January 29th
2016
University of Amsterdam
Graduate School of Communication
Master’s programme Communication Science Specialization in Corporate Communication
Author: Aletta A.M. Hoedjes 10101373 Supervisor: Christine C. Liebrecht Master’s Thesis
Let them entertain, inform
and influence you
Abstract
The term “influencer marketing” has been buzzing in the business sphere, yet how this does work? The current research investigates influencer marketing on YouTube, combining theories of UGC, eWOM, influencing, credibility and community feeling. Two separate content analyses were conducted, one analyzing reviews given by a creator, the second analyzing the comments to these reviews given by the audience. The results show that a positive review leads towards a positive product attitude amongst the audience, and the purchase intention will be influenced by the review when purchase information is given. The creator on YouTube is perceived as credible, which makes creators interesting influencers for practitioners. Influencer marketing is here to stay and provides organizations with a lot of interesting possibilities to communicate their message.
Let them entertain, inform and influence you
The term ‘influencer marketing’ has been buzzing in the business sphere and companies are embracing their influencers (Harder, 2015; Morin, 2015; Smithuijsen, 2015). Influencers can be seen as ordinary people that are popular online, who gained almost an online celebrity status and who are seen as credible sources amongst their followers. These online trendsetters became popular by producing content about a certain area like beauty or fashion, amongst other, and can therefore reach a specific target group through their online communication channels (Ranga & Sharma, 2014). In the digital world we live in now it is al about reaching the brand’s target audience and receiving credible electronic word-of-mouth (eWOM). Influencers can help brands to achieve this (Bambauer-Sachse & Mangold, 2013; Bickart & Schindler, 2001; Misopoulos, Mitic, Kapoulas, & Karapiperis, 2014).
In their article, Ranga and Sharma (2014) investigated influencer marketing and concluded that there are different types of influencers. It has also been investigated how this kind of
influencing works differently from traditional influencing strategies (Eccleston & Griseri, 2008; Lyons & Henderson, 2005) and how influencing works in this digital era (Bodendorf & Kaiser, 2010). Influencer marketing is based on user-generated content (UGC) (Fournier & Avery, 2011; Misopoulos et al., 2014). UGC is content that is produced by users and differs from brand-generated content (BGC) (Van Noort & Willemsen, 2012). UGC about a brand is seen as more credible (Bambauer-Sachse & Mangold, 2013; Bickart & Schindler, 2001). However, it is
not only the content, but also the creator that can be perceived as credible (Lafferty, Goldsmith, & Newell, 2002). Due to the amount of followers influencers have, they can be perceived as a credible or expert source in a certain segment and therefore become powerful influencers (Ranga & Sharma, 2014).
Influencers can be active on almost every medium where UGC can be published
(Constantinides & Fountain, 2008; Kaplan & Haenlein, 2010). This content can be published on social media like Facebook, Instagram and YouTube. YouTube is a successful content
community, which is attractive for content creators and is one of the largest UGC communities consisting of videos (Cha, Kwak, Rodriguez, Ahn, & Moon, 2007; Susarla, Oh, & Tan, 2012). YouTube is, other than for instance Facebook, also a relatively rich medium (Daft & Lengel, 1986) as it combines sound and moving images in videos with the possibility of direct communication between a creator and the audience in the form of comments. It is a medium where the creator posts their own content, on their own channel, and it has less to do with reposting content that is not necessarily created by the creator, as is for instance the case on Facebook. Consumers visit YouTube for entertainment and it is a place for self-representation (Cheong & Morrison, 2008; Kruitbosch & Nack, 2008). Over the years the mediascape changes (Macnamara, 2010), and YouTube as a medium changed as well. Nowadays billions of people watch YouTube everyday (YouTube, n.d.) and it became clear that people can even earn a royal salary of this medium (Mandle, 2015).
With these changes in the mediascape and the new opportunities YouTube offers, some gaps in the literature appear. The growth of the medium and its business opportunities show a big change, and have not yet been accounted for in previous research. Previous research also only focuses on one asset or group on YouTube, by for instance only focussing on the content of the videos created by a creator, or how the audience use the medium (Cheong & Morrison, 2008; Kruitbosch & Nack, 2008). Research that combines both content of a creator and the audience
responses towards this has not been done before, and this possibility of communication between these two groups is a great part of why the medium is so rich (Daft & Lengel, 1986; Kaplan & Haenlein, 2010). Therefore the current research will investigate a combination of these different assets. Furthermore there is existing literature on subjects like UGC (Cheong & Morrison, 2008; Kaplan & Haenlein, 2010), eWOM (Cheung & Lee, 2011; Van Noort & Willemsen, 2012) and influencers (Shoham & Ruvio, 2008; Ranga & Sharma, 2014), but no research has yet
combined these theories together in one study. These theories also have not been investigated in the light of YouTube. This investigation is necessary as YouTube is a very interesting research object due to its interactive nature where the power of moving images is combined with direct two-way communication feedback options. As the proposed subjects of this research are closely related further investigation is needed.
In order to fill the gaps in the literature, influencer marketing on YouTube will be investigated in the current research. This will be investigated by conducting two content analyses. The YouTube videos provide UGC for the first part of the analysis. In the first content analysis the reviews and opinions about products of brands, as voiced by an influencer will be investigated. The influencer will be referred to as “the creator” in the following part of this research. In this research it will be argued how the creator can influence consumers and in order to investigate this a second content analysis will be conducted. This second analysis will analyze the
comments that the audience left in response to the YouTube videos, investigated in the first analysis. The aim of this research is to investigate whether the opinion of a creator about a product matters to consumers, and how these consumers voice this in the comments. In the following part of this research the consumers will be referred to as “the audience”, or “audience members”. The research question proposed below will be investigated in this research.
RQ: What kind of relations can be seen between the product reviews of the creator and what is mentioned in the comments by the audience?
By investigating this research question, gaps in the academic field can be filled. The combination of these theories is a valuable addition to the existing academic knowledge on these topics, and this research will provide some more up to date information on how YouTube works nowadays. This research is also important for corporate companies. Online business articles are buzzing about influencer marketing (Harder, 2015; Morin, 2015; Smithuijsen, 2015), but proof that this kind of marketing actually works has not yet been found. Therefore this article will be a source of information on influencer marketing. With these findings
practitioners can understand how influencing on YouTube works and develop strategies on how they might invest in these influencers in the future. Besides influencing, this research will also shed light on what is discussed in the videos. This makes the research a source of information on what product attributes are important and mentioned by the creator, and which product attributes the audience mentions. With this information corporate companies have the opportunity to improve their products in order to better match their target group’s needs.
This article will continue with discussing the relevant theories on the subjects of UGC, eWOM, YouTube, influencers marketing and credibility theories. In this argumentation a clear
distinction between the creator and the audience will be made. The process of the content analyses will be discussed, followed by the results and the discussion of these results.
