the effect of influencer marketing in a highly involved ...1445862/fulltext01.pdf · effect of...

86
The effect of Influencer Marketing in a highly involved product An experiment conducted in the golf industry Julia Halin Business Administration, master's level (120 credits) 2020 Luleå University of Technology Department of Business Administration, Technology and Social Sciences

Upload: others

Post on 31-Dec-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

The effect of Influencer Marketing in a highly involved product

An experiment conducted in the golf industry

Julia Halin

Business Administration, master's level (120 credits) 2020

Luleå University of Technology Department of Business Administration, Technology and Social Sciences

Page 2: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century
Page 3: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

Acknowledgement This master thesis represents the final project for me as a student at Luleå University of

Technology. The thesis was conducted during the spring of 2020 and I would like to thank

everyone who has been involved in the development of this research study.

First of all, a sincere thanks to the company I got the opportunity to create this study together

with. They have supported me in many different ways during this journey and I am grateful for

the knowledge and insight in the company that I got during this time. Without their help, I

would not have been able to reach the results I present in this thesis. Also, a sincere thanks to

my supervisor Kerry Chipp who has contributed with feedback, recommendations and helped

me move forward in my job constantly.

I would also like to express my appreciation to all participants that made the effort to answer

my inquiries. Both the participants of my questionnaire and the participant in my interview.

Further, I would like to thank all course members who have sustained me with useful comments

and feedback along the journey. Finally, I would like to thank family and friends that have

supported me during this spring!

Stockholm, 23th of June 2020

_____________________

Julia Halin

Page 4: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century
Page 5: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

Abstract Purpose – The purpose of this research is to create an understanding of how influencer

marketing could be used to increase the purchase intention in an area of a high involvement

product in a niche market. Since this study is collected together with a Swedish golf company,

the golf industry was evaluated. In order to fulfill the purpose, the social media influencer value

(SMIV) model was adopted in this study.

Method – A quantitative experiment was conducted through the use of experimental vignette

methodology (EVM) together with a qualitative validity check-in form of an interview with a

key person from the industry. The main data was collected through a combination of exposure

to the independent variable (through a stimulus) and a set of following questions (a

questionnaire). The questionnaire was sent out to the company’s consumer through email and

to analyze the collected data an analysis of covariance (ANCOVA) was used.

Findings – In the collected data, evidence indicating that golf is a high involvement activity

where the participants put down a lot of time and effort in the sport was found. In the SMIV

model three significant paths was found between expertise and purchase intention, similarity

and purchase intention, and usefulness and purchase intention. All three independent variables

had a positive impact on purchase intention which tells us that when expertise, similarity, or

usefulness increases, consumers purchase intension increases as well.

Theoretical and practical implications – The results in this study contributes to current

literature in the area of influencer marketing by confirming some relationships but also by

extending the literature. The study also contributes to important insights for marketing

practitioners since the results present important variables to consider when using influencer

marketing to create purchase intention. It also creates meaningful insights for the influential

person itself and open up for further research in the area of the effect of influencer marketing

in a high involvement product.

Keywords: Influencer marketing, source credibility, high involvement, purchase intention,

sports industry.

Page 6: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

Sammanfattning Syfte – Syftet med denna studie är att skapa en förståelse för hur man, med hjälp av influencer

marketing, kan öka köpintensionen bland konsumenter av en högengagemangs produkt som

finns på en nischad marknad. Då denna studien är genomförd tillsammans med ett svenskt

golfföretag, är influencer marketing applicerat på en golfprodukt. För att uppnå syftet med

studien, användes en model ”social media influencer value (SMIV)” modellen för att försöka

förstå de fundamentala mekanismerna av influencer marketing i golfindustrin.

Metod – Denna studien är en kvantitativ studie baserad på ett experiment, experimental vignette

methodology (EVM), och en kvalitativ validitetskontroll i form av en intervju tillsammans med

en person från branschen. För att samla in primärdata skickades en enkät ut, via mail, till

företagets kunder. Enkäten bestod av en bild följt av ett antal frågor kopplade till bilden. All

insamlade data var sedan analyserad i SPSS med hjälp av ANVOCA.

Upptäckter – Baserat på resultatet från ANCOVA analysen visade sig tre variabler (likhet,

expertis och användbarhet) ha en signifikant nivå i SMIV modellen. Baserat på

bakgrundsinformationen av varje deltagare kunde man utläsa att Golf är en högengagemangs

aktivitet där de som spelar golf lägger ner mycket tid och engagemang i sporten. Det visade sig

också att när denna målgruppen exponeras för ett Instagram-inlägg där personen känner likhet

till influencern, känner att inlägget är användbart eller att influencern är expert inom området

ökar deras köpintension av produkten som presenteras av influencern.

Bidragande upptäcker till teorin samt praktiken – Resultatet i denna studien bidrar och stärker

vad tidigare forskning har hittat samtidigt som den expanderar tidigare forskning, både med

nya variabler som visar sig ha en påverkande faktor samt nya upptäckter. Studien skapar även

nya, viktiga insikter som kan vägleda marknadsförare i beslutet om att använda influencer

marketing i mer nischade branscher. Studien bidrar även till nyttig information för influencern

den själv som får kunskap av vad som skapar värde i sammarbeten. Studien öppnar även upp

för vidare forskning inom effekterna av influencer marketing i en högengagemangsprodukt.

Nyckelord: Influencer marketing, source credibility, högengagemangsprodukter, köpintention,

sportindustrin.

Page 7: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

Table of Content

1. Introduction ..................................................................................................................................... 1 1.1 Background ........................................................................................................................................... 1 1.2 Problem discussion ................................................................................................................................ 2 1.3 Purpose .................................................................................................................................................. 4 1.4 Delimitations ......................................................................................................................................... 4 1.5 The international aspect ........................................................................................................................ 4

2. Literature review .............................................................................................................................. 5 2.1 Influencer marketing ............................................................................................................................. 5 2.2 Source credibility .................................................................................................................................. 6

2.2.1 An overview of the Social Media Influencer Value (SMIV) model ................................................ 8 2.3 Factors in the adapted model .............................................................................................................. 10

2.3.1 High- and low-involvement ............................................................................................................ 10 2.3.2 Purchase intentions ......................................................................................................................... 11 2.3.3 Perceived trust ................................................................................................................................ 12 2.3.4 Trustworthiness .............................................................................................................................. 13 2.3.5 Expertise ......................................................................................................................................... 14 2.3.6 Similarity ........................................................................................................................................ 15 2.3.7 Usefulness ....................................................................................................................................... 16 2.3.8 Moderators ...................................................................................................................................... 17 2.3.9 Variables excluded from the model ................................................................................................ 18

2.4 Golf versus fashion .............................................................................................................................. 20

3. Methodology ................................................................................................................................... 23 3.1 Research purpose ................................................................................................................................ 23 3.2 Research approach .............................................................................................................................. 23 3.3 Research strategy ................................................................................................................................ 24 3.4 Data collection .................................................................................................................................... 26

3.4.1 Measurement instrument ................................................................................................................ 26 3.4.2 Experimental Vignette Methodology ............................................................................................. 29

3.5 Credibility ............................................................................................................................................ 40 3.5.1 Reliability ....................................................................................................................................... 40 3.5.2 Validity ........................................................................................................................................... 41

3.6 Summary of methodology .................................................................................................................... 42

4. Data results .................................................................................................................................... 43 4.1 Description of the final sample ........................................................................................................... 43 4.2 Testing assumptions ............................................................................................................................ 46

4.2.1 Reliability of covariates .................................................................................................................. 46 4.2.2 Correlations among the covariates ................................................................................................. 47 4.2.3 Homogeneity .................................................................................................................................. 49

4.3 Significant of the data results .............................................................................................................. 50

5. Analysis of data .............................................................................................................................. 53 5.1 High involvement ................................................................................................................................. 53

Page 8: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

5.2 Analysis of H1 ..................................................................................................................................... 54 5.3 Analysis of H2a, H2b, H2c H2d, and H2e .......................................................................................... 55 5.4 Analysis of H3a, H3b, H3c, H3d, and H3e ......................................................................................... 56 5.5 Analysis of the moderators .................................................................................................................. 57 5.6 Significant paths .................................................................................................................................. 57

6. Findings and conclusions ................................................................................................................ 59 6.1 Answer to research question ............................................................................................................... 59

6.1.1 Specific findings toward the collaborating company ..................................................................... 60 6.2 Contribution to the theory ................................................................................................................... 60 6.3 Implications for practitioners .............................................................................................................. 61 6.4 Limitations and further research ......................................................................................................... 61 6.5 Conclusions ......................................................................................................................................... 63

Reference list ........................................................................................................................................... 64

Appendix A ............................................................................................................................................. 70

Appendix B ............................................................................................................................................. 72

Appendix C ............................................................................................................................................. 75

Appendix D ............................................................................................................................................. 76

Page 9: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

List of figures Figure 2.1: Original SMIV model………………………………………………………………………… 8 Figure 2.2: Proposed hypothesis H1………………………………………………………………………. 13 Figure 2.3: Proposed hypothesis H2a-d & H3a-d added……………………………………...................... 17 Figure 2.4: Proposed SMIV model……………………………………………………............................... 18 Figure 3.1: Adapted model after the qualitative validity check of the model....………………………….. 28 Figure 3.2: Steps and decisions points in conducting an EVM study…………………………………….. 29 Figure 3.3: The vignettes used in the questionnaire………………………………………………………. 33 Figure 3.4: Summary of methodology…………………………………………………………………….. 42 Figure 4.1: Significant paths………………………………………………………………………………. 52

List of Tables Table 3.1: Types of variables…...…………………………………………………………………………. 25 Table 3.2: Source’s for each variable……………………………………………………………………... 28 Table 3.3: Question construction………………………………………………………………………….. 39 Table 4.1: Dummy variables…...………………………………………………………………………….. 43 Table 4.2: Participants gender…………………………………………………………………………….. 44 Table 4.3: Participants age……………………………………………………………................................ 44 Table 4.4: Participants country of residence……………………………………………………................. 45 Table 4.5: When the participants mainly play golf………………………..…………................................. 45 Table 4.6: Number of golf rounds participants play each year ……….…………………………………... 46 Table 4.7: Years of interest/played golf among the participants.…….…………………………………... 46 Table 4.8: Cronbach Alpha…………………………………...……….…………………………………... 47 Table 4.9: Correlations…..…………………………………...……….…………………………………... 48 Table 4.10: Distribution among the variables ……….……...……….…………….……….……………... 48 Table 4.11: Homogeneity of regression slopes……………………………………………………………. 49 Table 4.12: Levene’s Test of Equality of Error Variances………………………………………………... 50 Table 4.13: Significant level in ANCOVA…...….….……...…….….…………….……….……………... 51 Table 4.14: Estimated Marginal Means….…...….….……...…….….…………….……….……………... 51

Page 10: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

1

1. Introduction This chapter will introduce the reader to a background and an understanding of why this research area is important. From the literature in the background, a problem discussion will be presented which focuses on the lack of theory in the area of influencer marketing in the high involvement category. Thereafter, the purpose of the thesis, a research question, and delimitations will be presented.

1.1 Background Despite a growing volume of research on influencer marketing, a lack of understanding of the

effect of influencer marketing in high involvement products is still partial. In the 21st century

a focus on the personalities, termed influencers, has arisen and the term source credibility has

been shown to be an important factor to be considered in the influencer marketing literature

(Lou & Yuan, 2019; Munnukka, Uusitalo, & Toivonen, 2016; Xiao, Wang & Chan-Olmsted,

2018). Previous use of celebrity endorsers has during the years moved over to the use of

influencers as endorsers instead of celebrities. Traditional endorsers are normally public figures

or celebrities that have gained their publicity through traditional media, when influencers are

“regular” people that have become famous through their social media channels (Lou & Yuan

2019). The use of social media has had an expanding phase the last years and in a recent report

from Statista, numbers showed that in June 2018, Instagram reached one billion users with a

monthly activity on this channel (Statista, 2019a). Data also shows that all ages are slowly

becoming members of the social media platform Instagram (Statista, 2019b), which has a huge

impact on the ability of companies to reach out to potential customers through this platform.

The success of Influencer marketing has changed the way companies reach out to consumers

nowadays, and statistics show that 88% of customers trust online reviews by strangers as much

as they would trust a recommendation by a friend (Ahmad, 2018). Recent studies have also

revealed that influencer marketing can drive yearly incremental sales at a rate 11 times greater

compared to conventional digital marketing (Ahmad, 2018; Kemp, Randon McDougal, &

Syrdal, 2019). The exponential growth of influencer marketing has influenced more companies

to considering the use of this strategy. Recent reports show that marketers intend to invest even

more on influencer marketing in the near future (Hughes, Swaminathan, & Brooks, 2019; Lou

& Yuan, 2019; Reinikainen, Munnukka, Maity & Luoma-aho, 2020). This growth increases the

importance of understanding the underlying mechanisms of the strategy’s success in a world

Page 11: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

2

where consumers nowadays are facing thousands of commercials on a daily basis (Ganguly,

2015).

The phenomenon has been investigated from various perspectives and researchers have made

attempts to define factors that need to be considered to achieve an effective outcome when

considering influencer marketing as a strategy (De Veirman, Canberghe & Hudder, 2017;

Djafarova & Trofimenko, 2019; Munnukka et al., 2016; Reinikainen, et al., 2020). However,

Lou and Yuan (2019) recently discussed that even if there has been research on influencer

advertising, there still is a gap in the knowledge of what the fundamental mechanisms of the

strategy are. As a result of this gap, Lou and Yuan (2019) created a theoretical model, the social

media influencer value (SMIV) model, that aimed to explain the important mechanisms that

makes influencer marketing effective.

Although the SMIV model aimed to cover the existing gap, the gap of analyzing the

effectiveness of influencer marketing in more high involvement product categories still exist.

Lou and Yuan’s (2019) study, together with the majority of previous literature in this area, has

mainly analyzed influencer marketing from two perspectives, the low involvement product

perspective and the phenomenon more in general (Lou & Yuan, 2019; Kunkel, Walker &

Hodge, 2019; Reinikainen, et al., 2020). Recently there has been some updated literature testing

the effect of influencer marketing in different product categories (Kunkel, et al., 2019; Lee &

Watkins, 2016; Foti and Devine, 2019) or based on the consumers’ involvement level in the

product (Handriana, 2017; Lee & Watkins, 2016), however, these studies are relatively new

and in need of further research. Since influencer marketing as a marketing strategy increases,

new industries and product categories implementing the strategy increases as well. Therefore,

there is a need for extended literature regarding the use and effectiveness of influencer

marketing in different contexts.

1.2 Problem discussion

Based on influencer marketing’s growth previous research has been conducted and the topic

are commonly represented in literature already. However, the phenomenon has mainly been

researched in a more general aspect or in the context of low involvement products (Lou & Yuan

2019; De Veirman et al., 2017; Evans, Phua, Lim, & Jun, 2017; Djafarova & Trofimenko,

2019). Due to the lack of literature, covering empirical evidence of how effective influencer

marketing is in other product category, such as high involvement products, the maximal

Page 12: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

3

effectiveness of this strategy, in more complex and niche markets, is hard to create without

assuming what will work. Since there is a difference in consumers indention-behaviour between

low- and high-involvement products (Foti & Devin’s, 2019), previous findings may not be

applicable in other contexts. This lack of knowledge could, in the long run, harm the

development of using influencer marketing in those areas.

One area that has recently explored the effect of influencer marketing is the sports industry.

The use of influencer marketing in the industry has increased during the past years and it is now

considered an important marketing tool (Maderer, Parganas, & Anagnostopoulos, 2018).

However, sports include different activities and difference age profiles which makes it hard to

draw only a few conclusions of what is working in the sports industry. Since people usually

puts down a lot of time and engagement in sports, sports could be considered as a high

involvement activity. But some sports needs more engagement than others and based on the

different level of engagement a general conclusion of the effect of influencer marketing in the

sports industry is not enough. One example of a sport, that might not get the same result when

using general findings in this area, is Golf. Golf is a sport where the engagement from the

practitioners is high, it is practiced by all ages and the practitioners put down a lot of time and

money into the sport. Golf has for many become an established community and culture, and

today’s literature is not covering this context.

In today’s literature, that discuss influencer marketing in the sports industry, the findings mainly

focus on the effect sports team could gain from using social network as branding-enhancement

tools (Maderer et al., 2018; Anagnostopoulos, Parganas, Chadwick & Fenton, 2018), or they

have focused on consumer perspective of the endorsers used in influencer marketing for a sports

brand, such as Nike (Kunkel, et al., 2019). But if a high involvement activity, such as Golf,

launches a product/services that facilitates for the practitioner, and the company behind it wants

to promote this product through influencer marketing, there is today no literature explaining

how to use influencer marketing to increase the consumers purchase intension of this product.

Since influencer marketing is growing, further research needs to be done in the area to creating

an understating for different product categories. Due to the lack in literature, this study, in

behalf of a golf company, will test and adopt the SMIV model to evaluate the fundamental

mechanisms in a high involvement participatory product in the golf industry. Managing to

establish a deeper understanding of the fundamental mechanism that explains what makes

influencer marketing effective, managers could potentially implement more effective strategies

Page 13: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

4

in the researched area (Lou & Yuan, 2019). For researchers, this insight could contribute with

a deeper understanding of if there is any need for further research in the area of influencer

marketing in different product categories or if one general strategy is enough.

