The effect of micro-influencer advertising versus celebrity-influencer advertising on brand loyalty, and how trust mediates this relationship.
Bachelor’s Thesis Business Administration 2017/2018
Marjolein Duivenvoorden 10782125
Supervisor: Antoon Meulemans
Words: 9880
Statement of originality
This document is written by Marjolein Duivenvoorden who declares to take full
responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is
Abstract
Influencers promote products and brands on their personal social media accounts where they create valuable relationships with consumers. This enhances different features of brand loyalty such as satisfaction, brand trust, a positive attitude towards the brand and repeated purchase intentions. Two types of influencers are used for promoting products and brands, micro-influencers and celebrity-influencers. Non-academic resources suggested that there is a difference in the effectiveness of these different types of influencers. This difference seems to be caused by the different levels of trust that micro-influencers and celebrity-influencers generate. Because the lack of academic evidence, this study aims to test whether micro- influencers generate higher levels of trust than celebrity-influencers and therefore have a larger positive effect on brand loyalty. To investigate this, a questionnaire was distributed. A sample of 96 cases was collected. Data was analyzed by the simple mediation model of the PROCESS macro in SPSS developed by Hayes. Results indicated that micro-influencers do not directly have a larger positive effect on brand loyalty than celebrity-influencers, but that they do have an indirect larger positive effect on brand loyalty than celebrity-influencers, because they generate a higher level of trust.
Table of Contents 1. Introduction ... 5- 8 2. Literature review ... 9 - 16 2.1 Influencer advertising ... 9 - 14 2.1.1 Word-of-mouth ... 9 - 10 2.1.2 Influencers ... 10 - 11 2.1.1 Social media: Instagram ... 11 - 12 2.1.1 Micro-influencers ... 12 - 13 2.1.1 Celebrity-influencers... 13 - 14 2.2 Brand loyalty ... 14 - 15 2.3 Trust ... 15 - 16 3. Conceptual framework ... 17 - 21 3.1 Influencer advertising on brand loyalty ... 17 - 18 3.2 Influencer advertising on trust ... 19 - 20 3.3 Trust on brand loyalty ... 20 4. Methodology ... 22 - 28 4.1 Research design ... 22 - 23 4.2 Measurements ... 23- 25 4.3 Procedure ... 25 4.4 Sample ... 25 4.5 Analysis ... 26 – 28 5. Results ... 29 - 37 5.1 Reliabilities & Factor analysis ... 29 - 30 5.2 Means, standard deviations and correlations ... 31 - 33 5.3 Simple mediation analysis ... 33 - 37 6. Discussion ... 38 - 40 6.1 Summary of results ... 38 - 39 6.2 Limitations and recommendations for future research ... 39 - 40 7. Conclusion ... 41 - 42 References ... 43 - 48 Appendix 1. SPSS Output ... 49 - 57 Appendix 2. Questionnaire ... 58 - 69
1. Introduction
In 2016, Instagram introduced its newest feature: Instagram Business. Now, two years later, it provides over 25 million businesses with access to user analytics, contact information,
Instagram ads, and more. It gives businesses the opportunity to connect with over 800 million users, from which 80% follows at least one business profile (Osman, 2018). The need for Instagram Business came from the fact that businesses increasingly started using Instagram and other social media platforms as a marketing resource. Social media evolved from a place where people could connect with their friends and family, to a shopping channel, where consumers can gather information about products, get inspired and can connect with their favorite brands (Shankar, Mantrala, Kelley & Rizley, 2011). This means that in past years, it became important for businesses to learn how to use this rise in social media platforms as a marketing tool (Mangold & Faulds, 2009).
Instagram Business started using algorithms, which ensure that users get exposed less to ads. Algorithms made it more difficult for businesses to advertise their products directly to the right target groups, so businesses started looking for alternative advertising methods (Carbone, 2018). Influencers seemed to be an alternative way of advertising in which businesses could still target their customers and reach the right audience.
Targeting the right audience seems to be an important factor in creating brand loyalty, as consumers prefer to be exposed to relevant content on social media platforms (Erdoğmuş, & Cicek, 2012). There are multiple predictors of brand loyalty. One important and closely related predictor of brand loyalty is repeated purchase behavior, or purchase intention. When consumers have a high purchase intentions, this can be seen as a sign of brand loyalty (Jacoby & Kyner, 1973). Another predictor of brand loyalty is a positive attitude towards the brand (Oliver, 1999). Finally, brand satisfaction and brand trust are also important predictors that
be very useful in increasing predictors like purchase intentions, positive brand attitudes, brand satisfaction and brand trust (Cruz, 2018).
Influencers promote or endorse products and brands on social media in exchange for money (De Veirman, Cauberghe & Hudders, 2017). Influencer advertising has proven to be a less obtrusive way of targeting an audience than traditional advertising ways (Boerman, Willemsen & Van der Aa, 2017). Influencers are individuals who have social accounts with a sizable amount of followers, usually with mutual interests and preferences. On these accounts, influencers post about products or brands in exchange for money. Influencers are perceived as highly trustworthy and are therefore effective in promoting products or brands (De Veirman et al., 2017).
Two types of influencers can be distinguished: Micro-influencers and celebrity- influencers (Cruz, 2018). Micro-influencers are known from their social media accounts where they have built their personal brand. This creates a niche audience (De Veirman et al., 2017). Celebrity-influencers are individuals who enjoy personal recognition, often due to their profession, and who use this recognition to endorse brands through advertisements
(McCracken, 1989). Apart from the nature of micro-influencers and celebrity-influencers, the difference between them is that micro-influencers tend to build long-term relationships with consumers which are built on trust , and which create consumer engagement. Celebrity- influencers, on the other hand, are more likely to enhance short-term brand awareness (Cruz, 2018).
The perceived differences between micro-influencers and celebrity-influencers have led to some discussion in literature. McGrath (2018) states that micro-influencers are more effective in creating consumer engagement than celebrity-influencers. They can create high quality leads. Another aspect that makes micro-influencers more effective than celebrity influencers is that they can help businesses with creating brand trust. Creating brand trust can
be challenging for many businesses. Because micro-influencers have already built a relationship with consumers which is partly based on trust, consumers will also trust the products and brands these influencers promote.
Another point of discussion in literature on influencer advertising and brand loyalty is trust. Wottrich, Verlegh & Smit (2017) explain that trust is an important factor in creating strong relationships and thus brand loyalty. When applying this knowledge to advertising, it is important that consumers trust the advertising resource. For influencers, it is important to be perceived as trustworthy and credible. When consumers perceive influencers as trustworthy and credible, purchase intentions for the promoted brand seem to be higher (Chu & Kamal, 2008). Forrest & Cao (2010) state that micro-influencers are often perceived to be more trustworthy and credible than celebrity-influencers. Micro-influencers seem to be more
trustworthy and credible because they promote products and brands they seem to be genuinely passionate about. Also, micro-influencers are more identifiable because they are ‘normal’ people with normal lives instead of celebrity-influencers who are mostly famous from certain professions that regular people cannot imagine for themselves (Jang & Stefanone, 2009).