Literature review
A closer look at UGC and eWOM
UGC can be labelled as such when it is created by users (Van Noort & Willemsen, 2012). These
users are non-brand related people, that created content outside of professional routines and practices, and the content has been published online (Cheong & Morrison, 2008; Daugherty,
Eastin, & Bright, 2008; Kaplan & Haenlein, 2010). This content is publically available on a website, or can be posted on a social media networking site where it will be accessible to a selected group of people (Kaplan & Haenlein, 2010). Also the created content should show
some creative effort. Content created by practitioners is sometimes referred to as producer-generated content (Cheong & Morrison, 2008), but will be referred to as brand-producer-generated content (BGC) in the current study (Van Noort & Willemsen, 2012). The big difference
between BGC and UGC is the lack of control that organizations have when it comes to UGC, as the brand is not the main concern of users. Also online people are not bound by standards of objectivity like this is for instance the case with newspapers. Online creators can therefore express their opinions freely about products and brands.
Closely related to UGC is electronic of-mouth (eWOM). Looking at traditional word-of-mouth communication, this is a form of interpersonal non-commercial communication amongst acquaintances (Cheung & Lee, 2011). Due to electronic evolvements and the rise of Web 2.0 this consumer-to-consumer communication exchange is now taking place online (Cheung & Lee, 2011; Hornik, Satchi, Cesareo, & Pastore, 2015; Van Noort & Willemsen,
2012). EWOM refers to positive or negative online statements about a product or an
organization made by potential, actual or former customers (Hennig-Thurau, Gwinner, Walsh, & Gremler, 2004; Willemsen, Neijens, & Bronner, 2013). Different from the traditional form of WOM, eWOM communication is a lot faster and has a bigger reach. This form of
communication is more persistent and more accessible due to the fact that it is publically
available online. The downside however, is that it is harder for people to judge the credibility of the source that is communicating eWOM online (Cheung & Lee, 2011).
UGC and eWOM are closely related as they are both communicated by users and they are both publically available online. The difference however is that UGC is about content created by users, and eWOM is about content being conveyed by users. Applying these theories to YouTube, the videos can be seen as UGC. When these videos contain reviews, the reviews are a form of eWOM that can be conveyed through the content that the creator created. Responses by
the audience on these created contents are a form of consumer-to-consumer communication exchange and can therefore be seen as eWOM.
The influence of UGC and eWOM
After publishing UGC or eWOM online it is publically available for the audience and can therefore be of influence on them. UGC about products is more influential than BGC, and UGC will be seen as more credible by the audience (Bambauer-Sachse & Mangold, 2013; Bickart &
Schindler, 2001). Online reviews, either given through UGC or eWOM, will give chatter about a brand, product or service, and that chatter matters (Dhar & Chang, 2009; Duan, Gu, &
Whinston, 2008a; 2008b). When looking at the amount of reviews it was found that multiple negative reviews would increase the consumers’ negative product evaluation (Kim & Gupta, 2012). This also works in the other direction as multiple positive reviews will be seen with positive product evaluations. Positive online reviews drive sales, whereas negative online reviews reflect on sales (Ren, 2015). An important factor of eWOM’s influence is the reach it can have. In a research focussing on Twitter followers it was found that eWOM created by a person with a lot of followers had a stronger eWOM effect, than eWOM communicated by a person with less followers (Rui, Lui, & Whinston, 2013). The medium of choice in the current research differs from Twitter, as this current research focuses on YouTube. The results of Rui et al. (2013) are however of use as the amount of followers indicates the reach a message can have. On YouTube people can also follow specific creators, which will also be of influence on the reach a video has. Considering all this, an overall conclusion on the influence of UGC and eWOM can be drawn from the work of Hornik et al. (2015); bad news travels faster than positive news, and has a bigger impact on consumers.
As can be seen from the previous discussed studies, there are several aspects that play a role when it comes to UGC and eWOM. Duan et al. (2008a) recognized eWOM as one of the most influential resources for the transmission of information. It is also an alternative resource for
customers, as the main purpose is not based on persuading customers. Several sources point to the possibility that positive reviews will influence the consumer in a positive way, whereas negative reviews will influence the consumer in a negative way (Dhar & Chang, 2009; Duan et al., 2008a; 2008b; Kim & Gupta, 2012; Ren, 2015). Though none of these studies have been applied to YouTube, there is no reason to believe these results will not be applicable in the current study as well. Therefore it is expected that a positive or negative review will have the same kind of results on YouTube, as has been found in previous research. To investigate this the first hypothesis is proposed (H1). A further look will be given towards negative reviews, as bad news seems to travel faster, and is stated to be of more influence on consumers than positive reviews (Hornik et al., 2015). Therefore it is assumed in the current research that the audience will be more likely to agree with negative reviews, than with positive reviews. In order to investigate this the second hypothesis is proposed (H2).
H1: A positive product review given by the creator will lead to a positive product attitude amongst the audience, and a negative product review given by the creator will lead to a negative product attitude amongst the audience.
H2: A negative product review given by the creator will lead to a higher review agreement amongst the audience than a positive product review given by the creator.
Besides the review attitude, the purchase intention can also be influenced by the creator’s reviews. Duan et al. (2008a) advises practitioners to embrace eWOM activities as they have a significant impact on sales. This advice is also supported by Dhar and Chang (2009). In these researches a correlation was found between positive reviews and sales. Due to this finding it can be assumed that positive reviews are of influence on the purchase intention. A clear relation between the two variables has not yet been found, but as the previous findings hint in that direction this possible relation is assumed to exist in the current research. It is assumed that a
positive review will lead to a higher purchase intention. Therefore, the third hypothesis (H3) is proposed as follows.
H3: The sentiment of the review will be of influence on the purchase intention; a positive review will lead to a high purchase intention, and a negative will lead to a lower purchase intention.
In the current research online UGC and eWOM reviews will be investigated on YouTube. The following paragraph will shed more light on this medium and will further discuss how this medium works.
Broadcast yourself on YouTube
YouTube is one of the largest online Video-on-Demand (VoD) systems where UGC can be published (Cha et al., 2007). In 2010 research showed that every minute, ten hours worth of content was uploaded to YouTube (Kaplan & Haenlein, 2010). Current YouTube analytics state the medium has over a billion users, that everyday billions of views are generated, and people watch hundreds of millions of hours worth on YouTube daily (YouTube, n.d.).
YouTube is a content community (Arguello et al., 2006; Cha et al., 2007; Kaplan & Haenlein, 2010; Smith, Fischer, & Yongjian, 2012), where users are self-publishing consumers (Cha et al., 2007). The community empowers users to be more creative, and they are creating new viewing patterns, social interactions and with that, new business opportunities (Cha et al., 2007). Besides UGC, BGC can also be found on YouTube (Kruitbosch & Nack, 2008).
Interestingly, it was found that BGC videos received more views than UGC videos on YouTube. But more recent research by Smith et al. (2012) states that UGC videos tend to receive the most comments from viewers. Thus, even though BGC videos might have gotten more views in the past (Kruitbosch & Nack, 2008), research shows that much more interaction can be found with UGC videos (Smith et al., 2012). Smith et al. also concluded that the most popular content on YouTube are vlogs, which can be in the form of a visual diary, music
videos, live material such as performances, informational content such as reviews, and scripted performances like entertaining sketches.
