A fashion perspective on digital influencers : does modality of content matter?

(1)

A fashion perspective on digital influencers: does modality of content

matter?

Milou van Mastrigt

11419709

Master Thesis

University of Amsterdam – Amsterdam Business School Msc. Business Administration – Marketing track

Supervisor: Dr. Mossinkoff

(2)

Thesis – Milou van Mastrigt – University of Amsterdam

2

Statement of Originality

This document is written by Milou van Mastrigt who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Acknowledgements

This thesis was written as part of the last semester of the Master’s programme in Business Administration – Marketing track at the University of Amsterdam – Amsterdam Business School. Over the past 6 months I was given to opportunity to dig deeper into a subject that interests me a lot, namely digital influencers. While digging deeper into this subject by extensively reading and analyzing existing literature, I created and extended the current body of literature on this topic by conducting an experiment, whilst learning more about digital influencers, their content and their status than ever expected.

Nevertheless, I was not able to perform this research without the help of others. Therefore, I would like to start off by showing my gratitude to my thesis supervisor Dr. Marco Mossinkoff for being a source of inspiration. I would like to thank him for all his support, enthusiasm and valuable insights. Furthermore, I would like to thank the professors of the University of Amsterdam for their interesting lectures and in particular Eloisa Federici, who provided me with valuable feedback regarding the statistical and more technical side of this research.

Last but not least, I would like to thank my family and friends for supporting me throughout the writing process of this thesis. The process has not been easy and without your support I would not have been able to finish this thesis. Finally, I would like to show my gratitude to all respondents for participating in the pre-test, pilot study and main experiment, which has been extremely valuable. I hope that you, as a reader, will enjoy reading this thesis and that it will add to your current knowledge on the rising topic of digital influencers.

(4)

4

Abstract

Digital influencers have grown in numbers tremendously over the past years, however academic research has lagged the popular growth of these phenomena. This study examines the differences in modality of digital influencer content and the moderating role of digital influencer status on brand awareness, brand attitude and product attitude amongst the most frequent viewers of digital influencer content in the fashion context: females aged between 18-35 years old. An online experiment was conducted (N=276) with a 2 (static vs. moving content) x 2 (low vs. high status) between subjects factorial design. The aim was to extend current literature on digital influencers and to provide managers with a more clear understanding of what the most effective way of collaborating with a digital influencer is. The results show that there is a significant difference between (static vs. moving content) groups in effect on brand awareness. Moving digital influencer content creates higher brand awareness, i.e. higher brand recall rates and higher brand recognition rates, compared to static digital influencer content. However, brand attitude and product attitude are not significantly different when comparing the influence between static and moving digital influencer content. Furthermore, it was expected according to multiple theories (e.g. the Celebrity Endorsement theory, Source Credibility theory, Social-Dominance theory and the Social Capital theory) that a high digital influencer status would moderate the relationships between digital influencer content and brand awareness, brand attitude and product attitude. However, the results show that the moderating role of status was not supported. Digital influencer status does not influence the relationship between moving digital influencer content and brand awareness, nor brand attitude, nor product attitude, so that these relationships are not stronger when the digital influencer has a high status.

Keywords: digital influencer content, digital influencer status, brand awareness, brand attitude and product attitude

(5)

5

1. Introduction

Vloggers on the rise! There has been a surge in the number of vloggers over the past years and this number is still rising (Gao, Tian & Huang, 2010; Chris, 2017), with as most popular video sharing platform YouTube, also known as the content community (Smith, Fischer & Yongjian, 2012), which has over a billion viewers and over a million content creators worldwide1 that are earning money from their video’s posted (Sanchez-Cortes, Kumano, Otsuka & Gatica-Perez, 2014). YouTube reaches more adults (aged between 18-35 years old) than any cable network in the United States1_{. The content that is being created on such}

platforms can be referred to as user-generated content (UGC), which describes the “various forms of media content that are publicly available and created by end-users” and content that is “created outside of professional routines and practices” (Kaplan & Haenlein, 2010, p. 61). Another example of UGC refers to blogs, short for ‘web log’ and is a website that is a publicly accessible personal journal including pictures and text (Lee, Hwang & Lee, 2006). Technological advancements have contributed to a recent development of blogging, namely the concept of vlogging, which refers to video blogging and does not include text and photos but solely video created content (Gao et al., 2010). Vlogging has become a popular topic of debate, as it is more expressive than blogs and reaches a large amount of viewers on a daily basis (Gao et al, 2010). The proliferation of vlogs presents a great opportunity for brands to increase brand awareness and shape attitudes of their target audience towards their brand and products (Freberg, Graham, McGaughey & Freberg, 2011). Brands have recognized the power and potential of collaborating with vloggers (Berryman & Kavka, 2017; Liu, Jiang, Lin, Ding, Duan & Xu, 2015), especially since it is becoming more difficult to reach the young and tech savvy generation via traditional channels2.

1_{YouTube, YouTube in cijfers. Retrieved from: https://www.youtube.com/yt/about/press/ Date accessed: 15}

October 2017

(8)

8

Vloggers are referred to as digital influencers, since some have persuasive power over the viewers of their online content. Whilst continuously sharing their experiences through videos, these content creators are gaining followers and broadening their reach (Lee & Watkins, 2016). Digital influencers are experts in a specific field, knowledgeable about one or more subjects and continuously engaging their target audience (IMA, 2017). Companies are exploiting the effects of influencers for sales and advertising (Liu et al., 2015) and some replace traditional marketing practices such as TV advertisements with digital influencers. One of the reasons that digital influencers are interesting for brands is that consumers trust each other, peers, over brands, and therefore influencers have become an interesting communication vehicle. The personal recommendation and the power of trusted individuals make the influence of digital influencers different than that of brands themselves (Freberg et al., 2011).

