Beauty vloggers and their influence on

consumer-buying intentions
The case of the Netherlands

Master thesis

Name: C.E. ten Have

Student number: 427548

Supervisor: Dr. M.N.M. Verboord

Master Media Studies: Media, Culture & Society

Erasmus School of History, Culture and Communication,
Erasmus University Rotterdam

June 20, 2017

2
 Beauty vloggers and their influence on consumer-buying intentions
The case of the Netherlands

Abstract
This study offers an analysis of the influence of a beauty vlogger’s persona on the buying intent of
consumers. Through an experimental survey design Dutch consumers are asked how they perceive the
persona of two beauty vloggers, after they are presented with two shortened YouTube clips per vlogger.
Then they are asked how likely they are to buy the product recommended or discussed by the beauty
vlogger in the videos. The beauty vlogger persona has been conceptualized with the help of theories
concerning source credibility, para-social interaction and the Big Five Personality model, and hypotheses
were formulated accordingly. The results showed that the dimension of social attractiveness, which is
concerned with how kind someone is perceived to be, has the largest influence on consumer buying
intentions, and is therefore the most important aspect of a beauty vlogger’s persona in light of this study.
The character traits extraversion and conscientiousness closely follow the dimension social attractiveness
and are also found to be important influencers. They are mainly concerned with perceived enthusiasm and
responsibility of the beauty vlogger. However, where social attractiveness and extraversion influence
buying intent positively, conscientiousness leads to a decrease in buying intent. This suggest that the
more responsible a beauty vlogger is perceived to be, the less consumers are intended to buy products
discussed by this person. As a beauty vlogger that expresses this personality trait will negatively influence
any brand, it is important to consider these effects as well. Additionally, the familiarity of the consumer with
the product a beauty vlogger discusses prior to watching the video is found to be important across all
analyses conducted. Finally, the findings show that even though scientific research organizations point
towards 18 to 34 year-olds as the most frequent watchers of online vlogs, a large group of younger and a
small group of older people are interested in beauty videos as well. This can again have important
implications for brands in creating marketing strategies. The study is mainly limited in its size. However,
this study brings innovation to the debate on social influencing by researching a particular type of social
influencer that has not been the focus point of any study before, the beauty vlogger, and thereby adds to
existing knowledge on the exact workings of interpersonal relations online in this particular context.

KEYWORDS: [Social influencers, eWOM, Vloggers, Interpersonal influence, consumer-buying intent]

3
4
 Table of contents

Preface 7

1. Introduction 9

2. Theoretical framework 13
2.1 Beauty vloggers as social influencers 13
2.1.1 Interpersonal influence online 13
2.1.2 The market mavens of social media 14
2.1.3 Professionalization of social influencing 14
2.2 Previous research on social influencers 15
2.3 Consumer-buying intent 16
2.3.1 Defining buying intent 16
2.3.2 Consumer characteristics 16
2.4 The impact of the social influencer persona 18
2.4.1 Source credibility 18
2.4.2 Para-social interaction (PSI) 19
2.4.3 The Big Five model 20

3. Research design 23
3.1 Method 23
3.1.1. An experimental survey 23
3.1.2 Sampling 25
3.2 Operationalization 26
3.2.1 Consumer characteristics 26
3.2.2 Beauty vlogger characteristics 28
3.3 Design 29
3.3.1 Stimulus selection and pretest 29
3.3.2 Survey design 31

4. Results 33
4.1 Descriptive results 33
4.2 Source credibility 36
4.3 Social and physical attractiveness 37
4.4 Homophily 39
4.5 Openness to experience 40
4.6 Conscientiousness 41
4.7 Agreeableness 42
5
 4.8 Extraversion 44
4.9 Relative impact of explanatory variables 45

5. Conclusion 49

References 55

Appendix A: survey 59

Appendix B: results of the pretest 62

Appendix C: grouping of beauty vloggers 62

Appendix D: overview correlations of all persona dimensions and buying intent 62

Appendix E: significant variables from the overall OLS regression 63

Appendix F: results of the linear mixed model 63

6
 Preface

One of my greatest passions is studying the world of online media and communications and thereby
exploring its commercial opportunities and weaknesses. The topic for this thesis was therefore easily
decided on by combining this passion with a large interest in consumer-behavior, and especially,
interpersonal influence. I find the way that the internet has created a new way of communication between
people across the world, facilitating not only commercial businesses to grow but also for people to find,
contact and relate with each other extremely fascinating. I would however have never been able to take a
deep-dive into my greatest passion for almost six months without the support of my thesis supervisor. I
would hereby like to thank Marc Verboord for supporting my topic and own initiative, allowing me to
develop this piece of work I am very proud of. Great thanks also goes out to my friend Lisette Nientker
who was always there when I needed her advice.

Above all I enjoyed writing this thesis. I have enjoyed the entire process, from struggling with what names
I would give my different literature review sections, to repeating my statistical tests over and over to get
them just right. I hope you will enjoy reading my thesis as much as I enjoyed writing it.

Charlotte ten Have

Rotterdam, June 13, 2017

7
8
1. Introduction
The amount of beauty content available on the Internet is increasing steadily, and especially on the video
platform YouTube. In April 2015 YouTube had 1.8 million beauty videos online that together generated
45.3 billion views (Pixability, 2015). With the rise of the Internet the traditional path between companies,
their brands and consumers has been disrupted. Especially social media has changed the way consumers
are faced and interact with brands (Fiorella & Brown, 2013). These media are distinguished from other
social networks by the phenomenon of social influence (Li, Cui, & Ma, 2015). This is apparent in the
beauty industry where beauty bloggers and vloggers have established their influence over the years, and
has been recognized by the business world (Coursaris & Van Osch, 2016). Of all the marketing buzz that
is created within the beauty world on YouTube, 97% is created by individual vloggers and other content
creators, as opposed to major brands (Marshall, 2014). These social influencers are not only important to
business and profit related goals, but the way that these influencers interact with consumers, which is
defined as the interpersonal influence, and the extent of their influence is of societal importance as well.
Interpersonal influence is concerned with the change in belief, attitude or behavior in the target of
the influence. The person exercising action to create this change in the target of the influence is called the
influencing agent, and is the social influencer (Raven, 2008). The term interpersonal influence is well
known within not only the field of sociology and consumer behavior but also psychology (Buchegger &
Hui, 2009). Since it has recently become possible to gather, store and retrieve the data from social media
networks such as YouTube, firms and governments are increasingly interested in what these networks
can contribute to their media- and marketing strategies. The availability of this data on social influencers
has raised questions concerning the implication of the interaction and interpersonal influence between
these social influencers and consumers (Galeotti & Goyal, 2009).
However, not all content creators are also capable of influencing people with their online videos.
Multiple studies focus on how brands can identify the right social influencers for their marketing strategies
and how to leverage their power (Cakim, 2009; Freberg, Graham, McGaughey & Freberg, 2010). When
studies are looking to identify the type of actor that is exercising more influence than others, and can
therefore be distinguished as a social influencer, the study is referred to as ‘influence persona discovery’
(Cai, et al., 2011). The notion of persona as researched in sociological studies acknowledges this
difference. ‘Persona’ is the way through which people express and present themselves, and is therefore
not grounded in biological features but rather in one’s character traits (Marshall & Barbour, 2015).
This study was inspired by the notion of social influence varying across different personas and
aims to develop a better understanding of the influence of persona on consumer-buying intent. Current
literature is missing knowledge on which particular aspects of a social influencer persona influence the
buying intentions of the consumers. Lee and Watkins (2016) have studied how vlogs influence consumer’s
perceptions on luxury brands in the United States, whereby they focused on brand perceptions of
consumers after watching a vlog.

9
The results of their study showed that consumers who feel similar to a vlogger will have the same positive
associations with the brands they discuss. They found that a relationship has to be established between a
vlogger and a consumer for luxury brand perceptions to be positive. Lee and Watkins (2016) incorporate
the notion of persona in their study as they examine several character traits of vloggers. However, their
focus is mainly on relationship establishment between the vlogger and the consumer. They therefore fail
to incorporate the difference between personas. Additionally, by focusing on brand perceptions they do
not include the buying intentions of consumers in their study.
Building on the research by Lee and Watkins (2016) my study will examine beauty vlogger
personas, the effect of these different personas and its different dimensions on the intentions of
consumers to buy the products discussed by the vloggers. That is why I propose the following research
question: to what extent are the buying intentions of consumers influenced by the perceived persona of
beauty vloggers? This question will be answered in this study and will aim at finding out to what extent
female consumers are influenced by social influencers in the context of YouTube make-up videos. Data is
gathered using a quasi-experimental method, in which seven versions of an online administered
quantitative survey are distributed. Each version contains two different beauty vloggers.
Over the years the concept of social influence has risen in importance. This is mainly due to the
fact that social influence can predict and determine our behaviors, which are both consciously and
unconsciously embedded in every activity we perform (Simonson, 2005). Even though we know that social
influence exists by for example when we look at for product reviews on online web shops and let it
influence our purchase decision, the exact workings of social influence are less clearly visible. With the
rise of social media channels such as YouTube researches have been given access to large chunks of
data that allow for the analysis of social problems. This research will socially contribute to this line of
research and will help understand not only the benefits of social influence as companies use social
influencers in their marketing strategies, but also help us understand why we behave a certain way and
how social influence works in the context of YouTube and social influencers (Huszár, 2012). The social
relevance can thus be found in the contribution that this study will bring to the current literature on online
interpersonal influence, a concept that has emerged with the rise of the Internet, the changing
environment of social networks in which businesses, brands and their consumers participate as well as
the buying intentions of consumers within this new environment of online interpersonal influence (Cheung
& Thadani, 2012).
Consequently, findings from this study will add to the debate on how new media and the Internet
guide and change behavior based on interpersonal influence, defining its scientific relevance. The findings
can also be of help in assisting both media organist ions and the government to determine the impact of
the new media. My study is therefore socially relevant in that it explores the phenomena of social influence
in the changing environment of the Internet and the new form of interpersonal influence that has emerged
on social media networks such as YouTube, linking it to the notion of persona and how this influences the
effects of interpersonal influences by social influencers on social media networks.

10
 The literature review in chapter 2 commences with a discussion on interpersonal relations online,
incorporating the concepts of word-of-mouth (WOM) communications online called electronic
word-of-mouth (eWOM), whereby the concept of the social influencer is introduced as a person using
eWOM to influence consumers. Subsequently the social influencer is discussed as to who they are and
how they can be defined, based on existing theories on market mavens as researched by Clark and
Goldsmith (2005). Furthermore the professionalization of social influencing, as well as the reasons for this
development, are discussed in light of the YouTube platform and its monetization. The concepts of
consumer-buying intent and the social influencer persona are explored and based on the reviewed
literature hypotheses have been formulated.
Chapter 3 will subsequently discuss the process of how research was conducted, including how
the survey has been created. The results of the analysis conducted on the gathered data are reported and
discussed in chapter 4. These results are continuously compared to the literature reviewed and
hypotheses formulated in earlier chapters. The thesis will be concluded in chapter 5 where all findings,
results, implications and limitations of the study are highlighted.

11
12
2. Theoretical framework
This chapter starts with a discussion on what social influencing entails, supported by the concepts of
electronic word-of-mouth communications and user-generated content. Following social influencers are
discussed in light of theory on market mavens, following to Clark and Goldsmith (2005). Social influencing
has developed itself as a profession, which is discussed accordingly. What follows is a close look at
previous studies on social influencing. The third section in this chapter is dedicated to the concept of
consumer-buying intent, which will first be defined, after which it is discussed how other studies have
conceptualized it. Discussion will then flow into the characteristics of the consumer that, as according to
Khaniwale (2015), can also influence the buying intent of consumers, finalized with arguments by Gillin
(2007) on the importance of source credibility. The final section will discuss the impact of the social
influencer persona in relation to the concept of para-social interaction (Lee & Watkins, 2016) and the Big
Five personality model, a well-established model in persona studies (Gosling, Rentfrow, & Swann Jr.,
2003). Throughout the chapter hypotheses are formulated, based on the main research question as posed
in chapter 1, which will be answered in this thesis.

2.1 Beauty vloggers as social influencers

2.1.1 Interpersonal influence online

Word-of-mouth (WOM) is concerned with the informal communication of information on products and
brands from consumers to other consumers (Hennig-Thurau & Walsh, 2004). Electronic word-of-mouth
(eWOM) is exactly this, however communication is performed online. Even though eWOM can take place
in many forms and shapes such as on blogs, forums and in news groups, it is often distributed via social
media. The effect eWOM has on consumer-buying behavior is recognized by scholars and marketers in
different sectors (Vermeulen & Seegers, 2008; Chu & Kim, 2011; Fan & Miao, 2012). The amount of
interpersonal influence that can be achieved through eWOM makes it an interesting field of study. EWOM
is often exercised through the use of user-generated content (UGC), which refers to content that is
created by consumers, rather than by corporations, and allows anyone with access to the Internet to
create and upload his or her own content (Molyneaux, O’Donnell, Gibson, & Singer, 2008). A form of UGC
is for example online vlogs. Vlogs are video blogs created and uploaded to social media channels. It is a
form of online publication by consumers and therefore user-generated communications.
Interpersonal influence refers to a network of influence between multiple people in which one
person exercises influence over another (Friedkin & Johnsen, 2011). New forms of communications such
as eWOM and UGC have led to the emergence of online interpersonal influence, whereby the concept of
social influencers comes in (Walther, et al., 2011). Social influencers are people that act as a third party
and influence consumer attitudes through the use of social media, facilitated by the features of UGC
(Freberg, Graham, McGaughey, & Freberg, 2010). It is the notion of interpersonal influence through which
13
social influencers attempt to influence consumers (Friedkin & Johnsen, 2011).

2.1.2 The market mavens of social media

Social influencers are consumers that offer added value over other consumers, to for example brands,
due to their influence in interpersonal communications. They can be categorized into three groups: the
innovator, the opinion leader and the market maven. Clark and Goldsmith (2005) define the innovator as
someone who adopts new products early, the opinion leader as someone who influences consumer
opinions in a particular product sector, and the market maven as someone who has a lot of information on
many aspects of the market and talks both with and to consumers whereby he or she acts on questions
concerning market information. Market mavens are most attractive to brands because they are the most
likely to spread WOM communications (Clark & Goldsmith, 2005). First, when comparing this
categorization to the concept of social influencers it can be said that a social influencer is relatively similar
to what Clark and Goldsmith (2005) call a ‘market maven’. However, a social influencer can also be an
opinion leader. Clark and Goldsmith (2005) acknowledge this as well and argue that market mavens are
actually generalized opinion leaders. Secondly, while a social influencer cannot be seen as an innovator
only, he or she does adopt products and/or services early to be able to become a social influencer. This is
in line with what the authors argue on all three types of influencers possessing some degree of early
adopter behavior. The main difference between market mavens and social influencers can then be found
in the fact that social influencers operate on social media, while this is not necessarily true for market
mavens. This group is not defined to operate in one single space, and could therefore for example also be
writing columns in a magazine. Social influencers are then defined as the market mavens of social media.
In this study I will focus on a particular group of social influencers: the beauty vloggers.

2.1.3 Professionalization of social influencing

Social influencers have become recognized as a separate social group both by online channels
such as YouTube, but also by companies in search of the best marketing strategies for their brands.
Social influencers across all market segments are increasingly signing contracts with media agencies that
help them organize their personal brands and assist them in establishing promotional contracts and
advertising campaigns with brands; the social influencer thereby becomes an entrepreneur (Duffy & Hund,
2015). Being a social influencer can therefore be seen as a profession rather than as an ordinary
consumer who creates and publishes content. This development is referred to as the professionalization
of social influencing (Fiorella & Brown, 2013).
The development of social influencing as a profession came from two-ways: on the one hand it is
facilitated by the YouTube platform, while on the other hand brands have helped the social influencers to
emerge. The monetization of YouTube started when Google purchased the platform in 2006. A new e-
commerce model was adopted where advertisements were introduced before and during videos, and
revenues of these advertisements were shared with the creators of the video (Kim, 2012). The purchase

14
of YouTube represented a new era in the world of online video; it led to the inclusions of advertising
whereby the lines between advertising and UGC became blurred. From here on YouTube became an ad-
friendly environment where video uploaders started to make money. The economic potential of YouTube
grew larger and attracted the interest of more brands as they recognized a new opportunity to target their
potential customers.
Secondly, marketers actively approach social influencers on YouTube as brand ambassadors for
their own videos (Lee & Watkins, 2016). For example: a make-up brand just released a new type of
mascara. The brand then searches for a social influencer with a similar target group as the brand, and
approaches this person. These partnerships can be based on using the mascara in a video by for example
including it in a tutorial on a make-up look, or the product can be tested by the beauty vlogger in a
product-dedicated video. The first example refers to what we know as product placement in movies. The
aim of product placement is for consumers to unconsciously acknowledge the presence of a brand and
therefore remember it (Gupta & Lord, 1998). This is usually achieved by a product or brand being used by
characters in the movie, without explicitly mentioning it.
In short, the more views a social influencer gets, the higher the revenue one receives from the ads
shown before and during the videos. But also, the more views a social influencer gets, the more interest
this person attracts from brands to partner up for collaborative campaigns.