1.3 Purpose The purpose of this research is to create an understanding of how influencer marketing could

increase the purchase intention in the area of high involvement products in a niche market.

Since this study is based on a collaboration with a Swedish golf company, with the aim of

creating an understanding of how the company can use influencer marketing to generate

purchase intent toward one of their products, the SMIV model will be adopted in the golf

industry. Grounded on the problem discussion above, together with the aim from the company,

the following research question has been developed in order to answer the purpose of this study:

How is the purchase intention generated through the use of social media networks in a highly

involved, niche market?

To answer the research question, three hypotheses will be developed and presented in the

following chapter.

1.4 Delimitations One limitation of this study is that the data will be collected on consumers that are already

involved in the product and therefore have high brand awareness, and a relationship with the

product since earlier. Consequently, the findings of this study may have been different if they

were collected on participants that had not had a relationship with the brand earlier.

1.5 The international aspect This study is conducted in collaboration with a company where its consumers can be found all

over Europe. Hence, the data in this study will be collected among different nationalities, and

therefore will the conclusions be drawn with a perspective focusing on strategies working for

participants living around Europe.

Page 14: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

5

2. Literature review To be able to understand the use of social media networks to increase purchase intention in a high involvement category the literature will first turn to what influencer marketing is, and later cover the main aspects of the model and how it will be adopted. Lastly, the chapter will be concluded with a review of the high involvement participations in the category of golf in contrast to the original study on fashion.

2.1 Influencer marketing Influencer marketing has become an advertising and marketing technique that utilizes the

impact of influential individuals or opinion leaders that online try to influence the brand

awareness and/or purchasing decisions of consumers’ (Lou & Yuan, 2019). De Veirman et al.,

(2017) describes opinion leaders as persons sharing content that focuses on their areas of

expertise and their specific interests instead of providing content focusing on themselves.

Another explanation, that is built on Corey’s (1971) definition of opinion leaders, is that those

are said to influence consumers through their expertise and knowledge. During the previous

years, opinion leaders have in literature been equal with the term influencers (De Veirman et

al., 2017; Corey’s, 1971), that can be defined as “a new type of independent third party endorser

who shape audience attitudes through blogs, tweets, and the use of other social media” (Freberg,

Graham, McGaughey, & Freberg, 2011. p.90). McCracken (1989) also defines influencers as

"endorsers" which are "any kind of individual that delights in public recognition as well as that

utilizes this recognition on behalf of a customer good by appearing with it in an ad" (p. 310).

During the years, the use of influencers as endorsers for companies have become more and

more popular and the result of this was the phenomenon influencer marketing. Sigala (2017)

defined influencer marketing as “... the practice of identifying individuals who have influence

over a target audience” (p.94), Sudha and Sheena (2017) defined it as “... a process of

identifying and activating individuals who have an influence over a specific target audience or

medium, in order to be part of a brand's campaign towards increased reach, sales, or

engagement“ (p.16), and, De Veirman et al. (2017) describe influencer marketing as “rather

than pushing their (whether or not personalized) ads to their target audience, brands are turning

to trusted online personas to get their products and messages out to the consumers.” (p.813).

The main purpose of influencer marketing is to use these famous social media personalities to

influence consumers in different aspects. To reach the desired effect of this strategy, literature

have found factors such as trustworthiness, attractiveness or credibility, only to mention a few

among others, to be important elements to consider (De Veirman et al., 2017; Lou & Yuan

Page 15: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

6

2019; Evans et al., 2017; Djafarova & Trofimenko, 2019; Lee & Watkins, 2016). In the end,

the importance for companies lies in finding influencers that, with their personality, in an

effective way can impact the purchase decision of their follower-basis (Liu, Jiang, Lin, Ding,

Duan, & Xu, 2015).

Even though the positive outcomes of influencer marketing have been well highlighted in

literature and articles there is a concern regarding how genuine the influencer actually is in all

their different collaborations. Ahmad (2018), relished statistics showing that 60% of people

seeks advice from social media posts, vlogs or blogs before considering purchasing a product,

and 90% of consumers have been shown to trust recommendations from an influencer.

However, the question arises if the influencers recommendations regarding the products

actually are genuine and real, or if they mainly do it for money and that covert (misleading)

advertising might take place instead. Since the phenomenon is rather recent, there is still some

uncertainty regarding rules and guidelines in this area (Gurkaynak, Kama, & Ergün, 2018).

Another important aspect is the lack of internet infrastructure around the world that could still

be considered an obstacle (Van-Tien Dao, Nhat Hanh Le, Ming-Sung Cheng & Chao Chen,

2014).

Although, the majority of the literature has discussed the positive outcomes of the phenomenon

and De Veirman et al. (2017) discussed different reasons of why there is an importance in

forming a strong relationship with the right influencer from both the companies’ and the

influencers’ viewpoint. First of all, the chosen influencers’ own characteristics will be reflected

on the brand and therefore impact the association among the company’s consumers. Secondly,

the communication from the influencer needs to be relevant for the followers and therefore a

fit between the brand and the influencer needs to be existent (De Veirman et al., 2017). Since

this fit between the brand and influencer have become more important during the years, the

factors impacting the credibility of influencer used in the marketing need to be investigated and

understood by the company using influencer marketing.

2.2 Source credibility Perceived credibility, usually referred to as source credibility of the influencer, is one of the

factors determining the effectiveness of influencer endorsements (Munnukka et al., 2016). “

‘Source credibility’ is a term commonly used to imply a communicator’s positive

characteristics that affect the receiver’s acceptance of a message” (Ohanian, 1990. p.41)

Ohanian (1990) developed a scale in order to measure three factors that celebrity endorsers

Page 16: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

7

could be perceived with. These three factors were expertise, trustworthiness, and attractiveness.

By time, literature has extended the phenomenon source credibility with other factors as well

(Lafferty, Golfsmith, & Newell, 2002; Lim, Radzol, Cheah, & Wong, 2017; Reinikainen, et al.,

2020). One extended factor is similarity (Munnukka et al., 2016), that Ohanian (1990) discussed

in context of the regular use of celebrities that usually were “closely associated with both the

product and the target audience” (p.39). However, similarity, was only “employed as a measure

for nomological validity” (p.41) back then.

The phenomenon source credibility has been used in literature form many decades and has with

time also become an important part of the influencer marketing literature (Lim, et al., 2017;

Xiao et al., 2018; Reinikainen et al., 2020; Lou & Yuan, 2019). the phenomenon has been

evaluated in different contexts and Wheeler (2009) did a study where he tested the effect of

source credibility in high versus low involvement products with a result of a non-proved

hypothesis. Wheeler’s (2009) hypothesis tested if the source credibility was greater under low

involvement than under high involvement, which could not be proved. However, more recent

literature testing the relationship between source credibility and involvement level in a product

is still thin. Instead, literature have focused on testing source credibility in relationship to trust

in branded trust and purchase intention. Reinikainen et al. (2020) tested the relationship

between source credibility and branded trust where the relationship was found positive and

source credibility was found to have an indirect impact on the purchase intention. This was in

line with Lafferty’s et al. findings (2002) that indicated that an endorser’s credibility seemed to

influence consumers’ attitudes toward the brand and their purchase intentions indirectly. Lim

et al. (2017) though tested source credibility, only containing the factors of expertise and

trustworthiness of the influencer, in relationship with purchase intention, and found this

hypothesis rejected.

Common to hose studies mentioned above is that all of them have measured source credibility

as one variable in the hypothesis. However, there have also been literature looking at each

dimension of source credibility as separated variables instead of measured together to one

(Munnukka et al., 2016; Xiao et al., 2018; Lou & Yuan, 2019). Munnukka et al. (2016) looked

into a peer endorser’s credibility based on the factors of trustworthiness, expertness, similarity,

and attractiveness and the advertising effectiveness. Lou and Yuan (2019) continued to explore

the source credibility in their study where trustworthiness, expertise, similarity, and

Page 17: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

8

attractiveness were tested toward consumers’ trust in branded content. In their study they

created the SMIV model that will be further explained in the following paragraphs.

2.2.1 An overview of the Social Media Influencer Value (SMIV) model

As mentioned above, Lou and Yuan, (2019) created a model, see figure 1 below, built on

Munnukka et al. (2016) study where source credibility included four variables: expertise,

trustworthiness, attractiveness and similarity. Lou and Yuan (2019) measured source credibility

together with advertising value containing of two other factors, informative value and

entertainment value. These, in total six different factors, were tested towards three outcome

variables; trust in influencers’ branded posts, brand awareness, and purchase intention. The

results showed that the source credibility variables (expertise, trustworthiness, attractiveness,

and similarity) and the advertising value variables (informative value and entertainment value)

all had different effects on the three outcome variables.

Figure 2.1: Original SMIV model

Source: Lou and Yuan, 2019, p.63

In Lou and Yuan, (2019) study all variables were tested in separated hypotheses and research

questions. In total they had five hypotheses and two research question. H2, H5, and RQ 1 aimed

Advertising value

Source credibility

Informative value

Entertainment value

Expertise

Trustworthiness

Attractiveness

Similarity

Trust in branded posts

Brand awareness

Purchase intentions

Involvement Age Gender

H1&H2

H3 & RQ1

H4 & H5

RQ2a RQ2b

Page 18: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

9

to test the four factors of source credibility (expertise, trustworthiness, attractiveness, and

similarity) toward the three outcome variables (trust in influencers’ branded posts, brand

awareness, and purchase intention), and H1, H3, and H4 aimed to test the two factors of

advertising value (informative value and entertainment value) toward the three outcome

variables (Lou & Yuan, 2019).

In Lou and Yuan’s (2019) first research question (RQ1) the effect of the four source credibility

factors on brand awareness were tested. The findings showed that out of the four source

credibility variables, attractiveness and expertise had a positive impact. Similarity did not have

any relationship at all, and trustworthiness was shown to have a negative impact on brand

awareness. In the hypothesis regarding expertise, trustworthiness, attractiveness and

similarity’s effect on trust in influencer-generated branded posts (H2) all of the variables,

except expertise, had a positive impact. However, in the hypothesis regarding effect on

purchase intention (H5) trustworthiness actually had a negative effect on the consumers

purchase intention whereas expertise, attractiveness and similarity did not have any impact at

all. However, the authors also found that consumers that have trust in branded posts are

significantly influencing their brand awareness and their purchase intention. Thereby,

trustworthiness, attractiveness, and similarity have an indirect impact on purchase intentions

since they have a positive relationship with the trust in branded posts. Additionally, expertise

also had an indirect impact on purchase intention since it was found that brand awareness

impacted the purchase intention positively.

In hypotheses H1, H3, and H4 Lou and Yuan (2019) tested the advertising values variables,

informative value and entertainment value, toward the three outcome variables. For the factor

entertainment value, no relationship with any of the depended variables was found. The findings

regarding the information value was shown to be the only factor that actually had a positive

relationship with the dependent variable purchase intention. It was also shown to have a positive

relationship with the dependent variable trust in branded posts. However, none of the

advertising value factors was found to have a relationship with the variable brand awareness.

So, Lou and Yuan’s (2019) findings were spread, and the most unexpected finding was the

negative effect of trustworthiness. Something Lou and Yuan (2019) said deserved to be further

researched. Although, Lou and Yuan’s study is only one study with these results and recently

conducted so therefore in need of further research.

Page 19: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

10

2.3 Factors in the adapted model Based on Lou and Yuan’s (2019) findings and prior research have found in the context of source

credibility in the context of influencer marketing, a model for this study have been formed. The

standpoint for the model is Lou and Yuan’s (2019) SMIV model. However, the model has been

adapted to contain purchase intention, perceived trust, trustworthiness, expertise, similarity and

usefulness. All different factors will be discussed and presented below together with those

factors that have not been used in this study.

2.3.1 High- and low-involvement

The phenomenon of high- and low-involvement could be found in many different contexts. To

mention a few, it could be found in the context of work practices, decision making, management

and products. In an early stage, involvement was viewed as the relationship between a person’s

ego and an objective (Aldlaigan & Buttle, 2001). Later, researchers have developed the area

and Haynes, Pipkin, Black, and Cloud, (1994) said that; “Involvement, in general, is determined

by the perceived relevance of an object or activity to an individual. Involvement can occur with

an object (product involvement) or an activity (process involvement)” (p.23). In the context of

activity, purchase intention has been identified as an activity which people have been said be

able to involve with. The involvement level in purchase intentions is usually connected to the

information processing a person puts down in a product. In those products where a person does

not process much information and the product choice is not that important, a low involvement

is discussed. When the situation is opposite, high involvement is instead considered (Aldlaigan

& Buttle, 2001).

When it comes to the context of products, key elements to attract and maintain existing

consumers is to make distinguishing elements in a product that makes consumers want to

identify and buy the product (Radder & Huang, 2008). In prior literature, high-involvement

products have been said to usually have a symbolic meaning, psychological satisfaction or

image reinforcement, where in low-involvement products the consumption is less subject, and

consumers buy these products more on routine (Radder & Huang, 2008). Connecting back to

Aldlaigan and Buttle’s article 2001, Radder and Huang’s explanation of high versus low

involvement indicates different information processes and importance in the different levels as

well. A product bought on routine is less in need of a more extended information process than

a product with a higher meaning and satisfaction. Another finding of Radder and Huanger

(2008) was that high- and low-involvement products could need different types of marketing

Page 20: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

11

channels to get the desirable outcome of a purchase intention. However, Radder and Huanger’s

study was made in the early 2000, so the use of social media (Instagram, Facebook, Twitter)

was not that common during this timeframe and therefore these types of channels were not

discussed. In a more recent study made by Lee and Watkins (2016), looking into the impact of

the social media channel YouTube’s impact on the purchase intention on luxury products, the

result indicated that those consumers that had watched vlogs (on YouTube) promoting the

luxury products (usually considered high-involvement products) had a higher purchase

intention than consumers that had not watched the vlogs. Interesting conclusions, a couple of

years later than Lee and Watkins (2016), was Foti and Devin’s (2019) that said that “since the

intention–behaviour gap differs between low- and high-involvement products, it is necessary

to study high-involvement ethical consumption directly, rather than simply interpolate low-

involvement findings onto high-involvement consumption” (p.8). This really indicates the

importance to understand the difference between the high- and low-involvement products when

marketing a product.

Another study, discussing the effect of the consumer’s product involvement on a consumer’s

attitude toward an advertisement (further abbreviated as ad) was Munnukka et al. (2016),

(though not specific on social medias). According to Munnukka et al’s (2016) previous

literature had found that a consumer’s involvement with a product impacted their attitude

formation toward an ad and depending on the consumers’ level of involvement, an endorser

would be differently evaluated. The result of Munnukka et al’s (2016) study showed that the

effect of product involvement on attitude toward ad was significant. From previous studies it

was told that for consumers with low involvement, expertise and attractiveness was two

important factors and Munnukka et al (2016) extended those findings with their results

indicating a relationship between the consumers’ involvement in the product and perceived

similarity to the endorser.

2.3.2 Purchase intentions

Regardless if a company has a high- or low-involvement product, the outcome for the company

are the same; to create purchase intention among their consumers and sale their products.

Purchase intention have been a popular research topic in the prior marketing/advertising

literature (Ohanina, 1990; Lafferty et al., 2002; Amos, Holmes & Strutton, 2008; Lee &

Watkins 2016; Reinikainen et al., 2020; Lou & Yuan, 2019). The term has been referring to a

consumer’s possibility or likelihood to purchase a certain product (Lou & Yuan, 2019). Spears

Page 21: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

12

and Singh (2004) defined purchase intentions as “an individual’s conscious plan to make an

effort to purchase a brand” (p.56). Consumers’ attitudes toward ads have been found to have an

impact on the purchase intentions in different types of context and have lately also been tested

in the context of influencer marketing (Lee & Watkins, 2016; Reinikainen et al., 2020; Lou &

Yuan, 2019). Those factors that have had a direct impact on purchase intention have varied.

However, previous literature has found that the dimensions of source credibility have had a

positive effect on the consumers’ purchase intentions (Lafferty et al., 2002).

Although, prior studies have mainly been measured purchase intention in the area of low-

involvement products this gap made Foti and Devine (2019), to tried to understand consumers’

decision-making process in high-involvement ethical products in their study. They found that

two driving factors in this process were trust and information. “This is important as consumers’

cognitive dissonance is reduced when they trust the agent and are provided with reassuring

information” (Foti & Devine, 2019. p.9). Furthermore, Hung et al. (2010) studied the

antecedents of luxury brand purchase intention and argued that there is “considerable research”

in the area of luxury brand but few studies explaining the relationships influencing purchase

intention. Based on the authors’ results, they suggested marketers, working with luxury brand,

to concentrate on “maintaining the factors of authenticity (e.g historical background, overall

image & quality expectations)” and “strengthen symbolical meaning-construction attached to

the product” (p.473). Although, literature talking about antecedents of purchase intention in

high involvement products are still limited.