The digital influencers, their content and the modality of the content that they are sharing are the main focus of this study. Vlogging is a new phenomenon and a result of recent developments, though still relatively unexplored in terms of academic literature (Uzunoglu & Kip, 2014; Lee & Watkins, 2016). Literature regarding digital influencers within the context of vlogging and examining the effects of modality of content within this context is scarce. Some authors state that moving content is able to create better brand perceptions, compared to static content, whereas other authors reject these findings. For example, Lee & Watkins studied how vlogs influence consumer perceptions, in the context of luxury fashion brands. Results of their study show that luxury brand perceptions significantly increase after watching a vlog and that vlog viewers report better luxury brand perceptions and even higher purchase intentions compared to non-viewers (Lee & Watkins, 2016). But there are ongoing debates on the actual influence of digital influencers as the report of the https://www.pixability.com/industry-insights/l2pixability-study-fashion-youtube-2014/ Date accessed: 20 October 2017.

(9)

9

Global Web Index3_{states that digital influencer content is solely consumed as a form of}

entertainment and not applicable to function as a tool to alter consumer perceptions or influence purchase decisions. Multiple authors stated that social media and its influence are still at the beginning stage of exploration and that there is only a limited amount of academic research on social media marketing (Uzunoglu & Kip, 2014; Lee & Watkins, 2016). Mainly for that reason this study aims to extend the body of literature on digital influencers regarding the modality of produced content and its influence.

Another point of debate is the status of a digital influencer. Research has shown that consumers are seeking information from influencers with a high online status (Liu et al., 2015). According to these authors, online status can be defined through the amount of followers of a digital influencer. Digital influencers that have a high amount of followers can be compared to celebrities, also acknowledged as celebrities of the Internet, micro-celebrities or self-created celebrities (Senft, 2008; Marwick, 2015; Sykes & Zimmerman, 2014). Because of increasing audience sizes in the form of followers, some digital influencers have achieved a celebrity status. Researchers claim that it is the amount of followers, thus celebrity status, of digital influencers what makes their persuasive power so high (Morris & Anderson, 2015). However, it can be questioned if the status of the digital influencers indeed matters most regarding the influence they exert over their followers.

1.1 Research question

After introducing the topics at hand, the following questions arise: what is the influence of digital influencer content that has the persuasive power to enhance brand awareness and to alter attitudes of viewers regarding brands and products presented? What makes digital influencers so advantageous, that a rising number of brands are willing to pay digital

3_{Global Web Index, Vloggers Trend Report Q1 2015. Retrieved from:}

(10)

10

influencers to wear their brands and products? To answer these questions, this research will focus on the modality of content and the digital influencer status. The sample of this study will be the most frequent consumers of digital influencer fashion content: females aged between 18-35 years old4 (Berryman & Kavka, 2017). This study will quantitatively research how the modality of digital influencer content (static vs. moving) influences brand awareness, brand attitude and product attitude of viewers. Moreover, this study will look into how the status of the digital influencer moderates this effect. The research question is therefore as follows:

“What is the effect of moving digital influencer content (vs. static) on brand awareness, brand attitude and product attitude of viewers, and how is this effect moderated by the digital influencer status?”

1.2 Theoretical relevance

In terms of theoretical contributions, this research will be the first to shed light on the different types of content that can be shared by a digital influencer. Researching this topic will therefore expand current literature on digital influencers, by shedding light on the relatively new phenomenon of vlogging while examining the differences in modality of digital influencer content. Furthermore, combining the modality of content and the status of digital influencers has not been researched before and therefore this research will be the first to combine these two factors, which are of vital importance in influencer marketing. Moreover, multiple authors have mentioned that digital influencer marketing and literature on social media is still in the beginning stage of exploration (Uzunoglu & Kip, 2014; Lee & Watkins, 2016) and thus more research should be dedicated to understand the effectiveness of this relatively new marketing tool.

4_{Multiscope, Een op de drie nederlanders kijkt naar vlogs. Retrieved from: http://www.multiscope.nl/}

(11)

11

1.3 Managerial relevance

In terms of managerial implications, the findings will enable brands to forecast what effect collaborating with a digital influencer will have on brand awareness, brand attitudes and product attitudes of consumers. Furthermore, according to the hierarchy of effects model, cognition and affection are the foregoers of conation, hence important predictors for consumers’ actual behavior (Lavidge & Steiner, 1961). Therefore, it is crucial for brands to have an influence on the first layers of the model, in order to reach the last phase of actual behavior. Digital influencer content, and especially moving content, can play an increasingly powerful role for brands for several reasons. First of all, vlogs reach members of particular communities gathered around similar interests and those members are interested in the type of content covered by the vlogger on the platform, such as fashion, beauty and lifestyle (Uzunoglu & Kip, 2014). For managers, collaborating with the right digital influencer results in directly influencing the right target group for a specific brand or product. Secondly, it is the viewers’ choice to watch a specific video, implicating that whatever the digital influencer is communicating, interests them. Vlogs are highly contextual, issue-focused, personalized and have a conversational, two-way communication nature (Segev, Wang & Fernandes, 2014). Thirdly, combining the modality of digital influencer content and status in this specific context is valuable on a managerial level as the higher the amount of followers, the higher the price that companies have to pay in order to collaborate with a digital influencer (Jolique, 2017a). It is thus valuable for managers to know whether they have to collaborate with a high status influencer, which is most likely equal to a high price or with a medium-sized influencer or micro-influencer, which is probably less expensive. This research will provide brands with information on what the most effective way of reaching their target audience is, in terms of digital influencer content. It helps understanding how different types of influencer

(12)

12

content may influence or shape brand awareness, brand attitudes and product attitudes, which helps managers in their decision-making on using digital influencers in their marketing plan.

The report will proceed as follows. Firstly, literature on the main concepts of this study is extensively researched, leading to the research gap and research question. Afterwards the conceptual framework is established and hypotheses are developed, followed by explaining the context of this research. Subsequently, the methodology of this study will be explained thoroughly, followed by the results of the preliminary analysis and the main analysis. After discussing the results extensively, the theoretical and managerial implications of this study will be shared and conclusions will be drawn, ending with limitations and further research suggestions.

2. Literature Review

In this chapter the present study is placed within the context of current literature and the main concepts of this study, digital influencers, digital influencer content and digital influencer status, will be defined. The literature review will start off by introducing and defining digital influencers and will continue with analyzing what digital influencers share on social media platforms, namely digital influencer content and the different forms it may have. Furthermore, it will analyze what effect digital influencer content has on the consumers of the content. Afterwards, the literature review will continue by defining what status comprehends in the context of digital influencers and there will be elaborated on the influence that status might have on the relation between the produced content of digital influencers and the content consumers.