2.2 Previous research on social influencers

Most studies conducted on social influencers have focused on the social influencers themselves, rather
than taking the perspective of the consumer (Cakim, 2009; Booth & Matic, 2010; Wiedmann, Hennigs, &
Langner, 2010). They focus on the source and the message, the social influencer and their content.
However, as established earlier, the potential influence social influencers exert over consumers, and the
increasing awareness of marketers about associated opportunities, has created a need for knowledge on
the consequences and implications of this influence on the end of the audiences (Lee & Watkins, 2016).
Booth and Matic (2010) have focused on identifying social influencers online by looking at aspects such as
number of views per video, the posting frequency of the videos, and the engagement by looking at the
number of comments and likes the videos received. Wiedmann, Hennigs, and Langner (2010) aim to
identify social influencers in the fashion market, and take a different approach by categorizing them into
three groups based on their level of word-of-mouth referrals: (1) the fashion super spreaders, (2) the
narrative fashion experts and (3) the helpful friends. While both studies by Booth and Matic (2010) and
Wiedmann, Hennigs, and Langner (2010) are beneficial to brands that are looking to incorporate social
influencers in their marketing strategies, it remains unclear why some social influencers are more effective
in influencing consumers than others. Both studies focus on identifying social influencers, however do not
determine their actual influence but study the influencers rather than incorporating consumer responses,
attitudes and reactions. These studies have only conducted content analysis on online available data on
social influencers, such as messages on an influencer’s Facebook page. They hereby do not incorporate

15
the opinions of the target group by asking consumers how they perceive the people that are supposedly
influencing them, and collecting their perspectives. This study will therefore attempt to close this literature
gap through researching social influence directly from the perspective of the consumer by asking the
consumer how they perceive the persona of social influencers and how these affect their purchase
intentions. Consequently it will yield findings that are not just grounded in data accumulation on perceived
influence, but on rather on customer perceptions and empirical data. Furthermore the focus will be on one
specific type of social influencer: the beauty vlogger.

2.3 Consumer-buying intent

2.3.1 Defining buying intent

The aim of this study is to explore to what extent buying intentions of consumers are influenced by the
perceived persona of beauty vloggers. However to determine whether buying intentions are influenced by
the persona of social influencers, it is important to understand what influences buying intent, and how the
concept has been used in previous studies. The Internet has caused a severe increase in information for
consumers to browse through and interpret. This does not only make it difficult for consumers to decide
which information path to follow, but also provides a challenge for researchers to determine how
consumers make their buying decisions and which pieces of information they deem relevant and therefore
use in their decision-making process (Khaniwale, 2015).
Baker and Churchill Jr. (1997) conducted research on the perception and reaction of a consumer
about another person, in relation to physical attractiveness. The authors argue that buying involves three
different decisions: (1) to try a product, (2) to buy a product when you come across it in a store, and (3) to
actively seek out to buy a product in a store. They posed these three questions to their respondents by
asking them to rate their intent on a five-point likert scale. The results of their study show the effectiveness
of these three measurement levels including the reliability of the scale, and therefore their argumentation
is important to consider in assessing buying intentions in this study.

2.3.2 Consumer characteristics

In her study on the theoretical aspect of consumer-buying behavior and the associated factors that
influence this behavior, Khaniwale (2015) identifies four factors that influence buying behavior: (1) cultural
factors, (2) social factors, (3) personal factors and (4) psychological factors. The first two factors are
external, meaning that they are factors within the environment in which a consumer exists that cannot be
controlled by the consumer itself, but that do impact buying behavior. The last two factors are internal and
are concerned with features of a consumer that impact his or her buying behavior, such as motivation or
perception, and that are unique to every individual.
The cultural factor is made up of two different indicators: social class and culture. These factors
are concerned with the consumer’s way of life, and will not be considered for this study. The social factor

16
is mainly concerned with the influence of reference groups on buying behavior; the perceived relationship
with others and the influence of social norms and values. Langner, Hennigs and Wiedmann (2013) agree
with Khaniwale (2015) on the impact of social factors on buying behavior, and in particular stress
reference groups and an important notion. They argue that when consumers experience a relatively high
level of in-group salience, and therefore define themselves as members of a particular social group, they
will adapt to the behavior of this social group.
This self-identification with a social group provides the basis for how people make buying
decisions, and how they look at their social group when establishing buying intensions. Some individuals
within this social group are able to intensify the in-group salience, the social influencers. They can do this
directly by recommending a certain product, or just use a certain brand or product so that other group-
members acknowledge its usage (Langner, Hennigs, & Wiedmann, 2013). Important to note here is that
the social group that is being referred to is not an established group in which participants are familiar with
each other such as one would be in the case of a classroom. It however refers to a sense of community
people feel due to their shared interests in a topic, in this case beauty videos (Higgens, 2000). In this
study the larger social group is the YouTube community in which social influencers post vlogs, and
audiences consume the content. The smaller social group can be defined as the beauty vlogs on
YouTube, which people can identify with and adapt to the behavior of the social influencers in this group. I
will come back to the social factor in chapter 2.6 when the concept of homophily is discussed.
The personal factor that Khaniwale (2015) discusses as impacting buying behavior is concerned
with the demographics of the consumer. Income, age and occupation she argues are important
determinants in the consumer-buying process. Additionally, Coursaris and Van Osch (2016) emphasize
the importance of the characteristics of the viewer as well, but also include the notion of prior knowledge.
This is especially important in this study because the consumers will be faced with stimuli that are
originally posted on YouTube. It then needs to be determined to what extent the respondent is familiar
with YouTube in general, as well as with beauty videos.
The psychological factor includes the perception, motivation, beliefs and attitudes of a consumer.
The perception of a product, as Khaniwale (2015) argues, can differ from person to person, and is
therefore important to consider in establishing buying behavior. The motivation of a consumer to buy
something is part of the psychological factor as well, however will not be considered further in this study
since the aim is to determine the extent of the influence by beauty vloggers, whereby additional reasons
for making a purchase are left out of scope. Finally, beliefs and attitudes are about someone’s positive or
negative evaluations of something, which in the case of this study refers to the attitude of the consumer
towards the beauty vlogger that will be showed. It argues that attitudes towards a person can have an
impact on buying behavior. It is therefore important to establish whether the respondent is familiar with the
beauty vlogger shown to them, as well as the product that is discussed.

17
2.4 The impact of the social influencer persona
The word persona has been derived from the Latin language in which it directly refers to a mask. In
ancient Greek times actors convinced the audience of a particular identity of a played character by using a
mask. It also allowed one person to play multiple characters in the same play (Marshall & Barbour, 2015).
This historical meaning can directly be translated into what the notion persona means today in online
communication environments such as YouTube: the strategic presentation and communication of the self.
Harré (1985) discusses the notion of strategic rhetoric as a form of performance wherein choices of how
someone presents him or herself are made based on the way of talking, writing, expressed behavior and
appearance. This form of performance is strategically chosen by a beauty vlogger in constructing a
persona. It is possible for a vlogger to strategically create a persona because the interpersonal
communication on YouTube is asynchronous, meaning not live communication as in face-to-face
interaction, allowing someone to selectively present him or herself. Also, vlogs are unidirectional, meaning
that the audience cannot directly interact with the vlogger (Frobenius, 2004). A persona is then
deliberately constructed by a vlogger, rather than an expression of whom someone really is. As mentioned
in the introduction, a persona is therefore not grounded in biological features, because these cannot be
adapted, but rather in the character traits that are conveyed and expressed in in this case, beauty vlogs
(Marshall & Barbour, 2015). The persona of a beauty vlogger is thus a construct of different character
traits whereby he or she presents itself in vlogs on YouTube.

2.4.1 Source credibility

Additionally to the factors Khaniwale (2015) argues to impact buying behavior, studies on the influence of
eWOM, such as on beauty vloggers on YouTube, have acknowledged another factor that influences
buying intentions by consumers based on the source of the influence: source credibility (Martawilaga &
Purwanegara, 2016; Ananda & Wandebori, 2016). The credibility of the source refers to the perceived
ability and motivation of the beauty vlogger to produce information that is both truthful and accurate.
Martawilaga and Purwanegara (2016) argue that the more credible a beauty vlogger is perceived to be by
consumers, the stronger the impact on consumer-buying behavior is. Additionally, Martawilaga and
Purwanegara (2016) discuss the impact of perceived source credibility on information acceptance and
perceived usefulness of the information that is communicated, which is out of the scope of this study and
will therefore not be considered further.
But how can it be determined whether a source is perceived as credible? According to Gillin
(2007), who wrote a book on the new influencers that emerged with the rise of social media, influencers
possess five characteristics: they (1) have great market knowledge, (2) want to be involved, (3) are on top
of transparency towards their viewers, (4) look for discussion and (5) are concerned with their audience. In
his book ‘The Enthusiasts’ Gillin (2007) assumes social influencers are found a credible source of
information. Therefore it can be said that when an influencer possesses the above characteristics, he or
she is a credible source. It should be noted that Gillin (2007) is referring to influencers that do not earn
18
money, and are therefore not social influencers by profession. Despite this fact it can be assumed that his
results are applicable to professional social influencers since they are assumed to be more active then
non-professional influencers.
As established in this chapter, source credibility is related to buying behavior. Based on
arguments by Martawilaga and Purwanegara (2016) and Ananda and Wandebori (2016), supported by
Gillin (2007), the following hypothesis has been formulated:

Hypothesis 1: Perceived source credibility has a positive influence on consumer-buying intent

2.4.2 Para-social interaction (PSI)

As mentioned earlier, Lee and Watkins (2016) have studied how vlogs influence consumer’s perceptions
on luxury brands in the United States. In their study they used the concept of para-social interaction (PSI)
to conceptualize the relationship between vloggers and their audiences, which amongst others is indicated
by social and physical attractiveness. Social attractiveness refers to a vlogger being perceived as
interesting to talk to or interact with, while physical attractiveness refers to a vlogger having appealing
looks.
Another important notion Lee and Watkins (2016) raise awareness of is the need for a vlogger’s
persona to be close to the persona of the viewer. This notion was included in their study by the concept of
homophily, indicating the degree to which the vlogger in question has similar traits, and can therefore be
perceived as the reference group of the consumers that experience high levels of in-group salience, based
on Khaniwale’s (2015) arguments that influence intensifies when the consumers feels comparable to the
social influencer, and is therefore an important factor. A question posed by Lee and Watkins (2016) to
measure this dimension was for example how the respondent believed the beliefs of the vlogger and him
or herself were similar. For example the believe that testing make-up products on animals is wrong. When
a consumer perceives a vlogger to have the same believes, he or she will score high of the dimension on
homophily. All measurements were taken on a seven-point likert-scale. Based on the results from the
study by Lee and Watkins (2016) the following hypothesis has been formulated:

Hypothesis 2: Perceived (a) social attractiveness and (b) physical attractiveness both have a
positive influence on consumer-buying intent
Hypothesis 3: Perceived homophily has a positive influence on consumer-buying intent

19
2.4.3 The Big Five model
A well-known and often used model for assessing personality and therefore character traits is the Big Five
model (Becker, 1999; Bilsky & Schwartz, 1994). The model organizes personality traits around five
dimensions: extraversion, agreeableness, conscientiousness, neuroticism and openness to experience.
Each dimension refers to its own associated personality traits such as active, energetic and outgoing for
the dimension of extraversion (Biel & Gatica-Perez, 2013). An overview of the associated adjectives as
Biel and Gatica-Perez (2013) formulated them can be found in table 2.1.

Table 2.1: overview personality dimensions and associated adjectives

Personality dimension Adjective
Extraversion Enthusiastic, energetic, active and assertive
Agreeableness Kind, sympathetic and forgiving
Conscientiousness Efficient, organized, responsible and thorough
Neuroticism Anxious, tense, touchy and unstable
Openness to experience Curious, imaginative and original

The Big Five model has found considerable support within the research community as being
relevant in studying personalities (Gosling, Rentfrow, & Swann Jr., 2003). Recently several studies have
applied the model to the online environment of social media to study the personality of vloggers on
YouTube and the audience responses to these personalities (Biel & Gatica-Perez, 2013; Farnadi, et al.,
2014), indicating its applicability in this new social context as well. Biel and Gatica-Perez (2013)
researched personality impressions on vlogs obtained from YouTube to determine the impressions by the
audience, the social attention, and the audiovisual behavior of the vloggers, with the use of the Big Five
model. Biel and Gatica-Perez (2013) focused on the amount of attention that a vlogger received from the
audiences as a determinant of influence, whereby they assume that more attention leads to more
influence. For the purpose of this study, and in line with previously discussed literature on social
influencers, I subsequently assume that more influence leads to increased buying intent.
The respondents in the study by Biel and Gatica-Perez (2013) were shown an edited YouTube
clip, containing one minute in which the vlogger talks, and were asked to assess the vlogger based on the
dimensions of the Big Five model. For each dimension the respondents were given the adjectives
associated with that particular dimension, as displayed in table 2.1, and were then asked to rate the
vlogger on these adjectives, on a seven-point likert scale. Biel and Gatica-Perez (2013) subsequently
compared the dimensions of the Big Five model against the average ratings vloggers received on their
videos. The results of their study showed that vloggers who scored high on the dimensions of openness to
experience and conscientiousness received more attention from their audiences. Additionally, the results
showed that vloggers who were perceived as agreeable received more attention when they were rated in
extremes. This implies that vloggers who are rated either very agreeable or very disagreeable get more
attention. The results also showed that extraversion is not a determinant because of the weak correlation
with the attention measures (r = .24).
20
 Both high and low extraverts received high ratings, so no difference between levels of
extraversion was found. Perceived neuroticism, which Biel and Gatica-Perez (2013) present as emotional
stability, was not found to have significant impact on, and a negligible association with, increased attention
for vloggers. This dimension is therefore out of the scope of this study and will not be considered further.
As mentioned earlier, it can be assumed that increased attention and therefore influence leads to
increased buying intent. Applying the conclusions of the study by Biel and Gatica-Perez (2013) the
following hypotheses can be drawn:

Hypothesis 4: Perceived openness to experience has a positive influence on consumer-buying

intent
Hypothesis 5: Perceived conscientiousness has a positive influence on consumer-buying intent
Hypothesis 6: Perceived agreeableness only positively impacts consumer-buying behavior when it
is perceived either very agreeable or very disagreeable
Hypothesis 7: Perceived extraversion has no impact on consumer-buying intent

21
22
3. Research design
In this chapter the method used in this study will be explained, including a description of how the data was
collected, why this design was chosen as well as its advantages and disadvantages. Following is a
discussion on the research units and sampling. The second section is concerned with the
operationalization of the concepts of the research question, and is divided into the two largest
stakeholders in this study: the consumer and the beauty vlogger. Finally the design is discussed in terms
of the pretest that was conducted prior to the main study, and an overview of the survey is given.