2.3.3 Perceived trust

One area that is not limited in literature regarding marketing and influencer marketing is trust.

Trust is important to consider in both high- and low-involvement products and researches have

suggested that if people perceive a higher degree of trust, they are more likely to purchase as

well (Corbitt, Thanasanki & Yi, 2003). Moorman, Deshpande and Zaltman (1993), describe

trust as “a willingness to reply on an exchange partner in whom one has confidence” (p.82). In

marketing, Racherla, Mandviwalla, and Connoly (2012) examined consumers’ trust in online

product reviews and argued that perceived background similarity (reflecting a social element)

and message argument quality (a content element) were two factors that contributed to

increased trust. It has also been said to be based on consumer beliefs and impacted consumers’

knowledge of the brand and their decision making about the brands (Reinikainen et al., 2020).

Page 22: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

13

Reinikainen et al. (2020) also argued in their literature that trust could be formed through the

interaction in context of relationship, impact of consumers’ previous experience, built through

consumers’ engagement and relationship with the brand and that it could be transferred, to

mention a few factors. Both Reinikanien et al. (2020) and Lou and Yuan (2019) tested brand

trust/trust in branded posts since prior literature have said that trust could impact consumers’

purchase loyalty. The findings of Reinikanien et al. (2020), reveal that trust is an important

factor when it comes to how consumers feel about a brand. Additionally, Lou and Yuan (2019)

found a positive impact of trust in branded posts on purchase intention. Therefore, this study

predict that trust is an important factor to consider in influencer marketing to generate purchase

intention and the following hypothesis is:

H1: Trust in branded posts mediates the effect of the source credibility variables

(trustworthiness, expertise, similarity, and usefulness) on purchase intention.

Figure 2.2: Proposed hypothesis H1

Source: Author’s own work, adopted from Lou and Yuan, 2019, p.63

2.3.4 Trustworthiness

When it comes to the different variables that have been shown to impact trust in branded posts

and to have an indirect impact on purchase intention, trustworthiness is one of them.

Trustworthiness is one of the dimensions of source credibility that have been a key factor since

an early stage (Ohanian, 1991). Prior research has supported the trustworthiness effect on

attitude change and Ohanian (1991) cited Hovland, Janis and Kelley (1953) that defined

trustworthiness as “the degree of confidence in the communicator’s intent to communicate the

assertions he considers most valid” (p.41). The role of trustworthiness in the context of

marketing effectiveness has got some different outcome during the years. Xiao et al. (2018)

found in their study that trustworthiness had a positive influence of the perceived credibility of

the influencer and was the only factor that was fully supported to have a positive correlation

Trust in branded posts

Purchase intentions

Page 23: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

14

with both the consumers’ attitude toward the posted material and the attitude toward the brand.

These findings were in line with Munnukka et al’s. (2016) findings that showed a strong

relationship between trustworthiness and consumers’ attitude toward the brand.

As mentioned, prior studies have also investigated the effect of source credibility as one factor,

that has contained trustworthiness as one of the keystones. These studies have found out that

source credibility, (i.e trustworthiness) have an indirect impact on consumers’ purchase

intention but not a direct impact (Lafferty et al., 2002; Reinikainen et al., 2020). Something that

Lou and Yuan (2019) partly found in their study as well. What Lou and Yuan (2019) found was

though that trustworthiness was the only factor that had a negative impact on the consumers

purchase intentions. A result which, according to the authors themselves, was unexpected.

However, there have been previous literature finding trustworthiness as an important variable

when measuring the impact of source credibility on purchase intentions, and Lou and Yuan

(2019) themselves say that this result needs further research. Therefore, trustworthiness will be

considered as an important variable in this study and it is purposed that trustworthiness (concern

to the receivers’ perception of the influencer) will affect the trust in branded posts and the

consumers’ purchase intention in a high involvement product.

2.3.5 Expertise

Expertise is another dimension that prior studies have included in the phenomena source

credibility. Ohanian (1991) cited Hovland, Janis and Kelley (1953) who defined expertise as

“the extent to which a communicator is perceived to be a source of valid assertions” (p.41).

Together with trustworthiness, expertise also has been a key component of source credibility

(Ohanian, 1991). Munnukka et al. (2016) also found that expertise together with trustworthiness

correlated strongly with each other and formed a pair among the credibility dimensions. But

even if expertise as a factor has a strong relationship with the component trustworthiness, the

effect of expertise in ads has been more varied. In those studies where the phenomenon source

credibility has been tested as one variable, the difference between expertise and trustworthiness

impacts have been hard to compare since they have been measured together as one variable

(Lafferty et al., 2002; Reinikainen et al., 2020). However, in those studies where source

credibility has been divided into different variables, the impact of each variables has varied.

Munnukka et al. (2016) tested the effect of expertise on consumers’ attitude towards the ad and

found a positive relationship. However, the effect of expertise was not as strong as the effect of

Page 24: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

15

trustworthiness. Later on, Xiao et al. (2018) tested the effect of expertise on influencers’

credibility to investigate if there was a correlation between expertise and attitude towards the

content and brand. Xiao et al (2018) findings showed that expertise had a positive influence on

the perceived credibility of the influencer and the correlation between attitude and expertise

was only partly supported. Furthermore, Lou and Yuan, (2019) found that expertise of an

influencer positively affected brand awareness so it was only affecting the purchase intention

indirectly since brand awareness had a positive relationship with purchase intention.

Even if several studies have tested the impact of expertise in different scenarios with different

outcome the factor is still important to consider since it has shown to have an impact on factors

that impact purchase intention and therefore has an indirect impact (Lou & Yuan, 2019;

Munnukka et al., 2016; Xiao et al., 2018). According to Lou and Yuan (2019) source expertise

is “a source’s competence or qualification” (p.61), and therefore expertise could be an important

variable to consider in high-involvement products since, as the name indicates, these products

have a higher involvement from its consumers which may drive up the requirement of expertise

of the influencer promoting the product. And since a luxury brand, that Lee and Watkins (2016)

mention, could be identified as expensive, high in quality, uniqueness and rare, could be

categorized under the high-involvement category, source expertise would play an important

role in advertising of this products.

2.3.6 Similarity

More recently, similarity has also been included as a factor in the phenomenon on source

credibility. Ohanian (1990) did not include similarity as one of the factors in the scale, however,

she employed it as a measure for nomological validity in her study. It was found that a high

degree of similarity to the endorser could impact the attitude towards the advertised brand

positively. The term similarity has been defined as “perceived similarity of the source to the

recipient (e.g. similarity in terms of demographic or ideological aspects)” (Munnukka et al.,

2016. p.184) and “has been proposed as an important antecedent to attitudes toward an

advertisement and ultimately, advertising effectiveness” (Lafferty et al., 2002. p.2). Further

literature “propose that similarities between the recipient of a message and the endorser in terms

of values and cultural background mediate the credibility associated with the source”

(Munnukka et al., 2016. p.184). In Munnukka et al. (2016) study the authors tested the effect

of similarity on consumers’ attitude toward ads and found a positive relationship between these

two variables. Together with trustworthiness, similarity had the strongest influence out of four

Page 25: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

16

factors tested. Lee and Watkins (2016) tested YouTube Vloggers’ influence on consumers’

perception and intentions towards luxury product and found that consumers who viewed

influencers as similar contributed to establish long-term relationship with the brand and that

could lead to increased likelihood of repeated purchase intentions. However, Xiao et al. (2018)

tested homophily, as denotes the similarity between the influencer and the consumer, and found

that homophily (similarity) together with expertise was not as important as trustworthiness

when consumers evaluated the credibility of eWOM ads, but they also said that those results

might look different if other products were tested.

Lou and Yuan, (2019) also used similarity as a factor of source credibility and they did not find

a positive relationship between similarity and purchase intention. Instead, Lou and Yuan (2019)

found that similarity was having an indirect impact on purchase intention since it positively

impacted both brand awareness and trust in influencer generated branded posts. Although the

results of prior studies do not provide wholly consistent evidence, it generally suggested that

the greater the similarity between the audience and the endorser is, the greater the credibility of

the endorser and the more effective the ad will be (Munnuka et al., 2016). And in Xiao et al

(2018) study they argued for different results on similarity depending on the product that is

tested, so, therefore, the author of this report argue that similarity will be an important

dimension to consider in this study.

2.3.7 Usefulness

Lou and Yuan (2019) did not use the dimension usefulness in their study, as can be seen in

figure 2.1 above. In prior studies, usefulness has not been used in the context source credibility,

instead Lou and Yuan (2019) thought it could be interesting to look at other factors that could

be related to media channel, such as usefulness. Based on statistics, the use of influencer

marketing is found to be effective among 94% of marketers (Ahmad, 2018). However, the

usefulness of influencer marketing could differ from industry and product. Munnukka et al.

(2016) discussed in their study that previous studies have showed that “endorsers’ influence on

ad effectiveness varies between product categories” (p.188) and in this study the sample that

will be evaluated includes an older generation (which will be explained further down) and, in

an industry, where influencer marketing, is not, yet, well developed, the dimension of influencer

marketing usefulness could be an interesting aspect to look closer into.

Page 26: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

17

Based on the findings above, the author in this report argues that the source credibility variables

(trustworthiness, expertise and similarity) plus usefulness will have a positive impact on trust

in branded posts and purchase intentions.

H2: Source credibility, including (a) expertise, (b) trustworthiness, (c) similarity, and (d)

usefulness will positively influence the consumers trust in branded posts.

H3: Source credibility, including (a) expertise, (b) trustworthiness, (c) similarity, and (d)

usefulness will positively influence the consumers purchase intentions.

Figure 2.3: Proposed hypothesis with H2a-d & H3a-d added

Source: Author’s own work, adopted from Lou and Yuan, 2019, p.63

2.3.8 Moderators

Lou and Yuan (2019) used three moderators in their study. The choice of adding the moderator

“involvement” (see figure 2.1 above) was based on previous literature examining the role of

involvement in decision making and advertising effectiveness. In Lou and Yuan’s case, they

conceptualized involvements as “social media users’ involvement in following influencers’

posts or updates” (p.63). Lou and Yuan also added two more variables: age, and gender, which

was found crucial for classifying social media users. In the adapted model, involvement will be

left out as a moderator since this study aims to look at influencer marketing from the perspective

of a specific product. Lou and Yuan (2019) looked at the model from the perspective of a social

Expertise

Trustworthinesss

Similarity

Usefulness

Trust in branded posts

Purchase intentions

Page 27: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

18

media user who followed at least one influencer and all the interaction that influencer had with

different brands online. However, the two moderators age and gender, will still be used in this

study. The moderator age is chosen since the average age of the users of the product is above

55 years old, and previous studies have mainly used a sample size of students and the younger

generation (Munnukka et al., 2016; Lafferty et al., 2002; Reinikainen et al., 2020). The

moderator gender is chosen since the users of the product are mainly men and prior studies have

based their results mainly on female respondents (Lee & Watkins, 2016; Munnukka, et al.,

2016; Reinikaninen et al., 2019). In figure 2.4 below the proposed SMIV model is presented.

Figure 2.4: Proposed SMIV model

Source: Author’s own work, adopted from Lou and Yuan, 2019, p.63

2.3.9 Variables excluded from the model

As could be seen in figure 2.4, compared with figure 2.1, prior findings, and the chosen area,

have made the researcher taking the decision to exclude dimensions in the adapted model. In

the left column, the independent variables, attractiveness, information value, and entertainment

value have been excluded together with the outcome variable brand awareness. Regarding the

factor attractiveness, Ohanian (1990) states it as a well-explored area, however, the issue is

described to be far from clear. His findings became the base of the indication of adding

attractiveness as one of the dimensions included in the phenomena source credibility. Since

then, attractiveness has been tested in various literature afterward. Even if Lou and Yuan (2019)

found that attractiveness had a positive impact on both brand awareness and the consumers’

Expertise

Trustworthiness

Similarity

Usefulness

Trust in branded posts

Purchase intentions

Age

Gender

Page 28: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

19

trust in the branded posts, prior research has argued that attractiveness is one of the dimensions

that have less effect in this context (Amos et al., 2008). A few studies also suggest that

attractiveness may have less or even no effect on the credibility of the message or on buying

behavior. It has also been reported that there is a negative relationship between attractiveness

and consumers’ buying behavior (Munnukka et al., 2016). It is said to be considered as a key

component of source credibility in the case of peer endorsements of beauty products

(Munnukka et al., 2016), and it has been shown to positively influence the advertising

credibility when there are products that are related to physical appearance (Till & Busler, 2000).

Munnukka et al. (2016) tested attractiveness effect in advertising toward beauty products and

found a positive effect on the consumer attitude toward the ad, however, the effect was weaker

than the dimensions trustworthiness and similarity. Since this study will be conducted in the

sports industry and prior research has found evidence that attractiveness could be considered to

be left out as a variable of source credibility depending on the product category, this study will

leave attractiveness out from the model. Even if Lou and Yuan (2019) found a positive

relationship with attractiveness, their study was conducted on influencer marketing in general,

not a specific industry or product category which makes it hard to know in what context the

dimension attractiveness was tested. Munnukka et al. (2016) also suggest that the attractiveness

of the message source has the largest influence in low-involvement situations and this study

will examine the high-involvement category. Furthermore, since the factor similarity will be

included, which prior studies have suggested being a factor connected with attractiveness

(Munnukka et al., 2016), it could be said that dimension attractiveness will indirectly be tested

in this study anyway. So, in conclusion, attractiveness will be left out in this study since this

study is based on the sports industry and not in the context of beauty products and second, since

attractiveness have been indicating on importance in low-involvement product and in this study,

a high-involvement product will be evaluated.

Furthermore, as can be seen in figure 1, Lou and Yuan (2019) also tested the impact of

advertising value (informative value and entertainment value) in their study. Previous literature

had found that “advertising informativeness, entertainment, and credibility determined

consumers’ perceptions of advertising value, which in turn influenced their purchase

intentions” (Lou & Yuan, 2019. p.60). However, their results indicated no relationship between

the dimension entertainment value and brand awareness, trust in branded posts, or purchase

intentions. The information value was shown to have an impact on consumers’ purchase

Page 29: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

20

intention and trust in branded posts. Even if information value had an impact on two out of

three dependent variables in their study, both these dimensions of advertising value will be left

out in this study. Since entertainment value did not indicate on any relationship at all and the

source credibility dimensions have been shown to have an impact and indirect impact on trust

and purchase intentions (Lafferty et al., 2002; Munnukka et al., 2016; Reinikainen et al., 2020)

this study will focus on the source credibility phenomenon and not include advertising value.

Lastly, brand awareness will also be excluded from the adapted SMIV model. This decision

was based on the participants’ prior engagement with the product. “Brand awareness denotes

whether consumers know about a certain brand and whether they can recall or distinguish it”

(Lou & Yuan, 2019. p.61). High brand awareness has been found to have a positive impact on

consumers’ consideration of the purchase of a product or not. All four dimensions of source

credibility that Lou and Yuan (2019) tested, trustworthiness, expertise, similarity, and

attractiveness had a significant effect on the consumers’ brand awareness. However, since this

study will be tested on a sampling that already is involved with the product and users of the

product, the brand awareness is already considered as high among this sampling and therefore

excluded.

2.4 Golf versus fashion As discussed earlier, in 2.4.1 High- and low involvement, the phenomenon high- and low

involvement could be used in different contexts. As was mentioned, a consumer’s involvement

in a product depends on the importance of information needed and what type of products the

consumer is considering (Aldlaigan & Buttle, 2001). Ha and Hu, (2013) explained that

involvement in social media channels, or social network sites (SNS), could be divided into

active and passive users where the active users had more activity on SNS than passive users.

One really interesting finding in Ha and Hu’s study (2013), was that a difference in high versus

low SNS involvement users’ preference in topics could be found. Those who had a high SNS

involvement seemed to be more interested in “soft news” such as gossip and celebrity news

while people with low SNS involvement showed more interest in sports, politics, business, and

science.

Ha and Hu (2013) further discussed that high SNS involvement users, that had a higher

interaction and activity on their own channels, were more likely to post content from other news

media and comments on posts and linking to other media sites on their own sites, than what

users with low involvement would do. This could be one of the reasons why influencer

Page 30: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

21

marketing, in general, has focused more on expanding in the fashion industry in an earlier stage

than those areas with more profound importance of informative have done. Wolny and Mueller,

(2013) found influencers particularly effective in the fashion industry since it was shown to

have the ability to quickly follow upcoming trends in the industry. This made it possible to

rapidly spread those trends throughout their social media network and in that way impact those

consumers that had a strong willingness to embrace trends that were confirmed by others. They

meant that the fashion industry has a strong connection with the users’ norms, values, and

preferences that adapt fashion along the way. “The fact that the flow of new fashion or styling

choices is never-ending, with the kaleidoscope of products and trends (be it recycled) constantly

changing, means there is always a potential element of surprise in user-generated fashion

messages.” (Wolney & Muller, 2013. p.563). If the fashion industry in an early staged focused

on targeting those users with a high SNS involvement, Ha and Hu’s findings (2013) indicated

on a fast spread among new trends since these users were more likely to post and link to other

media sites. It was further discussed that younger people were spending much more time on

their SNS and had a higher frequency of updating status than what the older generation had.