Although there is some research about the trust levels of micro-influencers and celebrity-influencers, little to none academic research has been done about the effect of micro-influencers and celebrity-influencers on brand loyalty. Therefore, this study aims to investigate whether there is a difference in the effect of micro-influencers and celebrity- influencers on brand loyalty, and whether trust mediates this relationship. Therefore:
RQ: Does micro-influencer advertising on Instagram have a larger positive effect on brand loyalty than celebrity-influencer advertising, and is this relationship mediated by trust?
To answer this question, first a theoretical framework will be given in which the independent variable, the type of influencer, will be described. Secondly, the dependent variable, brand loyalty, will be described and finally the mediating variable, trust, will be described. The conceptual framework section will explain the predicted relationships between the variables and will interpret these predictions by analyzing literature. In the methodology section the research design, measurements, procedure, measurements and analyses will be described. Then the results will be presented and interpreted. Finally, a discussion of the results, limitations and recommendations will be given. Some concluding thoughts will be given at the end of this thesis.
2. Literature review
2.1 Influencer advertising
In this section the concept of influencer advertising will be discussed. Influencer advertising is the independent variable in this study. The first aspect of influencer advertising that will be discussed is word-of-mouth, then a definition of the influencer will be given and influencer advertising will be explained. Then the topic of Instagram will be discussed. Finally, the two sub-categories of influencers, micro-influencers and celebrity-influencers, will be explained.
2.1.1 Word-of-mouth
Although influencer advertising sounds like a modern phenomenon, people have been
practicing it for as long as the human kind exists. For example, the oral stories and fables that people passed on to each other from the Middle Ages on. They were meant to teach citizens about believes. These stories were presented by important people in the community. The citizens looked up to these people, and therefore adopted the ideas (Weiss, 2014). Weiss calls this phenomena word-of-mouth, or WOM. WOM is an interesting topic to understand for this study, because it lies at the heart of influencer advertising. Anderson (1998) states that word- of-mouth is an informal mode of communication about the evaluation of goods and services between consumers who are independent of the marketers. WOM is likely to be more trustworthy and relevant compared to traditional marketing tools, such as television
commercials or newspapers. It can create empathy and reduce consumer resistance (Bickart & Schindler, 2001).
Word-of-mouth evolved through the years. As technology developed, eWOM came up. Nowadays, technology is one of the most influential marketing instruments. Through technologic devices like mobiles and computers, consumer can share their experiences,
Consumers use all kinds of platforms to do this, like blogs, review websites, shopping websites and social media platforms (Cheung & Thadani, 2012). As social media platforms came up, several parties tried to make a job out of it. One of the parties who succeeded in this are influencers.
2.1.2 Influencers
In the previous section, WOM was describes as lying at the heart of influencer advertising. Another topic that has close links to the concept of influencer advertising is opinion
leadership. Rogers & Cartano (1962) describe opinion leadership as “individuals who exert an unequal amount of influence on the decision of others”. This means they impact others’ decisions more than other people do, and that they have an dominant influence on others’ decisions. Hayes (2008) describes influencers as “a third-party who significantly shapes the customer’s purchasing decision, but may never be accountable for it”. As you can see, the meanings of these two definition are similar. According to McQuarrie, Miller & Phillips (2012), there are some differences between influencers and opinions leaders. He states that influencers reach a mass audience of strangers on whom they have an impact. Opinion leaders, on the other hand, is more of an ongoing communication between consumers. An opinion leader for example can just be the most dominant person in a group of friends. An influencer is always more than that and speaks to a broader audience. But why do these people who we call influencers have the power to actually influence other people’s opinions and preferences?
The nature of influencer advertising is simple. People, or in the case of this study, consumers, like to compare themselves to others. They evaluate themselves on specific traits by observing the possessions of others (Shalev & Morwitz, 2012) . Influencer advertising is regarded as the same as influencer endorsement in this study. In both concepts, products or
brands are promoted and advertised by people who have an influence on consumers (Silvera & Austad, 2004). These concepts can be illustrated by an example about women who compare themselves to pictures on social media. In a study of Chatard, Bocage-Barthélémy, Selimbegović & Guimond (2017) pictures of thin women were showed to respondents which were afterwards asked to evaluate themselves. The respondents seemed to be affected by these pictures and evaluated themselves lower than before. This study also showed that this comparison takes place outside awareness. According to the self-comparison theory, people compare themselves to others to find a relative standard. This especially goes for possessions. Consumers want to live up to these standards (Chatard et al., 2017). Influencer advertising, or influencer endorsement as Silvera & Austad (2004) call it, causes consumers to want to live up to the same standards as the influencers set. The effectiveness of influencers changes between different types of influencers and is mostly caused by the attitudes consumers have towards them (Silvera & Austad, 2004). Influencers operate in different environments. One of their most important sources of work lies within social media, especially Instagram.
2.1.3 Social media: Instagram
As Osman (2018) says, Instagram is a powerhouse against social media rivals. Here are some facts about Instagram: The social media platform, established in 2010, has more than 800 million active users monthly. That is two times the active users of Twitter and three times the active users of Facebook and Whatsapp. A big source of advertising on Instagram comes from hashtags. They are often used by influencers and can link to brand profiles. Seven out of ten of these hashtags are branded. Examples of branded hashtags are #nikeairmax and
#hunkemollerambassadors. In 2017, approximately 71% of all US businesses used Instagram. A big rise in business users took place in 2017 because of the release of Instagram Business Profiles, which gives businesses access to powerful analytics data, contact information and
new ad opportunities. Having a business profile seems to be highly useful due to the fact that 80% of all users follow at least one business profile and at least 30% of these users bought a product they first discovered on Instagram. As research suggests, 80% of all influencers prefer Instagram as a source of brand collaboration (Osman, 2018).
The statistics prove that Instagram is a powerful and accessible tool. Influencers and business like to use it as a marketing tool (Chaffey, 2017). Boerman et al. (2017) describes the benefits of influencer advertising on Instagram. According to them, it provides businesses with a less obtrusive way to reach a target audience than traditional marketing tools. This has to do with tags where consumers search for. When an influencer ‘tags’ a brand in their post, it will pop up when an consumer searches for this name. This also works the other way around. When a consumer follows an influencers and this influencer posts a tagged picture of a product , the consumer can easily click on the tag if they like the product. By clicking on the tag they will directly be linked to the brand profile and sometimes even the brand website. On Instagram, two types of influencers can be distinguished: Micro-influencers and celebrity- influencers.
2.1.4 Micro-influencers
Due to the rise of social media, micro-influencers went from being ‘’nobodies’’, to
‘’somebodies’’ (Booth & Matic, 2011). Booth & Matic (2011) tried to define and categorize the new, upcoming phenomenon, micro-influencers - bloggers, as they call them -. They proposed three categories. The A tier, which are typically micro-influencers with a large reach. They share general, news related topics and are normally less social than B and C tier influencers. B tier influencers have a smaller, but still a sizable reach. They share posts about particular topics and often get monetary rewards for these posts. C tier influencers have a
small reach. They have share specific content and are passionate about the contents they share.