3.1 Method
This study seeks to find out to what extent the buying intentions of consumers are influenced by the
persona of beauty vloggers, and is therefore explanatory. The results are used to compare between the
personas of the different social influencers included in the study, and the buying intentions of the
consumer.
YouTube is chosen as the platform to best analyze social influencers due to the business potential
of the platform that is created by the different advertisement possibilities, which are discussed earlier in
relation to the professionalization of social influencing in chapter 2.1.3.
The research focuses on the Dutch market with the inclusion of Dutch social influencers as stimuli
in the survey, and by seeking out Dutch respondents. This decision was made because social influencers
primarily discuss products that are for sale in the country they reside in. By including social influencers
from multiple countries, or respondents from different countries, the risk would arise that the product
discussed is not for sale in the country the consumer resides in. The choice was therefore made to focus
on the country of residence of the researcher. Additionally, Lee and Watkins (2016) argue that the topics
of social influencing and consumer-buying behavior can best be studied when the social influencers in
question and the respondents speak the same language.
The study was performed on a cross-sectional basis and therefore conducted at one point in time.
Important here is to note that a snapshot of the situation was taken, and that any developments around
the beauty vloggers or the platform YouTube before or after the data was gathered has not been
incorporated into the results (Babbie, 2014).

3.1.1. An experimental survey

The research question was studied and finally answered with the use of a quantitative survey through
which data was systematically gathered and subsequently analyzed with the data analysis program SPSS.
The program SPSS allowed the researcher to calculate the influence of one variable over another, as well
as the extent of that influence, which is the purpose of this study. SPSS is chosen over other programs
due to its prevalence in the Erasmus School of History, Culture and Communication, for which this thesis
was written.
23
 By grounding the items within the survey in literate discussed in chapter 2, the survey was created
deductively. This increased the content validity of the research because of the proven effectiveness and
standardization of measurement procedures as they were used in previous studies on similar topics
(Hinkin, 1998).
This study was approached quantitatively due to the standardization ability of this type of study.
The aim was to conduct an experiment in survey design to discover which aspects within the persona of a
beauty vlogger influence consumer-buying intentions, and also to what extent. The survey design enabled
the standardization of questions so that every respondent was asked the exact same questions. This
enabled later comparison of the different experimental conditions to find patterns.
In the survey respondents were confronted with different conditions, whereby stimuli differed
between subjects, making the method quasi-experimental (Gravetter & Wallnau, 2013). By stimuli I refer
to the different beauty vloggers that were shown to the respondents in the form of shortened YouTube
videos. The survey design enabled the random assignment of the respondents to the different conditions.
This random assignment of the different conditions also ensured that all variation in the results could
possibly be attributed to the stimulus (Juster, 1966).
As it is a common method in market- and behavioral research, the survey was administered online
(Kaplan, 2004; Hinkin, 1998), whereby a geographically widely dispersed audience could be reached. This
advantage resulted in the inclusion of larger diversity in the respondents, increasing the quality of the
conclusions that are drawn from the findings (Van Selm & Jankowski, 2006).
However, administering the survey online can also be a disadvantage because the respondents
do not know who the researcher is. On the other hand this anonymity can be an advantage as well
because the online environment eliminates the social context in which the survey is created, and thereby
standardizes interaction between the researcher and the respondents. This creates an anonymous
environment in which respondents are less concerned with the impressions they make on others, and
therefore the responses gathered are more honest than responses would be from a paper survey that is
filled out in front of the researcher (Kiesler & Sproull, 1986).
Web-based surveys can however result in a low or poor response rate because of participants
that drop out halfway through the survey. An e-mail based survey would allow for sending the potential
respondents reminders, however because the survey used for this study is primarily distributed via social
media channels with an anonymous link, there is no list of names to send reminders to. This is a strong
disadvantage for this method and provides a risk to obtain enough valid responses (Brennan, Rae, &
Parackal, 1999). However, the topic of the research is concerned with YouTube, an online video
distribution channel. Administering the survey online enables the inclusion of YouTube videos as stimuli
as well. Together with the absence of an interviewer, who could have possibly had an impact on the
subject whereby answers are given differently, the online survey was found the most suitable method for
the purpose of this study (Couper, Traugott, & Lamias, 2001).

24
3.1.2 Sampling
According to the Dutch foundation for scientific research in commercial communications, online vlogs are
most frequently watched by youth between the ages of 18 and 34 (Stichting Wetenschappelijk Onderzoek
Commerciële Communicatie, 2017). Because the topic of this study is concerned with beauty vloggers as
social influencers, a requirement for the respondents is to be female. The group of males that are
interested in beauty-related videos is too small for consideration in this study and is therefore excluded.
The units of analysis for this study are then defined as Dutch women between the ages of 18 and 34.
Respondents that have participated in the pre-test will be excluded for participation in the survey. They
have prior knowledge on the different stimuli that are used in the survey, which could influence their
responses and affect the results.
Respondents were recruited in 3 ways: on a word-of-mouth basis via personal networks, via
forums and via social media. My personal connections were used as a distribution network for the survey.
Within this network several connections were asked to distribute the survey within their own networks; for
example within student associations, sport clubs and other universities. The survey was also distributed
via a range of forums on websites that focus primarily on females. For example, the survey was posted on
the forum of Girlscene.nl, a Dutch website aimed at females between the ages of 14 and 26 who are
interested in fashion, beauty and lifestyle (EEN Media, 2017). Another example of where the survey was
distributed is the VIVA forum, which is part of the VIVA magazine and aims at Dutch women between 25
and 45 (MediaBookers, 2017). Finally, respondents were recruited via a targeted Facebook
advertisement. This advertisement was created to appear on the Facebook pages of Dutch women
between the ages of 18 and 34, the units of analysis. It cannot be said how many respondents were
recruited via which of the three recruitment methods because an anonymous link was used to distribute
the survey on the different platforms.
Because the survey was distributed outside the network of the researcher it was not possible to
fully control the participation of respondents younger or older than the set units of analyses of 18 and 34.
However, the survey was distributed with the slogan: ‘what do you think about Dutch beauty vloggers?’.
This implies that people who clicked on the link and filled out the survey are at least interested in online
vlogs, even though they do not fall within the age category that Stichting Wetenschappelijk Onderzoek
Commerciële Communicatie (2017) stated as most frequently watching online vlog. Responses from
people that are either younger than 18, or older than 34, are therefore included in the study.
In determining the sample size for this study the requirements as stated in the ‘Methodological
guidelines Thesis Research 2015-2016’ were taken under advisement, which stated a sample of between
150 and 250 respondents. Comparable studies have acquired around 300-400 responses (Lee & Watkins,
2016; Chu & Kim, 2011). However, because this study is quasi-experimental the sample size has been set
at a lower level and aimed at reaching between 100 and 150 responses. Because each respondent was
asked to watch four YouTube clips, and answer 15 questions per beauty vloggers, the length of the survey
was expected to influence the response rate. The choice was made to include two beauty vloggers per
25
survey version to enable comparison between ratings for different beauty vloggers by the same person.
The aim was to reach the maximum number of respondents, despite this fact.
116 individuals filled out the survey. Out of these 25 stopped halfway through the survey, meaning
that they only rated 1 vlogger instead of two. 91 people filled out the full survey. It was decided to also
include the responses by the 25 people who only rated one beauty vlogger because they fully rated one
vlogger, making this a complete and valid piece of data. Responses that were not complete – those that
did not complete the full rating of one beauty vlogger – were excluded.
Each respondent was given a unique response ID to indicate which ratings the same respondent
produced. Subsequently the data was restricted to display the rating of one vlogger per row. This resulted
in a data sheet with 207 units of data.
It was not necessary for respondents to have any experience with beauty vlogs. However,
because it is relevant for the results whether they do, a topic in the survey was dedicated to the
experiences of the respondents with YouTube in general as well as beauty vlogs, as Coursaris and Van
Osch (2016) have established the importance of prior knowledge in predicting buying behavior.

3.2 Operationalization
This study aimed at finding out to what extent buying intentions are influenced by the perceived persona of
beauty vloggers. The dependent variable is the consumer-buying intention. The independent variables are
concerned with the persona of the beauty vlogger, as well characteristics of the consumer that were
discussed in the chapter 2 and defined to impact buying behavior.
Buying intent is defined as how likely an individual consumer is to purchase a product, which is
measured by the consumer’s likeliness to buy a product. Following Baker and Churchill Jr. (1997) in how
they operationalized the concept buying intention in their study, it is measured on three levels: trying the
product, intention to buy and actively seeking out to buy the product. Respondents are asked to rate their
likeliness of the three different questions on a five-point likert scale. To determine whether these three
separate items together form a reliable scale for buying intent a Cronbach’s alpha test was conducted. It
showed that the items together form a scale that is a reliable measure of buying intent, Cronbach’s alpha
= .915. Therefore, a new variable is created to reflect this scale, computed out of the means of the three
separate items measuring buying intent.

3.2.1 Consumer characteristics

All variables that will be discussed in this section are concerned with the characteristics,
experiences and levels of familiarity of the consumer. Additionally to the beauty vlogger persona and
based on arguments by Khaniwale (2015) and Coursaris and Van Osch (2016) as discussed in chapter 2,
these variables are expected to have an impact on how a consumer makes buying decisions. It is
therefore necessary to control for these variables when any analyses are conducted, to ensure these
effects are taken into account.

26
 The characteristics of the consumer are measured according to the factors that Khaniwale (2015)
argues to influence buying behavior: the personal factor and the psychological factor. The personal factor
refers to a consumer’s demographics: age, educational level, income and occupation.
Because the units of analysis are originally defined as Dutch women between the ages of 18 and
34, the respondents are expected to include people across all occupations, ranging from students without
jobs of the side to people who are self-employed. However, establishing income levels by asking
respondents to indicate their income in amount of money can here for prove difficult. Students without a
job on the side for example might not earn money, but borrow money to finance their studies, which might
officially not be a form of income. For the purpose of this study income level will therefore be measured
according to the perceived financial situation of a respondent. Respondents are therefore asked to
indicate whether they think they are (a) struggling, (b) maintaining, (c) just comfortable, or (d) very
comfortable in terms of their personal finances.
The psychological factor as Khaniwale (2015) discussed it refers to a consumer’s perception,
motivation, beliefs and attitudes about a person and a product. The psychological factor is therefore
operationalized as the familiarity of a consumer with (1) the beauty vlogger and with (2) the product
discussed by the beauty vlogger. Respondents are asked to indicate whether: (a) yes I am familiar with
the product, (b) I have heard about the product but that is all, and (c) no I am not familiar with that product.
Additionally, the dimension of prior knowledge that is argued by Coursaris & Van Osch (2016) as
relevant in determining buying intent is measured by asking the respondent what prior experience she has
with YouTube in general and with beauty vlogs on YouTube. Answering categories for both questions are:
I watch YouTube/beauty related videos on YouTube (a) everyday, (b) several times per week, (c) once per
week, (d) less than once per week and (e) never.
As discussed at the beginning of this section, the variables that measure the personal factor need
to be controlled for when the hypotheses are tested to take into account that these can also influence
consumer-buying intent, additionally to the persona of the beauty vlogger. Even though all variables seem
to be measured on a nominal level, where the answering categories represent group classifications, they
can also be perceived as ordinal variables. The variable age is measured on a ratio level. From this
variable four age categories will be created so that the variable can be used in analyses. Additionally, I
would like to assume that there no linear relationship between the different age categories. By assuming
non-linearity the strength of the effect of each separate age category on consumer-buying intent can be
tested without assuming that that the effect is dependent on age increasing or decreasing. To facilitate this
dummy variables are created for each age category. For each respondent it is then indicated whether they
fall within that age category or not. These dummy variables are then used as the control variables in the
analyses as conducted in chapter 4.
However, there is a problem with the variable occupation, which is not measured on an ordinal
level. There is no meaningful ranking possible between working for a boss or being self-employed. This
variable will therefore be excluded as a control variable and will only be used in the descriptive analyses.

27
The control variables therefore include: age (dummy variables), educational level, financial situation,
YouTube experience, beauty video experience, vlogger familiarity and product familiarity.

3.2.2 Beauty vlogger characteristics

The concept persona of the beauty vloggers is conceptualized based on findings by Gillin (2007), Lee and
Watkins (2016) and Biel and Gatica-Perez (2013) into the following dimensions: source credibility, social
attractiveness, physical attractiveness, homophily, openness to experience, conscientiousness,
agreeableness and extraversion. As mentioned before, because vlogs are used as stimuli in which a
vlogger has deliberately constructed his or her persona (Marshall & Barbour, 2015), the notion of persona
has for the purpose of this study been defined as the character traits one conveys, rather than possesses.
For each dimension respondents are given statements and are asked to indicate to what extent they
agree with them or not. The statements are formulated based on the associated adjectives per persona
dimension as formulated by Biel and Gatica-Perez (2013) and shown in Table 2.1. All persona dimensions
are measured on a seven-point likert scale, following Lee and Watkins (2016), and are therefore
measured on an ordinal level. This section discusses per persona dimension through which indicators
they are measured, and tests the reliability of the overall scale. If the reliability of the scale is proven
reliable, new variables are created to represent the overall dimensions/scales based on the means of the
separate items. Additionally, by following established procedure and valid indicators used by other
researchers in similar studies, as they were discussed in chapter 2, the validity of the measurements is
increased. It can then be assumed that the instruments that will be discussed next measure the concepts
they intent to (Field, 2013).
Source credibility is measured according to the five characteristics that Gillin (2007) found to be
present in influencers: (a) having great market knowledge, (b) wanting to be involved, (c) being on top of
transparency (d) looking for discussion and (e) be concerned with their audience. These items together
are found to be a reliable scale of source credibility, Cronbach’s alpha = .835, and therefore a new
variable is created to represent this scale.1
Following the procedures by Lee and Watkins (2016), the second dimension of social
attractiveness and physical attractiveness is measured using the determinants of perceived (a) kindness,
(b) being friends and (c) conversing with someone to measure the perceived social attractiveness, which
was tested on reliability and found reliable, Cronbach’s alpha = .851. Physical attractiveness is measured
on the indicators of someone (a) looking good, (b) being sexy, and overall being perceived as (c)
physically attractive, a scale that is found reliable as well, Cronbach’s alpha = .866. Due to the reliability of
both scales new variables were created to represent them.
Homophily is measured based on research by Lee and Watkins (2016) by asking the respondents
whether the consumer believes that the vlogger (a) thinks the way she would, (b) shares the same values,
(c) has a lot in common with her and (d) whether this person is a lot like the respondent. The four items

1
The test results showed that deleting the first item ‘market knowledge’ would lead to a slight improvement in reliability, Cronbach’s alpha = 0.843.
However, the correlation between item 1 and the others is moderate (r = .490). Item 1 is therefore included in the scale. 28
were tested for reliability and found a reliable measure of homophily, Cronbach’s alpha = .893, and
therefore a new variable is created the represent this scale.
Openness to experience is measured based on the perceived (a) curiosity, (b) imagination and (c)
originality of the vlogger, while conscientiousness is measured on the perceived (a) efficiency, (b)
organization, (c) responsibility and (d) thoroughness of the vlogger by the respondent, as these are the
associated adjectives for these dimensions in the big five model according to Biel and Gatica-Perez
(2013). Both openness to experience, Cronbach’s alpha = .801, as well as conscientiousness, Cronbach’s
Alpha = 0.855, are found reliable scales and therefore new variables are created based on these scales.
The dimension of agreeableness, following the big five personality model (Ananda & Wandebori,
2016; Biel & Gatica-Perez, 2013), is measured on the perceived (a) kindness, (b) sympathies, and (c)
forgivingness of the beauty vlogger by the respondent. Results of the reliability test show that the scale is
not a very reliable measure for agreeableness, Cronbach’s Alpha = .644. Deleting any of the three items
will however not result in a larger alpha and therefore the new variable is computed based on this scale.
Also based on this model is the measurement by the final dimension of extraversion, using items
on the perceived (a) enthusiasm, (b) energy, (c) activity and (d) assertiveness of the vlogger by the
respondent to measure it. This scale was tested and found a reliable measure of extraversion, Cronbach’s
Alpha = 0.872. Subsequently a new variable is computed based on the mean scores of extraversion.

3.3 Design

3.3.1 Stimulus selection and pretest

The beauty vloggers were selected based on two criteria: (1) the beauty vlogger needs to upload Dutch-
spoken videos and (2) the beauty vlogger needs to have at least 60,000 subscriptions to ensure the
professionalism. Entering the following search strings on YouTube resulted in a list of beauty vloggers
matching these criterions: ‘beauty vlogger Nederland’ and ‘beauty vlogger Nederlands’. For each beauty
vlogger on the list it was then decided whether she indeed discusses beauty topics, by looking at the
‘about’ sections on their YouTube channel pages. When it stated that beauty topics are discussed in the
video, the beauty vlogger was included as a stimulus. This resulted in a selection of 14 Dutch beauty
vloggers. Table 3.1 shows an overview of the selected beauty vloggers including the real names of the
vloggers, the names of their YouTube channels, and the number of subscribers. Please note that the
number of subscribers was retrieved in February 2017.