In comparison to users with low SNS involvement, influencer marketing, in the conclusion of

Ha and Hu’s findings (2013) were, therefore, indicating not being so effective back then since

these users seemed to have a lower activity on their own, as well as on general media channels.

Although, a lot has changed during the years and influencer marketing has increased in those

areas, where Ha and Hu (2013) said people with low SNS involvement had high interest, such

as sports and politics. For the sports industry, social media platforms have become an important

communication tool lately (Madere et al., 2018) and Chan-Olmsted and Xiao (2019) mentioned

in their article that “[a]s more and more sports content is migrated to social media, social media

is now an essential media platform of fan engagement and even content distribution. Sports

fans can express themselves while obtaining instantaneous information, interacting with

favorite athletes, and socializing with other fans online” (p. 185). The importance of influencer

marketing in the sports industry is confirmed by other authors (Maderer et al., 2018;

Anagnostopoulos et al., 2018; Jang, Ko, Kim, & Jeong, 2015; Kunkel et al., 2019), as well.

Since this study is conducted in collaboration with a company, active in the golf industry, the

question arises if those general findings that have been made in the sports industry so far could

be applicable in the golf industry as well. This since golf is a sport where the participants put

down a lot of time and effort and the involvement could be seen as very high. Golf is a sport

Page 31: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

22

that during many years has been associated with business people and which companies spend

a lot of money on each year (Ceron-Anaya, 2010). During its history, rules have slowly

transformed the etiquette of the sport and it has become a more expensive activity (Ceron-

Anaya, 2010). In the year 2018, golf had over 4.11 million registered golfers in Europe (Statista,

2020). However, even if the sport is being practiced by a lot of people, changes that need to be

made in order to follow the development in the world are slow and technology is, for example,

considered to be threatening the sport’s future according to Forsyth (2020). Furthermore,

Forsyth (2020) discusses the still low acceptance of women in the sport, which for many people

would not even be a subject in the year 2020.

Page 32: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

23

3. Methodology Chapter three gives the reader a presentation of the chosen methodology for the research study. It begins with the purpose, approach, and strategy, which later on is followed by how the data was collected, the choice of sampling selection, and the method of analysis. At the end of the chapter there is a summary and a “methodology map” presented to give the reader an overview of the author’s decisions.

3.1 Research purpose A study will always have the nature of either an exploratory, descriptive, explanatory, or a

combination of these three purposes. What characterizes an exploratory study is that these type

of studies tries to create an understanding of phenomena that has not been investigated before

by asking open questions about the topic. The focus of an exploratory study could be perceived

as broad since the aim is to clarify an understanding of a problem (Saunders, Lewis & Thornhill,

2012). Another alternate purpose of a study could be a descriptive purpose. Descriptive studies

usually aim to answer questions such as, whom, what, where, and when and are conducted since

a research area could lack investigation, the same as with exploratory studies (David & Sutton,

2016). The aim with a descriptive purpose is to profile an event, person or situation as accurately

as possible (Saunders et al., 2012). The third nature of a study could be an explanatory nature.

This is usually laid over the exploratory and descriptive nature where the aim usually is to

explain relationships between variables (Saunders et al., 2012). Since this study aimed to test

the relationship in the adapted SMIV model throughout the hypothesis, the nature of this thesis

was explanatory.

3.2 Research approach

The design of a research study could be either quantitative, qualitative, or multiple methods

research. The quantitative research design is often characterized by numeric data (numbers)

where questionnaires or data analysis procedures, such as statistics and graphs, often are used

as a synonym for the quantitative approach (Saunders et al., 2012). Qualitative research is often

characterized by non-numeric data such as words, video clips, or other material that does not

consist of numbers. The numerical differences are often one way to differentiate these two

research designs from each other (Saunders et al., 2012). Another way is that a quantitative

study often tends to have a higher focus on generalizability when qualitative study tends to have

a more subjective focus (David & Sutton, 2016). However, sometimes it could be necessary to

have a mix of these studies. For example, a quantitative study may need follow-up questions to

explain the findings in the questionnaire.

Page 33: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

24

To answer the research question for this study a quantitative method was used where an

experimental vignette methodology (EVM) was used (will be further explained in 3.4.2). This

seemed to be the most suitable tool to find the relationships between the dependent,

independent, and cofounding variables presented in figure 4. Since the sample of interest was

in the company’s consumers (e.g. the whole population) a quantitative method gave the

researcher the opportunity to collect a higher number of respondents.

Furthermore, a quantitative and qualitative research design could have three different

approaches, deductive, inductive or a mix of these two called, abductive. A deductive approach

could be said to start with theory to, later on, move over to data collection, whereas an inductive

approach starts with collecting data in order to later go over to theory. The abductive approach

is a mix of deductive and inductive where the author moves back and forth between theory and

data (Saunders et al., 2012). The aim of this thesis was to explore how companies could generate

purchase intention in the area of a high involvement product with the use of influencer

marketing and the question was answered through using prior literature to develop the study to

later on test the theory with collected data. Therefore, the deductive approach was judged to be

the most suitable approach for this study.

3.3 Research strategy

When the nature and approach of the study are set, the study needs a roadmap of how to achieve

the goal. To achieve a goal, you usually need a strategy, and when conducting a study, the

author needs to know how to go forward to be able to answer the research question, in the end,

a so-called research strategy (Saunders et al., 2012). According to Yin (1994), a research

strategy describes the way empirical data is collected and analyzed. The research approach and

purpose are usually good guides for the choice of a suitable research strategy for the study.

Since there are several strategies to choose between, the author should keep in mind that it is

okay to combine a number for strategies if that is needed to answer the research question.

However, some strategies are more suitable for quantitative research versus qualitative research

(Saunders et al., 2012). Strategies that could be considered in a quantitative study is

experimental and survey research strategies. In a qualitative study, action research, case study

research, ethnography, grounded theory, or narrative research could be considered appropriate

strategies. However, a variety of different strategies are often what qualitative research is

associated with (Saunders et al., 2012).

Page 34: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

25

In this study, an experimental strategy was used. “The purpose of an experiment is to study the

probability of change in an independent variable causing a change in another, dependent

variable. An experiment uses predictions, known as hypotheses, rather than research questions.

This is because the researcher anticipates whether or not a relationship will exist between the

variables (Saunders et al., 2012, p.174). Since this study aims to answer the research question

through testing hypotheses, it will contain variables where the relationship between the

variables will be tested. According to David and Sutton (2016), a hypothesis is a statement

about a relationship between two variables, a dependent and independent variable. The

dependent variable is the variable that changes as a response to change in the independent

variable. In this study the independent variables were measured in terms of responses to

questions as well as exposure to a stimulus. Since it is an experiment that is conducted, it allows

the researcher to use more than only two variables. Table 3.1 shows the types of variables that

could be used and what variables that is used in this study.

Table: 3.1 Types of variables

Variable Meaning Variables in this study

Independent Affecting the dependent variable if changes in this variable happens.

Trustworthiness, expertise, similarity, usefulness, and quality

Dependent Could be impacted of the changes in the independent variables

Purchase intention

Mediating Lays in-between the independent and dependent variables explaining the relationship.

Trust in branded posts

Moderator The moderator is a variable, outside the dependent and independent variable and could affect the nature of their relationship.

Age, gender

Control Variables that could avoid influencing the effect of the independent and dependent variable if it is kept constant.

Cofounding Could potentially undermine the inferences drawn between the independent and dependent variable and needs to be considered when discussing results.

Source: Sounders et al. (2012), p.174

Page 35: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

26

3.4 Data collection After the research strategy is decided, the next step is to choose a data collection method.

Saunders et al. (2012) describe two ways of collecting data: primary data, and secondary data.

With primary data, the author collects “new” data that is specific for the purpose of the study

where secondary data means that previous data that have been collected for another purpose

will be further analyzed in the study. For this study, the use of primary data was chosen since

the area that was explored did not contain enough prior data to answer the research question.

When using primary data in a study, there are three different ways of collecting the data,

observation, interviews, or questionnaires (Saunders et al., 2012). To answer the research

question and purpose, this study used an EVM which made use of a combination of exposure

to the independent variable (a stimulus) and a set of following questions (a questionnaire). The

use of a questionnaire could be characterized by the same set of questions that are sent out to a

decided sample; however, the produce of good questions could be much harder than many

authors think. This since your questions really need to be so précised so they answer your

research question the first time you ask the respondents, since the opportunity to go back to the

respondents does not exist in this way of collecting data (Saunders et al., 2012).

3.4.1 Measurement instrument

Before the questionnaire was put together and sent a pilot check, in the form of a qualitative

validity check, the model was mead through an interview with a key industry person. This

interview was made to justify the alterative of the questionnaire measurement instrument and

to get the key person’s input on the model. In the end the final measurement instrument is

presented as well.

Pilot check

The interview guide was based on McCracken’s, (1988) “the long interview”, see appendix A

for the full interview guide. As mentioned above, the interview was made with a key person

from the industry that worked with influencer marketing in the sports industry, together with

other product categories as well. One of the core findings of the interview was that the product

category was said not to have any specific impact on the effectiveness of influencer marketing.

This was based on an own experience from the industry person that has worked with influencer

marketing in many different product categories. Furthermore, the industry person confirmed

Page 36: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

27

that purchase intention is the main outcome that companies want to achieve when using

influencer marketing.

Regarding the different factors (trustworthiness, expertise, and similarity) that have been said

to impact both trust in branded posts and purchase intention, the only thing that came up in the

discussion was the idea of quality in the branded posts, which is a factor that is not included in

the model since before. However, after some more directed questions about source credibility,

the person mentioned that expertise is an important factor to consider as well. It was also

mentioned that expertise is considered more important in some products.

Expertise I would say it is important if the product is in need of it. To come up

with one example, what mobile phone you chose are maybe not that critical, and

in that type of product you as a company rater would like to get access to an

influencer having more followers and are inspiring their followers. So, in these

cases I do not think expertise is the most critical factor, however, if you are going

into the golf area, it might be more important with an influencer that is really good

at what he/she is doing.

Final measurement instrument.

Based on the findings in chapter 2 together with the findings in the interview, the final SMIV

model for this project was adopted and is presented in figure 3.1 below. Based on the findings

in the interview, an adaption of the hypothesis H2 and H3, inserting quality as an independent

variable was made as well.

H2: Source credibility, including (a) expertise, (b) trustworthiness, (c) similarity, (d)

usefulness, and (e) quality will positively influence the consumers trust in branded posts.

H3: Source credibility, including (a) expertise, (b) trustworthiness, (c) similarity, (d)

usefulness, and (e) quality will positively influence the consumers purchase intentions.

To visualize the background of each variable in the final SMIV model, table 3.2 presents the

source of each variable. Each variable has been named with an abbreviation, where trust in

branded posts is (TB), trustworthiness is (TRUST), expertise is (EXP), similarity is (SIMI),

usefulness is (USE), quality is (QUA) and purchase intention is (PI).

Page 37: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

28

Table 3.2: Source’s for each variable

Measure Source

TB Reinikainen et al., (2019); Lou & Yuan, (2019)

TRUST Ohanina, (1991); Munnukka et al., (2016); Lafferty et al., (2002); Lou & Yuan, (2019)

EXP Ohanina, (1991); Munnukka et al., (2016); Xiao et al., (2018); Lou & Yuan, (2019)

SIMIL Munnukka et al., (2016); Xiao et al., (2018); Lou & Yuan, (2019)

USE Lou & Yuan, (2019)

QUA Key person from industry, personal communication, 3th of Marsh, 2020

PI Ohanina, (1991); Lafferty et al., (2002); Lou & Yuan, (2019)

In figure 3.1 quality is marked with bold to illustrate the addition of this variable. The questions

created to measure each variable are presented further down in figure 3.1.

Figure 3.1: Adapted model after the qualitative validity check of the model

Source: Author’s own work, adopted from Lou and Yuan, 2019, p.63

Expertise

Trustworthiness

Similarity

Usefulness

Trust in branded posts

Purchase intentions

Age

Gender

Quality

Page 38: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

29

3.4.2 Experimental Vignette Methodology

One way to collect experimental research is through the use of experimental vignette

methodology (abbreviated EVM). “EVM studies consist of presenting participants with

carefully constructed and realistic scenarios to assess dependent variables including intentions,

attitudes, and behaviors, thereby enhancing experimental realism and also allowing researchers

to manipulate and control independent variables” (Aguinis & Bradley, 2014. p.351). When the

researcher uses the EVM, a result with a high level of confidence regarding internal validity is

to be expected. However, the external validity could be a bigger challenge in this way of

collecting data. To be able to conduct a successful EVM, there are three steps the author needs

to consider and follow. In each of the steps, there are in total 10 decision points as well, see

figure 3.2 below.

Figure 3.2: Steps and decisions points in conducting an EVM study.

Source: Aguinis and Bradley (2014). p.360

x x

x

x

x

x

x

x

x

Step 1 Planning

Specify the type of EVM study and its design features

Address decision points # 1-6

Plan on maximizing both internal and external validity

Step 2 Implementation

Assess how, where, and when data will be collected

Address decision points #7-9

Specify the how, where, and when of data collection

Step 3 Reporting of Results

Report results obtained from conducting previous steps

Address decision point #10

Full disclosure & transparency to enhance replicability

Page 39: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

30

In the first step, planning, the author has six decision points to consider. The focus of this step

is to maximize internal and external validity.

1. Deciding whether EVM is a suitable approach

2. Choosing the type of EVM

3. Choosing the type of research design

4. Choosing the level of immersion

5. Specifying the number and levels of the manipulated factors

6. Choosing the number of vignettes.

In the second step, implementation, the author needs to consider decision points 7-9. In this

step, the author determines when, where, and how the data will be gathered and analyzed.

7. Specifying the sample and number of participants

8. Choosing the setting and timing for administration

9. Choosing the best method for analyzing the data

In the final step, the reporting of results, the author has the last decision point to consider.

10. Choosing how transparent to be in the final presentation of results and methodology.

(Aguinis & Brandly, 2014).

3.4.2.1 Step 1- Planning

As shown in figure 3.2 above, the first step in implementing EVM is to plan and choose whether

or not EVM is a suitable approach, and if it is suitable, what type should be implemented. Below

decision points 1-6 till be discussed.

Decision point 1: Deciding whether EVM is a suitable approach

Regarding EVM, there are some situations that are more suitable to implement the method than

in others. When the aim is to exercise “control of independent variables to gather evidence

regarding causation” (Aguinis & Brandly, 2014. p.357), the EVM is especially useful. EVM

was chosen as a suitable approach in this study since the author gets the opportunity to include

relevant factors in the study which increase the control over the study. It also allows the author

to test causal hypotheses (Aguinis & Brandly, 2014).

Page 40: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

31

Decision point 2: Choosing the type of EVM

After the author has decided that EVM is suitable for the study, the next decision lies in which

type of EVM to use, paper people studies or policy capturing and conjoint analysis studies. In

the paper people study, the aim is to measure explicit responses in hypothetical scenarios

wherein policy capturing, and conjoint analysis studies could increase the understanding of

single individual’s decision-making processes. It is useful when the researcher needs to collect

a large amount of information on a smaller sample and it gives the researcher implicit outcomes

(Aguinis & Brandly, 2014). In this study, a paper people study was used since the participants

were responding to a set of hypotheses and it allowed the author to focus on collecting explicit

processes and outcomes.

Decision point 3: Choosing the type of research design.

In the third decision point, the researcher has three different research designs to choose between

which are 1) between-person, 2) within-person, and 3) a mixed research design (Aguinis &

Brandly, 2014). In a between-person design, the author has several vignettes, but the

respondents will only be introduced to one vignette and then a comparison across the

participants is made. In this type of research design, the researcher must provide the participants

with sufficient information to give the participants as much context as possible since they will

only be introduced to one vignette. In comparison with a within-person design, the number of

vignettes will be shown for all participants and the comparison is made between the vignettes

(Aguinis & Brandly, 2014).

In this study, a between-person design has been used due to the fact that each participant only

was exposed to one out of four vignettes (see figure 3.3 below). In two of the vignettes, the

participants were exposed to an introduction text regarding the influencer’s background where

the variable expertise was highlighted in the text and the participants also took part in the

Instagram post information, see appendix B. In the other two vignettes (in total there were four

vignettes), the participants were only exposed to the information in the Instagram post before

answering the questions. This was made to create an understanding whether the variable,

expertise and similarity, had a significant impact on how the participants perceived the

collaboration between the influencer and the company. A between-person design was also

chosen since it shortened the time it would take for each respondent to answer the questionnaire.

Since the questionnaire was sent out online, an overly long questionnaire could have impacted

the respondent number even more.

Page 41: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

32

Decision point 4: Choosing the level of immersion

Even if the EVM has been considered as a good method in experimental research, there are also

some criticisms toward it, and in the early year of 2000 one criticisms was that EVM “is

unrealistic and results are not easily generalizable” (Aguinis & Brandly, 2014. p.361). Through

increasing the level of realism, prior researchers have tried to improve external validity. One

way to do this has been to use virtual reality technology (VRT) where the participants have

been able to smell, see, hear, and touch in the experiment (Aguinis & Brandly, 2014). Based on

a short time edge and budget restriction, the author in this study did not have the opportunity to

conduct a video to increase the validity, instead, the vignettes were based on pictures with an

image of a golf person and written text in two of the vignettes describing the background of the

person they saw in the picture (see figure 3.3 below).