Videos on YouTube differ from other forms of UGC, as for instance blogs. Blogs are low in media richness and social presence, whereas content communities, such as YouTube are already medium in media richness and social presence (Kaplan & Haenlein, 2010). Social presence refers to the contact that can be achieved on an acoustic, visual, and physical level, between two communication partners. This is influenced by the intimacy and immediacy of the medium. With intimacy it is meant that social presence is higher in interpersonal communication, than in mediated communication. Immediacy means that social presence is lower in asynchronous communication, like in e-mails, than in synchronous communication, which can be live chats. Media richness theory (MRT) is based on ambiguity and uncertainty in communication (Daft & Lengel, 1986). Some media are more effective for communicating ambiguous issues than others, which is based on the amount of information that can be transmitted in a certain time interval and the possibility of transmission of social cues such as gestures. Blogs only allow a simple exchange, where as YouTube videos can already transmit much more, like visual cues, body langue, and tone of voice. This richness makes YouTube more influential than blogs when it comes to reviews about products and services.
Kaplan and Haenlein (2010) state that YouTube is a medium that is low in self-presentation and disclosure, but this classification should be critiqued. YouTube is about
self-presentation which can be seen as a way to express a certain image and identity to others (Smith et al., 2012). YouTube’s slogan is “Broadcast Yourself” and serves as a platform that facilitates self-expression through UGC (Kruitbosch & Nack, 2008, YouTube, n.d.). With this focus on self-expression and broadcasting oneself this medium is a perfect place for self-presentation and self-disclosure and should therefore be classified as a medium that scores high on these points.
As seen in previous research the amount of followers a person has, has a lot of influence on how influential their UGC or eWOM can be (Rui et al., 2013). This power makes them interesting partners for corporate companies, as they can influence the target audience with their content. This topic of influencing will be further discussed in the following paragraph.
Online influencers
Creators active on media like YouTube have generated a sphere of influence where they can reach millions of users (Cheong & Morrison, 2008). Influencers are defined as “individuals who disproportionately impact the spread of information or some related behaviour of interest” (Bakshy, Hofman, Mason, & Watts, 2011, p. 66). Earlier research about influencing people mostly focussed on opinion leaders (Lyons & Henderson, 2005; Shoham & Ruvio, 2008). Influence by opinion leaders is a process that happens, and people’s tendency here to, when people try to influence other consumers’ purchase behaviour (Shoham & Ruvio, 2008). Opinion leaders can informally influence a potentially large audience of consumers, and can be of influence on how other consumers seek, purchase and use products or services (Lyons & Henderson, 2005).
There are a handful of articles that try to explain how influencers and influencer marketing works (Brown & Fiorella, 2013; Shoham & Ruvio, 2008; Ranga & Sharma, 2014). Ranga and Sharma (2014) describe influencers as popular personalities that have influence over
prospective buyers from the target market. They state that these influencers have the audience factor, meaning they will have a lot of followers and people they can reach with their message. Also, people follow influencers for a reason, because they are for instance interested in the creator or the produced content, which makes the audience much more reactive towards the shared content.
Ranga and Sharma (2014) identify four types of influencers, of which influencer by connection and influencer by topic are applicable to creators on YouTube. Influencers by
connection are people with hundreds of online followers and this group is specifically interesting for brands to create brand action. Influencers by topic are opinion leaders on a certain topic. If a brand matches the topics discussed these influencers can be interesting for them. Besides the connections and the topics, expert and referent power are also important elements that can be seen on YouTube (Shoham & Ruvio, 2008). Creators with expert power are those who have certain knowledge, or expertise in a certain area. Referent power creators are influencers whose followers look up to them.
The previous discussed literature makes it clear how influential creators on YouTube can be. As seen in the Fisherman’s Marketing Model influencers can amplify brand messages or offer recommendations to their audience (Brown & Fiorella, 2013). This with hopes that some followers will embrace the messages, and possibly share these messages with other people, which will lead to eWOM. Through this creators can in turn generate more sales (Brown & Fiorella, 2013; Lyons & Henderson, 2005). The information that the creator gives about a product in a review is therefore important. In the current research it is assumed that when the creator shares purchase information, this will be of influence on the purchase intention.
Purchase information is operationalized as information concerning where the product or brand can be purchased, what the price of the product is, and if some of the original brand message is given. The more elements mentioned by the creator, the higher the purchase information will be. When this is combined with a high level of product specific information, which can be given through mentioning a high amount of product attributes, this effect on the purchase intention is suspected to be even higher. In order to investigate these assumptions the following hypothesis is proposed (H4).
H4: (a) When a high level of purchase information is given by the creator, this will lead to a higher purchase intention amongst the audience. (b) When a high level of overall product
information is given by the creator, this will lead to a higher purchase intention amongst the audience, than when only a high level of purchase information is given.
Considering the power that influencers have (Brown & Fiorella, 2013; Shoham & Ruvio, 2008; Ranga & Sharma, 2014), it is also assumed that what the creator says will be reflected on in the comments by the audience members. As the research on influencer marketing is quite limited still, there is no proof yet that this relationship is possible. However, as creators are very influential, there is enough reason to assume this relation is present. Therefore the fifth hypothesis (H5) is proposed as follows.
H5: There will be a relation between the product attributes mentioned by the creator, and the product attributes mentioned by the audience in the comments.
As briefly pinpointed in the previous paragraph, expert power is an important aspect of influencing (Shoham & Ruvio, 2008). This will be further discussed in the credibility
paragraph, before there will be a switch over to the audience side, where the community feeling will be explained.
The credibility factor
In the previous parts of this theoretical analysis it had been touched upon how reviews created by users are seen as more credible than brand-generated information (Bickart & Schindler, 2001). Mostly because this concerns reviews by fellow customers. As fellow customers have no intention of manipulating readers it can be expected that the provided information reflects the product performance. However, reviews are not always honest and people can have persuasion knowledge about the possibility that reviews can be manipulated, which influences the effect of the review on consumer reactions (Bambauer-Sachse & Mangold, 2013). This has a lot to do with the credibility of the source providing the review (Bambauer-Sachse & Mangold, 2013), as it is very hard for a consumer to differentiate which reviews are real and which are not
(Bambauer-Sachse & Mangold, 2013; Cheung & Lee, 2011; Duan et al., 2008b). Therefore the credibility of the source is important for the attitude towards product information and its effectiveness (Lafferty et al., 2002). High credible sources are more persuasive than low credible sources (Pornpitakpan, 2004). Persuasive messages of sources that are credible will have a stronger effect on the product evaluations (Bambauer-Sachse & Mangold, 2013). When a source is perceived as credible and a review is positively framed, this leads to elaboration of the information (Jones, Sinclair, & Courneya, 2003). When this is the other way around and the source is perceived as non-credible, people fail to engage in elaboration of the information. A source’s credibility is dependent on the perceived expertise, authenticity and
transparency (Birnbaum & Stegner, 1979; Buda & Zhang, 2000; Fournier & Avery, 2011). Expertise about a certain topic is important in reviews, and this amplifies the effect of source’s bias (Birnbaum & Stegner, 1979). Expertise is amongst consumers expected to be based on things such as training, experience and ability (Birnbaum & Stegner, 1979; Shoham & Ruvio, 2008). Due to this expected expertise people can be biased about the source’s report and the true ability the source has (Birnbaum & Stegner, 1979). When people expect someone to know a lot about a subject they will be more likely to see the source as credible. Buda and Zhang (2000) concluded that besides expertise, trustworthiness also determines the source credibility. This ties in with the importance for authenticity and transparency (Fournier & Avery, 2011).