29
 Table 3.1: selected beauty vloggers
Name person Name YouTube channel Number of subscribers
1 Mascha Feoktistova Beautygloss 552,626
2 Joy BeautyNezz 475,103
3 Diana D for Dazzle 259,910
4 Serena Verbon. Beautylab 238,696
5 Vera Camilla VeraCamilla 226,226
6 Jiami LifeSplash 191,774
7 Bibi Breijman Bibi Breijman 166,848
8 Laura Ponticorvo LauraPonticorvo 123,096
9 Anna Nooshin Anna Nooshin 118,539
10 Jessie Maya Jessie Maya 96,766
11 Willemijn & Martine LiveLifeGorgeous 70,268
12 Sara Ras Saar 66,664
13 Cynthia Schulz MissLipgloss 66.236
14 Sheling Kamkes ShelingBeauty 62.076

7 Survey versions were created in total, each including 2 different beauty vloggers. Following the
method used by Freberg, Graham, McGaughey, and Freberg (2010), 2 videos were selected from each
beauty vlogger. In one video the social influencer introduces herself. In the other video the beauty vlogger
is testing a product. If a beauty vlogger had an introduction video available on their channel it was used.
When an introduction video was not present a Q&A video was used. In a Q&A video vloggers answer
questions about themselves and their lives, that are send in by their viewers. This gives a good impression
of who they are. Therefore this type of video is an appropriate substitute for the introduction video. Both
videos were edited and shortened to last between 2 and 3 minutes. This was done to ensure that the
respondent would get a good impression of the beauty vlogger while at the same time avoiding
respondents to drop out off the survey because it takes them too long to watch the full videos.
Each survey version includes two beauty vloggers whereby the aim was to achieve diversity
between them. This was achieved by a pretest that was conducted prior to the creation of the survey in
which 5 female subjects between the ages of 18 and 34 assessed each beauty vlogger based on their
personality traits by watching 1 video of each of them. The introduction video, or a Q&A video, was used
for this. In the pretest the respondents were asked to rate the vloggers on the same dimensions of
persona as used in the main research: social attractiveness, physical attractiveness, homophily, openness
to experience, conscientiousness, agreeableness, extraversion and source credibility. For the purpose of
the pretest the personal factor was been left out entirely, because it was out of the scope of the pretest.
The pretest in which 5 respondents rated 14 beauty vloggers resulted in a dataset of N = 70.
Based on the data that was collected in the pretest, the beauty vloggers were grouped together
according to their difference in persona. Following it was determined how the beauty vloggers scored on
average per dimension by calculating index scores. For each dimension it was then determined which
vlogger scored the highest and the lowest. For example, which beauty vlogger was perceived as the most
socially attractive, and who the least. An overview of the seven different groups of beauty vloggers can be
found in appendix C. These highest and lowest scoring beauty vloggers were grouped together, whereby
diversity between the personas of the beauty vloggers in each survey version was achieved.
Subsequently diversity was then also created in the different conditions to which the respondents in the
main study were going to be exposed. However, sometimes the highest or lowest scoring beauty
30
vlogger was already grouped with another beauty vlogger. Then the second highest or lowest scoring
beauty vlogger was selected for grouping, and so on. An overview of the results from the pretest can be
found in appendix B.
A pretest is a valid method to prepare the data collection process and in this case, determine
categories of stimuli (Ghauri & Grønhaug, 2005). This pretest resulted in a persona impression for each of
the beauty vloggers, which was then used to create diverse groups. Due to this design each respondent in
the main research was faced with different types of vloggers, which allowed for the comparison between
perceived persona and buying intentions of different personas.

3.3.2 Survey design

The purpose of the survey was to obtain spontaneous impressions of the respondents, and therefore the
only instructions given prior to filling out the survey was to watch the full videos and to answer the
questions based on the videos (Biel & Gatica-Perez, 2013). To capture any possible prior knowledge of
the respondent with the product or the beauty vlogger questions on familiarity were included in the survey,
which will be discussed later.
The survey commenced with questions concerning the personal factor. Questions were asked
about the respondent’s age, educational level, occupational status and perceived financial situation. Even
though the sample is supposed to include only females, a question on gender was included to enable the
exclusion of any males who filled out the survey afterwards. Additionally, it was important to establish the
experience of the respondent with YouTube videos in general and specifically beauty videos. Therefore,
each respondent was asked to indicate how often she watches YouTube videos general, and how often
she watches beauty-related videos. Following, the respondent was asked whether she is familiar with the
first vlogger she was going to see, and whether she is familiar with the first product that was going to be
discussed.
The respondents were then shown the first clip: the introduction video of the first beauty vlogger.
After this video the respondent was asked to rate the vlogger’s social attractiveness (3 items), the physical
attractiveness (3 items), and the homophily (3 items).
Following, the second clip was shown: the product testing video of the same beauty vlogger as in
clip 1. After the second video questions of openness experience (3 items), conscientiousness (4 items),
agreeableness (3 items) and extraversion (4 items) followed. Subsequently respondents were asked to
rate the perceived credibility of the source (5 items), and finally questions of buying intent were posed (3
items).
The process described in this section is then repeated for the second beauty vlogger. The survey
as distributed into the population can be found in appendix A.

31
32
4. Results
This chapter will discuss the results of the research as well as linking them to the theory that was reviewed
in chapter 2. First an overview of the type of consumers that responded to the survey is offered by
providing descriptive information on the dimensions that measure the personal factor (Khaniwale, 2015):
age, educational level, occupational status, financial situation, experience with YouTube, experience with
beauty videos, vlogger familiarity and product familiarity. Secondly, each hypothesis as it was formulated
in chapter 2 is tested. The aim of the study is to find to what extent buying intent is influenced by
perceived persona of the beauty vloggers, and therefore OLS regression analyses are run for each
hypothesis. However, prior to this correlation analyses are conducted to find whether there is any
association between the persona dimensions and the variable buying intent to determine whether OLS
regression analyses are required. Additionally, the data was checked against the following assumptions
for OLS regression according to Field (2013): (1) the variables used are continuous, (2) there is a linear
relationship between the two variables, (3) there are no significant outliers, (4) observations are
independent, (5) the data is homoscedastic and (6) the residuals are normally distributed. The
assumptions were checked by examination of the histogram, normal p-p plot and scatterplot and were
found to be satisfied. After all persona dimensions are separately tested according to the hypotheses, an
additional OLS regression model including all persona dimensions is conducted to determine the relative
impact per dimension. Finally, to take into account that most of the respondents rated two different beauty
vloggers, a linear mixed model with the respondent as the subject is run.

4.1 Descriptive results

The data showed that the variable age was widely dispersed, and has a standard deviation of 6.25. Most
respondents are 23 years old (18 respondents), followed by the respondents who are 21 years old (12
respondents) and who are 17 years old (11 respondents). 4 respondents did not indicate their age.
Because the deviation is so large, age categories were created to be able to determine in the statistical
analyses whether the influence of this control variable is larger for
one age category than for another. The age categories are not
proportionally developed, which means that in the youngest age
category 13 – 17 there is a difference of 4 years, while the
difference between the oldest age category 35 – 28 is 13 years.
This is done on purpose because the largest group of respondents
is between 13 and 23, and by dividing this group up into two
categories the difference between respondents in these age
categories becomes clear. Additionally, only a small proportion of
respondents fall into the largest age category making it
unnecessary to further specify them into smaller categories.
33
Figure 4.1: age distribution of all
respondents per age category
Additionally, I would also like to assume that there is no linear relationship between the different age
categories. To account for this assumption dummy variables were created out of the four different age
categories. These dummies will be used in the analyses to determine per age category, independently
from each other, whether there is a significant effect or not.
Looking at the education level the results show that most respondents indicated to have only
finished their high school (35.3%). However, what should be taken into account is that the question asks
the respondent about the highest level of education that has been completed. A respondent can therefore
still be studying. To find how many of the respondents are still studying a crosstab was created with the
independent variable education, and the dependent variable occupation. The results show that 85.3% of
the respondents that indicated to have a high school diploma as their highest obtained degree, are still
studying.
It is also important to consider the financial situation of the respondents when looking at buying
behavior and what the factors that influence it. Someone might not be able to afford the product, and
therefore be less intended to buy. The data however shows that most respondents are ‘just comfortable’
(39.7%) or ‘maintaining’ (34.5%).
The level of experience a respondent has with YouTube in general could affect buying intent as
well, as well as the experience with beauty videos on YouTube prior to participating in this study. The
results show that the majority of the respondents watch YouTube everyday (36.2%), or several times per
week (30.2%). Beauty videos are however watched less often, and the majority of the respondents
indicated that they watch these types of videos less than once per week (35.3%). But when looking at the
overall picture we see that out of the total 116 respondents, 93 (80.1%) do watch beauty videos.
Each respondent was confronted with two different beauty vloggers. Because as mentioned in
chapter 3.1.2., out of the 116 individuals that responded to the survey, 25 only rated one vlogger. Also,
there is a slight difference between the number of times each vlogger is rated. BeautyGloss, BeautyNezz,
ShelingBeauty, D for Dazzle, BeautyLab, Vera Camilla, LifeSplash, Anna Nooshin, Jessie Maya and Saar
are rated 15 times, Bibi Breijman is rated 16 times, and Laura Ponticorvo, LiveLifeGorgeous and
Misslipgloss 14 times. Each beauty vlogger was rated on the eight persona dimensions: source credibility,
social attractiveness, physical attractiveness, homophily, openness to experience, conscientiousness,
extraversion and agreeableness. Table 4.1 shows how each vlogger was rated on average per dimension.
The table shows that beauty vlogger Saar is perceived as the least credible source (4.38) and Vera
Camilla the most credible (2.84). BeautyNezz is rated highly for social attractiveness (4.40), physical
attractiveness (5.11) and homophily (5.25), which means that on average respondents do not believe
BeautyNezz to possess these character traits. A noticeable point of difference is however that the average
ratings for the dimension homophily often deviate largely from the ratings for the same beauty vlogger on
other dimensions. D for Dazzle for example, is rated an average of 2.76 on agreeableness, while scoring
4.12 for homophily.

34
Table 4.1: overview average scores per dimension and beauty vlogger (N = 207)
Vlogger Source Social Physical Homophily Openness to Conscientiousness Extraversion Agreeableness
credibility attractiveness attractiveness experience
Beautygloss 3.17 3.71 4 3.93 3.27 3.27 3.09 3.11
BeautyNezz 4.15 4.40 5.11 5.25 3.73 3.52 2.95 4.13
Sheling Beauty 3.09 3.02 3.51 4.05 3.04 3.63 2.6 2.64
D for Dazzle 2.91 2.47 2.87 4.12 3.84 3.23 2.82 2.76
BeautyLab 3.41 2.76 2.8 4.42 3.56 3.6 2.87 3
Vera Camilla 2.84 3.22 4.2 4.02 3.38 3.12 2.7 2.29
LifeSplash 4.35 4.11 3.38 5.13 3.76 4.12 2.88 4.18
Bibi Breijman 3.09 3.46 2.17 4.42 3.54 3.36 2.63 3.04
Laura Ponticorvo 3.41 3.52 2.74 4.5 3.19 3.23 2.45 3.21
Anna Nooshin 2.95 3.38 2.22 4.3 3 3.2 2.72 3.51
Jessie Maya 3.64 3.38 4.53 4.65 2.62 3.02 3.2 2.91
LiveLifeGorgeous 3.17 3.21 4.5 3.63 3.98 2.71 3.09 2.67
Saar 4.38 4.13 4.09 4.9 4.22 4.03 3.7 4.42
MissLipgloss 3.22 3.74 3.49 4.5 3.74 2.98 2.94 3.18
Note1: the table represents the average scores based on a scale ranging from 1 – ‘strongly agree’ to 7 – ‘strongly disagree’.
Note2 the dark and light grey shading highlight the highest en lowest scores per variable

After respondents were asked how they perceived the persona of the beauty vlogger based on the eight
dimensions, they were asked to indicate their buying intent of the product the beauty vlogger discussed.
Table 4.2 shows the averages per beauty vlogger of how likely respondents is to try, buy or actively seek
out the product(s). Additionally, overall buying intent is represented. The higher a respondent rated the
vlogger on this dimension, the less likely she is to buy the product. From table 4.2 it is therefore clear that
the product that D for Dazzle discussed in her videos is most likely to be bought by consumers, which is
indicated by the low score on the total buying intent scale. Compared to the other beauty vloggers,
products that were discussed by BeautyNezz are least likely to be bought by the consumer. Additionally,
consumers are most likely to try and products discussed by D for Dazzle, but they are most likely to
actively seek out products discussed by Bibi Breijman.

Table 4.2: overview consumer-buying intent per beauty vlogger (N = 207)

Beauty vlogger Try Buy Seek out Total buying intent (av)
BeautyGloss 3.07 3.33 3.6 3.3
BeautyNezz 3.8 3.53 4 3.78
Sheling Beauty 2.4 2.73 3.27 2.8
D for Dazzle 2.13 2.33 3 2.49
BeautyLab 2.87 3.33 4 3.4
Vera Camilla 2.73 2.93 3.53 3.07
LifeSplash 3 2.93 3.87 3.27
Bibi Breijman 2.25 2.75 2.75 2.58
Laura Ponticorvo 2.64 2.57 3.36 2.86
Anna Nooshin 2.2 3 3.67 2.96
Jessie Maya 3.13 3.2 3.73 3.36
LiveLifeGorgeous 3 3.07 3.71 3.26
Saar 3.47 3.6 3.6 2.56
MissLipgloss 3.15 3.46 3.77 3.46
Note1: the table represents the average scores based on a scale ranging from 1 – ‘Definitely yes’ to 5 – ‘Definitely not’.
Note2: the dark and light grey shading highlight the highest en lowest scores per variable

35
As discussed in the beginning of this chapter the average ratings for the dimension homophily often
deviate largely from the ratings for the same beauty vlogger on other dimensions. This raises the question
of the extent to which the persona dimensions are associated with each other, but also with the dependent
variable buying intent. To look into this a little closer Pearson’s correlation test was conducted for all
persona dimensions and for buying intent. It was found that all dimensions, including buying intent,
correlate positively and significantly with each other, except for the dimension conscientiousness. This
dimension correlates very weakly and not significantly with buying intent (r = .132, P = .058). Thus, even
though the separate scores per dimension and beauty vlogger seemed to indicate that the dimension
homophily varies independently from the other dimensions, this variable correlates positively and
significantly with the others. I will come back to the associations between the persona dimensions and
buying intent in the following chapters when the hypotheses are tested. An overview of all associations
can be found in appendix D.
The survey also asked respondents to indicate to what extent they were familiar with the beauty
vlogger they were going to be confronted with in the survey, as well as their familiarity with the product this
beauty vlogger discussed. The results showed that most people (52.7%) had at least heard about the
vlogger they were about to watch. Additionally, 40.6% had at least heard about the product the vlogger
was going to discuss in the clip. This indicates differentiation in the level of familiarity between the
respondents.