2.1 Digital influencers

(13)

13

who researched the influence of brand communication through digital influencers. These authors used the two-step-flow theory of Katz & Lazarsfeld (1995) to explain that opinion leaders interpret information from the media and pass it on through others. They state that media moves in two distinct stages, whereas the first stage is when opinion leaders, who pay close attention to media, receive information and the second stage is when these opinion leaders pass on their own interpretations in addition to the actual media content, and thus increase the influence of this media (Katz & Lazarsfeld, 1955). This increasing influence plays a central role in this research, because digital influencers have the power to influence decisions of others in their roles of opinion leaders (Song, Chi, Hino & Tseng, 2007; Goldsmith, Lafferty & Newell, 2000). Digital influencers can share different types of content, but the most common types are in the form of pictures or videos. Recent research by Woods (2016) denotes that “Instagram as a social media channel is one of the first to come to mind when a topic is mentioned […] due to the visually engaging nature of the platform” (Woods, 2016, p. 11). Instagram is an example on which static digital influencer content is created and shared. YouTube is an example on which moving digital influencer content is created and shared. Tolson (2001) argues that through YouTube and especially by vlogging, people can achieve a celebrity status. Such celebrities can also be referred to as a micro-celebrities or self-created celebrities (Senft, 2008; Sykes & Zimmerman, 2014). Senft defines a micro-celebrity as “a new style of online performance in which people employ webcams, video, audio, blogs and social networking sites to ‘amp up’ their popularity among readers, viewers, and those to whom they are linked online” (Senft, 2008, p. 25). In this specific study, micro-celebrities and self-created micro-celebrities through social media are referred to as digital influencers. A recent study of Lee and Watkins (2016) examines how vlogs, moving digital influencer content, influence consumer perceptions of luxury brands. Their findings indicate that brand perceptions of viewers are increased after watching vlogs and that purchase

(14)

14

intentions were higher for people watching vlogs compared to people who were not watching vlogs. Next to these findings, the study shows that vlogs for unknown luxury brands had a larger positive impact on brand perceptions than vlogs for known luxury brands (Lee & Watkins, 2016). Meaning that if a brand is unknown and a vlogger is recommending the brand, the positive feeling of the vlogger is transferred to the viewers of the vlog. The reason why vlogs have such a positive effect on brand perceptions and purchase intentions is because people feel related to the vlogger, who vlogs in a natural format and shares whatever he or she is thinking, wearing, doing and feeling. People feel that they can identify with the vlogger, which leads to a friendship feeling that influences the consumers’ perceptions of a brand or product that are communicated in a vlog (Lee & Watkins, 2016).

In contrast to the above-mentioned findings, the Global Web Index report, published in 2015, questions the actual influence of vloggers. Their findings indicate that vlogs are the least effective source of product or brand discovery and that only two percent of Internet users are watching vlogs when they would like to get more information about brands3_{. The}

study shows that vlogs are valued more for entertainment rather than for brand promotion, which is agreed upon by researchers Sanchez-Cortez et al. (2014), as they as well state that vlogs are a popular form of entertainment, which contrasts the findings above. However, the finding that vlogs are being valued only for entertainment is arguable, as this also depends on whether viewers are actively searching for specific information or not. Vlog viewers can also be triggered to buy something without previous intentions. Furthermore, Freberg et al., (2011) indicate that influencers can also negatively impact a brand when sharing negative reviews of a specific product or brand, which appears to be the downside of digital influencers.

It can be concluded from reviewing digital influencers’ influence in general, that there seems to be inconsistency in the findings on the effects of digital influencer content on

(15)

15

consumers’ brand awareness, brand perceptions and purchase intentions. On the one hand, many brands have embraced the power of digital influencers and are collaborating with them, whereas on the other hand this power is questioned, as research showed that digital influencer content is considered as a form of entertainment. To dig deeper into the actual influence of digital influencer content, the next section will review the static vs. moving nature of digital influencer content and its effects.

2.2 The effects of modality of digital influencer content

The development of the relatively new phenomenon of vlogging has created opportunities for brands to collaborate with digital influencers who are frequently sharing video content. However, whilst measuring the effectiveness of digital influencers, it is necessary to study whether there is a difference in effect between digital influencers sharing static content, in the form of photo’s through e.g. Instagram, and digital influencers sharing moving content, in the form of video’s through e.g. YouTube. This in order for brands to use the wide reach of the digital influencer they are collaborating with in the most effective way. It is not only relevant for brands to find out what the most effective modality of content is, but also for digital influencers themselves. For them it can be extremely valuable to know whether there is a difference in the influence of their content posted, might it be either static or moving. Especially since it can be assumed that producing a video requires more time than taking a picture.

To start off generally, professor McLuhan was the first to state that the medium through which a message is sent is the message, instead of the message itself (1964). His theory explains that the medium through which content is carried plays a vital role in the way it is perceived. This could as well be the case with static and moving digital influencer content, because is it the medium, might it be static Instagram content or moving YouTube

(16)

16

content, that is carrying the message that matters more than the message itself? The theory of McLuhan explains that it is the medium that “controls and shapes the scale and form of human association and action” (McLuhan, 1964, p. 2). Linking this back to digital influencer content, it is the visual aspect of the media that is creating the message and although McLuhans’ theory was not created in the era of the Internet, it can still be applied to nowadays media. Both Instagram and YouTube function as a platform to share visual content, which creates brand awareness and fosters perceptions (David, 2016). However, the theory of McLuhan has been criticized in the context of social media. Scientists mention that it is not the medium that is changing the message, but that it is the connectivity of social media that has the power to change the message (Malik, 2011). McLuhan also distinguishes media in ‘hot’ and ‘cool’ media (McLuhan, 1964). According to his theory, ‘hot’ media is whenever the media is a one-way communication, such as radio and TV, and ‘cool’ media is whenever the media offers the possibility of interaction, which is the case with social media. Both static and moving digital influencer are thus cool media, following McLuhan’s line of thought. It can be concluded that both static and moving digital influencer content include the connectivity and visual aspects of these platforms and therefore should not differ in terms of the message spread.