This study will focus somewhere in the middle of B and C tier influencers. It will focus on micro-influencers who get a monetary reward for the products and brands they promote, but share specific contents and seem to be genuinely passionate about these contents.
The biggest benefit of working with micro-influencers is that they established a personal image. Their blogs each have an unique identity. This is convenient for brands who like to reach a specific target group because they can choose the influencer that fits and reaches this target group the best (Dinesh, 2017). Think about a brand that wants to promote their product to young, sporty women. They will look for an influencer who has this image herself. Williams (2017) states that micro-influencers bring relevance to consumers and that they are a forward-looking marketing tool. He describes that the consumers engaged with the micro-influencer are engaged and passionate about the content the influencer shares. Taking all literature together, micro-influencers can be described as small opinion leaders on social media who build long lasting relationships with consumers based on loyalty and trust. This leads to great consumer engagement, something that businesses can benefit from in promoting their products through influencer advertising.
2.1.5 Celebrity-influencers
The other type of influencers are celebrity-influencers, sometimes also described as celebrity endorsers. In this study we will use the term celebrity-influencers, but for the theory section the term celebrity endorsement will also be used because this term is widely used in previous literature. McCracken (1989) defines celebrity endorsement as: ‘’Any individual who enjoys public recognition and who uses this recognition on behalf of a consumer good by appearing
with it in an advertisement.’’. Um & Kim (2016) go a little further and describes celebrity endorsers as individuals with a powerful position, often due to their career, like sports, television or politics. This description illustrates how this study distinguishes celebrity- influencers from micro-influencers: they have a powerful position from a different source than only their personal blog. Hovland & Weiss (1951) proposes a model that states that the effectiveness of a celebrity depends on their expertness and trustworthiness. Most celebrities possess these characterizes and are therefore credible and persuasive.
2.2 Brand loyalty
In the following section, the dependent variable in this study, brand loyalty will be discussed. A definition will be given and predictors will be explained.
Brand loyalty is one of the main factors for businesses to gain competitive advantage , so it is important to understand the underlying antecedents (Dick & Basu, 1994). According to Jacoby & Kyner (1973) brand loyalty consists of six necessary conditions. They define: ‘’Brand loyalty is a biased or non-random behavioral response that is made over time by any decision-making unit with respect to one or more alternative brands out of a set of such brands and is a function of a psychological process.’’. They also suggest that brand loyalty and repeated purchasing behavior are closely related. Previous research about brand loyalty suggests that the relationship between a consumer and a brand relies strongly on brand-self connection on significant life-themes. A loyal brand relationship can be based on several things that consumers find important. The themes that consumers find brand-self connection with can differ per person. For example, this can be love, a personal revelation and the feeling that the brand adds something to the consumer’s life. It can build consumers’ self- esteem and give them a sense of belonging or independence (Fournier & Yao, 1997).
Brand loyalty has two important aspects. The behavioral aspect and the attitude aspect. The behavioral aspect focuses on repeated purchase while the attitude aspect focuses on a deep pledge that consumers make to their preferred services and products (Oliver, 1999). Brand satisfaction and brand trust have been identified as the footstones for creating brand loyalty. Brand satisfaction and brand trust have both shown positive linear relationships with brand satisfaction (Anderson & Sullivan, 1993). Customer satisfaction is closely related to needs. When these needs are satisfied, customer satisfaction is reached. Customers make judgements about products or services. When brands have high levels of customer
satisfaction, this leads to higher profitability and market share (Fornell, 1992). The second footstone is brand trust, which will be explained in the next section. Veloutsou (2015) adds one more predictor of brand loyalty, positive attitudes towards the product or brand.
Brand loyalty has been interpreted in many different ways. In this study, brand trust, satisfaction, positive attitudes and purchase intentions will be used as predictors for brand loyalty.
2.3 Trust
In the following section, the mediating variable in this study, trust, will be discussed. There are many different types of trust and many interpretations of trust. This study will focus on trust in the influencer, so to understand this topic, a general insight in de concept of trust will be given. As micro-influencers can also be seen as their own personal brand (Dinesh, 2017), the concept of brand trust will also be explained in this section.
Chaudhuri & Holbrook (2001) define trust as: ‘’A feeling of security based on the belief that his/her behavior is guided and motivated by favorable and positive intentions towards the welfare and interests of his/her partner.’’. Trust is delicate and subjective, as it is based on consumers’ beliefs rather than on hard facts (Yannopoulou, Koronis & Elliot, 2011).
Trust develops over time and comes from interactions and previous experiences. In marketing, trust is an important concept because it can enhance relationships. These
relationships can be found between consumers and brands, companies, and in the case of this study, influencers. Because of the trust consumers have in them, it is easy to advertise and to persuade them to buy. Consumers believe that they act in their best interest (Wottrich et al., 2017).
Many predictors of trust have been explored and tested in previous research. For this study, it is assumed that there are four predictors. Fist, reliability seems to be an important predictor of trust. When people feel like they can rely on a person or thing, they are more likely to trust it (Pirzada & McDonald, 2004). Another predictor of trust is expertise. The perceived expertise seems to be important for trust because people with more expertise are perceived as having more knowledge and being more selective in the information they spread (Eiser, Stafford, Henneberry & Catney, 2009). In a study of Moorman, Deshpandé &
Zaltman (1993) expertise is also tested as a predictor for trust, together with sincerity. Both expertise and sincerity have proven to be strong predictors of trust. Finally, research proved that identifiability is a strong predictor for trust. People find other people more trustworthy when they feel like they are similar to them, and they can identify with them (Tanis & Postmes, 2005).
Taking all literature together, trust is a complex concept with many explored predictors. In this study, reliability, expertise, sincerity and identifiability will be used as predictors for trust.
3. Conceptual framework
In the following section the conceptual model of this study will be presented and explained. First, the expected relationship between influencer advertising and brand loyalty will be explained and hypothesis 1 will be given, then the expected relationship between influencer advertising and trust will be explained and hypothesis 2 will be given and finally the expected relationship between trust and brand loyalty will be explained and hypothesis 3 will be given. A visual representation of the conceptual model will also be given.
3.1 Micro-influencer advertising versus celebrity-influencer advertising on brand loyalty
Many studies about brand loyalty and advertising have been done, but little academic research is done about the effects of social media advertising on brand loyalty, especially the effect of influencer advertising on brand loyalty. Laroche, Habibi & Richard (2012) found that social media influence strengthens the relationships between customers and brands, as well as customer-to-customer relationships. They explain that these relationships lead to higher brand trust, which is one of the footstones of brand loyalty. Because of the positive linear
relationship between brand trust and brand loyalty, an increase in brand trust trough social media influence will also lead to an increase in brand loyalty.