4.2 Source credibility

The first hypothesis was stated as: perceived source credibility has a positive influence on consumer-
buying intent. Earlier the association between source credibility and buying intent showed to be moderate
and positive as well as significant (r = .381, N = 207, P < .000), justifying the application of an OLS
regression model.
The OLS regression model includes buying intent as the dependent variables, source credibility as
the independent variable, and is controlling for the seven variables that are concerned with the consumer
characteristics: age (3 dummies), education, financial situation, experience with YouTube, experience with
beauty videos, vlogger familiarity and product familiarity. As explained in chapter 3.2.1, age category 35-
48 is not included in the analysis because it serves as the reference group for the other age categories.
The model has a good fit with the data, F(10, 196) = 4.474, P < .000). The predictive power is
moderate: 18.6% of the total variation in the buying intent can be explained by source credibility (R2 =
.186), which has a moderately positive and significant effect on buying intent. For every extra point that a
beauty vlogger is found a more credible source, buying intent is increased with .319 points. This means
that it can be predicted that the more a beauty vlogger is found a credible source, the more the consumer
is likely to buy the products recommended by that particular beauty vlogger. From the seven control
variables included in the analysis only product familiarity showed to have a significant, but weak effect on

36
buying intent. Hypothesis 1 that stated that perceived source credibility to have a positive influence on
consumer-buying intent is therefore corroborated.
As reviewed in chapter 2, Martawilaga and Purwanegara (2016) argue that the more credible a
beauty vlogger is perceived to be by consumers, the stronger the impact on consumer-buying intent will
be. Gillin (2007) took this argumentation a step further and said that a social influencer can only be a
social influencer if he or she is found a credible source. The results of this study confirm the statements by
both Martawilaga and Purwanegara (2016) as well as by Gillin (2007), by showing that source credibility
has a positive and significant effect on consumer-buying intent.

Table 4.3: OLS regression model consumer-buying intent versus source credibility (N = 207)
B SE Beta P CI 95% (lower bound) CI 95% (upper bound)

Source credibility .319 .063 .348 .000 .195 .444

Age 13 -17 .172 .363 .051 .635 -.543 .887
Age 18 - 23 -.008 .284 -.003 .978 -.568 .552
Age 24 - 34 .048 .313 .017 .878 -.569 .666
Education .039 .050 .059 .442 -.060 .138
Financial situation -.010 .094 -.008 .912 -.195 .175
YouTube experience .038 .086 .037 .656 -.132 .209
Beauty video experience .022 .077 .024 .777 -.129 .173
Vlogger familiarity -.009 .101 -.007 .930 -.207 .189
Product familiarity .283 .110 .177 .011 .066 .499
Constant 1.089
R2 .186
Note: control variable age 35 – 48 was excluded because it serves as a reference group for the 3 remaining dummy categories

4.3 Social and physical attractiveness

The impact of social and physical attractiveness has been tested according to hypothesis 2: perceived (a)
social attractiveness and (b) physical attractiveness have a positive influence on consumer-buying intent.
Pearson’s correlation was tested for both social attractiveness (r = .420, N = 207, P < .000) and physical
attractiveness (r = .324, N = 207, P < .000) and found that both variables significantly, but moderately
weakly associate with buying intent. Therefore a regression analysis is run to determine whether buying
intent can be predicted based on social attractiveness and/or physical attractiveness.
The OLS regression model was first run with buying intent as the dependent variable and social
attractiveness as the independent variable, while controlling for the seven variables that measure the
personal factor. The model is found to statically significant predict the outcome variable and is therefore a
good fit for the data, F(10, 196) = 5.494, P < .000. With a high predictive power of 21.9% (R2 = .219),
social attractiveness significantly, but moderately and positively impacts buying intent. Every extra point
on the social attractiveness scale leads to .304 points extra buying intent. It can thus be predicted that the
more socially attractive a consumer finds a beauty vlogger, the more likely she is to buy the products
recommended by that person. Similar to what was found in the previous section, the only control variable

37
that seems to have a significant, but very weak positive effect on buying intent, is product familiarity.
Because an increase in social attractiveness is found to have a positive effect on buying intent, hypothesis
2a is corroborated.

Table 4.4: OLS regression model for predicting consumer-buying intent versus perceived social
attractiveness (N = 207)
B SE Beta P CI 95% (lower bound) CI 95%(upper bound)

Social attractiveness .304 .051 .393 .000 .203 .405

Age 13 -17 .176 .355 .052 .620 -.523 .875
Age 18 - 23 -0.60 .277 -.025 .830 -.606 .487
Age 24 - 34 -.140 .305 -.049 .648 -.742 .463
Education .091 .049 .137 .067 -.006 .188
Financial situation -.082 .093 -.060 .376 -.266 .101

YouTube experience .032 .084 .031 .708 -.135 .198

Beauty video experience -.005 .075 -.005 .952 -.153 .144

Vlogger familiarity .057 .097 .042 .558 -.134 .247

Product familiarity .273 .108 .171 .012 .061 .485

Constant 1.177
R2 .219
Note: control variable age 35 – 48 was excluded because it serves as a reference group for the 3 remaining dummy categories

To answer the second part of hypothesis 2 the OLS regression model is run including buying
intent as the dependent variable, physical attractiveness as the independent variable, and controlling for
the seven variables that deal with personal characteristics of the consumer. The model is found to
statically and significantly predict the outcome variable and is therefore a good fit for the data, F(10, 196)
= 3.894, P < .000. The predictive power is mediocre, whereby 16.6% of the total variation in buying intent
can be explained by physical attractiveness (R2 = .166). The model shows a moderate significant impact
of physical attractiveness on consumer-buying intent. Additionally, a significant but weak influence of
product familiarity is visible, which is one of the seven control variables included in the analysis. For every
extra point on the physical attractiveness scale, there is a predicted .226 extra buying intent. This means
that the more physically attractive a consumer finds a beauty vlogger, the more buying intent she is
expected to have, indicating a positive influence. Therefore hypothesis 2b, that stated perceived physical
attractiveness to have a positive influence on consumer-buying intent, is corroborated.
As Lee and Watkins (2016) raised awareness of the importance of social and physical
attractiveness in establishing influence on consumer-buying intent, the results of this study agree with
them. Both social attractiveness and physical attractiveness were found to positively and significantly
influence consumer-buying intent, even though social attractiveness seems to have a larger influence.

38
Table 4.5: OLS regression model for predicting consumer-buying intent versus perceived physical
attractiveness (N = 207)
B SE Beta P CI 95% (lower bound) CI 95%(upper bound)
Physical attractiveness .226 .050 .303 .000 .127 .326
Age 13 -17 .037 .365 .011 .920 -.683 .756
Age 18 - 23 -.128 .286 -.054 .656 -.692 .436
Age 24 - 34 -.152 .316 -.053 .631 -.775 .471
Education .068 .051 .103 .182 -.032 .168
Financial situation -.011 .095 -.008 .909 -.198 .176
YouTube experience -.014 .087 -.014 .871 -.187 .158
Beauty video experience .046 .078 .051 .554 -.107 .199
Vlogger familiarity .043 .100 .032 .670 -.155 .241
Product familiarity .319 .111 .200 .004 .101 .538
Constant 1.234
R2 .166
Note: control variable age 35 – 48 was excluded because it serves as a reference group for the 3 remaining dummy categories

4.4 Homophily
The effect of perceived homophily on buying intent is tested with the use of the third hypothesis. It was
stated as: perceived homophily has a positive influence on consumer-buying intent. A correlation analysis
showed a moderate but significant association between homophily and buying intent (r = .335, N = 207, P
< .000), justifying the application of an OLS regression analysis.
The OLS regression model includes buying intent as the dependent variable and homophily as the
independent variable, while controlling for the seven variables measuring the personal factor. The model
has a good fit with the data, F(10, 196) = 3.897, P < .000. The predictive power is moderate: 16.6% of the
total variation in the buying intent can be explained by homophily (R2 = .166). The impact of homophily on
buying intent is moderate but statistically significant and positively influences buying intent. For every extra
point on the homophily scale there is .262 extra buying intent. This means that it can be predicted the
more a consumer feels homophilistic towards a beauty vlogger, the more likely she is to buy products
recommended by this vlogger. In this case product familiarity is again found weakly statically significant,
as the only control variable out of the seven that are included. Because hypothesis 3 was posed as:
perceived homophily has a positive influence on consumer-buying intent, the hypothesis is found to be
true and is therefore corroborated.
Not only Lee and Watkins (2016) argued that the persona of a vlogger needs to be close to the
persona of the viewer for the vlogger to have an impact on buying intent, but Khaniwale (2015) argued
that influence intensifies when a consumer feels comparable to a vlogger. The concept of homophily is
concerned with this notion, and the associated hypothesis that argued homophily to positively influence
consumer-buying intent to be true was corroborated. The results are therefore in line with arguments by
Lee and Watkins (2016) as well as by Khaniwale (2015).

39
Table 4.6: OLS regression model for predicting consumer-buying intent versus homophily (N = 207)
B SE Beta P CI 95% (lower bound) CI 95%(upper bound)
Homophily .262 .058 .305 .000 .147 .377
Age 13 -17 -.015 .364 -.004 .968 -.733 .704
Age 18 - 23 -.175 .286 -.074 .542 -.739 .389
Age 24 - 34 -.090 .316 -.032 .775 -.713 .532
Education .056 .051 .085 .268 -.044 .156
Financial situation -.057 .096 -.042 .552 -.246 .132
YouTube experience .048 .087 .047 .582 -.124 .221
Beauty video experience .038 .078 .042 .629 -115 .190
Vlogger familiarity .047 .100 .035 .640 -.151 .244
Product familiarity .259 .112 .162 .022 .038 .480
Constant 1.088
R2 .166
Note: control variable age 35 – 48 was excluded because it serves as a reference group for the 3 remaining dummy categories

4.5 Openness to experience

Hypothesis 4 is concerned with whether perceived openness to experience has a positive influence on
consumer-buying intent. The openness to experience scale correlates weakly but significantly with buying
intent (r = .246, N = 207, P < .000). Even though the correlation is weak, it is significant and therefore not
likely to occur by chance. Therefore a regression analysis is appropriate to find the extent to which buying
intent can be predicted based on openness to experience.
The regression model contains buying intent as the dependent variable, openness to experience
as the independent variable, and controls for the seven personal variables. It is found to be a good fit with
the data, F(10, 196) = 2.686, P = .004. The predictive power is however mediocre, and 12.1% of the total
variation in the buying intent can be explained by openness to experience (R2 = .121). According to the
model as displayed in table 4.7, openness to experience positively and significant impacts buying intent.
However, this influence is weak. For every extra point on the openness to experience scale, there is .177
extra buying intent. This indicates a positive influence on buying intent, and suggests that the more a
consumer believes a beauty vlogger is open to experiences, the more she intents to buy the product
discussed by the beauty vlogger. The hypothesis argues that perceived openness to experience has a
positive influence on consumer-buying intent, which is proven by the data and therefore hypothesis 4 is
corroborated. These results are in line with what Biel and Gatica-Perez (2013) found in their study on
vloggers that are perceived as open to experiences receiving more attention from their audience.

40
Table 4.7: OLS regression model for predicting consumer-buying intent versus openness to experience (N
= 207)
B SE Beta P CI (lower bound) CI (upper bound)
Openness to experience .177 .059 .208 .003 .061 .293
Age 13 -17 .019 .275 .006 .960 -.720 .758
Age 18 - 23 -.071 .295 -.030 .809 -.653 .510
Age 24 - 34 -.051 .324 -.018 .874 -.691 .588
Education .052 .052 .078 .323 -.051 .154
Financial situation .002 .097 .002 .981 -.190 .194
YouTube experience .004 .090 .004 .967 -.173 .181
Beauty video experience .056 .080 .062 .485 -.102 .213
Vlogger familiarity .060 .103 .044 .560 -.143 .263
Product familiarity .303 .114 .190 .009 .078 .528
Constant 1.322
R2 .121
Note: control variable age 35 – 48 was excluded because it serves as a reference group for the 3 remaining dummy categories

4.6 Conscientiousness
The fifth persona dimension is tested in hypothesis 5: perceived conscientiousness has a positive
influence on consumer-buying intent. As discussed in the beginning of this chapter, conscientiousness
shows a very weak association, which is not statistically significantly associated with buying intent (r =
.132, N = 207, P = .058). Despite the fact that the association is not significant, the OLS regression model
is applied to determine the influence of conscientiousness on buying intent because the level of statistical
significance for the correlation analyses was set at P < .050, only slightly lower than P = .058.
The regression model which includes buying intent as the dependent variable conscientiousness
as the independent variable, and is controlling for the seven personal variables, is found not to be
statically significant and is therefore not a good predictor and not good fit for the data, F(10, 196) = 1.861,
P = .053. Only 8.7% of the total variation in the buying intent can be explained by conscientiousness (R2 =
.087), indicating a low predictive power. The impact of conscientiousness on buying intent is positively
low, and not significant. For every extra point on the conscientiousness scale there is .085 extra buying
intent. However, the control variable product familiarity is found to be a significant and positive influence
on buying intent. Because hypothesis 5 stated that conscientiousness has a positive influence on
consumer-buying intent, it is rejected. The data showed that there is no statistically significant impact of
conscientiousness on buying intent. Rather, product familiarity does seem to influence buying intent.
According to what Biel and Gatica-Perez (2013) found in their study, conscientiousness should
have been found to positively and significantly influence consumer-buying intent. However, the opposite is
true. Even though the influence is present, it is not significant. This means that random chance could
explain these results, and therefore the hypothesis has to be rejected.

41
Table 4.8: OLS regression model for predicting consumer-buying intent versus conscientiousness (N =
207)
B SE Beta P CI (lower bound) CI (upper bound)
Conscientiousness .085 .069 .089 .218 -.051 .221
Age 13 -17 -.053 .281 -.016 .889 -.805 .698
Age 18 - 23 -.187 .300 -.079 .535 -.779 .406
Age 24 - 34 -.098 .330 -.034 .767 -.749 .553
Education .057 .053 .087 .284 -.048 .162
Financial situation -.001 .099 -.001 .994 -.197 .195
YouTube experience .010 .091 .009 .917 -.171 .190
Beauty video experience .031 .081 .034 .706 -.130 .191
Vlogger familiarity .080 .105 .059 .445 -.127 .288
Product familiarity .328 .116 .205 .005 .099 .556
Constant 1.691
R2 .087
Note: control variable age 35 – 48 was excluded because it serves as a reference group for the 3 remaining dummy categories

4.7 Agreeableness
Different from the other hypotheses that were formulated, hypothesis 6 states that agreeableness only
positively impacts consumer-buying intent when it is perceived either very agreeable or very disagreeable.
Because the hypothesis asks to test whether extreme answers result in increased buying intent, dummy
variables are created that indicate whether a vlogger is rated as very (dis)agreeable (1), or anything else
(0).
The OLS regression model with buying intent as the dependent variable, dummy variable extreme
(dis)agreeableness as the independent variable, and controls for the seven personal variables, is found be
a good fit for the data, F(10, 196) = 2.398, P = .010. 10.9% of the total variation in buying intent can be
explained by extreme (dis)agreeableness (R2 = .109), which indicates that the predictive power is
mediocre. Table 4.9 shows that a weak negative, significant influence was found of the dummy variable
extreme (dis)agreeableness on buying behavior. This is indicates that people who indicated to perceive
the beauty vlogger either as extremely agreeable or extremely disagreeable, are less likely to buy the
products discussed by that beauty vlogger. Also, product familiarity is again found to be a significant
influence on buying intent, which is weak but positive. People that rated the agreeableness of the beauty
vloggers in extreme values score .521 points less on the buying intent scale. This goes against the
hypothesis as it stated that people who find beauty vloggers either extremely (dis)agreeable would be
more likely to buy the products discussed. The hypothesis is therefore rejected. These results are not in
line with what Biel and Gatica-Perez (2013) found in their study, and display the opposite. People who
perceive a beauty vlogger as either extremely agreeable or extremely disagreeable, are less intended to
buy the product they recommend. The results of the study by Biel and Gatica-Perez (2013) are therefore
not confirmed but rather disproven.

42
Table 4.9: OLS regression model for predicting consumer-buying intent versus extreme
(dis)agreeableness (N = 207)
B SE Beta P CI (lower bound) CI (upper bound)
Extreme (dis)agreeableness -.521 .205 -.174 .012 -.924 -.117
Age 13 -17 .002 .377 .001 .995 -.741 .746
Age 18 - 23 -.132 .296 -.056 .655 -.715 .451
Age 24 - 34 -.109 .326 -.038 .739 -.752 .534
Education .069 .052 .104 .191 -.035 .172
Financial situation .009 .098 .007 .927 -.184 .202
YouTube experience .038 .091 .037 .674 -.140 .217
Beauty video experience .039 .080 .044 .624 -.119 .197
Vlogger familiarity .075 .103 .055 .470 -.129 .278
Product familiarity .328 .114 .206 .005 .103 .554
Constant 1.893
R2 .109
Note: control variable age 35 – 48 was excluded because it serves as a reference group for the 3 remaining dummy categories

Because the results showed that extreme ratings on the agreeableness scale negatively impact
consumer-buying intent rather than positively, and therefore goes against what was expected based on
literature reviewed in chapter 2, an additional OLS regression analyses is run with the regular continuous
scale variable agreeableness, which runs from 1 – strongly agree – to 7 – strongly disagree, to find
whether this would yield similar results. The results of this model show that the predictive power of this
model is moderate: 16.2% of the total variation can be explained by agreeableness, (R2 = .162). The
model does seem to be a good fit for the data as well F(10, 105) = 2.032, P = .037. There now is a
positive but mediocre significant influence of agreeableness on consumer-buying intent. For every extra
point on the agreeableness scale, a consumer is .203 points more likely to buy the products
recommended by the beauty vlogger. On the contrary, product familiarity does not significantly influence
consumer-buying intent anymore.