To understand the difference in influence between static and moving content and because of the lack of empirical research into this difference in the context of digital influencers, the following section will describe the differences in these modalities of content within the context of advertisements and product placements. A study of Yoo, Kim and Stout (2004) assesses the effects of animation in advertisements on each level of the hierarchy of effects model. The hierarchy of effects model consists of three levels: cognitive effects, affective effects and conative effects, and shows how consumers respond to advertisements in a very sequential way (Lavidge & Steiner, 1961). Cognitive effects refer to the beliefs about a

(17)

17

brand or product structured into a network and consists of knowing, recalling and recognizing a specific brand. Affective effects refer to the feelings, emotions and attitudes attached to a specific brand and the final stage of the model is conation, which refers to actual behavior, such as purchasing a specific product or brand (Lavidge & Steiner, 1961). The results of the study of Yoo et al. (2004) show that animation in banner advertisements results in better advertising effects compared to static advertisements. When animation is used in banner advertisements, viewers have a higher recall of the advertisement; a more favorable attitude compared to static advertisements and higher purchase intentions. Furthermore, Reeves and Nass (1996) stated: “when objects or people in pictures move, attention will be higher than during segments with no motion” (p. 220). In line with these findings, Watt & Welch (1983) discovered that dynamic images (moving) increase attention and affect an individual’s memory, such as their ability to recall or recognize a brand. Additionally, scientists Lutz & Huitt (2003) mention that an external stimuli should consist of attention grabbing features such as motion, in order to break through the clutter of advertisements, as motion triggers memory and enhances the effectiveness of persuasion.

Russell (2002) as well researched the effects of modality of content, but in the context of product placements. The starting point of his research was that visual stimuli and auditory stimuli differ in the amount of meaning that they carry. Meaningfulness of visual stimuli in product placements refers to creating the context in which the story is shown, e.g. a TV show, and meaningfulness of auditory stimuli in product placements refers to carrying the script of the story, e.g. the script of the TV show. Using meaningful stimuli is important as these stimuli are more deeply processed and have the ability to integrate better in a person’s cognitive structure, which as a result enhances brand recall (Russell, 2002). When determining whether a product placement will be effective, it is of vital importance to use the right modality of presentation. Russell’s research consisted of an experiment investigating

(18)

18

under which conditions a brand name was better recalled when used in a specific show and his results show that auditory placements were better recalled than visual placements (Russell, 2002). This means that product placements including a verbal expression, such as talking about a specific product, are better recalled than product placements in which a brand is solely shown visually and not verbally expressed. Linking this back to digital influencer content, both static and moving content include visual stimuli, however only moving digital influencer content, e.g. vlogs, include both auditory and visual stimuli. In line with the results of Russell’s experiment, moving digital influencer content would thus create better brand awareness compared to static digital influencer content.

Though animation is one of the most popular attention-grabbing tools according to the former researchers (Reeves & Nass, 1996; Watt & Welch, 1983; Lutz & Huitt, 2003, Yoo et al., 2004; Russell, 2002), other studies reported inconclusive results about whether awareness is increased by animation in advertisements (Bayles, 2002; Burke, Hornof, Nilsen, Gorman, 2005; Kuisma, Simola, Uusitalo & Oorni, 2010). The study of Bayles (2002) found no correlation between animation and recall, however the study did find a positive effect of animation on recognition. Burke et al., (2005) also researched recall and recognition of animation in advertisements, and they found unpredicted results: animation in advertisements resulted in poor recognition rates and static advertisements had a higher memorization rate compared to animated advertisements. Kuisma et al. (2010) proved with their study that animation had little to no effect on attention grabbing of viewers, but that animation did lead to improved recognition of brands.

A large amount of studies have researched attitude towards advertisements and the majority of these studies showed that advertisements and attitudes are related in a way that attitude towards a brand or product is strongly influenced by the attitude towards the advertisement (Yoo et al., 2004). One of the major differences between animated and static

(19)

19

advertisements in terms of influencing attitudes are the identifiable ad elements of animated advertisements (Rossiter & Percy, 1978). These identifiable ad elements, such as motion and visual stimuli, are positively affecting the attitude formation of viewers (Yoo et al., 2004; Babin & Burns, 1997). It is therefore of vital importance for marketers to create positive attitudes from viewers towards an advertisement and to make sure that there are no negative attitudes from viewers towards an advertisement, to ensure that only positive attitudes towards an advertisement are transferred to the product or brand itself. Freberg et al., (2011) studied audience perceptions of social media influencers and the benefits of social media influencers for brands. Their results show that influencers indeed have the persuasive power to alter attitudes of viewers and that social media influencers are perceived as relevant to turn to for advice. Additionally, the study of Liu et al. defines effective influencers as “those who not only can maintain their high online status in a user trust network during a period but also have the ability to affect their follower’s acceptance of recommendations, product choices, and purchase decisions in specific domains” (Liu et al., 2015).

A host of findings on the effect of animation in advertising has proven to be divisive. Positive, but also neutral and even negative effects of animation on brand memory and attitudes have been reported by previous studies as mentioned above. Additionally, the effect of modality of content on brand awareness and attitude is almost unexplored in online environments such as social media (Kuisma et al., 2010), which implies the need for further research.

2.3 The effects of digital influencer status

As briefly mentioned in the previous section, effective influencers are the ones who are able to maintain their high online status (Liu et al., 2015). The popularity and the status of members that are active on social networks are important aspects, mainly to determine which

(20)

20

content receives most attention. Few authors have defined the concept of online status in the current literature, however authors Romero, Galuba, Asur and Huberman (2011) define online status as: “One aspect is the popularity and status of given members of these social networks, which is measured by the level of attention they receive in the form of followers who create links to their accounts to automatically receive the content they generate” (Romero et al., 2011, p. 19). The research of Marwick is in line with this definition, as this author defines online status as follows: “The presence of an attentive audience may be the most potent status symbol of all” (Marwick, 2015, p. 141). In addition to the above-mentioned definitions, Berriman and Thompson (2015) state that status on YouTube “emerges not at the moment of production but rather accumulates through the attention of their audience” (p. 11). Taking all three definitions together, status can be defined by the amount of followers of a digital influencer on YouTube, Instagram or any other social media channel given. Because of the scarce amount of literature on the topic status within the digital influencer context, it is interesting to consider theories regarding status in other contexts.

