer type, message sidedness, and brand engagement is not possible, as well as that credibility does not play a mediating role. The results simply show that in this specific setting the effects are not found, but there could be different reasons for this. Below are explanations of the findings compared to literature, what they signify, and possibilities for future research.
The results of this study show that there is no significant different effect of influencer type (paid or unpaid) on brand engagement. This means that the first hypothesis of this thesis, “An unpaid influencer will have a larger positive effect on brand engagement than a paid influencer” has been rejected. In the beginning of this paper it was explained that little research had been done specifically on the difference between paid and unpaid influencers (Kapitan & Silvera, 2016), also referred to as earned or paid influencer marketing by Hall (2016), economically motivated or socially motivated by Zajc (2015), and monetized or not monetized by Thomson (2017). Despite the fact that these papers agreed that there was a difference between these two groups of influencers, the different effects of these two groups
Page 35 of 58 on brand engagement was the gap this paper aimed to fill. Taken from the literature of paid influencers, it has been shown that influencers increase brand engagement (Ang, 2011; Hall, 2016; Lu et al., 2014; Thomson 2017; Yoshida et al., 2018; Zajc, 2015), and the results from this study show that it does not matter if an influencer is paid or not, they will have the same effect on brand engagement. The reason that an increase in brand engagement is important for organizations, is because it can lead to better organizational performance outcomes, as well as customer retention over time (Ferreira & Coelho, 2015; Hollebeek et al., 2014; Kumar et al., 2010; Van Doorn et al., 2010; Vivek et al., 2012).
One main factor, however, that could have explained a difference in the effect of influencer type on brand engagement is the use of disclosures. According to literature, disclosures activate persuasion knowledge (Asbeek Brusse et al., 2015; Kapitan & Silvera, 2016) which can lead to resistance strategies (Van Reijmersdal et al., 2016), which have a negative effect on brand engagement. As explained in the methodology section about the video creation, only the paid influencer videos had a clear disclosure that was both
specifically stated by the influencer in the video, as well as present underneath the video the whole time. According to the aforementioned literature, the inclusion of these disclosures could have been a reason that there was a significant effect. However, as no significant effect was found in this thesis, it is more supportive of the literature by Van Reijmersdal et al. (2016) who explain that even though disclosures activate persuasion knowledge, as long as the source seems credible, disclosures have no effect. This could be a reason as to why there was no significant difference found in the effect of unpaid influencers versus paid influencers on consumer brand engagement. Future research on the effects of influencer type (paid versus unpaid) could consider the use of explicit disclosures for both types of influencer.
Secondly, the results of this study also show that there is no significant interaction effect between influencer type (paid or unpaid) and message sidedness (positive or
two-Page 36 of 58 sided). This means that the second hypothesis of this thesis “An unpaid influencer will have a more positive effect on brand engagement than a paid influencer but this effect will be more pronounced for two-sided messages than one-sided messages” has been rejected. This is contradictory to the findings in literature, where it has been stated that two-sided messages are more effective than one-sided messages (Pierro et el., 2013; Yao et al., 2018).
Additionally, Majmundar et al. (2017) explain that specifically in the case of influencers, two-sided messages are more effective because they increase the credibility of the influencer as they are seen as more transparent and sincere for giving both sides of the argument.
Consequently, according to Eisend (2006) this increase in credibility of the influencer due to two-sided messages should lead to an increase in brand engagement. The results of the participants of this thesis, however, showed no interaction effect of message sidedness on the relationship between influencer type and brand engagement. This means that if the influencer is paid, it does not matter whether the message is one or two-sided, it will still have the same effect on brand engagement. This result was found to be the same for the case of the unpaid influencer review. The reason for this could be that the negative effects of a two-sided message are countered by the credibility of the source, as explained by Kapitan and Silvera (2016), and therefore no difference is found between a positive only, or positive and negative review. On the other hand, there is also literature (Majmundar et al., 2017; Uribe et al., 2016) that explains that it is the sidedness of the message that has an effect on the credibility of the influencer. Uribe et al. (2016) explain that a two-sided message increases credibility because it is offering the audience both sides of the argument, as opposed to only one. The reason that this is the case with two-sided messages is already explained in the persuasion knowledge model, where both sides of the argument are given to decrease counterarguing which in turn leads to an easier acceptance of the message. The effect of two-sided messages on credibility is further backed by Eisend (2006) who explains that increased source credibility due to the