Table 4.10: OLS regression model for predicting consumer-buying intent versus agreeableness (N = 207)
B SE Beta P CI (lower bound) CI (upper bound)
Agreeableness .203 .060 .319 .001 .084 .321
Age 13 -17 -.144 .496 -.048 .773 -1.128 .841
Age 18 - 23 -.432 .407 -.197 .291 -1.239 .375
Age 24 - 34 -.488 .442 -.189 .272 -1.365 .389
Education -.019 .066 -.031 .777 -.151 .113
Financial situation -.041 .121 -.033 .738 -.281 .200
YouTube experience .021 .113 .022 .856 -.205 .247
Beauty video experience .092 .101 .111 .366 -.109 .293
Vlogger familiarity .124 .130 .098 .342 -.134 .382
Product familiarity .234 .144 .156 .108 -.052 .520
Constant 1.722
R2 .162
Note: control variable age 35 – 48 was excluded because it serves as a reference group for the 3 remaining dummy categories

43
4.8 Extraversion

The final hypothesis argues that perceived extraversion has no impact on consumer-buying behavior.
Pearson’s correlation indicates that there is an association, even though moderate; it is significant (r =
.358, N = 208, P < .000). These results already go against the hypothesis as it states that extraversion
does not influence buying intent at all. To determine to what extent extraversion then influences buying
intent, and whether it can be used as a predictor, an OLS regression analysis is performed.
The regression model includes buying intent as the dependent variable, extraversion as the
independent variable, and controls for the seven personal variables. The model found to be a good fit for
the data, F(10, 196) = 4.480, P < .000. 18,6% of the total variation in the buying intent can be explained by
extraversion (R2 = .186); predictive power is therefore moderate. As shown in table 4.11, the influence of
extraversion on buying intent is positive, moderately strong, and significantly impacts buying intent. For
every extra point on the extraversion scale, there is .316 extra buying intent. Product familiarity positively
influences buying intent as well, with a weak but significant impact. According to the results it can be
predicted that the more a consumer perceives a beauty vlogger as extravert, the more she is likely to buy
the product discussed by this vlogger. Extraversion is then found to have an impact on consumer-buying
intent, opposing hypothesis 7 and thereby rejecting it. These findings are, similar to the previous
hypothesis that was tested, opposite to what Biel and Gatica-Perez (2013) found in their study. While they
found a very weak correlation (r = .24), these results indicate a slightly stronger but significant impact.
Therefore extraversion here is found to be a determinant of consumer-buying intent because the results
show a significant influence. Also, results from Biel and Gatica-Perez (2013) indicate that both high and
low extraverts received higher ratings, which is the case in this analysis.

Table 4.11: OLS regression model for predicting consumer-buying intent versus extraversion (N = 207)
B SE Beta P CI (lower bound) CI (upper bound)
Extraversion .316 .062 .339 .000 .193 .439
Age 13 -17 .077 .361 .023 .832 -.635 .788
Age 18 - 23 .044 .285 .019 .879 -.519 .606
Age 24 - 34 .030 .313 .011 .923 -.587 .647
Education .044 .050 .066 .383 -.055 .143
Financial situation -.006 .094 -.004 .950 -.191 .179
YouTube experience -.049 .087 -.048 .576 -.221 .123
Beauty video experience .086 .077 .096 .264 -.066 .239
Vlogger familiarity .070 .098 .051 .479 -.124 .264
Product familiarity .285 .110 .178 .010 .068 .501
Constant 1.018
R2 .186
Note: control variable age 35 – 48 was excluded because it serves as a reference group for the 3 remaining dummy categories

44
4.9 Relative impact of explanatory variables
In the previous sections all the hypotheses as they were formulated in chapter 2 have been tested. This
section provides overview of the relative impact of all the persona dimensions on consumer-buying intent.
Hereby the following variables are included: source credibility, social attractiveness, physical
attractiveness, homophily, openness experience, conscientiousness, agreeableness and extraversion.
Additionally the seven variables concerned with the characteristics of the consumer, as discussed in the
introduction to chapter 3.2.1 and used in previous OLS regression analyses, are controlled for.
In chapter 4.6 it was explained that a dummy variable was created for the variable agreeableness
to represent the cases in which agreeableness was rated in the extremes. Because the aim here is to
discover the relative impact of the Big Five model, as theory revised in chapter 2 argued it to influence
consumer-buying intent, these dummies are again used in this analysis.
The regression model including buying intent as the dependent variable and the aforementioned
independent- and control variables, is found to be a good fit with the data, F(17, 189) = 4.623, P < .000.
29,4% of the total variation in buying intent can be explained (R2 = .294). From table 4.12 it is visible that
only the persona dimensions social attractiveness and physical attractiveness significantly impact buying
intent. However, the control variable product familiarity also significantly impacts buying intent.
Finally, to determine the relative impact of only the significant variables, the OLS regression model
is run one more time including the variables social attractiveness, physical attractiveness and product
familiarity. Results show that the model does have a good fit with the data, F(3, 203) = 19.564, P = .000.
However, the variance that is now explained by these variables is 22.4%, which is lower than in the model
that included both the significant and the insignificant variables. Results show that compared to social
attractiveness (b* = .305, P = .000) and physical attractiveness (b* = .175, P = .012), the control variable
product familiarity has the largest influence on consumer-buying intent, even though the correlation is
weak (b* = .179, P = .005). For each extra point of product familiarity, there is an extra .285 points of
buying intent. An overview of the results from this test can be found in appendix E.

45
Table 4.12: OLS regression model for predicting consumer-buying intent versus persona dimensions and
control variables (N = 207)
B SE Beta P CI 95% (lower bound) CI 95% (upper bound)
Source credibility .089 .089 .097 .317 -.086 .264
Social attractiveness .169 .079 .219 .033 .014 .325
Physical attractiveness .110 .054 .148 .044 .003 .218
Homophily .102 .075 .119 .176 -.046 .251
Openness to experience .023 .062 .027 .716 -.100 .145
Conscientiousness -.135 .072 -.142 .062 -.277 .007
Extraversion .154 .080 .165 .054 .003 .311
Agreeableness -.051 .058 -.075 .381 -.164 .063
Age 13 – 17 .238 .346 .071 .492 -.445 .921
Age 18 - 23 .070 .275 .030 .798 -.472 .612
Age 24 - 34 -.042 .299 -.015 .888 -632 .548
Education .075 .049 .114 .129 -.022 .172
Financial situation -.076 .091 -.049 .460 -.246 .112
YouTube experience 2.961E-5 .085 .000 1.000 -.168 .168
Beauty video experience .054 .075 .061 .469 -.094 .203
Vlogger familiarity .019 .097 .014 .842 -.171 .210
Product familiarity .227 .106 .142 .034 .018 .436
Constant .572
R2 .294
Note: control variable age 35 – 48 was excluded because it serves as a reference group for the 3 remaining dummy categories

The survey asked each respondent to rate two different vloggers. As reported earlier, 25 out of the
116 individuals dropped out halfway through the survey, and have therefore only rated vlogger. The other
91 respondents completed the full survey and have therefore rated two vloggers. To take this matter into
account a linear mixed model test is conducted. This is a random intercept model with fixed effects,
comparable to an OLS regression model as applied in previous sections. However this linear mixed model
allows for the inclusion of the respondents as subject variables, whereby the analysis takes into account
that the answers are nested in the respondents. Since the majority of the respondents have rated two
beauty vloggers, these ratings will share a certain level of communality, simply because of the person who
produced them. Please note that I only conduct this analysis is only conducted to check the robustness of
the results since linear mixed models are not part of the MA curriculum.
Buying intent is included in the analysis as the dependent variable and source credibility, social
attractiveness, physical attractiveness, homophily, openness to experience, conscientiousness,
agreeableness and extraversion are included as the independent variables or predictors. The seven
control variables are added to the analysis as well.
First an empty model was run, meaning that only the dependent variable buying intent is included.
This resulted in a -2 Log Likelihood = 651.143, with 3 parameters. This statistic indicates the amount of
error within the model. The estimation method was set at ‘maximum likelihood’ because the model will
only differ in its fixed effects. Following the model was run including all predictors. This resulted in a -2 Log
Likelihood = 270.688, with 36 parameters, indicating a decrease in the amount of error compared to the
empty model. Therefore this model better fits the data (Field, 2013).

46
 Table 4.13 shows the estimates of the fixed effects persona variables. It shows that only social
attractiveness (b = .194, t(2.150), P = .034), conscientiousness (b = -.173, t(-2.234), P = .027) and
extraversion (b = .200, t(2.196), P = .030) significantly impact consumer-buying intent. Additionally, social
attractiveness and extraversion have a positive estimate, where conscientiousness has a negative
estimate. This means that the more a beauty vlogger is perceived as conscientious, the less buying intent
is expected. Additionally, all seven control variables used in the analysis were found to be statistically
insignificant. An overview of the results including, the estimates of the control variables, can be found in
appendix F.
Because the same variables were used in the linear mixed model as in the OLS regression model
the results can be compared, which shows a slight difference. Where in the OLS regression model the
variable conscientiousness was only close to significance (b* = -.142, P = .062), it was found significant in
the mixed model (b = -.173, t(-2.234), P = .027). Social attractiveness is found significant in the OLS as
well as in the linear mixed model. However, physical attractiveness was found significant in the OLS
regression (b* = .148, P = .044), but not in the linear mixed model (b = .057, t(.811), P = .419).
Extraversion was also found significant in the linear mixed model (b = .200, t(2.196), P = .030), while
results showed that it crossed the boundary of significance only slightly in the OLS regression model (b* =
.165, P = .054).

Table 4.13: Linear mixed model for buying intent versus persona dimensions and control variables (N =
207)
b SE P CI 95% (lower bound) CI 95% (upper bound)
Source credibility .142 .102 .167 -.060 .344
Social attractiveness .194 .090 .034 .015 .372
Physical attractiveness .057 .071 .419 -.083 .198
Homophily .080 .079 .311 -.076 .236
Openness to experience -.116 .070 .099 -.253 .022
Conscientiousness -.173 .077 .027 -.326 -.020
Extraversion .200 .091 .030 .020 .381
Agreeableness .029 .066 .668 -.103 .160
Intercept 2.389
Note1: in the model the following variables were have been controlled for: age, education, financial situation, YouTube experience, beauty vlog
experience, vlogger familiarity and product familiarity
Note2: control variable age 35 – 48 was excluded because it serves as a reference group for the 3 remaining dummy categories

47
48
5. Conclusion
This study was build on the question to what extent buying intentions of consumers are influenced by the
perceived persona of beauty vloggers. With a focus on Dutch beauty vloggers and Dutch consumers it has
used concepts of para-social interaction and the Big Five model to define the impact of the social
influencer persona. The characteristics of the consumer were not overlooked and several were defined to
also influence the buying intent of consumers. The literature review in chapter 2 resulted in the definition
of eight persona dimensions: source credibility, social attractiveness, physical attractiveness, homophily,
openness to experience, conscientiousness, extraversion and agreeableness. With the use of an
experimental survey design consumer perceptions on Dutch beauty vloggers were collected. Subsequent
analyses were guided by the hypotheses as formulated based on the results of previous studies
concerning the topics of interpersonal influence and social influencers.
As according to Khaniwale (2015) the personal factor of consumer’s is important to consider in
determining buying intent, the Dutch foundation for scientific research in commercial communications
(Stichting Wetenschappelijk Onderzoek Commerciële Communicatie, 2017) wrote that vlogs are most
frequently watched by youth between the ages of 18 and 34. However, the results of this study portray a
different picture by showing that there is a substantial group of mainly younger, but also older consumers
interested in beauty vlogs. The survey was distributed online and asked people what they thought about
Dutch beauty vloggers. It can therefore be assumed that when a consumer responded to the survey, she
at least has some interest in beauty vlogs. This raises new questions concerning the YouTube audience;
not only about who they are, but also what content is watched by which people. Additionally, these
findings create new opportunities for brands in reaching their audiences through beauty vloggers, since
the audience is dispersed wider then what was previously thought.
Five out of the seven hypotheses that were tested were corroborated; three were rejected. Source
credibility, social attractiveness, physical attractiveness, homophily and openness to experience were all
found to positively influence consumer-buying intent, which is in line with the associated literature review
in chapter 2. Current literature has however focused either on the impact of eWOM (Cheung & Thadani,
2012), the effects of eWOM on consumer-buying intent (Fan & Miao, 2012) or on determinants of how
customers get involved through eWOM (Chu & Kim, 2011). This study therefore focuses specifically on
the influence of persona’s of social influencers on consumer-buying intent, whereby the results show how
online interpersonal influence differs from its offline equivalent by looking at relations to earlier studies by
for example Clark and Goldsmith on market mavens (2005).
Findings from previous studies indicated that agreeableness was a special personality trait in
which extreme ratings (strongly agree or strongly disagree) would positively impact consumer-buying
intent (Biel & Gatica-Perez, 2013). However, findings from this study show this to be the other way
around, and indicate a negative influence on consumer-buying intent when a beauty vlogger is rated
extremely (dis)agreeableness. When the regular agreeableness scale was tested the results showed