The Social Dominance theory implies that most societies have status hierarchies in which some people have more social dominance compared to others (Sidanius & Pratto, 1999). Linking this with recent developments in online societies, the ones with more social dominance are the digital influencers of nowadays with their content consumers as being their followers. According to this theory, followers are vulnerable to the peer influence of their leader, which implies that whatever the leader is doing, wearing or promoting in published content, the follower would copy as a result of being influenced by this leader. Additionally, the theory highlights that the bigger the differences between the follower and the leader, the stronger the social dominance of the leader (Allen, Porter & McFarland, 2006). This theory is of vital importance, as the leader has the power to alter attitudes of the followers and this power increases as the distance between the leader and the follower

(21)

21

increases (Allen et al., 2006). In the terms of digital influencers, distance increases when influencers get more followers, thus gain a higher status. Status could also be compared to ‘social capital’, which is described by Bourdieu as an individual exerting power and influence over a privileged group because of the social capital the individual owns (Bourdieu, 1979). Each individual holds a position in the social space and social capital refers to an individual having power in society, by mutual recognition and acknowledgement from others who feel a sense of belonging to a specific group (Lin, 1999). When someone has social capital, he or she has the power to determine what constitutes taste within society (Bourdieu, 1979). This can be compared to the status and power that digital influencers have over their followers on social media.

In traditional marketing plans, celebrities are commonly used as endorsers for multiple reasons: to promote products and brands to a wide audience, to create a positive attitude in the minds of consumers towards a specific product or brand and to generate sales (Erdogan, 1999). Erdogan refers to this as the Celebrity Endorsement theory. Not only are these celebrities used to create a positive attitude towards a brand, but also for brands to break through the clutter of advertisements and to be more easily recalled and recognized. However, not every celebrity has the power to be an efficient endorser for a brand. There are criteria that must be met by celebrities in order to become a credible source and the theory that elaborates on these criteria is the Source Credibility theory (Ohanian, 1990). The tri-component celebrity endorser credibility scale consists of attractiveness, trustworthiness and expertise (Ohanian, 1990). Attractiveness refers to how likeable and physically attractive a source is to the audience and trustworthiness refers to the ability of a source to provide information that is non-biased, reliable and sincere in audience’s beliefs. The third component, expertise, refers to the ability of a source to provide accurate information as a result of possessing the right knowledge, skills and/or experience (Ohanian, 1990). The

(22)

22

higher the extent of the celebrity meeting these three requirements, the more efficient the celebrity can be in influencing receivers of the message.

Celebrities can be considered as the fore-goers of digital influencers. The shift in aesthetics of marketing communications because of the rise of the digital era has shifted the power away from traditional celebrities to micro-celebrities (Senft, 2008; Marwick, 2015), also referred to as self-created celebrities (Sykes & Zimmerman, 2014). Micro-celebrities and self-created celebrities can function in the same way ‘traditional’ celebrities used to do by transferring positive attitudes towards a brand or product and to create a stronger relationship between a brand and consumer. Previous section highlighted the status of ‘traditional’ celebrities and how this status affects consumers. But when considering the status of digital influencers, the study of Romero et al. (2011) plays an important role. As expected, considering the above-mentioned theories, high status and popularity would generate a high influence. Though, researchers Romero et al., (2011) proved differently in their research, as they state that high popularity is not the same as high influence and vice versa. By evaluation a 2.5 million dataset of users and analyzing influence propagation of web links on Twitter, they demonstrate that the relation between influence and popularity is weaker than expected. The difference with previous research is thus that high popularity does not equal high influence, implying that the status of a digital influencer does not have an effect on the influence they can have over their followers. On top of that, Agarwal, Liu, Tang & Yu (2012) have studied how to identify influential bloggers in virtual communities and their results show that active bloggers are not necessarily the most influential bloggers and influential bloggers can be inactive. Active bloggers are most of the times the ones with more followers, as the more frequently they post, the more familiar content consumers get with the blogger and the more followers it gains (Agarwal et al., 2012). In line with the aforementioned findings, a recent trend in influencer marketing is that more brands are collaborating with

(23)

23

micro influencers (Jolique, 2017b; Bernazzani, 2017). In terms of followers, micro influencers have between 5.000 - 25.000 followers, macro influencers have between 25.000 – 100.000 followers and top influencers have over 100.000 followers (Jolique, 2017b). Bernazzani describes the followers of micro-influencers as follows: “Unlike traditional influencers, micro influencers have a more modest number of followers, but they boast hyper-engaged audiences” (Bernazzani, 2017). She states that it is not the number of followers that matters most, but the level of engagement of followers that makes a difference. These contradictory findings and recent developments are debatable, because does digital influencer status, defined by the amount of followers, than still matter?

2.4 Literature gap and research question

The literature above offers a broad framework to build on when considering traditional theories on marketing and status. However, academic research on digital influencers, modality of digital influencer content and the status of digital influencers has lagged the popular growth of these phenomena. Therefore, this study aims to fill the gap between extensively researched animation in the context of traditional marketing practices and the poorly researched concept of digital influencer marketing and the modality of shared content. Influencer marketing is part of social media marketing (Li, Lai & Chen, 2011) and within the area of influencer marketing two main streams can be identified. The first stream focuses on the advantages of influencer marketing (e.g. Uzunoglu & Kip, 2014) and the second stream focuses on identifying effective influencers (e.g. Li et al., 2011; Liu et al., 2015). However, this current paper is positioned within the second stream of literature, as it focuses on evaluating the differences in effectiveness of modality of digital influencer content. Contradictory findings were found on the effectiveness of static vs. moving content and the role of digital influencer status and therefore the research question that this study addresses

(24)

24

is: “What is the effect of moving (vs. static) digital influencer content on brand awareness, brand attitude and product attitude of viewers and how is this effect moderated by the status of the digital influencer?”