Looking at source credibility on YouTube, it can be seen that the creator’s identity is more present. Due to YouTube’s focus on self-expression and its visual capacities, the person giving the review is much more visible (Cha et al., 2007; Kruitbosch & Nack, 2008; Smith et al., 2012), and their credibility can be judged on more cues. When a creator posts videos in a certain segment they can easily be seen as an expert on that topic. When a creator is for instance
reviewing a lot of beauty products, it can be assumed that this creator tried a lot of products, which is why the creator can possibly give better-formulated conclusions about the products
discussed. Trustworthiness can be achieved when the creator is giving both positive and negative reviews, which makes the reviews seem more honest. This can also help with the authenticity as there are not only positive reviews about certain brands. Authenticity can also be shown when the creator is formulating its own opinions and is not following the brand
messages or claims. Tying these factors together with the public identity of the creator, creators are potentially credible sources.
As there are a lot of benefits to using a credible source (Bambauer-Sachse & Mangold, 2013; Lafferty et al., 2002; Pornpitakpan, 2004), it is really interesting for organizations to arrange a partnership with creators on YouTube. However, knowledge that certain partnerships might occur can influence the credibility of the creator, and can even make the source less credible (Bambauer-Sachse & Mangold, 2013). In order for creators to maintain their
credibility, they can state a disclaimer. In the current research a disclaimer is seen as stating the reviews are based on personal opinions, and that the reviews are not sponsored by any brand. It is expected that stating a disclaimer will have a positive influence on how the audience
perceives the creator as credible, and will make the source seem more credible. To investigate this the sixth hypothesis is proposed (H6). This perceived credibility is important as reviews from high credible sources can have a strong effect on the product evaluations amongst the audience (Bambauer-Sachse & Mangold, 2013; Jones et al., 2003; Pornpitakpan, 2004). Therefore, creators on YouTube can potentially generate more sales (Brown & Fiorella, 2013; Lyons & Henderson, 2005). In order to investigate this it is assumed in the current research that when the creator is perceived as highly credible, this will lead towards a higher purchase intention. To investigate this the seventh hypothesis is proposed (H7).
H6: Stating a disclaimer will influence the audience’s perceived credibility; (a) a disclaimer stating the creator bought a reviewed product will make the creator to be perceived as more
credible by the audience. (b) a disclaimer stating the creator got paid for a product review will make the creator to be perceived as less credible by the audience.
H7: The perceived credibility will be of influence on the purchase intention; a high perceived credible creator will influence a higher purchase intention, and a low perceived credible creator will influence a lower purchase intention.
The community feeling
Reviews given by creators on YouTube can be of great influence (Bambauer-Sachse & Mangold, 2013). Yet, YouTube is not only a place for reviews, it is a content community (Arguello et al., 2006; Cha et al., 2007; Kaplan & Haenlein, 2010; Smith et al., 2012). People join online communities for reasons of seeking for information exchange, friendship or social support (Ridings & Gefen, 2004). Seeking for information exchange was found to be the most popular reason for joining a community, but the other motives closely follow. Friendship has been described as being together and feeling as a member of a group. This ties in with the social identity theory that states we divide our social world into groups of people who are similar to us (Hofhuis, Van der Zee, & Otten, 2015; Tajfel & Turner, 2004). People that we consider to be like us are part of the in-group. Being included in a group gives individuals self-identity, values, attitudes and shows them what is the accepted behaviour in the group (Ridings & Gefen, 2004). Everybody who does not belong to the in-group is categorized as the out-group (Hofhuis et al., 2015). People in the in-group are categorized as friends that are predictable, safe and pleasant. People in the out-group are seen as unpredictable, unreliable, inferior and sometimes as the enemy.
In the current research the in-group is seen as the group amongst whom a community feeling is present. The social identity theory ties in with credibility (Birnbaum & Stegner, 1979; Buda & Zhang, 2000; Fournier & Avery, 2011). As in-group members are seen as predictable, safe and pleasant, it is more likely that they will be perceived as credible. In the current research
the creator is seen as the influencer of the in-group, and it is assumed that the in-group is based around this person. When audience members actively show appreciation and support for the creator, they will be grouped as the in-group. When audience members post a comment without directing them at the creator or anything related to the creator, they will not be actively involved in the in-group, and will therefore be seen as a member of the out-group. As being part of the in-group shows what is the accepted behaviour in the group (Ridings & Gefen, 2004), people who show unaccepted behaviour can be seen as the enemy (Hofhuis et al., 2015). This
unaccepted behaviour can be directed at the creator or at the community, and is assumed to be more negative than messages of in-group audience members. In the current research it is
suspected that when audience members want to be part of the community this can be seen in the sentiment of their message. It is assumed that in-group member will frame their message more positively than out-group members. To investigate this the eighth hypothesis is proposed (H8).
H8: Audience members who show signs of in-group behaviour will frame their messages more positive than audience members showing signs of out-group behaviour.
Methods
For this research two quantitative content analyses were conducted. One coder coded YouTube reviews of the creator (analysis 1) and responses to those reviews by the audience (analysis 2). For this research the Dutch channel “VeraCamilla” was used (VeraCamilla, n.d.). Of this channel monthly product review videos discussing likes and dislikes of beauty products were selected. All the reviews mentioned in the YouTube videos served as research objects for the first analysis. All the responses given by the audience to these videos served as research objects for the second analysis. All objects of this research are publically available online, and have been documented for research purposes.
Sample
The sample consists of 46 reviews given by the creator (analysis 1), and 949 comments given by the audience (analysis 2). The reviews given by the creator were taken from 6 videos, published in the period between April 2015 and September 2015. In these videos beauty
products were discussed on a monthly basis. The creator selected a few products for each video that she was positive about (“like”) and a few she was negative about (“dislike”). This way of reviewing products in the videos made the reviews good research objects as both positive and negative reviews were given. See appendix A for a reference list of videos that have been used. Of these videos all the reviews (100%) were coded for analysis 1. All the comments (N=1,006) of the 6 videos used in analysis 1, were gathered for analysis 2. These comments were
published between April 2015 and December 1, 2015. In the process of coding comments that were posted more than once, were unreadable (due to using another language or due to a high amount of spelling errors), or were considered as a form of spam, were excluded from this research which gave a dropout rate of 5,7%. The channel of “VeraCamilla” was selected due to its activity, availability of review-only videos, and its reach. The channel has been active since September 2009 and is currently providing its audience with two videos a week (VeraCamilla, n.d.). The channel has 164,633 followers, and has had almost 27,1 million views. All the gathered content is in Dutch, which is the native language of the coder.