49
agreeableness to positively influence consumer-buying intent. The dimension of agreeableness is
concerned with how kind, sympathetic and forgiving a beauty vlogger is perceived to be by the
respondents. It makes sense that when a beauty vlogger is perceived as extremely unkind, unsympathetic
and unforgiving, advice from this person is not adopted and a consumer is less likely to buy the products
discussed and/or recommended. The results of this study however show that when the opposite is true,
and a beauty vlogger is perceived as extremely kind, sympathetic and forgiving, the advice is not adopted
as well, and a consumer is also less intended to buy the product discussed. On the contrary, the regular
agreeableness scale showed that when a beauty vlogger is found more kind, sympathetic and forgiving,
the buying intent of the consumer does increase.
The dimension of extraversion also showed to positively influence consumer-buying intent. Based
on previous studies it was however expected that the perceived enthusiasm and assertiveness of a beauty
vlogger would not have an impact on how likely someone is to buy the product discussed by that person,
and therefore the hypothesis was rejected. The hypotheses that were formulated concerning the
dimensions of agreeableness and extraversion were based on a study by Biel an Gatica-Perez (2013),
who looked at the amount of attention a vlogger received, whereby I subsequently assumed that
increased attention simultaneously leads to increased influence. What the results however show is that
this might not be true for all cases. It can be argued that the influence exercised by beauty vloggers as
social influencers can be different than the influence by other social influencers on YouTube. Biel and
Gatica-Perez (2013) have not focused on a particular type of vlogger, while this study has. This study
therefore adds to existing literature by showing that there is differentiation between different types of social
influencers on the YouTube platform. This opens up questions about the differences in influence exercised
by different types of influencers, but also about the differences in interpersonal influence by vloggers
beyond the YouTube platform.
Additionally, the current studies on social influencers and their potential impact on consumer-
buying intent apply the Big Five model, that has shown its effectiveness in the offline world of persona
studies, directly to the online environment of social media. The disproven hypotheses however suggest
that this model in its original form might not be the most suitable model for assessing personalities online.
It is therefore recommended for future studies to focus on adjusting existing, and identifying new methods
of assessing influencers online to allow for this new interactive environment, in which communication is
not only asynchronous, but also unidirectional (Frobenius, 2004), to be studied further.
Going back to the personality dimensions, no significant influence of perceived conscientiousness
on consumer-buying intent was found. Conscientiousness is one of the dimensions of the Big Five Model
and is as according to Biel and Gatica-Perez (2013) concerned with how efficient, organized, responsible
and thorough someone is. The results in this paper indicate that consumers are not bothered with this
personality trait and do not let it influence their buying intentions.
Examining the relative impact of all persona dimensions it was found that a consumer’s buying
intent increases when a beauty vlogger is perceived as kind, someone to be friends with, and as someone
to converse with (social attractiveness), but also as looking good, looking sexy and being physically
50
attractive (physical attractiveness). Being perceived as efficient, organized, responsible and thorough
(conscientiousness), and enthusiastic, energetic, active and assertive (extraversion) is important as well,
however the impact of these character traits is found barely insignificant. These results are quit different
from when the dimensions were tested separately from each other. When the mixed model was run to
take into account that some respondents rated two beauty vloggers, kindness and friendliness (social
attractiveness) continued to have a significantly positive impact, together with how outgoing someone is
perceived to be, the dimension of extraversion that has now become significant. Also, the influence of a
beauty vlogger’s perceived conscientiousness has a significant, but negative influence. Where in the
analyses run prior it seemed that consumer’s were not bothered with this personality trait, they now show
to be negatively influenced by an efficient, organized, responsible and thorough beauty vlogger.
The experience a consumer has with beauty vlogs also significantly impacts consumer-buying
intent in a negative manner, meaning that the more often a consumer watches beauty vlogs, the less
intended she is to buy the products. It could be speculated that this indicates that increased knowledge of
the consumer on marketing and promotional practices by beauty vloggers, which they have acquired by
watching many beauty vlogs, and therefore the influence they are trying to exercise, leads to decreased
effectiveness of this influence.
Finally, product familiarity seems to be an important factor throughout the study. In the linear
mixed model analyses its impact was found insignificant, however barely. Throughout all the analyses that
were conducted seven variables concerning the personal characteristics of the consumer were controlled
for. However, product familiarity was the only control variable that continuously showed to have both
positive and significant impact on consumer-buying intent. Results show that the more familiar a consumer
is with a product, the more intended she becomes to buy it when a beauty vloggers discusses it. However,
when someone has prior knowledge on the product that the beauty vlogger discusses, by for example
having an increased interest in that particular type of mascara, someone will already have increased
buying intent concerning that product, despite the influence of the beauty vlogger. Additionally, product
familiarity was measured by asking the respondent to indicate whether she knew the product, had heard
of it, or did not know the product. There was no follow-up question that asked how much the respondent
liked the product, when she indicated to know the product. This would have allowed for filtering out the
people that already extremely liked the product, and would therefore already have an increased buying
intent towards that product, the outliers. These findings are however important in that it is the only
dimension out of the nine consumer characteristics that were included in the study that was found to
significantly impact consumer-buying intent. It thereby implies that all effects that are found in this study,
all dimensions and factors that are found to influence consumer-buying intent, apply to everyone. Thus,
age, educational level, current financial situation, experience with YouTube, experience with beauty
videos, en familiarity with the vlogger do not influence the buying intent of a consumer watching a beauty
vlog. Looking at the commercial implications of these findings, it is important for brands to realize that
when they are focusing on finding the right beauty vlogger to cooperate with, they also realize that the
influence of any beauty vlogger increases largely when the consumer is familiar with the brand prior to
51
watching the beauty vlog. This can have strategic implications for their marketing campaigns because of
the additional knowledge on the buying-behavior of consumers.
Overall it can be said that the perceived kindness, and therefore attractiveness as a friend or
conversation partner is the most important influencer of consumer-buying intent in the persona of a beauty
vlogger. Throughout all analyses conducted this dimension is found to have a positively and significant
impact. It can therefore also be perceived as the most important for commercial businesses in determining
which beauty vlogger would best fit their brand, and would be perceived as socially attractive by their
target group. Additionally, finding a vlogger who is extravert and therefore enthusiastic, energetic, active
and assertive would, according to the results of this study, also have a positively influence on consumer-
buying intent and thereby increase it. Assuming commercial businesses base their marketing decisions on
academic studies conducted on the topic of social influencer personas (Lee & Watkins, 2016), these
results will require them to think differently. Where looks and appearance are often seen as a large
determinant of the success of a beauty vlogger, the real influence can be largely contributed to kindness
and attractiveness as a social partner. Also, it is important to think the other way around and realize there
are parts of personas that negatively influence buying intent. Vloggers portraying these characteristics
should therefore be avoided at all costs, since they will influence consumers to buy less of a product
rather than more. When a vlogger is perceived as being an efficient and thorough person who has their
work organized and acts very responsibly concerning their work-related appointments for example, should
be avoided. The negative effects of these character traits on the buying behavior of consumers should not
be overlooked. To answer the research question as repeated in the beginning of this chapter, social
attractiveness influences consumer-buying intent to the largest extent followed by extraversion and
conscientiousness.
Considering the results of this study in the larger context of society it can be argued that according
to these results, looking sexy and attractive is not found as important in general. As many studies argue
that feeling close to someone else in order to be influenced by this person is an important aspect of
interpersonal influence (Langner, Hennigs & Wiedmann, 2013; Khaniwale, 2015), this study shows the
opposite and finds that perceiving a vlogger to think the same way about for example testing products on
animals, or having a lot in common with the vlogger such as both being college students, does not matter.
Homophily is not found an important determinant in consumer-buying intent, which could make us
question how important this notion is in contemporary society where online communications are becoming
the standard, and as discussed in chapter 2.4, most communications are unidirectional (Frobenius, 2004).
Being a vlogger is not about whom one is, but about the character traits they express. Would this be the
reason why consumers do not find sharing the same values and having a lot in common with someone as
important any more? Might they be aware of the constructed persona of the beauty vlogger? More
research on this particular topic is necessary to determine why these results speak against researchers
such as Khaniwale (2015) and Clark and Goldsmith (2005) who have previously found that interpersonal
influence intensifies when a consumer feels comparable to the social influencer.

52
 The main reason to conduct this study was to fill the gap in literature concerning the consumer
perspective on social influencing online. However, even though consumers were presented with a survey
and asked to voice their opinions on beauty vloggers, the standardization component of a quantitative
survey, which is on the one hand a strength of this study through which the experimental design was
enabled, is on the other hand a limitation whereby the ‘why’ question cannot be answered. This study
raises questions on why for example beauty vlogger BeautyLab is perceived as being very organized and
efficient, while the BeautyNezz is perceived as almost the complete opposite and an irresponsible and
unorganized person. A qualitative study on the topic of online influence by social influencers on
consumers is necessary to be able to fully take the perspective of the consumer and get a full perspective
of their motivations.
Furthermore this study is limited in its size. On the one hand this matters because due to the
experimental nature of the study the results are not generalizable into the population of social influencers
as such the results are however relevant. Because the respondents were faced with two clips of two
different beauty vloggers, the data gathered and therefore the conclusions drawn from them are more
valid. The findings from this research therefore add to the debate on social influencing online and can be
used in future research by exploring whether the same can be said about for example different types of
vloggers, or vloggers from different countries. The findings are innovative in that research on this new type
of people that have recently emerged, the social influencer, has so far been very limited and has failed to
determine the actual impact of the content these people create. Additionally, by focusing on a particular
type of social influencer, the beauty vlogger, in-depth knowledge was gained which can in the future be
applied to other types of vloggers on YouTube, but also be extended into the larger online social network
providing material for comparison.
The set-up of the study requires a commentary note as well. The videos included in the survey
were long, and lasted between 2 to 3 minutes. As was visible in the number of dropouts halfway through
the survey, this has led to fewer respondents. The length of the videos did have the advantage that it
made the results more reliable, but it also allowed the respondents to obtain a clear image of the beauty
vlogger within the video. If imaginatively this study would be started over from scratch, the only different
decision I would make as a researcher is to extent the time that is scheduled for the survey distribution
and data collection. The validity of results, as discussed earlier, weighs out the low response rate obtained
due to the length of the study. Therefore the time should be extended, rather than the volume of the
survey reduced.
Finally, it should also be noted that the material used in the study was real. The videos concerned
actual content that is available on the YouTube channels of the selected beauty vloggers. Even though
the same types of videos were selected from all beauty vloggers, and all videos were edited to last
approximately the same amount of minutes, the content of the stimuli differed because different subjects
were discussed. One beauty vlogger for example discussed a hair product, while another discussed a
make-up product. This led to a reduced level of control on the different stimuli.

53
54
References
Ananda, A. F., & Wandebori, H. (2016). The impact of drugstore makeup product reviews by beauty vloggers on
youtube towards purchasing intention by undergraduate students in Indonesia. International Conference on
Ethics of Business, Economics, and Social Science, (pp. 255-263). Bandung, Indonesia. Retrieved from:
http://eprints.uny.ac.id/41794/1/22%20Azka%20Faranisa%20Ananda.pdf
Babbie, E. (2014). Research design. In E. Babbie, The Basics of Social Research (pp. 91-122). Canada: Wadsworth,
Cengage learning.
Baker, M. J., & Churchill Jr., G. A. (1997). The impact of physically attractive models on advertising evaluations.
Journal of Marketing Research, 14 (4), 538-555. DOI: 10.2307/3151194
Becker, P. (1999). Beyond the big five. Personality and Individual Differences, 26 (3), 511–530. DOI:10.1016/S0191-
8869(98)00168-8
Biel, J.-I., & Gatica-Perez, D. (2013). The youtube lens: Crowdsourced personality impressions and audiovisual
analysis of vlogs. IEEE Transactions on Multimedia, 15 (1), 41-55. DOI:10.1109/TMM.2012.2225032
Bilsky, W., & Schwartz, S. H. Values and personality. European Journal of Personality, 8 (3), 163-181. Retrieved from:
http://www.strandtheory.org/images/Schwartz_Value_Theory.pdf
Booth, N., & Matic, J. (2010). Mapping and leveraging influencers in social media to shape corporate brand
perceptions. Corporate Communications: An International Journal, 16 (3), 184-191. Retrieved from
http://www.corporatecomm.org/cci/ CCIProceedings2010.pdf#page=28
Brennan, M., Rae, N., & Parackal, M. (1999). Survey-based experimental research via the web: Some observations
Marketing Bulletin, 10, 83-92. DOI:10.1.1.579.3244
Buchegger, S., & Hui, P. (2009). Groupthink and peer pressure: Social influence in online social network groups.
International Conference on Advances in Social Network Analysis and Mining, 2009. (pp. 53-59). Berlin:
IEEE. Retrieved from: https://www.net.t-labs.tu-berlin.de/papers/HB-GPPSIOSNG-09.pdf
Cai, K., Bao, S., Yang, Z., Tang, J., Ma, R., Zhang, L., et al. (2011). OOLAM: An opinion oriented link analysis model
for influence persona discovery. Proceedings of the Fourth ACM International Conference on Web Search
and Data Mining. Hong Kong. Retrieved from: http://www.cs.cmu.edu/~ziy/pubs/WSDM11-Cai-et-al-
OOLAM.pdf
Cheung, C. M., & Thadani, D. R. (2012). The impact of electronic word-of-mouth communication: A literature analysis
and integrative model. Decision Support Systems, 54, 461–470. DOI: 10.1016/j.dss.2012.06.008
Chu, S.-C., & Kim, Y. (2011). Determinants of consumer engagement in electronic word- of-mouth (eWOM) in social
networking sites. International Journal of Advertising, 30 (1), 47-75. DOI: 10.2501/IJA-30-1-047-075
Clark , R. A., & Goldsmith, R. E. (2005). Market mavens: psychological influences. Psychology & Marketing, 22 (4),
289–312. DOI: 10.1002/mar.20060
Couper, M. P., Traugott, M. W., & Lamias, M. J. (2001). Web survey design and administration. Public opinion
Quarterly, 65 (2), 230-253. Retrieved from: https://goo.gl/3H7oRV
Coursaris, C. K., & Van Osch, W. (2016). Beauty brands versus vloggers: Exploring the effects of source credibility on
information adoption on youtube. ECIS 2016 Proceedings (pp. 1-9). Istanbul: AIS Electronic Library.
Retrieved from http://aisel.aisnet.org/cgi/viewcontent.cgi?article=1031&context=ecis2016_rip
Duffy, B. E., & Hund, E. (2015, September 30). “Having it all” on social media: Entrepreneurial femininity and self-
branding among fashion bloggers. Social Media + Society, 1-11. DOI: 10.1177/2056305115604337

55
EEN Media. (2017, May 09). Over Girlscene. Retrieved from Girlscene: http://www.girlscene.nl/page/over_girlscene_
Fan, Y.-W., & Miao, Y.-F. (2012). Effect of electronic word-of-mouth on consumer purchase intention: The perspective
gender differences. International Journal of Electronic Business Management, 10 (3), 175-181. Retrieved
from: https://search.proquest.com/docview/1240355575?accountid=13598
Farnadi, G., Sushmita, S., Sitaraman, G., Ton, N., De Cock, M., & Davalos, S. (2014) Multivariate regression
approach to personality impression recognition of vloggers. Proceedings of the 2014 ACM Multi Media on
Workshop on Computational Personality Recognition (pp. 1-6). Orlando, Florida: ACM.
DOI:10.1145/2659522.2659526
Field, A. (2013). Logistic regression. In A. Field, Discovering Statistics using IBM SPSS Statistics (pp. 760-813).
London: Sage Publications.
Fiorella, S., & Brown, D. (2013). Logic and reasoning (or how we learned to love influence marketing). In S. Fiorella, &
D. Brown, Influence Marketing: How to Create, Manage, and Measure Brand (pp. 1-2). Indianapolis: Que
publishing.
Freberg, K., Graham, K., McGaughey, K., & Freberg, L. A. (2010). Who are the social media influencers? A study of
public perceptions of personality. Public Relations Review, 1-3. DOI:10.1016/j.pubrev.2010.11.001
Friedkin, N. E., & Johnsen, E. C. (2011). Group dynamics: Structural social psychology. In N. E. Friedkin, & E. C.
Johnsen, Social Influence Network Theory: A Sociological Examination of Small Group Dynamics (Vol. 33,
pp. 3-27). New York: Cambridge University Press.
Frobenius, M. (2004). Audience design in monologues: How vloggers involve their viewers. Journal of Pragmatics, 72,
59-72. DOI:10.1016/j.pragma.2014.02.008
Galeotti, A., & Goyal, S. (2009). Influencing the influencers: A theory of strategic diffusion. Journal of Economics, 40
(3), 509-532. DOI: 10.1111/j.1756- 2171.2009.00075.x
Ghauri, P., & Grønhaug, K. (2005). Research design. In P. Ghauri, & K. Grønhaug, Research Methods in Business
Studies: A Practical Guide (pp. 56-74). Harlow: Pearson Educated Limited.
Gillin, P. (2007). The enthusiasts. In P. Gillin, The New Influencers: A Marketer's Guide to the New Social Media (pp.
33-56). Sanger, California: Quill Driver Books.
Gosling, S. D., Rentfrow, P. J., & Swann Jr, W. B. (2003). A very brief measure of the big-five personality domains.
Journal of Research in Personality, 37, 504–528. DOI:10.1016/S0092-6566(03)00046-1
Gravetter, F. J., & Wallnau, L. B. (2013). Introduction to statistics. In F. J. Gravetter, & L. B. Wallnau, Statistics for the
Behavioral Sciences (9th edition) (pp. 4-27). Belmont, CA, United States: Cengage Learning.
Gupta, P. B., & Lord, K. R. (1998). Product placement in movies: The effect of prominence and mode on audience
recall. Journal of Current Issues & Research in Advertising, 20 (1), 47-59. DOI:
10.1080/10641734.1998.10505076
Harré, R. (1985). Situational rhetoric and self-presentation. In J. P. Forgas, Language and Social Situations (pp. 75-
186). New York: Springer.
Hennig-Thurau, T., & Walsh, G. (2004). Electronic word of mouth: Motives for and consequences of reading customer
articulations on the internet. International Journal of Electronic Commerce, 8 (2), 51–74. DOI:
10.1080/10864415.2003.11044293
Higgens, T. E. (2000). Social cognition: Learning about what matters in the social world. Journal of Social
Psychology, 30, 3-39. DOI: 10.1002/(SICI)1099-0992(200001/02)30:1<3::AID-EJSP987>3.0.CO;2-I
Hinkin, T. R. (1998). A brief tutorial on the development of measures for use in survey questionnaires. Organizational