Chatard et al. (2017) look at the comparison theory to explain the positive relationship between influencer advertising and brand attachment, which is an aspect of the behavioral aspect of brand loyalty. He explains that consumers like to have a relative standard. Through self-evaluation and comparison consumers want to live up to this standard and want to have the same possessions as the people who they believe are the standard. Influencers present themselves as these standards and promote brands and products. When looking at the
comparison theory, it can be assumed that consumers like to live up to these influencers and like to have the products that are promoted.
When looking at the effect of influencer advertising on brand loyalty, it seems from academic research that there is a positive relationship. What is lacking is academic research about the effect of different types of influencers on brand loyalty. Nevertheless, there are some non-academic articles on this topic. According to McGrath (2018), micro-influencers are more compatible for enhancing consumer engagement than celebrity-influencers.
Enhanced consumer engagement can lead to higher brand loyalty. Cruz (2018) explains that micro-influencers build on long lasting relationships that enhance brand engagement and therefore trust, and that celebrity-influencers rather enhance brand awareness on the short term. Research has shown that one of the two aspects of brand loyalty is the behavioral aspect. One of the antecedents of the behavioral aspect of brand loyalty is the duration of the consumer-brand relationship (Oliver, 1999). When looking at this, it can be predicted that micro-influencers have a larger impact on brand loyalty than celebrity-influencers, due to the duration and the engagement that comes from the relationships micro-influencers build with consumers.
Also, previous research proved that when micro-influencers - bloggers, as they called them – shared promoted content like blogs or pictures, the attitudes from consumers towards the promoted content were highly positive. Because of these positive attitudes, both brand trust and purchase intentions also increased (Forrest & Cao, 2010). Positive attitude, brand trust and purchase intentions are all predictors of brand loyalty. Therefore:
Hypothesis 1: Micro-influencer advertising has a larger positive effect on brand loyalty than celebrity-influencer advertising.
3.2 Micro-influencer advertising vs. celebrity-influencer advertising on trust
Wottrich et al. (2017) explain that trust is an important driver of strong relationships. It has proven to be a powerful tool in marketing and brands can gain an advantage by building a trustworthy relationship with their customers. When looking at influencer marketing, one of the most important features of an effective influencer are credibility and trustworthiness. When an influencer is credible and trustworthy, consumers’ purchase intentions for the promoted brand are usually higher (Chu & Kamal, 2008). According to Everard & Galletta (2006), the importance of credibility and trustworthiness is especially important when
consumers do not get the physical experience of seeing and trying the product when they buy online. When the expertise of an influencer is higher, credibility is also likely to be higher (Silvera & Austad, 2004). The expertise between micro-influencers and celebrity-influencers is sometimes perceived different by consumers.
Forrest & Cao (2010) state that micro-influencers - bloggers, as they call them - are often perceived to be more credible and trustworthy. This is due to the fact that they created their own personal brand and image, unlike a celebrity who is just known for the thing they are famous for (Chu & Kamal, 2008). Micro-influencers promote brands and products they seem to actually have passion for. They only promote the things that fit their personal brand and style. They also seem to have more expertise because they are specialized in the products they promote, unlike celebrities who usually just promote brands and product because of the monetary reward they get (Forrest & Cao, 2010).
Micro-influencers reveal intimate information about themselves on their blogs. They show their personal traits and preferences. This makes them familiar and gives consumers the feeling that they actually know them. It makes micro-influencers identifiable, which gives consumers a feeling of positivity and trust (Jang & Stefanone, 2009). Micro-influencers seem to create stronger and long-term relationships with consumers, whereas celebrity-influencers
more often create short-term brand awareness (Cruz, 2018). According to Oliver (1999), a strong long-term relationship creates brand attachment and trust. From previous literature, micro-influencers seem to be perceived as reliable, sincere, identifiable, and have expertise. Those are all predictors of trust. Therefore:
Hypothesis 2: Micro-influencers generate higher trust levels than celebrity-influencers.
3.3 Trust on brand loyalty
Morgan & Hunt (1994) state that trust is one of the most important predictors of brand
commitment and brand loyalty. It builds a personal connection between consumers and brands (Hess & Story, 2005). Gilbert & Veloutsou (2006) support that trust is a key predictor for brand loyalty and add that results suggest that brand relationship acts as a mediating variable in the link between trust, satisfaction and loyalty. Brand trust is also studied as a mediating variable in loyalty studies. Brand trust has proven to act as a strong mediator between sensory experience and brand loyalty (Huang, 2017). Trust also acts as a mediator between risk aversion and brand loyalty (Matzler, Grabner-Kräuter & Bidmon, 2008).
Forrest & Cao (2010) state that micro-influencers are more trustworthy than celebrity- influencers. This seems to be an important factor for the effectiveness of the influencers because trustworthiness leads to stronger relations that tend to last longer. This enhances brand loyalty. Therefore:
H3 Trust Brand loyalty 3.4 A visual representation: H1 H2 Micro-influencer advertising vs. Celebrity-influencer advertising
4. Methodology
In the following section, the methodology of this study will be presented. First, the overall research design will be explained, then the measurements and constructs of the variables will be explained. The procedure and sample will explain how the sample is collected and what the sample consists of. Finally, the analysis will be explained.
4.1 Research design
In this study, the effect of micro-influencers versus celebrity-influencers on brand loyalty, mediated by trust was measured. To investigate this, a mono-method quantitative approach was used. This means that a single data collection technique was used (Saunders, Lewis & Thornhill, 2012, pp. 166-169). Data was collected through an online questionnaire among Dutch consumers who are familiar with Instagram. This method was chosen because
questionnaires are an easy way for collecting a large amount of respondents in a cheap way. It is easy to measure and compare variables from data collected in a questionnaire . A drawback of using questionnaires is that preparing and analyzing data can be difficult and time
consuming (Saunders et al., 2012, pp. 181-182). The questionnaire was distributed through Facebook and WhatsApp Messenger. This seemed to be the right platform because all
respondents met the criteria of being familiar with Instagram. The questionnaire can be found in Appendix 2.
The questionnaire consisted of a manipulation, which was performed through an experiment. According to Saunders et al. (2012, pp. 178-181) experiments are a good way to investigate the causal relationship between variables. To investigate trust and brand loyalty, two conditions were created. In the first condition, two examples of micro-influencers
advertising on Instagram were presented. To make the distinguish more clear for respondents, micro-influencers were called bloggers. The first micro-influencer promoted a watch from
Daniel Wellington. The second micro-influencer promoted L’Oréal hair. In the second
condition, two examples of celebrity-influencers advertising on Instagram were presented. For convenience, they were called celebrities. Similar to the first condition, the first celebrity- influencer promoted a watch from Daniel Wellington and the second celebrity-influencer promoted L’Oréal hair. The examples in which Daniel Wellington was promoted were similar posts, with similar persons and a similar way in which the product was presented in the picture. The same goes for the examples in which L’Oréal hair was promoted.