Decision point 5: Specifying the number and levels of the manipulated factors

Other criticisms against vignette methodologies are toward the theory the researcher decide to

use when picking the relevant factors for the specific research that will be made. Since EVM

requires prespecified variables, omitting important variables could be a threat in the study

(Aguinis & Brandly, 2014). To minimize the risk of leaving important variables out, there are

two approaches that could be used; attribute-driven design or “actual derived cases”. In an

attribute-driven design, the factors are orthogonal (have a similar approach) to one another and

the researcher could easily investigate the effects of the different factors. Where in “actual

derived cases” the researcher chooses variables representing real-life settings and this helps to

create results with a more generalized view (Aguinis & Brandly, 2014). In this study, the

derived case was used and based on the distribution in the collected data, the answer for each

variable was divided into different groups (low, medium, and high) through visual binning to

create a true picture of the data.

Decision point 6: Choosing the number of vignettes

To finish the first step, the researcher needs to decide how many vignettes to use. Why it is

important to get a correct number of vignettes is because too many vignettes could create an

overload of information for the respondent and exhaustion for the respondent, and too few could

instead limit the researcher’s ability to manipulate critical variables and result in responses

biased (Aguinis & Brandly, 2014). For this question no answer is more correct than another,

instead, each study needs to figure out what the most suitable number of vignettes is to answer

the study’s purpose (Aguinis & Brandly, 2014). In this study, four different vignettes were

Page 42: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

33

chosen, all with similar designs and approaches. The number of four was chosen since the

author wanted to measure two variables, expertise and similarity. To measure similarity the

author needed to have one picture with a female golfer and one with a male golfer. To measure

expertise the author needed to have one vignette with a text telling the respondents that the

golfer in the picture had a career and many years of experience of golf, and another picture

where no information at all about the golf persons background was given. The dissimilar factor

in the four vignettes was, therefore, two including a picture of a woman and two including a

picture of a man, and two vignettes including an information text about the golfers’

backgrounds and two vignettes without any background information. All other factors were the

same, see figure 3.3 below. (In the figure the company’s name has been left out).

Figure 3.3: The vignettes used in the questionnaire.

Source: Author’s own work

Page 43: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

34

3.4.2.2 Step 2-Implementation

In the second step in the process of EVM, the author’s decisions regarding the sample size,

setting, and timing of the questionnaire, and how the data will be analyzed are decided. In this

step, the author has the decision points 7-9 to consider.

Decision point 7: Specifying the sample and number of participants.

To be able to collect data it is important to find representative participants responding to collect

data from. This is a critical moment since the respondents are those that decide if the quality of

the data will be good enough (Aguinis & Brandly, 2014). It is also important to collect a sample

big enough to have the possibility to generalize the results of the collected data in a larger

population. However, to find a suitable sample size could be trickier than one might think.

Financial limitations, lack of access, or time restriction could impact the access to analyze the

entire population and therefore different sampling techniques can be implemented (Saunders et

al., 2012). Usually, there are two types of sampling techniques: probability sampling and non-

probability sampling. The difference between these two techniques is that in a probability

sampling, the whole population is known and could be chosen. In non-probability sampling,

the whole population is not known, and therefore there is not an equal chance for all to be

chosen. In this study, access to the company’s customers (e.g. the whole population) email

addresses was available. In this study the author chose to send the questionnaire to all available

email addresses and then used the answers that were received from those who voluntarily

answered the questionnaire. This means the author did not use neither probability nor non-

probability sampling, instead this study used a saturation survey (Chipp, 2004).

The choice of the population in this study was given based on the conduction if this project

together with the company. The company’s customers are located all over Europe which

resulted in different nationalities among the respondents. A total of 4196 questionnaires was

sent out through mail, where 398 clicked on the link to the questionnaire. In the end, there were

196 respondents completing the questionnaire which indicated on 5% of the total and 49% of

those who clicked on the questionnaire link. The distribution among the respondents was 84.7%

(166) males and 13.3% (26) females. 2.0% (4) participants did not want to disclose their gender

in the questionnaire.

Page 44: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

35

Decision point 8: Choosing the setting and timing for administration

Since EVM has been criticized for lacking realism, letting the respondents answer in their

natural setting should be considered (Aguinis & Brandly, 2014). When the setting is decided,

it is also important to consider the timing. By considering these two aspects, the validity of the

study could increase (Aguinis & Brandly, 2014). Since the respondents in this study received

the questionnaires through email (which left them the choice of using either a computer or a

smartphone), the location was not specially considered by the author. Regarding the timing, the

questionnaire was sent out to the respondent during the start of the season which made the

context in the questionnaire fit very well in time. To increase the external realism of the stimuli,

the influencers used in the vignettes were typical influencers’ that the company could have used

together with information text taken from an earlier collaboration on the company’s Instagram.

The choose of using information that the consumers, active on their Instagram, could have seen

before was to create vignettes as realistic as possible with the same language as the company

themselves use in their communication. Although, the background information, and the

information in the posted picture, was limited which could have impacted the ability to make a

judgement on.

To be sure that the questionnaire was measuring what is was supposed to measure and to check

for possible errors before sending out the questionnaire, a pilot test was conducted. A pilot test

could help the researcher to “refine the questionnaire” and “enable you to obtain some

assessment of the questions’ validity and the likely reliability of the data that will be collected.”

(Saunders et al., 2012, p.451). For this study, 13 people answered the pilot test to clarify if there

were any questions that could be misunderstood or needed to be rewritten. In the pilot test, four

extra questions were added to give the respondents the chance to give their feedback. Those

four questions focused on how many minutes it took them to answer the questionnaire, if any

of the questions were hard to understand, any questions that were repeated/unnecessary, or if

there was something else they wanted to improve in the questionnaire. The answers from the

pilot test were later evaluated by using Cronbach’s alpha and person correlation. Cronbach’s

alpha is usually used to measure the consistency of responses to questions built on scales. The

value of alpha is between 0 and 1 and if the value is over 0.7 or above, it indicates that the

questions within the scale are measuring the same thing (Saunders et al., 2012). The measured

alpha in prior studies regarding the different variable’s was all high, between 0,82, and 0.95.

Based on the results from the pilot test, the scale indicated on reliability since the Cronbach

alpha for the variables were between 0,754 and 0,917.

Page 45: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

36

Based on suggestions made by respondents in the four extra questions, minor changes in

wording and the introduction text were made before the final questionnaire was sent out.

Decision point 9: Choosing the best method for analyzing the data.

The last decision in step two is to decide what data analyzes methods to be used. Depending on

the researcher’s choice in decision point 3, choosing between- or within-person design the

analysis will differ. Since this study was conducted through a between-person design, the most

appropriate techniques to use are MANOVA, ANOVA, and ANCOVA (Aguinis & Bradley,

2014). To handle the analysis of the quantitative data there are different types of analysis

software tools ranging from simpler tools like Excel to more advanced data programs such as

Minitab or IBM SPSS Statistics (Saunders et al., 2012). Based on the suggested techniques

MANOVA, ANOVA, and ANCOVA when using EVM, this study used the program IBM SPSS

Statistics program to analyze the collected data through the ANCOVA technique. The

difference between MANOVA, ANOVA, and ANCOVA is that in ANOVA analysis you

compare the mean scores on two or more groups of variables. In a MANOVA analysis, a

comparison between a number of different dependent variables is made and in an ANCOVA

the researcher also analyzes the effect of a covariate (Pallant, 2010). The dependent variable in

the ANCOVA analysis was Purchase intention, where the fixed factors were all the virtual

binning variables (Trust in branded posts, trustworthiness, expertise, similarity, usefulness, and

quality) and the covariates were gender and age.

Since this study was based on collecting data through the use of the quantitative method in the

form of a questionnaire, the study resulted in containing missing data. According to Saunders

et al. (2012, p.485) are there four main reasons for missing data:

• The data was not required by the respondent.

• The respondent refused to answer the question.

• The respondent did not know the answer or did not have an opinion.

• The respondent may have missed a question by mistake.

Even the missing data should be coded since all variables need to have a code. However, it is

up to the person analyzing the data to decide what code the missing data should have (Saunders

et al., 2012). In this study, the missing data was based on those respondents that did answer “I

choose not to disclose” on the question “gender”. Number two (2) was used to code the missing

data.

Page 46: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

37

3.4.2.3 Step 3-Reporting the results

The final step in the EVM is to describe the process and administering the vignettes, as precisely

as possible, to create an opportunity for research in the studied area. In decision point 10, the

researcher needs to make a decision of how transparent to be in this area (Aguinis & Brandly,

2014).

Decision point 10: Choosing how transparent to be in the final presentation of results and

methodology.

To make it possible for further research to use the findings of the conducted study and extending

the findings in further studies, the researcher needs to reveal as much information as possible

about the process and vignettes (Aguinis & Brandly, 2014). As this study contains text and

picture vignettes, the transparency of this study will enhance if this is included in the

manuscript. In Appendix B a full example of the questionnaire for one of the vignettes is

presented. In all questionnaires, the respondents were first introduced to what to expect in the

upcoming questionnaire. Those vignettes including a background text of the influencer was

introduced with the following text (note that the company’s name is left out in this example):

Hi! In this survey, you will be introduced to a collaboration between an influencer and a golf

company. After you have read about the influencer's background and looked at the picture, in

the next slide, you will get a series of short questions to answer. Please answer on a scale of

one (1) to five (5). Note: this is just an illustration of how a collaboration could look like.

Those people that were only seeing the branded posts, see figure 3.3 above, were introduced

with the following text:

Hi! In this survey, you will be introduced to a collaboration between an influencer and a golf

company. After you have read and looked at the post, in the next slide, you will get a series of

short questions to answer. Please answer on a scale of one (1) to five (5). Note: this is just an

illustration of how a collaboration could look like.

For the two vignettes measuring expertise, the respondents also got introduced to a background

text on the same page as the branded post was shown. This text was following:

Page 47: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

38

Important to read this part. Anna has been playing golf since she was two years old and she is

the youngest pro player that has been winning the Amateur Golf Championship. In the year

2017, she won the Ladies European Tour in Australia. Since a couple of months back Anna has

been collaborating with a golf company and tried their products out. Now she is sharing her

experience of the product and a discount code on her Instagram. Please read Anna's text.

For the vignette with a male, the name was changed and some of the winning competitions was

changed as well. All vignettes used the same questions. To see an example of one of the full

questionnaires, see appendix B.

Since the questionnaire was adapted after the qualitative validity check the author needed to

add self-made questions to measure the new factor in the model. For all other questions, prior

literature was the guide to collect good questionnaire questions, see table 3.3 below. “The key

to developing good survey questions is to precisely define what is being measured.” (Dolnicar,

2013, p.553). The process of conducting well-articulated questions that will measure what is

intended to be measured are underestimated by many. The aim with the questionnaire is to

collect data that later will help the researcher to test the hypothesis in order to answer the

research question (David & Sutter, 2016). The questionnaire in this study, was mainly

constructed out of questions from Munnukka et al. (2016), Yuan and Jang, (2008) and Wu and

Lin’s, (2017) articles, which also has been used in some extent as references in the literature

review and to adapt the model, se figure 4 above. However, based on the validity check some

questions needed to be constructed by the author and not based them.

According to David and Sutton (2016) another important part to consider when constructing a

questionnaire is what type of questions that will be used; standardized or non-standardized

questions. Standardized questions are questions that already provided the respondent with

potential answers where non-standardized questions give the respondent the opportunity to

answer with their own words. Standardized questions have both pros and cons since they are

usually much faster to answer and it enables an easier process for the researcher to collect and

analyze the data. However, the use of standardized questions also increases the opportunity for

incorrect answers and the respondent could feel forced to answer something that was not on

their mind (David & Sutton, 2016). Since the main questions in this study were based on prior

studies using Liker scales, many questions were measured through Liker scales, all the

questions, expect the biographical questions in this study used a 5-point Liker scale.

Page 48: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

39

This means that this study only used standardized questions. The biographical questions were

not scale questions; however, they were also standardized questions. The decision only to use

standardized questions was to have a consistency in the questionnaire, decrees the time it took

to answer the questionnaire, and to make the analysis of the data easier and faster based on time

restriction. Since the questionnaire was sent out to all users of the app, the participants could

be from different countries around Europe. Therefore, the questionnaire was presented in

Swedish and English.

Table 3.3: Question construction

Measure Questions asked for each variable Source:

TB I consider this post as reliable Wu & Lin’s (2017) I consider this post as convincing

I consider this post as credible

TRUST I feel the endorser was honest Munnukka et al., (2016) I consider the endorser trustworthy

I feel the endorser was earnest

EXP I feel the endorser knows a lot about the product Munnukka et al., (2016) I feel the endorser is competent to make assertions about the product

I consider the endorser sufficiently experienced to make assertions about the product

SIMIL The endorser and I have a lot in common Munnukka et al., (2016) The endorser and I are a lot alike

I can easily identify with the endorser.

USE I consider this post as valuable Wu & Lin’s (2017) I consider this post as informative

I consider this post to be helpful

QUA I consider the quality in this picture as high Self-constructed

The quality in this post gives me a serious impression of the company

Branded posts with high quality give a more serios impression

PI I would listen to this person’s advice and follow this person on social media Yuan & Jang, (2008)

After seeing this post and knowing the influencers background, I considering using this product

Next time I will book a tee time, I will use this product

Page 49: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

40

3.5 Credibility When the researcher has decided what methods and tools that are going to be used to collect

the data, the researcher also needs to evaluate the quality of the research. Saunders et al. (2012)

mean that “the internal validity and reliability of the data you collect and the response rate you

achieve depend, to a large extent, on the design of your questions, the structure of your

questionnaire, and the rigor of your pilot testing” (p.428). The difference between validity and

reliability is that high validity indicates a study that has measured data that should be measured.

When high reliability indicates a consistent in the study which means that if the study was done

one more time the same result would be presented (David & Sutton, 2016). To achieve validity

and reliability, Saunders et al. (2012) discuss four important steps that must occur in the

questionnaire:

• Researcher is clear about the data required and designs a question

• Respondent decodes the question in the way the researcher intended

• Respondent answer the question

• Researcher decodes the answer in the way the respondent intended

“The question must be understood by the respondent in the way intended by the researcher and

the answer given by the respondent in the way intended by the researcher and the answer given

by the respondent must be understood by the researcher in the way intended by the respondent”

(Saunders et al., 2012, p.429).

3.5.1 Reliability

As was mentioned above, reliability is essential for a successful result of the study. To measure

reliability Golafshani, (2003) discuss three different types of reliability that the author needs to

refer to in a quantitative study: (1) the degree to which a measurement, given repeatedly,

remains the same (2) the stability of a measurement over time; and (3) the similarity of

measurements within a given time period. (p.598). The article also discusses the consistency in

between an individual’s answers which could be determined through a test-retest method. This

test measures stability and if the stability is high, the degree of reliability is indicated

(Golfashani, 2003).

Based on a time restriction, a test-retest was not made in this study. However, to contain the

reliability of this study, the selected scales in the questionnaire are from prior studies. The seven

items that were tested, were measured with a 5-point Likert scale ranging from 1 (strongly

Page 50: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

41

disagree) to 5 (strongly agree). Even if prior studies had a mix between a 5-point Liker scale

and a 7-point Liker scale, all questions in the questionnaire followed a 5-point Liker scale to

create a consistency in the questionnaire. When conducting a quantitative study, the researcher

needs to consider the reliability of the instruments that will be used to measure the collected

data. A common method to use then is Cronbach’s Alpha. The measured alpha in prior studies

regarding the different variable’s was all high, between 0,82, and 0.95. In this study, the

Cronbach alpha coefficient in the sharp questionnaire was measured to high as well, between

0,754 and 0,917. Since all of the variables had a Cronbach alpha above 0,7, the measured data

could be said to be reliable.

3.5.2 Validity

The validity, as discussed above, is about in which degree the study measures what it meant to

measure, or it could also be said how truthful the results are of the study (Golafshani, 2003). In

3.2.2 Experimental vignette methodology, Aguinis and Brandley, (2014) discuss two types of

validity in EVM, internal and external validity where the external validity could be the most

challenging validity to reach. To increase the external validity, the researchers using EVM

could put a focus on improving the level of realism in the vignettes (Aguinis & Brandley, 2014).

To increase the external validity in this study, the author made an identical post from Instagram

(see figure 3.3 above) with information taken from a prior collaboration between a professional

golf player and the company. The information texts regarding the golfer’s backgrounds were

also real information. The vignettes in this study also included both pictures and texts to

increase the realism, which affect the external validity. Aguinis and Brandley, (2014) also

discussed the importance of numbers of sessions. To increase the validity the vignettes should

preferably be made in a single session if not a specific area is measured such as history or if the

researcher uses videos. In this study, the author took this into account and all participants

responded to the vignettes in one session.