3. Research design

This chapter will include the design of this research and starts off with the conceptual model. Afterwards, the hypotheses are developed and accompanied by relevant theories and literature. Lastly, the context of this research is explained.

3.1 Conceptual model

The conceptual model, established based on the literature review, is presented in figure 1. In the following section, the developed hypotheses as shown in the conceptual model are described and accompanied by briefly explaining the relevant theories and literature, as presented and elaborated on in the literature review.

Figure 1: Conceptual model

Brand Awareness Brand Attitude Product Attitude H1 (+) H2 (+) H3 (+) H4a (+) H4c (+) H4b (+)

Digital Influencer Status Low / high Digital Influencer Content

(25)

25

Brand awareness can be distinguished as the possibility to recall a brand or recognize a brand (Keller, 1993). Social media has the power to build and raise brand awareness (Stephen & Toubia, 2010) as a vast amount of people is already spending a great deal of their time on these platforms. As a result, collaborating with influential players on social media platforms will ensure that a brand name is spread over those networks, which helps to notify people, thus creating brand awareness (Golding, 2015). Based on previous research examining animation in advertisements and product placements and above-mentioned findings, this study proposes the following hypothesis:

H1. Moving Digital Influencer Content creates a higher Brand Awareness compared to Static Digital Influencer Content

In the light of the preceding discussion on the importance of forming positive attitudes towards an advertisement in order to be transferred to the product or brand itself, and considering the reported findings that visual moving stimuli result in more positive attitudes towards the brand or product shown (Yoo et al., 2004; Babin & Burns, 1997), the following two hypotheses are proposed:

H2. Moving Digital Influencer Content creates a more positive Brand Attitude compared to Static Digital Influencer Content

H3. Moving Digital Influencer Content creates a more positive Product Attitude compared to Static Digital Influencer Content

The presented theories and the majority of studies in the literature indicate one direction of the expected influence of digital influencer status: namely the higher the status, the higher the influence of the digital influencer (e,g. Erdogan, 1999; Sidanius & Pratto, 1999; Bourdieu, 1979; Ohanian, 1990; Marwick, 2015; Berriman & Thompson, 2015). Digital influencer status will serve as a moderator in this study since it influences the relationship between

(26)

26

digital influencer content and brand awareness, brand attitude and product attitude. Therefore, the following three hypotheses will be researched in this study:

H4a. The positive relationship between Moving Digital Influencer Content and Brand Awareness is moderated by the status of the digital influencer, so that this relationship is stronger for High Digital Influencer Status

H4b. The positive relationship between Moving Digital Influencer Content and Brand Attitude is moderated by the status of the digital influencer, so that this relationship is stronger for High Digital Influencer Status

H4c. The positive relationship between Moving Digital Influencer Content and Product Attitude is moderated by the status of the digital influencer, so that this relationship is stronger for High Digital Influencer Status

3.2 Context

Industry wise, especially the fashion industry has recognized the power of digital influencer content. The symbolic meaning of fashion and its accessibility to most consumers make fashion an intriguing context (Banister & Hogg, 2004). Fashion consumption is an overt behavior and a way to distinguish oneself from others, but also to show belongingness to a specific group (Gronow, 1997). Furthermore, there is a high risk of social acceptance of fashion purchases (Gronow, 1997). Celebrities and digital influencers in their function of role models reduce the risk of social acceptance, because when they are seen wearing a specific product, a product becomes more socially accepted. Consumers, especially females, are continuously looking for fashion inspiration and are turning to vlogs for advice (Hoffman, 2013). By watching fashion vlogs, individuals learn about fashion trends and copy them to their own wardrobes. Luxury fashion brands lack empirical research regarding social media marketing (Lee & Watkins, 2016), but not only those brands lack empirical research, also

(27)

27

other types of fashion brands are behind in receiving empirical attention. Past years the fashion industry has made place for a new type of fashion brands, namely new luxury brands, also known as masstige brands. Masstige brands are offering products with a higher level of prestige to mass markets at a lower price compared to luxury brands (Truong, McColl & Kitchen, 2009; Silverstein & Fiske, 2005). It are especially the masstige fashion brands that are an interesting point of debate and context for this study, as these brands are not well known because of a well-developed name, quality or price, but because these brands are relatively new and rather unknown to mass markets and need digital influencers to create an image around the brand (Truong et al., 2009). Masstige brands revolve around the intrinsic feelings aroused when wearing a product of a specific brand and about showing belongingness to a relatively small group of people knowing the brand (Truong, et al., 2009). Furthermore, Lee and Watkins proved in their study that vlogs for unknown brands have a larger positive impact than vlogs for a known brand (Lee & Watkins, 2016). Therefore, the context for this study is the fashion industry, with a specific focus on masstige fashion brands.

4. Methodology

This study is explanatory in nature and executed in two phases. The first phase consists of a quantitative pre-test, conducted to verify the stimulus material to be used in the main experiment. The second phase consists of the main experiment, which means that by using manipulated conditions possible differences between scores on the dependent variables are evaluated. This is done to discover causal relationships between the variables (Boeije, Hart & Hox, 2009). The main experiment was a 2 (static vs. moving digital influencer content) x 2 (high vs. low digital influencer status) between subjects factorial design. An online experiment was conducted in which respondents were assigned to one of the four conditions.

(28)

28

4.1 Phase one: Pre-test

In order to verify the stimulus material for the main experiment, a pre-test was conducted. The variable ‘content’ did not have to be verified, as the respondents were either presented with static content, e.g. a picture, or moving content, e.g. a video. However, to verify the variable ‘status’, a pre-test had to be conducted to find out what respondents would perceive to be low and high digital influencer status, based on the amount of followers. Perceived status of a digital influencer is not easily measured, as it is highly contextual and personally dependent. What one individual might consider as high status, might be completely different for another individual. Influencer marketing however does indicate that digital influencer status is highly correlated with its amount of followers (Romero et al., 2011; Liu et al., 2015; Marwick, 2015). In order to determine the ‘low’ amount and the ‘high’ amount of followers, three different and independent sources were used, namely L'Oréal, Jolique and IMA. L'Oréal, a company that frequently works with digital influencers, defines vloggers as follows: top vloggers have over 50.000 followers and medium sized vloggers have between 5.000 – 50.000 followers (L'Oréal, 2017). The second source was Jolique, influencer marketing specialist, who applies the following numbers on influencers: top influencers have over 100.000 followers, macro influencers have between 25.000 – 100.000 followers and micro influencers have between 5.000 and – 25.000 followers (Jolique, 2017b). The third source was influencer marketing agency IMA, which applies the following numbers: top influencers have over 100.000 followers and micro influencers have up till 30.000 followers (Tabor, 2017). Based on the numbers provided by L'Oréal, Jolique and IMA, it was decided to keep the low status below 5.000 and high status above 50.000 followers. For the pre-test, respondents were randomly assigned by Qualtrics to one of the following two conditions: one condition in which a picture of a digital influencer was shown with 1.212 followers, and one condition in which the same picture of the same digital influencer was shown, but with