Operationalization of constructs
To analyse the content from the sample two codebooks and a coding guide were designed (see appendix B till D). Both codebooks start with general variables to define the coder and which item was coded. The codebook for analysis 1 continues with identifying the brand and product that has been reviewed and whether this was a “like” or a “dislike”. For each review different product attributes are coded with “yes” or “no” based on whether they are mentioned or, and if “yes” the attributes sentiment is coded on a 5-point Likert scale, reigning from very negative (1)
to very positive (5), or can be coded as not mentioned (6). The different product attributes are package design/look of product, colour(s) of product, texture of product, smell of product, purpose of product, performance/quality of product, price for value, ease of purchase, and recommendation for use or reason to purchase product. Purchase information is measured by 3 “yes” and “no” questions determining whether the price, place of purchase, and/or marketing information is mentioned. For the brand evaluation variable it is coded how many times the brand name is mentioned, if a brand evaluation is given and what the sentiment is, the last one coded on a 5-point Likert scale reigning from very negative (1) to very positive (5), or can be coded as not mentioned (6). The mentioning of a disclaimer is coded for the video as a whole, and for the product, and is measured with 4 “yes” or “no” questions. An example question here is “Does the creator mentions she received the product?”.
The codebook for analysis 2 continues with determining if a brand or product previously mentioned in the video is also mentioned in the comment. If so, it is coded which product attributes the comment mentions, which is coded through “yes” and “no” questions. When a product is mentioned the review attitude and the product attitude is also coded for this
comment. In all cases the sentiment of the comment is coded on a 5-point Likert scale reigning from very negative (1) to very positive (5). The credibility variable is coded through 3 items determining if there is a perceived expertise, trust or authenticity, which could be coded as “yes”, “no”, and “not mentioned”. The in-group and out-group behaviour is coded through 7 items answered with “yes” and “no”, with an example question as follows: “Does the comment contain advice or tips for the creator?”.
Coding procedure
The codebook was created by the coding process leader, who also conducted the research. The codebook was tested extensively before it was finalized. To test the intercoder reliability of the variables, an external coder was used. The external coder received an hour of training and coded
8 reviews for the reliability test of analysis 1, and 63 comments for the reliability test of
analysis 2. A Krippendorff’s Alpha test was run to test the reliability. The results of this test can be found in the Table 1. The variables review attitude (α = .36) and product attitude (α = .36)
turned out to be unreliable. However when testing the rater agreement of these two variables through crosstabs it turned out they both have a 85.7% rater agreement, which is why the variables will be used in further analyses. A couple of items were combined to create the variables for the analysis, being the purchase information and disclaimer variable from analysis 1, and the purchase intention, perceived credibility and community feeling from analysis 2.
Table 1.
Krippendorff’s Alpha for reliability of coding
Variables Video (Analysis 1)
Krippendorff's Alpha Variables Comments (Analysis 2)
Krippendorff's Alpha
Product is like or
dislike α = 1
Mentioning of brand name α = .78 Product attribute:
Package design/look of product
α = .79 Sentiment α = .84 Product attribute: Package
design/look of product α = 1 Product attribute:
Colour(s) of product α = 1
Sentiment α = .86 Product attribute:
Colour(s) of product α = 1 Product attribute:
Texture of product α = 1
Sentiment α = .93 Product attribute: Texture
of product α = 1 Product attribute:
Smell of product α = 1
Sentiment α = 1 Product attribute: Smell of
product α = 1 Product attribute:
Purpose of product α = 1
Sentiment α = 1 Product attribute: Purpose
of product α = 1 Product attribute:
Performance/Quality of product
α = .74 Sentiment α = .95 Product attribute: Performance/Quality of product
α = 1 Product attribute:
Price for value α = .74
Sentiment α = .62 Product attribute: Price for
value α = .79 Product attribute: Ease
of purchase α = .74
Sentiment α = .75 Product attribute: Ease of
purchase α = 1 Product attribute:
Recommendation for use or reason to purchase product
α = 1 Sentiment α = 1 Product attribute: Recommendation for use or reason to purchase product
α = 1
Purchase Information α = .79 Review attitude α = .36
Brand attitude α = 1 Product attitude α = .36
Brand evaluation α = 1 Sentiment of the comment α = .78 Disclaimer α = 1 Purchase intention α = 1
Perceived Credibility α = .79 Community feeling α = .71
Results
Analysis 1 consisted of 46 reviews of which 33 (71.7%) were likes and 13 (28.3%) were dislikes. On average the sentiment of the product attitude was positive (M = 3.98, SD = 1.32), measured on a 5-point scale. In 28 (60.8%) of the product reviews the purchase information given was low, in 17 (37%) cases this was medium and in 1 (2.2%) case the level of purchase information given was high. Analysis 2 consisted 949 comments of which the review attitude on average was slightly positive (M = 3.47, SD = 1.35), the product attitude on average was positive (M = 4.11, SD = 1.07), and the overall sentiment of the comment on average was positive (M = 3.97, SD = .96), all measured on a 5-point scale. In cases where the purchase intention is mentioned (N = 81) the audience is on average thinking of buying the product (M = 2.33, SD = .63), measured on a 3-point scale. The audience perceives the creator with a medium level of credibility on average (M = 2.06, SD = .15), measured on a 3-point scale. In 603
(63.5%) of the cases the audience member is an in-group member, and in 346 (36.5%) of the cases the audience member is an out-group member.
In hypothesis 1 a relation between the product review given by the creator and the product attitude of the audience was assumed. This hypothesis was tested through a regression analysis with the product attitude from the audience as dependent variable and the product attitude from the creator as independent variable and is significant, F (1, 207) = 26.92, p < .001. The model can be used to predict the product attitude amongst the audience, but the prediction is weak: 11 percent of differentiation in the product attitude of the audience can be predicted based on the product attitude of the creator (R2 = .11). Product attitude of the creator, b* = .34, t = 5.19, p < .001, 95% CI [.23, .51] has a very weak significant relation with the product attitude of the audience. With every point extra on the scale of product attitude reigning from 1 (very negative) to 5 (very positive), the audience product attitude increases with .37. Hypothesis 1 is accepted.
In hypothesis 2 it was assumed that a negative review would lead towards a higher review attitude amongst the audience. This hypothesis was tested through a regression analysis with the review attitude from the audience as dependent variable and the product attitude from the creator as independent variable and is significant, F (1, 207) = 34.05, p < .001. The model can be used to predict the review attitude amongst the audience, but the prediction is weak: 14 percent of differentiation in the review attitude of the audience can be predicted based on the product attitude of the creator (R2 = .14). Product attitude of the creator, b* = .38, t = 5.84, p < .001, 95% CI [.27, .54] has a weak significant relation with the review attitude of the audience. With every point extra on the scale of product attitude reigning from 1 (very negative) to 5 (very positive), the audience product attitude increases with .40. Therefore hypothesis 2 is rejected, as the model shows the more positive a product attitude is, the higher the agreement with the review will be amongst the audience.