56
 Research Methods, 1 (1), 104-121. Retrieved from:
http://scholarship.sha.cornell.edu/cgi/viewcontent.cgi?article=1515&context=articles
Huszár, F. (2012, November 27). The rising science of social influence: How predictable is your online behaviour.
TechCrunch. Retrieved from: https://techcrunch.com/2012/11/27/the-rising-science-of-social-influence-how-
predictable-is-your-online-behaviour/
Juster, F. T. (1966). Consumer-buying intentions and purchase probability: An experiment in survey design. Journal of
the American Statistical Association, 61 (315), 658-696. DOI: 130.115.77.66
Kaplan, D. (2004). Preface. In D. Kaplan, The SAGE Handbook of Quantitative Methodology for the Social Sciences
(pp. 3-6). Thousand Oaks: Sage Publications.
Khaniwale, M. (2015). Consumer-buying behavior. International Journal of Innovation and Scientific Research, 14 (2),
278-286. Retrieved from: http://www.ijisr.issr-journals.org/abstract.php?article=IJISR-14-129-01
Kiesler, S., & Sproull, L. S. (1986). Response effects in the electronic survey. The Public Opinion Quarterly, 50 (3),
402-413. Retrieved from: http://www.jstor.org/stable/pdf/2748727.pdf
Kim, J. (2012). The institutionalization of youtube: From user-generated content to professionally generated content.
Media, Culture & Society, 34 (1), 53–67. DOI: 10.1177/0163443711427199
Langner, S., Hennigs, N., & Wiedmann, K.-P. (2013). Social persuasion: Targeting social identities through social
influencers. Journal of Consumer Marketing, 30 (1), 31-49. DOI: 10.1108/07363761311290821
Lee, J. E., & Watkins, B. (2016). Youtube vloggers' influence on consumer luxury brand perceptions and intentions.
Journal of Business Research, 69, 5753–5760. DOI:10.1016/j.jbusres.2016.04.171
Li, H., Cui, J.-T., & Ma, J.-F. (2015). Social influence study in online networks: A three-level review. Journal of
Computer Science and Technology, 30 (1), 184-199. DOI: 10.1007/s11390-015-1512-7
Marshall, C. (2014, Febuary 4). YouTube and the beauty industry: How brands are getting crushed [report].
Retrieved on January 9, 2017, from TubularInsights: http://tubularinsights.com/youtube-beauty-industry/
Marshall, P. D., & Barbour, K. (2015). Making intellectual room for persona studies: A new consciousness and a
shifted perspective. Persona Studies, 1 (1), 1-12. Retrieved from:
https://ojs.deakin.edu.au/index.php/ps/article/view/464
Martawilaga, A. I., & Purwanegara, M. S. (2016). Information acceptance of electronic words of mouth (ewom) and
purchase intention through haul videos youtube. Journal of Business and Management, 5 (5), 651-660.
Retrieved from: http://journal.sbm.itb.ac.id/index.php/jbm/article/viewFile/2006/1081
MediaBookers. (2017, May 09). Adverteren op VIVA.nl. Retrieved from Mediabookers:
https://www.mediabookers.nl/adverteren/vivanl.html
Molyneaux, H., O’Donnell, S., Gibson, K., & Singer, J. (2008). Exploring the gender divide on youtube: An analysis of
the creation and reception of vlogs. American Communication Journal, 10 (2), 1-14. Retrieved from:
https://www.researchgate.net/profile/S_Odonnell/publication/44070600_Exploring_the_Gender_Divide_on_Y
ouTube_An_Analysis_of_the_Creation_and/links/02e7e5379f25dea942000000.pdf
Pixability. (2015). Beauty on YouTube. Retrieved from: https://cdn2.hubspot.net/hubfs/11326/Industry_Study_PDFs/
Pixability_2015_Beauty_on_YouTube_Study.pdf
Raven, B. H. (2008). The bases of power and the power/interaction model of interpersonal influence. Analyses of
Social Issues and Public Policy, 8 (1), 1-22. DOI: 10.1111/j.1530-2415.2008.00159.x
Simonson, I. (2005). In defense of consciousness: the role of conscious and unconscious inputs in consumer choice.
Journal of Consumer Psychology, 15 (3), 211-217. DOI: 10.1207/s15327663jcp1503_5

57
Stichting Wetenschappelijk Onderzoek Commerciële Communicatie. (2017, February 2). Is influencer marketing via
vloggers eigenlijk wel effectief? Retrieved on February 4, 2017, from Maketingfacts:
http://www.marketingfacts.nl/berichten/is-influencer-marketing-via-vloggers-eigenlijk-wel
effectief?utm_content=buffer92a17&utm_v medium=social&utm_source=facebook.com&u
tm_campaign=buffer
Van Selm, M., & Jankowski, N. W. (2006). Conducting online surveys. Quality & Quantity, 40, 435–456.
DOI:10.1007/s11135-005-8081-8
Vermeulen, I. E., & Seegers, D. (2008). Tried and tested: The impact of online hotel reviews on consumer
consideration. Tourism Management, 1-5. DOI: /10.1016/j.tourman.2008.04.008
Walther, J. B., Carr, C. T., Choi, s. W., Deandrea , D. A., Kim , J., Tong , S., et al. (2011). Interaction of interpersonal,
peer, and media influence sources online: A research agenda for technology convergence. In Z.
Papacharissi, A Networked Self (pp. 17-38). New York: Routledge.
Wiedmann, K.-P., Hennigs, N., & Langner, S. (2010). Spreading the word of fashion: Identifying social influencers in
fashion marketing. Journal of Global Fashion Marketing, 1 (3), 142-153. DOI:
10.1080/20932685.2010.10593066

58
Appendix A: survey
Personal factor (Khaniwale, 2015)

1. Please specify your gender

o Male
o Female

2. Please specify your age

o [open]

3. What is the highest level of education you have completed?

o Primary school
o High school
o Vocational (MBO)
o Bachelor (HBO)
o Bachelor (WO)
o Master (HBO)
o Master (WO)

4. What is your current occupational status?

o I am a student with a job on the side
o I am a student without a job on the side
o I am self-employed
o I work for a boss
o I am unable to work
o I am out of work but currently looking
o I am out of work but currently not looking

5. How would you describe your current financial situation?

o I am struggling
o I am maintaining
o I am just comfortable
o I am very comfortable

Experience (Coursaris & Van Osch, 2016)

6. How often do you watch YouTube videos

o Every day
o Several times per week
o Once per week
o Less than once per week
o Never

7. How often do you watch beauty-related videos on YouTube?

o Every day
o Several times per week
o Once per week
o Less than once per week
o Never

59
Familiarity (Khaniwale, 2015)

8. Are you familiar with [Name beauty vlogger 1]

o Yes I am familiar with that person
o I have heard about that person before, but that is all
o No I am not familiar with that person

9. Are you familiar with [Name product 1]

o Yes I am familiar with that product
o I have heard about that product before, but that is all
o No I am not familiar with that product

PSI: social attractiveness, physical attractiveness and homophily (Lee & Watkins, 2016)

SHOW VIDEO 1: beauty vlogger introduces herself

10. Please indicate to extent you agree with the following statements:
Rate from 1 – strongly agree to 7 – strongly disagree

Social attractiveness
1. This person is kind
2. I could see myself be friends with this person
3. I would like to converse with this person

Physical attractiveness
4. I think this person looks good
5. I think this person looks sexy
6. I think this person is physically attractive

Homophily
1. This person thinks the way I do
2. This person shares my values
3. This person has a lot in common with me
4. This person is a lot like me

SHOW VIDEO 2: PRODUCT TESTING

Big Five Model: openness to experience, conscientiousness, agreeableness & (Biel & Gatica-
Perez, 2013; Ananda & Wandebori. 2016)

11. Please indicate to what extent you agree with the following statements:
Rate from 1 – strongly agree to 7 – strongly disagree

Openness to experience
1. I think this person is curious
2. I think this person is imaginative
3. I think this person is original

Conscientiousness
1. I think this person is efficient
2. I think this person it organized
3. I think this person is responsible
4. I think this person is thorough

60
 Agreeableness
1. I think this person is very kind
2. I think this person is sympathetic
3. I think this person is forgiving

Extraversion
1. I think this person is enthusiastic
2. I think this person is energetic
3. I think this person is active
4. I think this person is assertive

Source credibility (Gillin, 2007)

12. Please indicate to what extent you agree with the following statements:
Rate from 1 – strongly agree to 7 – strongly disagree

1. I believe this person has great market knowledge

2. I believe this person wants to be involved
3. I believe this person is transparent
4. I believe this person encourages others to voice their opinions
5. I believe this person is involved with her audience

Consumer-buying intent (Baker and Churchill Jr., 1997)

13. Would you like to try [Name product]?

o Definitely yes
o Probably yes
o Might or might not
o Probably not
o Definitely not

14. Would you buy [Name product] if you came across it in a store?
o Definitely yes
o Probably yes
o Might or might not
o Probably not
o Definitely not

15. Would you actively seek out to buy [Name Product] in a store or online?
o Definitely yes
o Probably yes
o Might or might not
o Probably not
o Definitely not

Questions 5 to 12 are repeated for another beauty vlogger

61
Appendix B: results of the pretest
Table B1: results of the pretest (N = 70)
Social Physical Homophily Openness to Conscientiousn Agreeableness Extraversion Source
attractiveness attractiveness experience ess credibility
BeautyGloss 2,87 3,8 4 2,47 2,3 2,67 2,45 2,64
BeautyNezz 5,2 5,13 5,9 2,6 4,1 4,07 1,9 3,64
D for Dazzle 2,2 3,27 2,85 2,27 3,35 2,2 2,15 2,64
BeautyLab 2,53 2,53 3,25 1,73 1,9 2,4 2,4 2,72
Vera Camilla 2,93 4 3,4 2,93 2,85 2,8 3,15 3,04
LifeSplash 3 2,47 3,8 2,93 2,5 3,07 2,65 3,52
Bibi Breijman 3,13 2,07 3,45 2,6 2,45 3,13 2,3 2,76
Laura 4,07 2,67 3,85 3 2,8 3,2 2,95 2,6
Ponticorvo
Anna Nooshin 2,27 2 3,45 1,93 2,1 2,4 2,1 2,16
Jessie Maya 3,6 4,47 4,4 1,8 3,2 3,33 2,7 2,16
LiveLifeGorgeo 3,67 4,07 4,05 2,87 2,5 3,2 3,55 3,24
us
Saar 3,73 3 4,25 2,4 3,05 3,07 2,1 3,08
MissLipgloss 2,87 3,2 3,45 2,67 2,85 2,13 2,55 2,52
ShelingBeauty 2,6 2,27 3,6 2,13 2,5 2,47 2,35 2,64
Note : the results of the pretest have been calculated with an index average. Therefore the above results reflect the average scores on a scale from 1 – strongly agree, to 7 –
1

strongly disagree.
Note2: the dark and light grey shading highlights highest en lowest scores per variable

Appendix C: grouping of beauty vloggers

Table C1: grouping of beauty vloggers
Beauty vlogger 1 Beauty vlogger 2
BeautyNezz D for Dazzle
Bibi Breijman Jessie Maya
ShelingBeauty Saar
BeautyLab Laura Ponticorvo
MissLipgloss LiveLifeGorgeous
BeautyGloss Lifesplash
Vera Camilla Anna Nooshin

Appendix D: overview correlations of all persona dimensions

and buying intent
Table D1: Correlations overview of all persona dimensions and buying intent (N = 207)
Source Social Physical Homophily Openness to Conscientiousness Extraversion Agreeableness Buying
credibility attractiveness attractiveness experience intent
R P R P R P R P R P R P R P R P R P
Source credibility X X .574 .000 .385 .000 .530 .000 .417 .000 .459 .000 .581 .000 .528 .000 .381 .000
Social .574 .000 .445 .000 .621 .000 .315 .000 .384 .000 .430 .000 .607 .000 .410 .000
attractiveness
Physical .385 ,000 .445 .000 X X .273 .000 .167 .016 .311 .000 .319 .000 .199 .004 .324 .000
attractiveness
Homophily .530 .000 .621 .000 .273 .000 X X .383 .000 .417 .000 .267 .000 .417 .000 .335 .000
Openness to .417 .000 .315 .000 .167 .016 .383 .000 .242 .000 .448 .000 .282 .000 .246 .000
experience
Conscientiousness .459 .000 .384 .000 .311 .000 .417 .000 .242 .000 X X .222 .001 .350 .000 .132 .058
Extraversion .581 .000 .430 .000 .319 .000 .267 .000 .448 .000 .222 .001 X X .448 .000 .358 .000
Agreeableness .528 .000 .607 .000 .199 .004 .417 .000 .282 .000 .350 .000 .448 .000 X X .230 .001
Buying intent .381 .000 .420 .000 .324 .000 .335 .000 .246 .000 .132 .058 .358 .000 .230 .001 X X

62
 Appendix E: significant variables from the overall OLS
regression
Table E1: overview of the significant variables from the overall OLS regression (N = 207)
B SE Beta P CI 95% (lower CI 95% (upper
bound) bound)
Social attractiveness .236 .054 .305 .000 .130 .342
Physical attractiveness .131 .052 .175 .012 .029 .233
Product familiarity .285 .100 .179 .005 .088 .482
Constant 1.171
R2 .224

Appendix F: results of the linear mixed model

Table F1: overview of all variables in the linear mixed model (N = 207)
b SE P CI 95% (lower CI 95% (upper
bound) bound)
Source credibility .142 .102 .167 -.060 .344
Social attractiveness .194 .090 .034 .015 .372
Physical attractiveness .057 .071 .419 -.083 .198
Homophily .080 .079 .311 -.076 .236
Openness to experience -.116 .070 .099 -.253 .022
Conscientiousness -.173 .077 .027 -.326 -.020
Extraversion .200 .091 .030 .020 .381
Agreeableness .029 .066 .668 -.103 .160
Age 13 – 17 .161 .711 .821 -.1.249 1.570
Age 13 – 17 0b 0b . . .
Age 18 – 23 .288 .674 .670 -1.048 1.623
Age 18 – 23 0b 0b . . .
Age 24 – 34 .610 .682 .373 -.742 1.961
Age 24 – 34 0b 0b . . .
Age 35 – 48 -.178 .759 .815 -1.681 1.325
Age 35 – 48 0b 0b . . .
Education (primary school) .187 .453 .680 -.710 1.084
Education (high school) -.174 .346 .616 -.861 .511
Education (vocational MBO) .402 .381 .293 -.353 1.158
Education (bachelor HBO) -.238 .400 .553 -1.029 .554
Education (bachelor WO) -.064 .346 .853 -.749 .620
Education (master HBO) -.619 .540 .254 -1.689 .451
Education (Master WO) 0b 0b . . .
Financial situation (struggling) .210 .351 .550 -.485 .906
Financial situation (maintaining) .344 .322 .289 -.295 .983
Financial situation (just comfortable) .074 .309 .812 -.538 .686
Financial situation (very comfortable) 0b 0b . . .
YouTube experience (every day) -.171 .653 .794 -1.463 1.121
YouTube experience (several times per week) -.403 .618 .515 -1.628 .821
YouTube experience (once per week) -.685 .617 .270 -1.907 .538
YouTube experience (less than once per -.562 .642 .384 -1.834 .710
week)
YouTube experience (never) 0b 0b . . .

63
Beauty video experience (every day) -.535 .415 .200 -1.357 .287
Beauty video experience (several times -.647 .307 .037 -1.256 -.039
per week)
Beauty video experience (once per week) -.994 .380 .010 -1.747 -.241
Beauty video experience (less than once per .039 .252 .877 -.460 .538
week)
Beauty video experience (never) 0b 0b . . .
Vlogger familiarity (yes familiar) -.127 .221 .565 -.564 .310
Vlogger familiarity (heard about this person) .194 .213 .365 -.229 .617
Vlogger familiarity (not familiar) 0b 0b . . .
Product familiarity (yes familiar) -.471 .260 .073 -.986 .045
Product familiarity (heard about it) -.362 .201 .073 -.760 .035
b b
Product familiarity (not familiar) 0 0 . . .
Intercept 1.788
b = This parameter is set to zero because it is redundant

64