Page 51: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

42

3.6 Summary of methodology Figure 3.4 below illustrates a summary of what chapter 3 has concluded and the methodology

choices the author thought was most suitable for this project. Those boxes that contain a blue

color are the final decisions of what methods to use. The line between the boxes guide the reader

through the choices that has been made.

Figure 3.4: Summary of methodology

Source: Author’s own work.

Page 52: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

43

4. Data results In this chapter, a presentation of the collected data will be presented. First, a brief background

of the participants will be presented, followed by the results of the ANCOVA analysis. The data

that will be presented have been collected through questionnaires that were sent out to

consumers of the collaboration company. All analysis has been made in the program SPSS.

When the questionnaire closed for more answers, the total number of respondents was 196.

Since all questions in the questionnaire were mandatory to answer, the amount of 196 was fully

completed questionnaires that could be used in the analysis. However, since four (4) of the

participants answered that they did not want to disclose their gender these answers were

categorized as missing data, see table 4.1 below.

After coding the data into numbers (male = 0, Female = 1) in Excel, a frequency test was run

on all 196 answers. In excel, a dummy variable for expertise and similarity was added as well.

Those that had seen the vignette with text were coded with one (1) and those that had seen a

vignette with the same gender as themselves were coded with one (1). Those that did not see a

vignette with text or saw a vignette with another gender were coded with zero (0).

Table 4.1: Dummy variables.

Expertise Similarity

With text No Text Similar Not similar

Scenario 1 48 0 5 43

Scenario 2 0 47 2 45

Scenario 3 56 0 10 46

Scenario 4 0 45 9 36

Total 104 92 26 170

4.1 Description of the final sample

After the data was entered into SPSS a frequency test was made on the background variable

such as, gender, age, resident of living, when the participants usually played golf, how many

golf rounds they played each year, for how many years they had played golf and their golf

handicap. Below each variable will be presented in tables (except golf handicap where instead

the range from lowest to highest will be presented). In the first table, table 4.2, on the next page,

gender is presented. The majority of the respondents were men: 84.7% compared to 13.3 %

Page 53: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

44

females. Since golf is a sport where men are the dominating gender, this numbers represented

the diversity in the sport quite well.

Table 4.2: Participants gender

Frequency Percent Valid Percent

Male 166 84.7% 86.5%

Female 26 13.3% 13.5%

Total 192 98% 100%

Missing 4 2.0

Total 196 100%

Table 4.3, below, represent the age of the respondents. The median was 65+ which indicates a

relatively high respondent age compared to what previous literature has presented.

Table 4.3: Participants age

Frequency Percent Valid Percent

35-44 year 2 1.0% 1.0%

45-54 year 18 9.2% 9.2%

55-64 year 70 35.7% 35.7%

65+ year 106 54.1% 54.1%

Total 196 100% 100%

In table 4.4, on the next page, the participants resident of living is presented. Surprisingly,

19.4% of the respondents came from a country outside those countries the company is active

in. Otherwise, the majority of the respondents came from Sweden 67.3% which is a big market

area for the company.

Page 54: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

45

Table 4.4: Participants country of residence

Frequency Percent Valid Percent

Sweden 132 67.3% 67.3%

Spain 3 1.5% 1.5%

Italy 2 1.0% 1.0%

Portugal 3 1.5% 1.5%

France 5 2.6% 2.6%

Germany 1 0.5% 0.5%

Belgium 5 2.6% 2.6%

UK 2 1.0% 1.0%

Finland 5 2.6% 2.6%

Other 38 19.4% 19.4%

Total 196 100% 100%

Regarding the check of when the participants play golf, most of the respondents answered that

they like to play golf on weekends at their local clubs 77.0 %. This was followed by golf during

holidays 17.9 % and during business on 5.1%., see table 4.5.

Table 4.5: When the participants mainly play golf

Frequency Percent Valid Percent

Weekends 151 77.0% 77.0%

Holidays 35 17.9% 17.9%

Business 10 5.1% 5.1%

Total 196 100% 100%

Table 4.6, on the next page, presents how active each participant is each year measured in how

many golf rounds they play. 20.9% plays more than 21-30 rounds each year but the majority

plays 30+ rounds, 62.8%, each year.

Page 55: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

46

Table 4.6: Number of golf rounds participants play each year

Frequency Percent Valid Percent

5-10 rounds 5 2.6% 2.6%

11-20 rounds 27 13.8% 13.8%

21-30 rounds 41 20.9% 20.9%

30+ rounds 123 62.8% 62.8%

Total 196 100% 100%

In table 4.7, statistics over how long each participant had played golf is shown. It could be seen

that the majority of the respondents are dedicated players that have been practicing golf for

many years.

Table 4.7: Years of interest/played golf among the participants

Frequency Percent Valid Percent

0-5 years 10 5.1% 5.1%

6-10 years 23 11.7% 11.7%

11-15 years 31 15.8% 15.8%

16-20 years 43 21.9% 21.9%

21-25 years 26 13.3% 13.3%

25+ years 63 32.1% 32.1%

Total 196 100% 100%

The last background question that was asked was regarding the participants golf handicap. The

range between the handicaps was 3,5 up to 36, where 36 is what you have as a beginner and 3,5

is representing a person with a good handicap that had played for a while. The mean was 20,2

and the median was 18,55.

4.2 Testing assumptions Before the data was analyzed through ANCOVA, a reliability test and correlation were run on

all variables to look for errors in the data. Below the results from these tests are presented.

4.2.1 Reliability of covariates

In table 4.8, on the next page, the reliability test is presented. As discussed in the methodology

chapter, the Cronbach alpha should be above 0.7 to indicate on a scale measuring what it should

be measured.

Page 56: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

47

Table: 4.8: Cronbach Alpha

Measure Items Cronbach’s 𝜶

TB I consider this post as reliable 0.961

I consider this post as convincing

I consider this post as credible

TRUST I feel the endorser was honest 0.966

I consider the endorser trustworthy

I feel the endorser was earnest

EXP I feel the endorser knows a lot about the product 0.925

I feel the endorser is competent to make assertions about the product

I consider the endorser sufficiently experienced to make assertions about the product

SIMIL The endorser and I have a lot in common 0.947

The endorser and I are a lot alike

I can easily identify with the endorser.

USE I consider this post as valuable 0.922

I consider this post as informative

I consider this post to be helpful

QUA I consider the quality in this picture as high 0.879

The quality in this post gives me a serious impression of the company

Branded posts with high quality give a more serios impression

PI I would listen to this person’s advice and follow this person on social media 0.888

After seeing this post and knowing the influencers background, I considering using this product

Next time I will book a tee time, I will use this product Since the Cronbach Alpha was above 0.7 (high reliability) in all seven variables, this made it

possible to merge the number of items into one item for all variables. This was made to, later

on, be able to run the ANCOVA test.

4.2.2 Correlations among the covariates

When it comes to the correlations among the different covariates, Pallant (2010) explain that it

should not be strong correlations among the variables. What the researcher would like to

achieve is “a group of covariates that correlate substantially with the dependent variable but not

with one another” (Pallant, 2010. p.300). Table 4.9 below, represent the correlation between

the different variables except for the dependent variable (purchase intention).

Page 57: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

48

Table 4.9: Correlations

1 2 3 4 5 6

1.TB r 1

2.TRUST r .857** 1

3.EXP r .794** .865** 1

4.SIMI r .715** .785** .706** 1

5.USE r .774** .803** .740** .771** 1

6.QUA r .775** .847** .800** .774** .887** 1

According to Pallant (2010), a strong relationship between covariates is all r above .80. If r >.80

the researcher should consider removing those covariates. When the covariates have a

correlation above .80 they overlapping each other and are not contributing to a reduction in

error variance (p.300). In this case, we have a couple of covariates above .80 (marked bold) in

table 4.9.

To understand why the correlations were as high as they were, a new frequency test was run to

look at the distribution (skewness and kurtosis) of the data. In Appendix C all frequency tables

for each variable are presented. The result shows that the majority of the respondent strongly

disagreed (1), disagreed (2) or was neutral (3) in all 21 statement that was shown in the

questionnaire. Regarding the skewness and kurtosis, these two factors give an indication of the

symmetry and “peakedness” of the data. A perfectly normal distribution should contain

skewness and kurtosis with a value of 0 (Pallant, 2010). The results from the collected data

indicated on some variables more skewed than others. However, none of the variables was

perfectly normal, presented in table 4.10 below, which also could be seen in the frequency

tables in Appendix C.

Table 4.10: Distribution among the variables.

TB TRUST EXP SIMI USE QUA PI

Skewness .313 .393 .208 .945 .464 .463 .864

Kurtosis -.831 -.693 -.888 -.223 -1.030 -.895 -.433

In table 4.10, similarity (SIMI) and purchase intention (PI) (marked as bold) was those two

variables that had the highest skewness. When the data looks as it does above, Pallant (2010),

explains that analyses with very skewed distribution could be divided into equal groups, called

Visual Binning. With Visual Binning, the respondents’ score is divided into equal groups (low,

Page 58: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

49

medium, and high scoring groups). Based on the skewness in the data, each variable was divided

into equal groups before running the ANCOVA test. Since the collected data was measured

through Likert scales (very basic ordinal data) and not as factors as ANCOVA is demanding,

the most appropriate way to manage the data was to use Visual Binning based on the frequency

in the data, see Appendix C. The majority of the respondents had answered equal to three (3)

or below (neutral, disagree, or strongly disagree) and because of the unequal distribution the

variables were divided with cut-off points. Since the outcome variable (purchase intention) is

continuous, Visual Binning was not made on that variable. For the variable usefulness, quality,

and similarities, one (1) cut-point was used, and for the rest of the factors, two (2) cut-points

were used. This to make it more true to the distribution in the data.

4.2.3 Homogeneity

To test the relationship between the covariate and the depended variable a final assumption test

is rune, homogeneity of regression slopes (Pallant, 2010), see table 4.11 below. The interest in

this test is the significant of the interaction term, (including both independent variables and

covariates) which should be greater than .05. If the significant level is greater than .05 the data

is not violated and no interaction between the covariate and the treatment could be found

(Pallant, 2010).

Table 4.11: Homogeneity of regression slopes

Source Type III Sum of Squares df Mean Square F Sig.

Partial Eta Squared

Correlated model 135.301a 65 2.082 10.540 .000 .845

Intercept 221.226 1 221.226 1120.171 .000 .899

BiUse*BiQua*BiTrust

*BiExp*BiSim*BiTB

118.550 50 2.371 12.005 .000 .827

Gender*Age .069 1 .069 .350 .555 .003

((BiUse*BiQua*BiTrust

*BiExp*BiSim*BiTB) *

(Gender*Age))

.835 10 .084 .423 .933 .032

Error 24.884 126 .197

Total 795.778 192

Corrected Total 160.185 191

a. R Squared = .845 (Adjusted R Squared = .765)

Page 59: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

50

As shown in table 4.11, the interaction team’s significant (marked with bold and circled) is

safely above the significant level on .05 (p = .933). Therefore, the author has not violated the

assumption of homogeneity in this study.

4.3 Significant of the data results To take part in the full ANCOVA analysis, the results are found in Appendix C. Below some

of the important numbers are presented in table 4.12 – 4.14. In table 4.12 the Levene’s result is

presented. The Levene’s table indicates on similar results as table 4.11, since the numbers in

the table tells if the assumption of equality of variance is violated or no (Pallant, 2010). As in

table 4.11 the significant level should be greater than .05 in the Leven’s test as well. In the

homogeneity of regression slopes the significant level came out fine, however, in the Levene’s

test there were a significant level since we had sig. <.05.

Table 4.12: Levene’s Test of Equality of Error Variancesa

Dependent Variable: Purchase Intension F df1 df2 Sig.

2.321 54 137 .000

Violated data could be the result of analyses with incorrect or misleading results which needs

to be taken into consideration when drawing the conclusions in this study.

In the main ANCOVA result, the column presenting the significant value is of interest. If the

value in this column (Sig.) is less than .05 there if a difference in terms between the groups

(Pallant, 2010). In table 4.13 the significant level for each variable is shown together with the r

square.

Page 60: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

51

Table 4.13: Significant level in ANCOVA

F Sig. Partial Eta Squared

Gender Purchase intention .102 .750 .001

Age Purchase intention .019 .892 .000

Trust in branded post Purchase intention .0466 .628 .007

Trustworthiness Purchase intention .151 .860 .002

Expertise Purchase intention 4.606 .012 .064

Similarity Purchase intention 28.787 .000 .176

Usefulness Purchase intention 11.410 .001 .078

Quality Purchase intention 2.498 .116 .018

a. R Squared = .839 (Adjusted R Squared = .773) Those numbers that are bold are indicating a significant number of less than .05. A significant

level below .05 indicates a significant difference in mean. In the collected data there were three

variables (expertise, similarity, and usefulness) with a significant level. The partial Eta Squared,

the results describe how much of the variance in the dependent variable that is described by the

independent variable (Pallant, 2010). In this case, the partial Eta Square results that are of

interest, are those for the independent variable at a significant level. Partial Eta Square results

are measured through 0.2 – small effect, 0.5- moderate effect, and 0.8.- large effect. Since the

adjusted R Squared are .773, we know that 77.3% of the variance in the dependent variable is

explained by the independent variables.

Lastly, the estimated marginal means for each variable with a significant level are presented in

table 4.14. What is given of the results in the table of estimated marginal means is the “adjusted

means on the dependent variable for each of our groups, split according to each of our

independent variables separately and then jointly (Pallant, 2010. p.316).

Table: 4.14 Estimated Marginal Means

95% Confidence interval

Usefulness Mean Std. Error Lower Bound Upper Bound

<=2.00 1.385a,b .068 1.250 1.520

2.01 + 2.126a,b .069 1.989 2.264

Expertise

<=2.00 1.459a,b .080 1.292 1.607

2.01 – 3.00 1.790a,b .074 1.644 1.936

3.01 + 2.120a,b .106 1.910 2.329

Page 61: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

52

Similarity

<=1.33 1.368a,b .080 1.210 1.526

1.34+ 2.026a,b .061 1.906 2.145

The estimated marginal means is read in conjunction with the significant level and in the

column to the left, the three variables (usefulness, expertise, and similarity) is divided into the

different visual binning groups as discussed above. As could be seen in the table, expertise is

divided into three groups and usefulness and similarity into two groups. The column “mean”

indicates on the mean of each groups purchase intension. As could be read in the table, a higher

agreement on the independent variable indicates on a higher purchase intention.

Based on the presented data, three significant paths were found in this study. These are shown

in figure 4.1 below.

Figure 4.1: Significant paths

Source: Author’s own work

Figure 4.1 also indicates that there were three hypotheses supported in this study, hypotheses

3a, 3c, and 3d. All others were rejected. In the following chapter an analysis of the results is

presented together with a comparison to findings in previous literature.

Expertise

Similarity

Usefulness

Purchase intentions

p = .000; r2 = .176

Page 62: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

53

5. Analysis of data In the following chapter, an analysis of the collected data will be presented. A comparison

between the collected data presented in chapter 4 and the conceptual framework in chapter 2

will be made in a similar structure of prior chapters to maintain consistency in the report.

5.1 High involvement High involvement, in general, is measured in the engagement and time a consumer puts down

in a product (Aldlaigan & Buttle, 2001). For companies, the importance to be able to create

high involvement among their consumers is to create a product (goods or services) that the

consumers want to identify themselves with (Radder & Huang, 2008). In the results, there is

evidence indicating that golf is a high engagement activity. The majority of the respondents

were shown to put down a lot of engagement and time in the sport which is in line with was

Aldlaigan and Buttle (2001) said indicates on a high involvement level. Among all the

respondents, 32.1% had played or have had golf as an interest for more than 25 years and a total

of 67,3% of the participants had have golf as an interest for 16 years or more, which is evidence

pointing at a high engagement level in the sport. Further, the time each participant spend on the

sport was also indicated on a high involvement level where 62.8% played more than 30 rounds

each year. That is more than two rounds each month, on average.

Comparing the findings in this study with Ha and Hu’s (2013), which said that users with a

high SNS involvement were mainly interested in gossip and “soft news” as those users with

low SNS involvement had a stronger interest in the news regarding politics, sports or science,

this seemed to be the same in this study. The level of SNS involvement was, according to Ha

and Hu’s (2013), an indication of how likely it would be that a person would like, commenting,

or sharing an interest in other’s channels, and based on the result in this study, the participants

did not seem to be as likely to share, follow the influencer, or commenting in this study.

Therefore, the results strengthen Ha and Hu’s (2013) findings.

However, Ha and Hu’s study (2013), was conducted on participants of a relatively young age

(99,8% was between 18-25 years old) and this study have a majority (54,1%) of participants

with an age above 65 years old. In a comparison between these two generations on a report

from Statista.com 2019, numbers, based on Instagram users in Sweden, show that 31% of

people above 66 years old was active on Instagram when 85% in the age between 16-25 years

old was active. The huge different in age needs to be considered in the conclusions and

Page 63: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

54

according to the findings, the respondents’ high age, most likely, are the impacted factor of the

result of the low engagement on Instagram. Although, the sample in this study worked out well,

since it shows that people are happily involved in the sport for many years and the study has

evidence, showing that golf is a high involvement activity.