In hypothesis 3 it was assumed that a review could influence the purchase intention. This hypothesis was tested through a regression analysis with the purchase intention from the
audience as dependent variable and the product attitude from the creator as independent variable and is insignificant, F (1, 80) = 1.04, p = .311. The model cannot be used to predict the
purchase intention amongst the audience based on the product attitude of the creator. Therefore hypothesis 3 is rejected.
In hypothesis 4a and 4b it was assumed that the level of purchase and product information would influence the purchase intention. This hypothesis was tested through a multiple regression analysis with the purchase intention from the audience as dependent variable and purchase information (H4a) and product information (H4b) from the creator as independent variables and is significant, F (2, 79) = 4.88, p = .010. The model can be used to predict the purchase intention amongst the audience, but the prediction is weak: 11 percent of
and product information of the creator (R2 = .11). Purchase information of the creator, b* = .35,
t = 3.10, p = .003, 95% CI [.15, .68] has a weak significant relation with the purchase intention
of the audience. Product information of the creator, b* = -0.15, t = -1.37, p = .173, 95% CI [-0.35, .06] has a weak insignificant relation with the purchase intention of the audience. Hypothesis 4a is accepted, with every increasing point on the scale of purchase information reigning from 1 (low purchase information) to 3 (high purchase information), the audience purchase intention increases with .42. Hypothesis 4b proved to be insignificant and is therefore rejected.
In hypothesis 5 a correlation between the product attributes mentioned by the creator and the product attributes mentioned by the audience was assumed. This hypothesis was tested through a correlation between all the product attributes mentioned by the creator and all the product attributes mentioned by the audience. There is a significant correlation for the attribute texture of product, r = .16, p = .023, smell of product, r = .63, p < .001, and there is a significant correlation for the attribute performance and quality, r = .15, p = .028. For the other product attributes no significant correlation is found, therefore hypothesis 5 is only partly confirmed.
Table 2.
Correlation results product attributes
Attribute r p Attribute r p
Package design/look of product
r = .00 p = .990 Price for value r = .07 p = .300 Colour(s) of product r = .01 p = .873 Ease of purchase r = -0.02 p = .823 Purpose of product r = .04 p = .591 Recommendation for use
or reason to purchase product
r = .05 p = .804
In hypothesis 6 it was assumed that different types of disclaimers would influence the perceived credibility of the audience. This hypothesis was tested through a multiple regression analysis with the perceived credibility as dependent variable and the disclaimers that the creator bought
the product (H6a) and got paid to review the product (H6b) as independent variables and is insignificant, F (1, 207) = .04, p = .846. The model cannot be used to predict this relation and therefore hypothesis 6 is rejected.
In hypothesis 7 it was assumed that a high perceived credibility would influence the purchase intention. This hypothesis was tested through a regression analysis with the purchase intention as dependent variable and the perceived credibility as independent variable and is insignificant,
F (1, 80) = .75, p = .389. The model cannot be used to predict the purchase intention amongst
the audience based on the perceived credibility, therefore hypothesis 7 is rejected.
In hypothesis 8 it was assumed that in- and out-group behaviour would be of influence on the sentiment of the comment. This hypothesis was tested through a regression analysis with the sentiment of the comment as dependent variable and in-group/out-group behaviour as
independent variable and is significant, F (1, 946) = 180.55, p < .001. The model can be used to predict the sentiment of the comment amongst the audience, but the prediction is weak: 16 percent of differentiation in the comment sentiment can be predicted based on the in-group/out-group behaviour (R2 = .16). In-group/out-group behaviour, b* = .40, t = 13.44, p < 0,001, 95% CI [0,68, 0,91] has a weak significant relation with the comment sentiment of the audience. In-group members will frame their message more positive with .80, then out-In-group members. Hypothesis 8 is therefore accepted.
Conclusion and discussion
The current research was conducted to shed light on the concept of influencer marketing on YouTube. YouTube is an UGC community and its videos with product reviews and responses to these videos were the objects of this study. The research question was stated asking if
relations can be seen between what the creator mentioned in the reviews and what the audience mentioned in the comments.The current research was an exploratory research focussing on
variables of the creator, and how these variables could be of influence on the audience.
In the current research it was found that the creator gave both positive and negative product
reviews and had an overall positive sentiment in the reviews. In some of the reviews purchase information was given, but in most cases the creator gave a low amount of purchase
information. The creator was overall perceived as credible by the audience. The results show that whether or not the creator is stating a disclaimer in the review or video was not of influence on the perceived credibility amongst the audience. When looking at the relations that can be seen between the creator and the audience, a first assumption in the current research was that the sentiment of the product reviews would influence the audience. The results align with this assumption; a positive product review lead to a positive product attitude amongst the audience, whilst a negative product review lead to a negative product attitude amongst the audience. Other than expected in this research, the results show that a positive product attitude given by the creator lead to a higher review agreement amongst the audience. In this research it was also assumed that the creator could have some influence on the purchase intention of the audience, yet the results show that the purchase intention only gets positively influenced when the creator gives purchase information. The amount of product information mentioned, the sentiment of the review and the perceived credibility of the creator did not seem to be of influence on the
purchase intention amongst the audience. When looking at the audience a conclusion can be drawn based on the sentiment of their comment and their community feeling. The results showed that audience members who showed signs of in-group behaviour framed their comments more positively than out-group members. By tying these findings together the research question can be answered. The current research shows that overall the creator on YouTube is perceived as credible. Reviews given by the creator have an effect on the product attitude of the audience, and being part of the in-group has an effect on how the audience
member frames its message. The reviews given by a creator will not influence the purchase intention, unless purchase information is given.
Relating the findings in the current research to findings from previous articles, the finding that reviews have a positive influence on the audience, and negative reviews will have a negative influence on the audience, confirms with previous findings (Dhar & Chang, 2009; Duan et al., 2008a; 2008b; Kim & Gupta, 2012; Ren, 2015). Hornik et al. (2015) proposed that negative reviews would be of more influence on the audience, yet the results in the current study do not support this. The results actually indicate that a positive review is not only of influence on a positive product attitude, but also leads towards a higher review agreement amongst the
audience. A negative review will in turn lead towards a more negative product attitude, but will also lead towards a lower review agreement amongst the audience. The research of Hornik et al. (2015) was done in the light of eWOM and did not focus its research on YouTube, which can explain the difference in results. As YouTube is a content community where most of the communication stays inside the community, the spread of eWOM will not be as big as with for instance media like Facebook and Twitter. Through the current research it therefore can be concluded that a negative review will not lead to a higher agreement amongst the audience on YouTube, but these findings should not be assumed to be applicable on other media. When focussing on the sentiment of the comments by the audience, the comments were as expected more positive if the audience member was part of the in-group, which relates to previous findings (Hofhuis et al., 2015; Ridings & Gefen, 2004).