(29)

29

266.862 followers. The number 1.212 is a fictive number, deliberately chosen to be below 5.000 and the number 266.862 was the exact amount of followers on YouTube of that point in time of the digital influencer used. To prevent confusion, another digital influencer was chosen than to be used in the main experiment. This in order to prevent confusion, because if by any chance the same respondents would participate in both the pre-test and the main experiment, they would not receive the low status condition of the digital influencer in the pre-test, and the high status condition in the main experiment. The digital influencer shown in the pre-test is a Swedish fashion blogger, Lovisa Wallin. A Swedish digital influencer was deliberately chosen, to decrease the chance of being recognized by any of the respondents. Furthermore, a full body shot picture was chosen on which the complete outfit of the digital influencer was shown and in which no specific brands were obviously visible. This because if the digital influencer would be wearing for example designer brands or high street brands, respondents might link this to the status of the digital influencer. Please refer to Appendix A for the stimulus material of the pre-test.

4.1.1 Pre-test sample

It was decided that the sample for the pre-test should be the same as the sample for the main experiment, which will be elaborated on later in this chapter (4.2.1). To briefly verify, the most frequent consumers of digital influencer fashion content were selected to participate in the pre-test, namely females between 18-35 years old4_{(Berryman & Kavka, 2017). As the}

digital influencer shown is a female, which is showing female fashion attributes, it would result in biased results if males would be asked to rate the digital influencer in terms of perceived status as they would probably focus on other attributes than intended. They would probably not be interested to analyze a fashion piece worn by the digital influencer, as it is a female wearing fashion items for females.

(30)

30

4.1.2 Pre-test results

The pre-test was conducted from 20 till 22 November 2017 by a self-administered online survey through Qualtrics and in total 38 respondents participated. Three responses had to be deleted because of incompletion, which resulted in 35 completed responses. Participants were approached through social media channels such as Facebook and Instagram, to ensure that the right target group was reached. After being assigned to one of the two conditions, respondents were asked: ‘The status of the digital influencer, based on the amount of followers, is…’. Respondents could answer on a 7-point semantic differential scale (1) ‘Low’ till (7) ‘High’. Additionally, a scale was established based to measure digital influencer status and influence. Based on characteristics an influencer must meet to become influential (Kapitan & Silvera, 2015; Li et al., 2011), five statements were developed. Respondents could rank these statements on a 5-point semantic differential scale from (1) ‘Strongly disagree’ to (5) ‘Strongly agree’. Please refer to Appendix B for the pre-test survey.

The results of the study show that indeed the low condition was perceived as low by the respondents (N=18), with a mean of 2.11 and Std. Deviation of 1.64. The high condition was indeed perceived as high by the respondents (N=17), with a mean of 6.29 and Std. deviation of 0.85. Please refer to table 1 for a visualization of the results. The established scale can be considered as reliable, as in the case of low status the Cronbach’s alpha was α = 0.875 (5 items) (α >.70). The scale also indicates that the item means of all five items together is 2.29 for the low status condition, which is at the lower side of the scale indicating that respondents did not agree with the statements on status of the digital influencer, as they scored between ‘Somewhat disagree’ and ‘Neither agree nor disagree’. In case of the high status condition the Cronbach’s alpha was α = 0.779 (5 items), (α >.70), indicating that the scale is reliable. Moreover, the item means is 4.3 and indicates that respondents indeed replied on the right side of the scale, which shows that respondents on average scored

(31)

31

between ‘Somewhat agree’ and ‘Strongly agree’. An independent sample T-test was done in order to compare the means of the two conditions. This showed that there was a significant difference in the scores of the low and the high condition. Respondents in the low condition (M = 2.11, SD = 1.64) indeed scored lower on status than respondents in the high condition (M = 6.29, SD = 0.85), t (33) =-9.39, p = 0.000 < 0.001, 95% CI [-5.09, -3.28]. These results show that the provided number of followers was indeed perceived to be low vs. high and would hence work as intended in the main experiment5, and therefore used. Furthermore, there was asked for age and gender in the pre-test, to ensure that the right target group was reached. Results show that all participants were aged between 18-35, thus fall in the right target group (M= 24.89, SD = 2.11) and that only female respondents participated in this study.

Table 1: Perceived digital influencer status

Condition N Mean Standard Deviation

Low (1.212 followers) 18 2.11 1.64

High (266.862 followers) 17 6.29 0.85

4.2 Phase two: Main experiment

A second quantitative study was performed, by means of an experiment. A self-reported survey was used as a data collection method to gather cross-sectional data, administered online. Four surveys were designed in order to create a 2 (static/moving digital influencer content) x 2 (low/high digital influencer status) between-subjects factorial design and each respondent was randomly assigned to one of the four conditions. Two of the conditions included static influencer content in the form of two pictures of an influencer with a product of a specific brand, whereas the other two conditions included moving content in the form of

5_{The assumption of equal variances has been violated: Levene’s F-test p = 0,047. ‘Status’ is not normally}

distributed, as proved by the Kolmogorov-Smirnov test; D(35) = 0.21 p < .001. Officially the T-test could not be performed, but to verify and approve the results of the independent T-test, a Mann-Whitney test was performed. This showed for ‘Status’ U(35) = 10, p <.001, the same as T-test results showed.