Through randomization, respondents were showed either the first condition or the second condition, meaning that respondents got to see either the micro-influencer examples or the celebrity-influencer examples.
4.2 Measurements
The dependent variable brand loyalty and the mediating variable trust were both measured through the answers to the experiment. For each of the four examples, from which
respondents got to see either the two celebrity-influencer examples or the two micro-
influencer examples, similar questions about brand loyalty and trust were asked. All questions were based on a five-point Likert scale, ranging from ‘’Strongly agree’’ to ‘’ Strongly
disagree’’. Using a Likert scale is appropriate for measuring variables and gives the opportunity to test reliability with Cronbach’s Alpha (Matell & Jacoby, 1971) .
The construct of the dependent variable brand loyalty originally consisted of four predictors, brand trust, purchase intention, satisfaction and positive attitude towards the brand. For each of the four examples, two questions about purchase intention were asked and one question about brand trust, satisfaction and positive attitude were asked. Satisfaction was later removed because it seemed from a pattern matrix that this question did not test brand loyalty. This will further be explained in the analysis section.
The construct of the mediating variable trust consists of four predictors, reliability, sincerity, expertise and identifiability. For each of the four examples, one question about these constructs was asked. All of these predictors seemed to test trust, according to a pattern
matrix. This will further be explained in the analysis section.
The independent variable micro-influencer versus celebrity-influencer was tested through an experiment. Both of these categories represented two examples which were randomly showed to the respondents. To assign the right type of influencer to the right answers, two dummy variables were created. 0 represented micro-influencer, 1 represented celebrity-influencer. For all the respondents that were assigned the micro-influencer
examples, answers were labelled 0 and for all the respondents that were assigned the celebrity-influencer examples, answers were labelled 1.
Some general questions about micro-influencers, celebrity-influencers, brand loyalty and trust were asked previous to the experiment. They were originally meant to test the predicting variables, but seemed to be more adequate as control variable. For this reason, this study contains a sizeable amount of control variables. The first control variable,
‘’InfluenceLevel’’, is a summary of answers of answers about to what extent respondents get influenced by promotion on Instagram in general. The second control variable,
‘’LoyaltyBlog’’, is a summary of answers about whether respondents believe their brand loyalty increases more after seeing micro-influencers – or bloggers – promote on Instagram than after seeing celebrity-influencers promote on Instagram. The third control variable, ‘’TrustBlog’’, is a summary of answers about whether respondents believe their trust in the influencer increases more after seeing micro-influencers – or bloggers – promote on
Instagram than after seeing celebrity-influencers promote on Instagram. The fourth and last control variable, ‘’Loy_Control’’, is a summary of answers about all the predictors of trust and their impact on all the predictors of brand loyalty and how respondents feel about this. All
of the control variables were measured through a five-point Likert scale, ranging from ‘’Strongly disagree’’ to ‘’Strongly agree’’.
4.3 Procedure
To perform this study, a questionnaire was developed with the website www.qualtrics.com . Qualtrics has made it easy to create and distribute the questionnaire, and answers did not have to be recorded manually, instead they were directly transferred to SPSS (Zikmund, Babin, Carr & Griffin, 2013, p. 206) The questions were written in Dutch, because the way of
distributing the questionnaire was likely to reach mostly Dutch respondents. The questionnaire was distributed through Facebook and WhatsApp Messenger, using the personal network of the author.
4.4 Sample
First, the demographics and some general information were analyzed. All the original tables can be found Appendix 1. From the respondents, 41 are women and 7 are men. 3 respondents are in the age group <20, 44 is are the age group 20-30 and 1 is in the age group 30-40.2 of the respondents enjoyed VWO as their highest level of education, 2 did MBO, 2 did HBO and 38 did WO.
All of the respondents were familiar with Instagram. 45 of them use Instagram on a daily base. 2 use Instagram weekly, and 1 never uses Instagram. 44 of the respondents follows at least one influencer who promotes products or brands on Instagram. 27 respondents have bought at least one product or brand after seeing it being promoted on Instagram.
4.5 Analysis
The outcomes from the questionnaire were automatically imported into SPSS. This program was used to transform and analyze the data.
First, reliabilities of all measurements were calculated. A useful tool to measure the internal consistence of a test or scale is Cronbach’s Alpha . Cronbach’s Alpha tests whether all the predictors really predict the construct and whether their inter-relatedness is correct (Tavakol & Dennick, 2011). To further examine all measurements, a factor analysis was performed. Factor analysis tests the construct validity of a scale, and is mostly used to
decrease variables or measurements (Thompson, 2004, pp. 2-5). The factor analysis provided a pattern matrix which indicated that satisfaction did not seem to fit the construct of brand loyalty, meaning that the questions of which satisfaction consisted were not adequate for measuring brand loyalty. For this reason, satisfaction was removed as predictor for brand loyalty.
Apart from satisfaction, all reliabilities and the factor analyses were appropriate for transforming the variables. First, the mean from the answers about brand loyalty about the micro-influencer examples and the celebrity-influencer examples was taken to create the dependent variable ‘’Loyalty’’. Then the mean from the answers about trust about the micro- influencer examples and the celebrity-influencer examples was taken to create the mediating variable ‘’Trust’’. The examples were given dummies where all micro-influencer examples were 0 and all the celebrity-influencer examples were 1 to create the independent variable. Then the mean from the answers about the general level of influence was taken to create control variable ‘’InfluenceLevel’’. Then the mean from the answers about the brand loyalty from micro-influencers advertising versus the brand loyalty from celebrity-influencer
advertising was taken to create control variable ‘’LoyaltyBlog’’. Also, the mean from the answers about trust in micro-influencers versus trust in celebrity-influencers was taken to
create control variable ‘’TrustBlog’’. Finally, the mean from the answers about the predictors of brand loyalty was taken to create control variable ‘’Loy_Control’’.
After al variables were transformed, data was restructured. Restructuring is allowed when each case is used to create a new case (Thompson, 2004, pp. 103-105). In the case of this study, the amount of data was limited, so restructuring was useful to gather twice as much data and therefore being able to do an analysis that has a better reliability. The answers from one respondent to the two examples in each condition - micro-influencer or celebrity- influencer - were restructured, as if there were two separate examples, assigned to two separate respondents. To validate the restructuring, each case was doubled. This means that for each respondent case, there were now two respondent cases and thus 96 respondents. The independent variable remained the same. 0 means micro-influencer, or ‘’blog’’, as it is called in SPSS. 1 means celebrity-influencer, or ‘’celeb’’, as it is called in SPSS.
After restructuring, frequencies have been analyzed. This contained the demographics and other nominal variables, like whether respondents are familiar with Instagram. Then the descriptives were analyzed to explore what the means and standard deviations of the
independent, dependent, mediating and control variables are.