The possible influence on the purchase intention has been mentioned in several previous researches that pointed in the direction that the sentiment of reviews could influence the purchase intention, but no clear results were found yet (Dhar & Chang, 2009; Duan et al., 2008a). Based on these studies, the current research tried to reveal that the relation between the sentiment of the review and the purchase intention did indeed exist, but also failed to do so.
This finding is quite surprising but might be explained as both the previous studies and the current study are based on reviews on social media. It can be that the audience members of these media are also influenced by other factors such as advertisements or promotions of certain products. This possible combination of factors can be an explanation of why there was no relation found when the focus is only on the review sentiment in relation to the purchase intention. From the results in the current research it can be concluded that when practical information, like purchase information, was given this had a positive influence on the purchase intention. A conclusion drawn upon this is that when more purchase information is given this will make it easier for the audience to generate a purchase intention. If they have concrete information about where they can purchase the product and at what cost, it is easier for them to decide if they feel the product is worth purchasing. When no purchase information is given they need to be interested enough by the review to further investigate the product, before they can decide to purchase or not. This asks for a lot more effort from the audience, which can explain why there is a relation between the given purchase information and the purchase intention. Directing attention towards credibility and the purchase intention, it was found in previous research that high credible sources have a strong effect on the product evaluations amongst the audience (Bambauer-Sachse & Mangold, 2013; Jones et al., 2003; Pornpitakpan, 2004). And with this, it was assumed that this could lead towards a higher purchase intention (Brown & Fiorella, 2013; Lyons & Henderson, 2005). In the current research however, a relation between the perceived credibility and the purchase intention was not found. This can be explained as previous research focussed mainly on the source, and how the source could be high or low in credibility. In the current research this was taking into account by testing if stating any form of disclaimer would influence the perceived credibility, but it turns out this is not of influence. Therefore the credibility in the current research is only based on how this is perceived by the audience, which is a different perspective than in previous studies. Therefore, based on the
current research it is concluded that the creator on YouTube is perceived as credible by the audience. Being perceived as credible and giving reviews on YouTube is not of influence on the purchase intention, yet giving practical information in the review can lead towards a higher purchase intention.
From these conclusions it is possible to draw several implications for practice. Firstly the fact that a review by a creator influences the product attitude means that it is wise for corporations to consider working with influencers. Negative reviews can also have a negative influence on the product attitude, yet this will not lead to a lot of negative eWOM. This is an important finding for practitioners as this indicates that a possible negative review will not weigh as much as a positive review, and will therefore not be that damaging for a brand. Also it can be assumed that giving both positive and negative reviews will make the creator seem more credible, and
therefore be of more influence on their audience. The community feeling that can be seen on YouTube is very important and indicates how influential a creator can be. This community feeling is almost as though the conversation on YouTube is a conversation amongst friends, which makes creators powerful as influencers(Hofhuis et al., 2015; Ridings & Gefen, 2004).
Besides the conclusions a few shortcomings of this study should also be discussed. In the current research two content analyses were conducted based on text. In the first analysis the spoken words of the creator were coded and in the second analysis the written words of the audience were coded. This research design made it possible to see if there were relations between what the creator communicated and what the audience communicated. However, a lot more variables could have been analyzed in these analyses. Facial expressions, tone of voice and the time the creator took for a product review have for instance been left out in analysis 1 and might also be of influence on the audience. Also the amount of times one audience member responded to each of the videos has not been accounted for in analysis 2. An audience member might not have framed the message as an in-group member, but by always responding to the
videos this might indicate the community feeling this member has. By not accounting for this some cases might have been wrongly grouped as out-group members in this research. Another limitation is the fact that this research only looks at the comments for the second analysis. These comments show what the audience chose to voice in response to the reviews in the video, but might not be accurate when it comes to certain variables. This could be a reason why there were hardly any significant results found to indicate the influence on the purchase intention of the audience. If the audience did not mention something about the purchase intention this does not necessarily mean the intention was not present. Due to the current research design it is not possible however to witness the presence of the purchase intention when it is not mentioned in the comments. Another limitation in this study is that there was no control over which content would be available. This meaning that it is possible that the creator had removed mean or hateful comments from the comment section under a video. An audience member could have also decided to delete a comment that was personally published. Furthermore, it can be that people that continuously troll the creator have been banned from commenting on the channel of the creator. These monitoring actions are not visible to anybody watching the videos on
YouTube, and therefore the results might seem to be slightly more positively framed than they are in reality.
In future research the phenomenon of influencer marketing should be further investigated. The current research is limited to the medium YouTube, but there are several other social media that can be used by influencers nowadays. Therefore it is interesting to investigate content created by a creator in combination with responses of the audience towards this on other media such as Instagram or Facebook. Also research into the combination of different social media used by influencers can be an interesting topic of choice. It is also worthwhile to look at content where more negative comments are visible on the surface to give a better overview of the sentiment of the comments. Another possible way to get a better overview might be to team up with a creator
in order to get access to comments that are hidden for the public on their media channels, such as mean comments. As this research shows that the community feeling is present on YouTube, it is interesting to further investigate how this community feeling occurs on other media.
Furthermore, as the current research is limited for findings on the purchase intention, it is useful to investigate this further as there are several assumptions that influencers can influence the purchase behaviour of consumers (Bambauer-Sachse & Mangold, 2013; Dhar & Chang, 2009; Duan et al., 2008a; Jones et al., 2003; Pornpitakpan, 2004). Research specifically designed to unravel this variable could be an interesting new research direction. Purchase information in reviews, as found in the current research, is of influence on the purchase intention. It is very interesting to further investigate how this works and if there is a difference in how influential certain purchase information is. Further research into the purchase intention is specifically valuable for the corporate world. Besides future research into a combination of social media used by influencers, it is also interesting to further investigate which kinds of brand messages will be the most effective when it is communicated by an influencer through a certain media channel. Further research into these topics will help to elaborate on the concept of influencer marketing and is very interesting for the academic field and the corporate world.
The current research was conducted to fill some gaps in the existing literature. Previous
research on YouTube had been conducted several years back, and a reinvestigation was needed, which has been done through the current research. Also these previous researches only focussed on one aspect of the medium (Cheong & Morrison, 2008; Kruitbosch & Nack, 2008), whereas the current research is combining the content created by the creator with the comments by the audience. Combining these aspects through different analyses had not been done before and gives interesting insights in how the creator can be of influence on the audience. Furthermore this made it possible to see how the audience was influenced by what the creator said. By these findings, the current research did not only provide valuable knowledge for the academic world,
but is also a great source of information for corporations.
Despite the shortcomings and limitations presented before, it should be concluded that online influencers are powerful online popularities that can influence their audience and are a credible source for many. This research provided interesting insights on how influencing works on YouTube. Influencers are here to stay and provide corporations with a lot of opportunities for reaching their target market. This research shows that influencers should be embraced. Further investigation is still necessary and this research is a great starting point for that.