(32)

32

a video with a product of a specific brand. The digital influencer, product and brand were consistent in all conditions however the only thing that was different and manipulated was the modality of content among the conditions (static/moving), together with the digital influencer status (low/high) indicated by the amount of followers. This leads to the following conditions: (1) static digital influencer content with low status, (2) static digital influencer content with high status, (3) moving digital influencer content with low status, (4) moving digital influencer content with high status. Please refer to table 2 for the experimental groups that were created by this design.

Table 2: Experimental groups

Manipulation: Digital Influencer Content

Static Moving Low Manipulation: Status High 4.2.1 Sample

It is vital for a quantitative research to select a sample that represents the population that the research aims to study (Bryman & Bell, 2011; Easterby-Smith, Thorpe & Jackson, 2015). The target population for this study are females in generation Y. Generation Y is born between 1982 and 20016 and according to previous research the females of this generation are the most frequent viewers of digital influencer fashion content (Berryman & Kavka, 2017), thus most suitable for this study. It is for this reason that this survey was conducted amongst the target audience that is most familiar with the created content by digital influencers: females between 18-35 years old. Previous research amongst internationally

6_{WerkXYZ, Meer over de generatie X-Y-Z, Retrieved from: https://www.werkxyz.nl/werkxyz/meer}

over-de-generatie-xyz/ Date accessed: 21 November 2017. Condition 1 Static content, Low status Condition 3 Moving content, Low status Condition 2 Static content, High status Condition 4 Moving content, High status

(33)

33

famous British fashion, lifestyle and beauty vlogger Zoella shows that fashion vlogs have a highly feminized target audience (Berryman & Kavka, 2017). They refer to fashion vlogging as a ‘female-driven Youtube industry’ and mention that Zoella’s viewers are a youthful femininity audience, consisting of females between 14-35 years old. Due to having a minor age, respondents below 18 were not asked to participate in this research. A non-probability convenience sample was used to conduct the research, as the population is large and because of time constraints. The respondents were reached through social media, such as Facebook and Instagram.

Before sending out the survey, a pilot test was conducted with a small sample (N=11) to ensure that the survey worked without complications, that the questions were understood properly and to verify how long participants took to finish the survey. After gaining the results of the pilot test and amending minor details in the survey such as spelling mistakes and survey taking time, the survey was launched. The experiment was conducted in the period of 4 till 14 December 2017 and 303 respondents were randomly assigned to one of the four conditions. The first step after data was collected online was downloading the results in SPSS. In order to eliminate the risk of data entry errors, data was directly exported from Qualtrics to SPSS. However, a risk of online self-administered data collection is a poor response rate (Malhotra, 2010). Response rates could be improved by providing incentives for participation (Bryman & Bell, 2011; Malhotra, 2010). Therefore it was decided to include an incentive that was appealing to the sample and would enhance the response rate for the main experiment. Participants got the possibility to participate in a raffle in order to win a €20 gift card of the Bijenkorf, largest premium department store in The Netherlands.

In total 303 respondents took part in the experiment, of which 18 were removed from the sample because they dropped out. Moreover, one additional participant was removed from the dataset as it happened to be a male respondent, while this study solely focuses on

(34)

34

females. Furthermore, the survey included one reversed question to find out whether respondents were actually paying attention and not randomly ticking the boxes beneath each other. The results of the counter-indicative scale showed that eight respondents rushed through the survey without reading the questions and statements properly. If the counter-indicative scale indicated that there was no attention paid, the survey taking time of these respondents was analyzed and results showed that they took less than 150 seconds to finish the study. Therefore it was decided to remove these eight respondents, which resulted in a remaining sample of 276 respondents on which statistical analysis could be performed. This research design requires at least 50 respondents per condition, so a minimum of 200 respondents was required in order to ensure the sample was large enough to analyze. With a sample of 276 respondents, this requirement was met.

4.2.2 Stimulus material

The stimulus material that was used for this research consisted of two pictures and a video of a Swedish digital influencer. Previous research shows that the influence of a source is the largest when there are no previously established attitudes and knowledge about a specific subject (Kumkale, Albarracin & Seignourel, 2010). Since the research was conducted through the researchers’ personal extended network within the Netherlands, it was expected that most of the respondents would be Dutch. Therefore, to exclude previously formed attitudes and knowledge of participants, a Swedish digital influencer was chosen, namely Janni Deler. The brand that the digital influencer was wearing was also deliberately chosen. The chosen brand is Daniel Wellington, a Swedish masstige fashion brand selling, amongst other products, watches. Few people are familiar with the brand, but the ones who are familiar with the brand feel a type of belongingness to this small group of people. Again, a

(35)

35

relatively unknown brand was chosen in order to rule out previously formed attitudes towards the brand and thus a biased evaluation of the stimulus material.

The chosen static stimulus material consisted of two pictures on which the digital influencer is shown while wearing a watch of the masstige brand Daniel Wellington on her left wrist in front of a plain wall. The chosen moving stimulus material was a video (vlog) of 1:21 minute, shot by Janni Deler in which she is seen wearing and talking about a watch of the masstige brand Daniel Wellington on her left wrist. Both pictures and the video show Janni Deler wearing the exact same black outfit, so that no attitude differences could be referred to other stimuli than the digital influencer, the content, the brand and the watch. The pictures and the video were of high quality and were shot on the same day in which the whole setting was the same. The stimulus material is shown in Appendix C.

4.2.3 Procedure

Data was gathered through an online web-based, self-administered survey by using the software tool Qualtrics, which is frequently used by academic researchers (Qualtrics, 2017). Furthermore, as digital influencer content is published online, it seemed the most appropriate data collection tool. Via the researchers’ extended network female participants aged between 18-35 years old were approached through social media such as Facebook and Instagram by providing them with the Qualtrics link. After informing the participants about their rights as a participant and the procedure of the survey (the informed consent), they agreed to participate in the research by clicking on the arrow and were afterwards directly presented with the stimulus material. Before being exposed to the stimulus material, participants were informed about the meaning of status in influencer marketing, introduced to the digital influencer and explained that the digital influencer was wearing a watch of the brand Daniel Wellington. After seeing the picture or video, participants proceeded with the survey by answering

A fashion perspective on digital influencers : does modality of content matter?