A correlation matrix was made to determine if there are correlation effects between all the independent, dependent, mediating and control variables. Then the final regression could be made. For this regression, the simple mediation model in the PROCESS macro developed by Hayes (2018, pp. 78-83) was used. This model is developed by Hayes and exists of an independent variable (X), a dependent variable (Y) and a mediating variable (M). To analyze the relationships between the variables, firstly, three models were tested.
a-path Trust b-path
M
Y
X
To clarify this test, a visual representation: Type of
Influencer Brand loyalty
c’-path
The first model tested the a path. It tested the difference in the effect of micro-influencers and celebrity-influencers on trust. Thus, the first model tested Hypothesis 2.The second model tested the b path, which tested the effect of trust in the influencer on brand loyalty and
controls for the difference in the effect of micro-influencers and celebrity-influencers on trust. The second model tested Hypothesis 3. The third model tested the c’ path, which tested the difference in the total effect of micro-influencers and celebrity-influencers on brand loyalty. The last part of the analysis consisted of a summary of the direct, indirect and total effects. The direct effect represented the c’ path, so the effect of micro-influencers versus celebrity- influencers on brand loyalty, leaving trust out of notice. The indirect effect, or the mediating effect, was represented by the c – c’ paths. Finally, the total effect was represented by the c path, which means the total effect of micro-influencers versus celebrity-influencers on brand loyalty, taking trust into notice. The mediating, direct and indirect effects were measured by the following linear models respectively (Hayes, 2018, pp. 78-83):
1) M = i + a X + e M 2) Y = i + c′X +bM +e Y
5. Results
In the following section, results of the analyses will be discussed. First, the descriptives about demographics and general questions will be given. Then, the Cronbach’s Alpha’s will be given to prove reliability. Also, the factor analysis will be given. Then the descriptives about the variables will be given, including the means, standard deviations and correlations. Finally the regression analysis will be explained.
5.1 Reliabilities & Factor analysis
To analyze the data, multiple questions have been put together and have been transformed to scales and then to variables. To ensure that all items in the variable measure the right
construct, reliability was tested with Cronbach’s Alpha. Afterwards a factor analysis was performed to ensure the validity of the constructs. Both the Cronbach’s Alpha’s and the pattern matrix from the factor analysis are in Appendix 1.
Table 1. Cronbach’s Alpha’s
Cronbach’s Alpha’s Number of items
1. Loyalty 0.850 4 2. Trust 0.849 4 4. InfluenceLevel 0.744 7 5. LoyaltyBlog 0.869 4 6. TrustBlog 0.880 5 7. Loy_Control 0.835 13
The dependent variable, Loyalty, exists of Loy1, Loy2, Loy3 and Loy5. It’s Cronbach’s Alpha is 0.850. The mediating variable, Trust, exists of Trust1, Trust2, Trust3 and Trust4. The Cronbach’s Alpha for Trust is 0.849. For the control variables InfluenceLevel, LoyaltyBlog,
TrustBlog, Loy_Cotrol, Cronbach’s Alpha’s are respectively 0.744, 0.869, 0.880 and 0.835. These variables consist of multiple general questions about respondents’ attitudes towards being influenced by micro-influencers and celebrity-influencers, loyalty and trust. All of the Alpha’s are above 0.7, which indicates that all variables are reliable.
Table 2. Pattern Matrix
Component 1 Component 2 Loy2 0.932 Loy3 0.785 Loy1 0.799 Loy5 0.759 Trust4 -0.904 Trust2 -0.836 Trust3 -0.819 Trust1 -0.690
Extraction method: Principal Component Extraction Method: Principal Component Analysis
Rotation Method: Oblimin with Kaiser Normalization a. Rotation converged in 8 iterations
To get further insight into the validity of the scales, a factor analysis has been done. As showed in Table 2 all scales fall into the right component. Component 1 represents Loyalty and Component 2 represents Trust. The scores presented are after the scale Loy4 was
removed. This scale existed of questions about satisfaction. The score for this scale was 0.506 in component 2 and 0.424 in component 1. This means Loy4 had a better fit with Trust, which it was not meant to measure. Therefore, the scale Loy4 was removed from the data. The old pattern matrix is presented in Appendix 1.
5.2 Means, Standard deviations & Correlations
Table 3 presents the means and standard deviations for loyalty and trust under the micro- influencer condition (0) and the celebrity-influencer condition (1). Results indicate that brand loyalty is higher for micro-influencer (M = 3.076, SD = 0.941) than for celebrity influencers (M = 2.730, SD = 0.789). Micro-influencers seem to create higher levels of trust (M = 3.038, SD = 0.925) than celebrity-influencers (M = 2.286, SD = 0.720).
Table 3. Means and standard deviations
M SD
1. Loyalty – Blog 0 3.076 0.941
2. Loyalty – Celeb 1 2.730 0.789
4. Trust – Blog 0 3.038 0.925
5. Trust – Celeb 2 2.285 0.720
Table 4 presents the means, standard deviations, correlations and Cronbach’s Apha’s of the independent, dependent, mediating and control variables. For loyalty and trust, the means of the micro-influencers condition (0) (M = 2.896, SD = 0.878) and the celebrity-influencer condition (1) (M = 2.646, SD = 0.903) are taken together. TypeInf indicates the difference between the conditions micro-influencer and celebrity-influencer. The mean indicates how many respondents got to see the first condition or the second condition. Respondents got to see the celebrity-influencer condition slightly more often (M = 0.520, SD = 0.502). All the control variables present high levels. InfluenceLevel indicates that respondents are positively influenced by the promotion of products and brands on Instagram in general (M = 3.190, SD = 0.585). LoyaltyBlog indicates that respondents have larger positive levels of brand loyalty when products or brands are promoted by micro-influencers than when they are promoted by celebrity-influencers (M = 3.120, SD = 0.903). TrustBlog indicates that respondents have
Loy_Control indicates that respondents are positively influenced by the predictors of trust - reliability, expertise, sincerity, identifiability -, and that they have therefore feel larger positive levels of the predictors of brand loyalty - brand trust, positive attitude, purchase intention-, and thus feel larger positive levels of brand loyalty (M = 3.828, SD = 0.405). Table 4 presents correlations between the independent, dependent, mediating and control variables. There is a positive relationship between brand loyalty and trust r (96) = 0.599, p < .001. The difference between micro-influencers and celebrity-influencers has a negative, but non-significant relationship with loyalty r (96) = p > .05. The difference between micro-influencers and celebrity-influencers has a negative relationship with trust r (96) = -0.419, p < .001. TypeInf indicates the change that occurs when a celebrity-influencer promotes the product or brand, instead of the micro-influencer. This indicates that relationship between the promotion and trust decreases with -0.419 when a celebrity-influencer promotes instead of a micro-influencer. Control variables InfluenceLevel r (96) = 0.319, p < .001, LoyaltyBlog r (96) = 0.380, p < .001 and TrustBlog r (96) = 0.259, p < .05 have positive relationships with brand loyalty. Control variables InfluenceLevel r (96) = 0.387, p < .00, LoyaltyBlog r (96) = 0.428, p < .001, TrustBlog r (96) = 0.253, p < .05 and Loy_Control r (96) = 0.353, p < .001 have positive relationships with trust.