5.2 Analysis of H1

Regarding the results of the first hypothesis: trust in branded posts mediates the effect of the

source credibility variables (trustworthiness, expertise, similarity, usefulness, and quality) on

purchase intention, the findings differ from previous literature. In previous studies consumers’

trust in a brand, when considering a purchase of a product, has shown to be important in both

high- and low-involvement products (Corbitt et al., 2003). Lou and Yuan (2019) tested the

relationship between trust in branded posts and purchase intentions and found a positive

relationship between the two variables. Reinikainen et al. (2020), tested the dimensions that

constitute source credibility in a similar context of trust (brand trust) and found that brand trust

is an important construct when using endorsers in social media (in their case YouTube) to

increase the purchase intention. Even Foti and Devin (2019), that looked at an ethical high

involvement product, also found trust as an important variable when marketing the product

(they did not conduct their study in the area of influencer marketing though). However, the

result in this study did not found a relationship between trust in branded posts and purchase

intension, and H1 was therefore rejected.

Why the result in this study differ from previous findings could be a result of adopting the

SMIV model in a non-visual high involvement product (a service), when the majority of

previous literature, in the area of influencer marketing, has mainly been conducted in the

context of visual low involvement products (in form of goods) or influencer marketing in

general. This study is an evidence of the great differences it is between product categories, not

only between high- and low-involvement products but also between products (goods/services)

itself. Further the results also indicate the importance of creating a bigger understanding of the

effect influencer marketing has on different generations since this study was conducted on an

older generation and prior literature has mainly conducted data on the younger generation.

Page 64: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

55

5.3 Analysis of H2a, H2b, H2c H2d, and H2e H2: Source credibility, including (a) expertise, (b) trustworthiness, (c) similarity, (d)

usefulness, and (e) quality will positively influence the consumers trust in branded posts.

Even though trust in branded posts did not have a mediation role in this study, prior literature

has found the variables that constitute source credibility to impact trust in branded posts. Lou

and Yuan (2019) got a result indicating that both trustworthiness and similarity had a positive

influence on consumers’ trust in branded posts and even if Xiao et al. (2018) did not exactly

compare the variables with trust in branded posts, instead, they used the factor of information

credibility in the context of the credibility of eWOM communication, they found

trustworthiness and expertise to have a positive impact on information credibility. Although,

trustworthiness was shown to be much more important for consumers to feel the eWOM

communication credible than expertise was. Further, Xiao et al. (2018) argued that the result

might have looked different based on what type of product category that was measured. And

that is exactly what is found in the empirical data in this study. When running the ANCOVA,

no significant levels could be found (see Appendix D) between the independent variables and

trust in branded posts. Therefore, hypotheses H2a-e was rejected and not in line with what

previous literature has found.

The findings in this study could be a result of the high involvement level and dept of

involvement each participant has in the sport. With so many years of experience, the

practitioners have developed internal expertise in the sport, and trust may not be as important

in these contexts as in low involvement products where expertise might not be as important.

Even if the key person from the industry said during the interview that they as a company have

not noticed any difference between the effectiveness of influencer marketing in the different

product categories, the empirical data in this study tells us there is. Lou and Yuan’s (2019)

SMIV model (see figure 2.1) is working well in the context of visual products where the

involvement level is low, but not in the context of a non-visual product where the consumers’

engagement and passion for the product are as high as it is in this study. The internal expertise

of the consumers, built from their solid background and age, is beyond what previous literature

has been able to explore.

Page 65: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

56

5.4 Analysis of H3a, H3b, H3c, H3d, and H3e Regarding trustworthiness, expertise and similarity’s impact on purchase intention, the result

from previous literature has had different results (Lou & Yuan, 2019; Munnukka et al., 2016;

Lafferty et al., 2002). As has been mentioned before, the way of testing source credibility has

differed in previous literature, although, the most common findings, between previous

literature, have been that the variables in source credibility have mainly had an indirect impact

on purchase intention and not a direct impact.

When testing hypotheses 3a-e; (a) expertise, (b) trustworthiness, (c) similarity, (d) usefulness,

and (e) quality’s impact on the consumers purchase intentions, three independent variables;

expertise (F = 4.606 and p = .012), usefulness (F = 11.41 and p = .001), and similarity (F =

28.787 and p = .00), had a direct impact on purchase intension. The partial Eta Squared number,

in table 4.13 above, tells that each variable themselves do not explain as much of the variance

in purchase intention (expertise 6,4%, similarity 17,8% and usefulness 7,8%) however, together

they explained 77,3% (Adjusted R Squared = .773) of the variance in purchase intention. With

the results in estimated marginal means, see table 4.14 above, we could tell that there is a

positive impact from the independent variable on purchase intension. The higher the usefulness,

similarity, and expertise are, the higher the consumers’ purchase intention will be. Based on

these results, hypotheses 3a, 3c, and 3d were supported where 3b and 3e were not.

In comparison to Lou and Yuan’s findings (2019), the only variable that had a relationship with

purchase intention in their study was trustworthiness, although trustworthiness had a negative

influence on purchase intention, which in this study did not have any relationship at all. The

two variables, expertise and similarity, did not, in Lou and Yuan’s (2019) study, have any direct

impact on purchase intention, which they had in this study. This study’s results were instead

partly in line with Ohanian’s result (1991) that found expertise to be the only variable (out of

trustworthiness, expertise, and attractiveness) to have a significant effect on consumers’

purchase intention. Further, expertise was one of the variables that the key person from the

industry said should be an important factor, especially in golf, which was shown to be correct

in this study.

Regarding the variable usefulness, Lou and Yuan (2019) recommended this variable in their

area of further research which, in this study, was found to have a significant level. Why

usefulness was significant in this study might be connected with Ha and Hu’s findings (2013).

Page 66: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

57

People with low SNS involvement was, according to Ha and Hu (2013), mainly interested in

sports, politics and “heavier” news, which requires useful content and “real” information. De

Veriman et al. (2017) also said that the communication from the influencer needs to be relevant

for the followers and that relevant communication is a result of a good fit between the brand

and the influencers.

Another finding in this study, that contradicts previous literature (Munnukka et al., 2016) saying

that similarity is said to be more important for inexperienced consumers, than what it is for

experienced consumers, which was not the case in this study. Since the experience among the

respondents in this study, (94,9% had played golf for 6 years or more) was very high, the

similarity would, according to Munnukka et al. (2016), not be as important as it was in this

study. However, the respondents’ high engagement and internal expertise could be objects for

way similarity is important in this study. Because when hiring an influencer to promote a

product for an audience with this level of experience and years of interest in the product, the

consumers’ own expertise needs to be considered and the influencer’s backgrounds need to

match with the consumers. Otherwise, the consumers own expertise might take over their

judgment of the advertisement. This may not be as important when the product is visual, and

the consumers involvement level in the product is not as high as it was for the respondents in

this study.

5.5 Analysis of the moderators

The two covariates in this analysis were gender and age, which Lou and Yuan (2019) found did

not have a significant effect on the consumers’ reaction. Same results as Ohanian found in her

study year 1990. The result from the collected data in this study indicated the same as prior

literature since there was an insignificant effect in both age (F = .019 and p = .892) and gender

(F = .102 and p = .750). Based on the high significant (p) numbers, age and gender did not

affect the collected data in this study.

5.6 Significant paths The significant paths in this study differ a lot from what Lou and Yuan (2019) present in their

study, see figure 4.1 above. However, these two studies are conducted on two completely

different cases. Lou and Yuan (2019) collected their data in the United States, based on a

younger generation with the majority of female participants responding to their general

experience and thoughts about an influencer they had followed on Instagram. The collected

Page 67: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

58

data in this study came from a quite sophisticated sample where the involvement level and their

internal expertise was high. The nationality in this study also differs from previous literature

where the majority of the participants had their residence in Sweden (or Europe except the

19,4% that marked “other”). Further, the participants were asked questions regarding their

thoughts toward the promoted product (which was a non-visual product in a very niche market),

and toward the influencer, they saw at the beginning of the questionnaire when prior literature

mainly have focused on evaluated visual products or the phenomenon in general.

Therefore, the three significant paths that were found in this study (expertise, similarity, and

usefulness toward purchase intention) compare to findings in previous studies, indicate

evidence on an unexplored area that needs to be further researched. The significant paths in this

study may not even be the same in other sports contexts since sports could not only be

considered as sports; sports have many different aspects where golf is only one of many.

Appling the SMIV model in, for example, sailing, or in the area of music, or IT could result in

completely different outcomes. What could be said is that based on the findings in this study,

the SMIV model seems to not be applicable in more niched and high involvement product

categories.

Page 68: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

59

6. Findings and conclusions In the final chapter, the findings and conclusions of this study will be presented. The chapter will give the readers an answer to the research question followed by theoretical and practical contributions. The chapter will end with the limitations of the study and suggestions for future research.

6.1 Answer to research question The overall purpose of this study was to create an understanding of how influencer marketing

could increase the purchase intention in the area of a high involvement product in a niche

market. Based on this purpose one research question was stated:

How is the purchase intention generated through the use of social media networks in a highly

involved, niche market?

Since this study was conducted together with a Swedish golf company, the area of this thesis

was focusing on the golf industry. For the company, the importance was in creating an

understanding of how they could use influencer marketing to generate purchase intention for

one of their products on the market. To be able to answer the research question, and understand

how the company could work with influencer marketing as effective as possible, the SMIV

model was adopted and three hypotheses were developed, all results presented in chapters 4

and 5. What was found was that hypotheses H3a, H3c, and H3d were supported when H1, H2a-

e, H3b, H3e were rejected. The supported hypotheses indicated that there was a positive

relationship between expertise and purchase intention, usefulness and purchase intention, and

similarity and purchase intention. Based on these findings, the answer to the research questions

is; to generate purchase intention in a highly involved niche market through the use of social

media network, where the audience is of great expertise themselves in the area, it is important

to consider three factors: the expertise of the influencer representing the product, the similarity

between the influencer and consumers and lastly, it is important to create posts containing a

value for the consumers and that the consumers feel that the post is helpful. In other words, it

is important that the consumers feel that the post is useful.

However, the findings in this study are not only answering the research question, explaining

that there are three important variables to considered when using influencer marketing to

promote a non-visual high involvement product toward highly dedicated consumers spending

much time and engagement in the area, this study also prevents literature with evidence on the

Page 69: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

60

great difference it is when promoting a low- or high involvement product. In this study, there

was no relationship between trust in branded posts and source credibility and therefore no

indirect impact on purchase intention either. The respondent’s internal expertise seems to make

them look for other important variables in the promoted posts and the influencer is instead in

need of so much more to convince the consumers to purchase the product than what an

influencer in a visual, low involvement product is.

6.1.1 Specific findings toward the collaborating company

As for the collaborating company, using influencer marketing, it is important to have in mind

that this study has guidelines for how to use the strategy toward an older generation having a

lot of experience themselves in the area. But the effect of influencer marketing on the younger

generation in this context could not be distributed since they were not represented in the study.

A conclusion for why there was such inequality of generations in the sample, could be that the

company, in their newsletter, (which was the channel the questionnaire was sent out through),

mainly reach the older generation, where the communication with the younger generation seems

to be in other channels. What also was found in this study, was that influencer marketing did

not seem to be as popular among the respondents (see Appendix C), and therefore, this study

not really present any conclusion of how influencer marketing could be used to increase the

purchase intention among the company’s consumers. What though could be concluded is that

since the company seems to have different age groups in their different communication

channels, they could customize the content in the different channels to fit better with the

generation reaching the message.

6.2 Contribution to the theory Some of the findings in this study support prior literature, such as that age, and gender do not

have any significant effect in this type of context (Lou & Yuan, 2019; Ohanian, 1990). It also

supports Ohanian’s (1990) findings on a significant effect between expertise and consumers’

purchase intention of a product. However, the major finding in this study is that there exists a

difference in the effect of influencer marketing depending on what product category and

audience they are targeting. Companies promoting a high involvement product could not relay

on findings in prior literature building their conclusions on a low involvement product, since

the important variables may not be the same. These findings are of importance for future

research to look further into, to understand what is needed in different contexts. This study is a

Page 70: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

61

great standing point for future research to use and apply in other product categories. Further,

this study contributes to theory by expanding the important variables that need to be considered

when explaining the fundamental mechanism of the effect of influencer marketing. This study

also fills the gap that is discussed in chapter one and brings evidence into literature regarding

the difference that exists between marketing a high and low involvement product. The current

study also builds further on the findings of Lou and Yuan (2019), where trustworthiness was

influencing purchase intention negatively, where trustworthiness in this study did not have any

significant level at all.

6.3 Implications for practitioners These results offer more than one important insight for marketing practitioners in the area of

influencer marketing for high involvement products in a niche market. To be able to build

purchase intention among consumers with internal expertise in the area and a high involvement,

marketers should pay attention to the expertise of the influencer, the similarity consumers could

create with the influencer, and the usefulness the posted post gives the consumer. The perceived

usefulness could, for example, be connected toward how helpful the consumers thought the

information was, or if the consumer found the post valuable. Therefore, could usefulness be

enhanced through using well-explained information easy to understand and including valuable

information for the consumers. Perceived expertise could be simply gained by letting the

influencer use the product and using influencers with a strong background in the area. For

example, professional players in golf when advertising a golf product. Moreover, this study

contributes marketeers with an understanding that there is a difference between the effect of

influencer marketing’s based on the product category and the engagement and knowledge of

the consumers. What is working for one company may not work for another. Further, this study

also gives some meaningful insight for influencers themselves, working with products in a more

niche area with highly involved products, of what they should mediate in the branded posts.

6.4 Limitations and further research

As in many other cases, the presented study has a number of limitations that opens the

opportunity for further research. To close the gap that was discussed in the first chapter

regarding literature lacking explanations toward the effect of influencer marketing in high

involvement products, this study has provided important factors to considered when using

influencer marketing in a more niche market presenting a high involvement product. However,

this study also creates an understanding of how important it is to continue the research in the

Page 71: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

62

area since the outcome of this study contributes to new findings and unexplored results. Further,

as discussed above, sport is not only sport, there is some much that differed one sport from

another and there are also different age profiles that need to be considered. Therefore, future

research should continue to study different niched sports where the visualization, as in fashion,

is not as easy to show and where the age groups difference. A good base for further research

could be to use the model, including similarity, usefulness, and expertise, and maybe develop

the model with adding, for example, information as a factor in other high involvement product

categories. As Lou and Yuan (2019) found a relationship with information in their study, this

could be further research in areas such as sailing or other sports and niche areas.

Regarding one of the study’s limitations, the background information in the questionnaires, and

the information in general, was quite thin for the participants to make a judgment on and the

participants were not really exposed to any of the influencer’s advice. This should for sure be

considered in further research where they instead could engage the participants more in the app

and let the influencer giving his/her advice on what is good with the app and what is not so

good. For further research, videos where the influencer navigates in the app, would have a

greater impact and include more useful information for the practitioners and therefore make it

easier to make a judgment in the end. Another suggestion for future research is to look closer

into how effective influencer marketing actually is on people that already have a lot of

experience and internal expertise in the area. As could be found in this study, the participants’

general experience toward the stimuli (Instagram post with an influencer promoting the

product) was not as high (mostly answers below neutral (3) to strongly disagree (1)) which

could be an outcome of the high engagement in the sport. To explore this further, the elaboration

likelihood model (ELM), presenting two routes to persuasion, one based on the thoughtful and

the other on affective associations could be interesting to implement. The ELM was developed

to describe changes in attitudes and different ways a person could use to processing stimuli

based on the elaboration level. According to Petty and Cacioppo’s (1984) results, a high level

of message elaboration leads the person into one of the routes, the central route (thoughtful),

and a low level of message elaboration leads a person into the other, peripheral route (affective

associations).

Additionally, the collected data in this study were skewed in some of the variables which made

it hard to analyze the data without grouping the data into low, medium, and high groups (visual

binning). This could be one reason for why the Levene’s test did not came out ok when the

Page 72: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

63

homogeneity test was. For further research the importance should be in conducting data without

violation of assumption. Last, this study is limited by the fact that the researcher tried to find

out how influencer marketing was accepted and, thought of by the consumers of a golf company

in the golf industry. However, the understanding of the general use of Instagram among the

respondents was left out. This may imply that many of the respondents in this study did not

even have an Instagram account or not even was familiar with influencers in general. Therefore,

further research should have this in mind when conducting a study in this field. There is still a

smaller percentage of the older generation that uses Instagram so far. Furthermore, future

research should continue to study the area of influencer marketing in high involvement products

in more niche markets, for example, extreme sports. Influencer marketing is still quite

unexplored in these areas and more literature is needed. Also, comparing the effect of influencer

marketing across different social media channels such as Facebook, LinkedIn, and Twitter

would be interesting.

6.5 Conclusions This study tests the SMIV model to create a bigger understanding of the fundamental

mechanism of influencer marketing effects on consumers in a high involvement product in a

niche market. Based on this study’s empirical data, three important variables were found,

expertise, similarity, and usefulness. However, this study is also evidence that a general

conclusion in this area is not enough and that the SMIV model is not applicable in all contexts.