Literature list
Arguello, J., Butler, B. S., Joyce, E., Kraut, R., Ling, K. S., Rosé, C., & Wang, X. (2006, April). Talk to me: Foundations for successful individual-group interactions in online
communities. Paper presented at the SIGCHI conference on Human Factors in
computing systems, New York, United States of America. Retrieved from http://www.cs.cmu.edu/~jaime/ArguelloCHI06.pdf
Bakshy, E., Hofman, J. M., Mason, W. A., & Watts, D. J. (2011, February). Everyone's an
influencer: Quantifying influence on twitter. Paper presented at the fourth ACM
international conference on Web search and data mining, New York, United States of America. Retrieved from http://dl.acm.org/citation.cfm?id=1935845
Bambauer-Sachse, S. & Mangold, S. (2013). Do consumers still believe what is said in online product reviews? A persuasion knowledge approach. Journal of Retailing and
Consumer Services, 20(4), 373-381.
Bickart, B. & Schindler, R. M. (2001). Internet forums as influential sources of consumer information. Journal of interactive marketing, 15(3), 31-40.
Birnbaum, M. H. & Stegner, S. E. (1979). Source credibility in social judgment: Bias, expertise, and the judge's point of view. Journal of Personality and Social
Psychology, 37(1), 48.
Bodendorf, F. & Kaiser, C. (2009, November). Detecting opinion leaders and trends in
online communities. Paper presented at the ICDS'10 Fourth International Conference
on Digital Society, St. Maarten, Sint Maarten. Retrieved from http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=5432810
Brown, D. & Fiorella, S. (2013). Influence marketing: How to Create, manage, and
measure Brand Influencers in Social media marketing (1st ed.). United States of America: Que Publishing. Retrieved from www.books.google.com
Buda, R. & Zhang, Y. (2000). Consumer product evaluation: The interactive effect of message framing, presentation order, and source credibility. Journal of Product &
Brand Management, 9(4), 229-242.
Cha, M., Kwak, H., Rodriguez, P., Ahn, Y. Y., & Moon, S. (2007, October). I tube, you
tube, everybody tubes: analyzing the world's largest user generated content video system..Paper presented at the 7th ACM SIGCOMM conference on Internet measurement, New York, United States of America. Retrieved from
http://dl.acm.org/citation.cfm?id=1298309
Cheong, H. J. & Morrison, M. A. (2008). Consumers’ reliance on product information and recommendations found in UGC. Journal of Interactive Advertising, 8(2), 38-49. Cheung, C. M. & Lee, M. K. (2012). What drives consumers to spread electronic word of
mouth in online consumer-opinion platforms. Decision Support Systems, 53(1), 218-225.
Constantinides, E. & Fountain, S. J. (2008). Web 2.0: Conceptual foundations and marketing issues. Journal of direct, data and digital marketing practice, 9(3), 231-244.
Daft, R. L. & Lengel, R. H. (1986). Organizational information requirements, media richness and structural design. Management science, 32(5), 554-571.
Daugherty, T., Eastin, M. S., & Bright, L. (2008). Exploring consumer motivations for creating user-generated content. Journal of Interactive Advertising, 8(2), 16-25. Dhar, V. & Chang, E. A. (2009). Does chatter matter? The impact of user-generated content on music sales. Journal of Interactive Marketing, 23(4), 300-307.
Duan, W., Gu, B., & Whinston, A. B. (2008a). Do online reviews matter? An empirical investigation of panel data. Decision support systems, 45(4), 1007-1016.
Duan, W., Gu, B., & Whinston, A. B. (2008b). The dynamics of online word-of-mouth and product sales—An empirical investigation of the movie industry. Journal of
retailing, 84(2), 233-242.
Eccleston, D. & Griseri, L. (2008). How does Web 2.0 stretch traditional influencing patterns. International Journal of Market Research, 50(5), 591-161.
Fournier, S. & Avery, J. (2011). The uninvited brand. Business Horizons, 54(3), 193-207. Harder, J. (2015). Bloggers zijn uit: Zo betrek jij ‘influencers’ bij je marketing. Retrieved January 27, 2016 from http://www.frankwatching.com/archive/2015/09/23/bloggers-zo-betrek-influencers-marketing/#more-355088
Hennig-Thurau, T., Gwinner, K. P., Walsh, G., & Gremler, D. D. (2004). Electronic word-of-mouth via consumer-opinion platforms: What motivates consumers to articulate themselves on the internet? Journal of interactive marketing, 18(1), 38-52.
Hofhuis, J., Van der Zee, K. I., & Otten, S. (2015). Measuring employee perception on the effects of cultural diversity at work: Development of the benefits and threats of diversity scale. Quality & Quantity, 49(1), 177-201.
Hornik, J., Satchi, R. S., Cesareo, L., & Pastore, A. (2015). Information dissemination via electronic word-of-mouth: Good news travels fast, bad news travels faster! Computers
in Human Behavior, 45, 273-280.
Jones, L. W., Sinclair, R. C., & Courneya, K. S. (2003). The effects of source credibility and message framing on exercise intentions, behaviors, and attitudes: An integration of the elaboration likelihood model and prospect theory. Journal of Applied Social
Psychology, 33(1), 179-196.
Kaplan, A. M. & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of social media. Business horizons, 53(1), 59-68.
Kim, J. & Gupta, P. (2012). Emotional expressions in online user reviews: How they influence consumers' product evaluations. Journal of Business Research, 65(7),
985-992.
Kruitbosch, G. & Nack, F. (2008, October). Broadcast yourself on YouTube: really? Paper presented at the 3rd ACM international workshop on Human-centered computing, New York, United States of America. Retrieved from
http://dl.acm.org/citation.cfm?id=1462029
Lafferty, B. A., Goldsmith, R. E., & Newell, S. J. (2002). The dual credibility model: The influence of corporate and endorser credibility on attitudes and purchase intentions.
Journal of Marketing Theory and Practice, 10(3), 1-12.
Lyons, B. & Henderson, K. (2005). Opinion leadership in a computer mediated environment. Journal of Consumer Behaviour, 4(5), 319-329.
Macnamara, J. (2010). ‘Emergent’media and public communication: Understanding the changing mediascape. Public Communication Review, 1(2), 3-17.
Mandle, C. (2015). Forbes names PewDiePie as highest-earning YouTuber with annual
income reaching $12m. Retrieved January 27, 2016 from
http://www.independent.co.uk/news/people/forbes-names-pewdiepie-as-highest- earning-youtuber-with-annual-income-reaching-12m-a6695536.html
Misopoulos, F., Mitic, M., Kapoulas, A., & Karapiperis, C. (2014). Uncovering customer service experiences with Twitter: The case of airline industry. Management Decision,
52(4), 705-723.
Morin, R. (2015). Companies are embracing influencer marketing in 2015. Retrieved January 27, 2016 from