Table 4. Means, standard deviations, correlations and Cronbach’s Alpha’s
M SD 1 2 3 4 5 6 7 1. Loyalty 2.896 0.878 (0.850) 2. Trust 2.646 0.903 0.599** (0.849) 3. TypeInf 0.520 0.502 -0.198 -0.419** 4. InfluenceLevel 3.190 0.585 0.319** 0.387** -0.188 (0.744) 5. LoyaltyBlog 3.120 0.903 0.380** 0.428** -0.127 0.317** (0.869) 6. TrustBlog 3.233 0.881 0.259* 0.253* 0.017 0.387** 0.773** (0.880) 7. Loy_Control 3.828 0.405 0.195 0.353** -0.028 0.286* 0.249* 0.345** (0.835)
5.3 Simple mediation analysis
To test hypotheses, a simple mediation analysis was performed. Model 4 in the PROCESS macro of Hayes (2013) was used. This model tests causal relationships between different paths. In the following section, three models will be presented. All models are significant (p = 0.000, p = 0.000, p = 0.005 respectively) and can be used to interpret. All original models can be found in Appendix 1. As presented in the method section, there are different paths which the models present. Model 1 presents the a path, which is the mediating effect of Y (type of influencer) on Z (trust). Model 2 presents the b path, which is the mediating effect of Z (trust) on X (brand loyalty), which controls for the effect of Y (type of influencer) on X (brand loyalty), which represents the c’ path. Model 4 presents the c path, which is the effect of Y (type of influencer) on X (brand loyalty) (Hayes, 2017, pp. 78-83). Finally, the total, direct and indirect effects will be given, where total effects represent the c path, direct effects represent the c’ path and indirect effects represent c – c’ (Hayes, 2017, pp. 78-83).
In Table 5, the mediating effect of the type of influencer on trust is presented. The result shows that micro-influencers have a significantly larger positive effect on trust than celebrity-influencers (β = -0.5510, t(96) = -3.8345, p = 0.0002). The result indicates that trust decreases by 0.5510 when a celebrity-influencer promotes a product or brand instead of a micro-influencer. From this result, it can be concluded that Hypothesis 1 is confirmed.
** Correlation is significant at the 0.01 level (two-tailed) * Correlation is significant at the 0.05 level (two-tailed) c. Listwise N=96
Table 5. Model 1: Mediating effect: Y > Z (N = 96) - a path Trust B SE T p Constant -1.1079 0.7089 -1.5629 0.1216 TypeInf -0.5510 0.1437 -3.8345 0.0002 InfluenceLevel 0.3297 0.1324 2.4900 0.0146 LoyaltyBlog 0.4720 0.1233 3.8286 0.0002 TrustBlog -0.2871 0.1331 -2.1570 0.0337 Loy_Control 0.5868 0.1840 3.1887 0.0020 𝑅𝑅2 = 0.484 F(6) = 13.889, p = 0.000
In Table 6, the mediating effect of trust on brand loyalty is presented. Model 2 presents both the b and the c’ path. The b path is controlling for the c’ path. When the b path is significant, and the c’ path is not, this is a sign of a mediating effect (Hayes, 2017, pp. 78-83).
Results indicate that trust has a significantly positive effect on brand loyalty (β = 0.6413, t(96) = 6.0393, p = 0.0000). This result means that brand loyalty increases by 0.6416 when trust increases by one unit. Therefore, it can be concluded that Hypothesis 3 is
confirmed. The type of influencer seems to have a positive, but non-significant impact on brand loyalty (β = 0.1758, t(96) = -1.1316, p =0.2609). This means that brand loyalty increases by 0.1758 when a celebrity-influencer promotes a product or brand instead of a micro-influencer, which would contradict Hypothesis 1. Although this result is contradicting, it is non-significant so it cannot be taken in notice, but Hypothesis 1 can still not be
confirmed.
Taking both of these results into notice, it can be concluded that there is some form of mediation.
Table 6. Model 2: Mediating effect: (N = 96) - b and c’ path Loyalty B SE T p Constant 1.3775 0.7198 1.9137 0.0589 Trust 0.6413 0.1062 6.0393 0.0000 TypeInf 0.1758 0.1554 1.1316 0.2609 InfluenceLevel 0.1031 0.1372 0.7514 0.4544 LoyaltyBlog 0.0924 0.1333 0.6934 0.4899 TrustBlog 0.0197 0.1368 0.1442 0.8857 Loy_Control -0.1853 0.1946 -0.9523 0.3436 𝑅𝑅2 = 0.4578 F(7) = 10.6151, p = 0.000
In Table 7, the total effect, which represents the c path, is presented. This represents the total effect of the type of influencer on brand loyalty. The result indicates that micro-influencers have larger positive, but non-significant effect on brand loyalty than celebrity-influencers (β = -0.1775, t(96) = -1.0428, p = 0.2999). This means that brand loyalty decreases by 0.1775 when a celebrity-influencer promotes a product or brand instead of a micro-influencer. This result is in line with Hypothesis 1, but is non-significant, so Hypothesis 1 is still rejected.
Table 7. Model 3: Total effect model: Y > X (N = 96) - c path Loyalty B SE T p Constant 0.6670 0.8398 0.7943 0.4292 TypeInf -0.1775 0.1702 -1.0428 0.2999 InfluenceLevel 0.3145 0.1569 2.0050 0.0480 LoyaltyBlog 0.3951 0.1460 2.7052 0.0082 TrustBlog -0.1644 0.1577 -1.0426 0.3000 Loy_Control 0.1910 0.2180 0.8761 0.3833 𝑅𝑅2 = 0.2331 F(6) = 4.5084, p = 0.005
-0.1775, t(96) = -1.0428, p = 0.2999). Contradicting the total effect, the direct effect indicates celebrity-influencers have a larger positive effect on brand loyalty than micro-influencers (β =
0.1758, t(96) = -1.1316, p =0.2609), controlling for trust. Both the total and direct effects are non-significant.
The result for the indirect effect tests the indirect effect of the type of influencer on brand loyalty. This means the mediating effect, which runs through trust. A normality test has been performed for the indirect effect (β = 0.3534, p =0.0013). Because of the significant normality of the test, indirect effects can be used for interpretation (Hayes, 2017, pp. 78-79). The result indicates that micro-influencers have a larger positive effect on brand loyalty than celebrity-influencers (β = -0.3534, LLCI = -0.6677, ULCI = -0.1360). Because the indirect effect is measured by c – c’, and no test was performed, there are no T and P values. The upper and lower bounds still indicate that the effect is significant because 0 is not in the interval.