The effect of influencer marketing in a low involvement product is not the same as it is in a

high involvement product. In product categories where consumers have the internal expertise

and are highly experienced, the original SMIV model could not be used. So, the findings in this

study open up for more questions to be examined in the area of influencer marketing in high

involvement product categories.

Page 73: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

64

Reference list Aguinis, H., & Bradley, K. J. (2014). Best practice recommendations for designing and implementing experimental vignette methodology studies. Organizational Research Methods, 17(4), 351-371. Ahmad, Irfan (2018), “The Influencer Marketing Revolution,” Social Media Today, Retrieved 2020 January 29 from https:// www.socialmediatoday.com/news/the-influencer-market- ing-revolution-infographic/517146/. Aldlaigan, A. H., & Buttle, F. A. (2001). Consumer involvement in financial services: an empirical test of two measures. International Journal of Bank Marketing. Amos, C., Holmes, G., & Strutton, D. (2008). Exploring the relationship between celebrity endorser effects and advertising effectiveness: A quantitative synthesis of effect size. International journal of advertising, 27(2), 209-234. Anagnostopoulos, C., Parganas, P., Chadwick, S., & Fenton, A. (2018). Branding in pictures: using Instagram as a brand management tool in professional team sport organisations. European Sport Management Quarterly, 18(4), 413-438. Cacioppo, J. T., & Petty, R. E. (1984). The elaboration likelihood model of persuasion. ACR North American Advances. Ceron-Anaya, H. (2010). An approach to the history of golf: Business, symbolic capital, and technologies of the self. Journal of Sport and Social Issues, 34(3), 339-358. Chan-Olmsted, S., & Xiao, M. (2019). Smart Sports Fans: Factors Influencing Sport Consumption on Smartphones. Sport Marketing Quarterly, 28(4). Charness, G., Gneezy, U., & Kuhn, M. A. (2012). Experimental methods: Between-subject and within-subject design. Journal of Economic Behavior & Organization, 81(1), 1-8. Chipp, K. (2004). E-Commerce: A southern African perspective. New Africa Books. Corbitt, B. J., Thanasankit, T., & Yi, H. (2003). Trust and e-commerce: a study of consumer perceptions. Electronic commerce research and applications, 2(3), 203-215.

Corey, L. G. (1971). People Who Claim to be Opinion Leaders: Identifying Their Characteristics by Self-report. Journal of Marketing, 35(4), 48-53.

David, M. & Sutton, C. (2016). Samhallsvetenskaplig Metod. 1st ed. Lund: Studentlitteratur AB

Page 74: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

65

De Veirman, M., Cauberghe, V., & Hudders, L. (2017). Marketing through Instagram influencers: the impact of number of followers and product divergence on brand attitude. International Journal of Advertising, 36(5), 798-828.

Djafarova, E., & Trofimenko, O. (2019). ‘Instafamous’ – credibility and self-presentation of micro-celebrities on social media, Information. Communication & Society, 22(10), 1432-1446.

Dolnicar, S. (2013). Asking good survey questions. Journal of Travel Research, 52(5), 551-574. Evans, N. J., Phua, J., Lim, J., & Jun, H. (2017). Disclosing Instagram influencer advertising: The effects of disclosure language on advertising recognition, attitudes, and behavioral intent. Journal of Interactive Advertising, 17(2), 138-149 Forsyth, P. (2020). Golf is slowly changing its attitude towards women – but there is still a long way to go. Retrieved 2020 Mars 25 from https://www.thetimes.co.uk/article/golf-is-slowly-changing-its-attitude-towards-women-but-there-is-still-a-long-way-to-go-t55kbgt0s Foti, L., & Devine, A. (2019). High Involvement and Ethical Consumption: A Study of the Environmentally Certified Home Purchase Decision. Sustainability, 11(19), 5353. Freberg, K., Graham, K., McGaughey, K., & Freberg, L. A. (2011). Who are the social media influencers? A study of public perceptions of personality. Public Relations Review, 37(1), 90-92.

Ganguly, S. (2015), Why Social Media Advertising Is Set to Explode in the Next 3 Years. Retrieved 2020, February 3 from https://marketingland.com/social-media-advertising-set-explode-next-3-years-121691

Golafshani, N. (2003). Understanding reliability and validity in qualitative research. The qualitative report, 8(4), 597-607 Gurkaynak, G., Kama, O., & Ergün, B. (2018). Navigating the uncharted risks of covert advertising in influencer marketing. Business Law Review, 39(1), 17-19. Ha, L., & Hu, X. (2013). Social media involvement among college students and general population: Implications to media management. In Handbook of Social Media Management (pp. 751-773). Springer, Berlin, Heidelberg. Handriana, T. (2017). Consumer attitudes toward advertisement and brand, based on the number of endorsers and product involvement: An experimental study. Gadjah Mada International Journal of Business, 19(3), 289.

Page 75: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

66

Haynes, J. L., Pipkin, A. L., Black, W. C., & Cloud, R. M. (1994). Application of a choice sets model to assess patronage decision styles of high involvement consumers. Clothing and Textiles Research Journal, 12(3), 22-32. Hughes, C., Swaminathan, V., & Brooks, G. (2019). Driving brand engagement through online social influencers: An empirical investigation of sponsored blogging campaigns. Journal of Marketing, 83(5), 78-96. Jang, W. E., Ko, Y. J., Kim, H. Y., & Jeong, S. H. (2015). The effect of world ranking on the selection of athlete endorsers: The case of the PGA. International Journal of Sport Communication, 8(4), 411-430. Kemp, A., Randon McDougal, E., & Syrdal, H. (2019). The matchmaking activity: An experiential learning exercise on influencer marketing for the digital marketing classroom. Journal of Marketing Education, 41(2), 141-153 Kunkel, T., Walker, M., & Hodge, C. M. (2019). The influence of advertising appeals on consumer perceptions of athlete endorser brand image. European Sport Management Quarterly, 19(3), 373-395. Lafferty, B. A., Goldsmith, R. E., & Newell, S. J. (2002). The dual credibility model: The influence of corporate and endorser credibility on attitudes and purchase intentions. Journal of Marketing Theory and Practice, 10(3), 1-11. Lee, J. E., & Watkins, B. (2016). YouTube vloggers' influence on consumer luxury brand perceptions and intentions. Journal of Business Research, 69(12), 5753-5760.

Liu, S., Jiang, C., Lin, Z., Ding, Y., Duan, R., & Xu, Z. (2015). Identifying effective influencers based on trust for electronic word-of-mouth marketing: A domain-aware approach. Information sciences, 306(2), 34-52.

Lim, X.J., Radzol, A., Cheah, J.H. & Wong, M.W. (2017). The Impact of Social Media Influencers on Purchase Intention and the Mediation Effect of Customer Attitude. Asian Journal of Business Research, 7(2), 19-36.

Lou, C., & Yuan, S. (2019). Influencer marketing: How message value and credibility affect consumer trust of branded content on social media. Journal of Interactive Advertising, 19(1), 58-73. Lup, K., Trub, L., & Rosenthal, L. (2015). Instagram# instasad?: exploring associations among instagram use, depressive symptoms, negative social comparison, and strangers followed. Cyberpsychology, Behavior, and Social Networking, 18(5), 247-252.

Page 76: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

67

Maderer, D., Parganas, P., & Anagnostopoulos, C. (2018). Brand-image communication through social media: The case of European professional football clubs. International Journal of Sport Communication, 11(3), 319-338 McCracken, G. (1988). The long interview (Vol. 13). Sage. McCracken, G. (1989). Who is the celebrity endorser? Cultural foundations of the endorsement process. Journal of consumer research, 16(3), 310-321. Moorman, C., Deshpande, R., & Zaltman, G. (1993). Factors affecting trust in market research relationships. Journal of marketing, 57(1), 81-101. Munnukka, J., Uusitalo, O., & Toivonen, H. (2016). Credibility of a peer endorser and advertising effectiveness. Journal of Consumer Marketing. Ohanian, R. (1990). Construction and validation of a scale to measure celebrity endorsers' perceived expertise, trustworthiness, and attractiveness. Journal of advertising, 19(3), 39-52. Ohanian, R. (1991). The impact of celebrity spokespersons' perceived image on consumers' intention to purchase. Journal of advertising Research. Pallant, J. (2010). SPSS survival manual (4. ed.). McGraw-Hill Education (UK). Racherla, P., Mandviwalla, M., & Connolly, D. J. (2012). Factors affecting consumers' trust in online product reviews. Journal of Consumer Behaviour, 11(2), 94-104.

Radder, L., & Huang, W. (2008). High-involvement and low-involvement products: A comparison of brand awareness among students at a South African university. Journal of Fashion Marketing and Management: An International Journal, 12(2), 232-243.

Reinikainen, H., Munnukka, J., Maity, D., & Luoma-aho, V. (2020). ‘You really are a great big sister’–parasocial relationships, credibility, and the moderating role of audience comments in influencer marketing. Journal of Marketing Management, 1-20. Saunders, M., Lewis, P., & Thornhill, A. (2012). Research methods for business students (6. utg.). Harlow: Pearson.

Sigala, M. (2017). Advances in Social Media for Travel, Tourism and Hospitality: New Perspectives, Practice and Cases. California: Routledge.

Spears, N., & Singh, S. N. (2004). Measuring attitude toward the brand and purchase intentions. Journal of current issues & research in advertising, 26(2), 53-66.

Page 77: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

68

Statista. (2019a). Number of monthly active Instagram users from January 2013 to June 2018. Retrieved 2020 February 6 from https://www.statista.com/statistics/253577/number-of-monthly-active-instagram-users/ Statista. (2019b). Share of Instagram users in Sweden in 2019, by age group. Retrieved 2020 February 10 from https://www.statista.com/statistics/544434/sweden-instagram-users-by-age-group/ Statista. (2020). Number of registered golfers in Europe from 1990 to 2018. Retrieved 2020 Mars 25 from https://www.statista.com/statistics/275308/number-of-registered-golf-players-in-europe/ Sudha, M., & Sheena, K. (2017). Impact of Influencers in Consumer Decision Process: The Fashion Industry. Journal of Indian Management, 14(3), 14-30. Till, B., & Busler, M. (2000) “The match-up hypothesis: physical attractiveness, expertise, and the role of fit on brand attitude purchase intent and brand beliefs”, Journal of Advertising, 29(3), 1-13. Van-Tien Dao, W., Nhat Hanh Le, A., Ming-Sung Cheng, J., & Chao Chen, D. (2014). Social media advertising value: The case of transitional economies in Southeast Asia. International Journal of Advertising, 33(2), 271-294. Wheeler, R. T. (2009). Nonprofit advertising: Impact of celebrity connection, involvement and gender on source credibility and intention to volunteer time or donate money. Journal of Nonprofit & Public Sector Marketing, 21(1), 80-107. Wolny, J., & Mueller, C. (2013). Analysis of fashion consumers’ motives to engage in electronic word-of-mouth communication through social media platforms. Journal Of Marketing Management, 29(5-6), 562-583. Wu, T. Y., & Lin, C. A. (2017). Predicting the effects of eWOM and online brand messaging: Source trust, bandwagon effect and innovation adoption factors. Telematics and Informatics, 34(2), 470-480. Xiao, M., Wang, R., & Chan-Olmsted, S. (2018). Factors affecting YouTube influencer marketing credibility: a heuristic-systematic model. Journal of Media Business Studies, 15(3), 188-213.

Yin, Robert. (1984/1994). Case Study Research: Design and Methods. Thousand Oaks, London, New Delhi: Sage.

Page 78: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

69

Yuan, J., & Jang, S. (2008). The effects of quality and satisfaction on awareness and behavioral intentions: Exploring the role of a wine festival. Journal of Travel Research, 46(3), 279-288.

Page 79: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

70

Appendix A Interview guide based on the long interview Biographical questions: Examinee: Gender: Age: Ethnicity: Education Level: Examiner: Evaluation date: Place: Reasons for referral: Typical Grand-tour questions:

- Could you describe what your background in influencer marketing is? - Could you describe how a typical day at work look like? - Could you describe what product categories you are working with influencer

marketing in? - Could you describe how a typical process, from the company contact you to the

collaborations are finished looks alike? - Could you describe for me how the process usually looks like between an influencer

and a company? Specific grand tour questions:

- Could you describe how your latest collaboration looked alike? From the beginning when the company contacted you until the collaboration was done.

In the answers, mini-tour questions could be asked. Example questions

- Could you give me an example of what problems that could arise in a collaboration between an influencer and a company?

- Could you give me an example on important factors that needs to be considered when starting a collaboration between an influencer and a company?

- Could you give me an example on companies that contact you, what are their primary purpose with using influencer marketing?

- Could you give an example on what you should think about to find THE right person for the collaboration?

Experience questions – (best used after grand- and mini tour questions)

- Could you tell me about your own experience of influencer marketing? - Have you gotten the feeling that influencer marketing has changed during the years

and how do you think it will develop in the further? Floating promts – based on the key terms the respondent is discussing.

- What do you mean with xxx?

Planned promts – Discuss phenomena that do not come readily to mind or speech

Page 80: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

71

“Contrast” promt – what is the different between categories x and y? (should firstly be restricted to terms that the respondent has introduced. First after that could the investigator introduce terms from the literature. “Category” questions - These questions allow the investigator to account for all of the formal characteristics of the topic under discussion. The investigator will want to know how the respondent defines the event’s key actors, central action, dramatic structure, important props, necessary audience, ascribed roles, designated critics, social significance, cultural significance, and the consequences of good and bad performances.

- Is there any special product category you think influencer marketing is more suitable? - What would you say is important factors the influencer should contain to be able to

create an effective marketing campaign? - Would you say there is areas where influencer marketing is less effective? - What would you say is the outcome of influencer marketing you would like to

receive? - Are you familiar with the phenomena source credibility? - What should you say is the most important factor that needs to be considered to create

trust among the customers/followers receiving the sponsored post? - What would you say are the important factors to make consumers/followers do a

purchase after seeing the sponsored post? - How would you define source credibility? - What do you think about trustworthiness? - What do you think about expertise? - What do you think about similarity? - What do you think about usefulness?

Page 81: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

72

Appendix B One of the questionnaires that was sent out to the company’s customers. Hi! In this survey, you will be introduced to a collaboration between an influencer and a golf

company. After you have read about the influencer's background and looked at the picture, in

the next slide, you will get a series of short questions to answer. Please answer on a scale of

one (1) to five (5). Note: this is just an illustration of how a collaboration could look like.

Important to read this part. Anna has been playing golf since she was two years old and she

is the youngest pro player that has been winning the Amateur Golf Championship. In the year

2017, she won the Ladies European Tour in Australia. Since a couple of months back Anna has

been collaborating with a golf company and tried their products out. Now she is sharing her

experience of the product and a discount code on her Instagram. Please read Anna's text.

Page 82: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

73

1. I consider this post as reliable 1, 2,3,4,5 2. I consider this post as convincing 1, 2,3,4,5 3. I consider this post as credible 1, 2,3,4,5

4. I feel the endorser was honest 1, 2,3,4,5 5. I consider the endorser trustworthy 1, 2,3,4,5 6. I feel the endorser was earnest 1, 2,3,4,5

7. I feel the endorser knows a lot about the product 1, 2,3,4,5 8. I feel the endorser is competent to make assertions about the product 1,

2,3,4,5 9. I consider the endorser sufficiently experienced to make assertions about the

product 1, 2,3,4,5

10. The endorser and I have a lot in common 1, 2,3,4,5 11. The endorser and I are a lot alike 1, 2,3,4,5 12. I can easily identify with the endorser. 1, 2,3,4,5

13. I consider this post as valuable 1, 2,3,4,5 14. I consider this post as informative 1, 2,3,4,5 15. I consider this post to be helpful 1, 2,3,4,5

16. I consider the quality in this picture as high 1, 2,3,4,5 17. The quality in this post gives me a serious impression of the company 1,

2,3,4,5 18. Branded posts with high quality give a more serios impression 1, 2,3,4,5

19. I would listen to this person’s advice and follow this person on social media

1, 2,3,4,5 20. After seeing this post and knowing the influencers background, I considering

using this product 1, 2,3,4,5 21. Next time I will book a tee time, I will use this product 1, 2,3,4,5

22. Gender:

Female Male I choose not to disclose

23. Age: 18-24, 25-34, 35-44, 45-54, 55-64, 65+

Page 83: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

74

24. Country of residence: Sweden, Czech Republic, Slovakia, Spain, Italy, Portugal, France, Germany, Belgium, Denmark, Poland, UK, Lithuania, Estonia, Finland, Norway, Other

25. When do you usually play golf? During weekends at local club On holidays During business

26. How many golf rounds do you play each year? 1-4 5-10 11-20 21-30 30+

27. For have many years have you been interested/played golf? 0-5 years 6-10 years 11 – 15 years 16 – 20 years 21 – 25 years 25 +

28. What is your golf handicap? Please state

Page 84: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

75

Appendix C Frequency table for all variables.

Page 85: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

76

Appendix D Full tables from the ANCOVA analysis.

Page 86: The effect of Influencer Marketing in a highly involved ...1445862/FULLTEXT01.pdf · effect of influencer marketing in high involvement products is still partial. In the 21st century

77