Taking all results together, it can be concluded that there is mediation. The positive effect of micro-influencers vs. celebrity-influencers on brand loyalty is not directly higher, but is significantly higher when trust is taken into notice. This means Hypothesis 2 and 3 are confirmed and Hypothesis 1 is rejected. Concluding, the positive effect of micro-influencer advertising on brand loyalty is not directly higher than the effect of celebrity-influencer advertising, but because micro-influencer advertising generates higher levels of trust than celebrity-influencer advertising, micro-influencers indirectly have a larger positive effect on brand loyalty than celebrity-influencers.
Table 8. Sobel test: Total, direct and indirect effects
B SE T P LLCI ULCI
Total effect of
the type of influencer on brand loyalty – c path
-0.1775 0.1702 -1.0428 0.2999 -0.5158 0.1607
Direct effect of the type of influencer on brand loyalty –
c’ path
0.1758 0.1554 1.1316 0.2609 -0.1330 0.4846
Indirect effect of the type of influencer on brand loyalty (mediating effect) –
c – c’ path
-0.3534 0.1328 - - -0.6677 -0.1360
Normality test for indirect
6. Discussion, recommendations and implications
The purpose of this study was to investigate whether micro-influencers have a larger positive effect on brand loyalty than celebrity-influencers, and whether this effect is mediated by trust. Three hypotheses have been tested, testing whether micro-influencers have a larger positive effect on brand loyalty than celebrities, whether micro-influencers generate higher trust levels than celebrity-influencers, and whether trust therefore leads to a higher level of brand loyalty. In the following section, results will be summarized and interpreted. Then, the limitations of this study and recommendations for future research will be given. Finally, the managerial implications of this study will be given.
6.1 Summary and interpretation of results
The results were partly as predicted. Hypothesis 1 stated that micro-influencer advertising has a larger positive effect on brand loyalty than celebrity-influencer advertising. This hypothesis is not confirmed. Results indicated that micro-influencer advertising has a larger positive effect on brand-loyalty than celebrity-influencer advertising does, but this result was not significant. This result is not surprising because no academic literature was written on this direct link, so the prediction was based on non-academic resources which did had no prove for their statements.
Hypothesis 2 stated that micro-influencers generate higher levels of trust than
celebrity-influencers. This hypothesis is confirmed. Results indicated that micro-influencers significantly higher levels of trust than celebrity-influencers. This result is in line with the statements of Forrest & Cao (2010). They stated that micro-influencers are perceived as more trustworthy and credible. This is due to the fact that they seem to be genuinely passionate about the products and brands they promote. Micro-influencers built long-lasting relationships with consumers and built a personal brand on their social media accounts. They seem to have
expertise in the products and brands they promote. Also, micro-influencers are more identifiable than celebrity-influencers (Jang & Stefanone, 2009). Taking all these things together, micro-influencers generate higher levels of trust than celebrity-influencers.
Hypothesis 3 states that a higher level of trust in the influencer leads to higher brand loyalty. This hypothesis is confirmed. Results indicated that a higher level of trust in the influencers leads to significantly higher brand loyalty. This is in line with literature about trust as a mediator for brand loyalty. Veloutsou (2006) states that trust is an important predictor of brand loyalty. When trust is high, brand loyalty is likely to also be high. Trust acts as a mediating variable in the link between trust, satisfaction and loyalty. Forrest & Cao (2010) state that micro-influencers are perceived as being more trustworthy and credible than celebrity-influencers. Taking these two statements together, the satisfied result can easily be explained. Because consumers have more trust in micro-influencers than in celebrity- influencers, brand loyalty is higher for micro-influencer advertising than for celebrity- influencer advertising.
6.2 Limitations and recommendations
The limitations of this study can mostly be assigned to the research design. One of the limitations of this study is that it is cross-sectional. The results are a snapshot of how respondents felt at the time they filled out the questionnaire (Saunders et al., 2012, pp. 181- 182). A longitudinal study might have been more appropriate for this research. Previous study has proved that trust is an important feature of creating long-term relationships (Wottrich et al., 2017). For this study, it might have been interesting to have the respondents follow both the celebrity-influencers and the micro-influencers for a while before testing how they feel about the influencers and about the products or brands the influencers promote. Due to a lack the short time-horizon of this study, and the complexity of letting a sizable amount of
respondents follow the right influencers for a while, this way of testing could not be
performed in the current study. For future study, it might be interesting to test the attitude of respondents towards micro-influencer advertising and celebrity-influencers advertising in a longitudinal study.
Another limitation of this study is the sample. Having a reprehensive sample is important for the generalizability of the study (Saunders et al., 2012, pp. 178-180). In the current study, the sample exist mostly of females between 20 and 30 years old who are in college. This means that results are not generalizable to other groups than the one in the sample. For future research, it might be interesting to collect a more reprehensive sample so results can be more generalized.
6.3 Managerial implications
The results of this study have some implications for managers of products and brands. The results indicated that micro-influencers can create larger positive levels of brand loyalty than celebrity-influencers, because consumers have more trust in micro-influencers. Managers can take their advantage from this in their future marketing campaigns. Firstly, they should look at the aim of their campaign. When looking at results from the control variables, influencers on general are effective in creating brand loyalty. Thus, using influencers in marketing
campaigns can be very useful for products and brands. When the aim of the campaign is to create short-term brand awareness, using a celebrity-influencer is most effective (Cruz, 2018). When the aim of the campaign is to create long-term brand loyalty, it is more effective to use a micro-influencer. It is important to use the right micro-influencer for the target group, because they are likely to feel a relationship with this influencer, and will therefore trust the micro-influencer more than a celebrity-influencer. Brand loyalty will therefore be increased, which is a desirable outcome for managers.
7. Conclusion
This study aimed to investigate whether there is a difference in the effectiveness of micro- influencer advertising versus celebrity-influencer advertising on brand loyalty, and whether trust mediates this relationship. The following research question was asked: ‘’Does micro-
influencer advertising on Instagram have a larger positive effect on brand loyalty than celebrity-influencer advertising, and is this relationship mediated by trust?’’. The reason for
this question was because of the lack of academic literature on this topic. Research has been done about the general effect of influencers on brand loyalty, and there seemed to be a positive relationship. Also, research has been done about trust, and how trust in the person promoting a product or brand could increase the effectiveness of the promotion. From other research, trust seemed to be an important predictor of brand loyalty. The current study aimed to investigate all of these aspects together.
To investigate the research question, a questionnaire was distributed. Results from the survey were statistically investigated. First, the descriptives of the demographics, general questions and the manipulation were investigated. Then, the reliabilities of the variables were tested and a factor analysis was performed. The correlations of the variables were tested. Finally, the models were tested by model 4 in the PROCESS macro developed by Hayes. This tested whether there were significant effects of the independent, dependent and mediating variables.
The results of this study were partly predicted. Micro-influencers did significantly generate higher trust levels than celebrity-influencers. The higher level of trust in the influencer also significantly led to a larger positive effect on brand loyalty. What was not predicted was that there was no significant direct difference in the effectiveness of micro- influencer advertising versus celebrity-influencer advertising.
Taking all things together, it can be concluded that micro-influencers do not directly have a larger positive effect on brand loyalty than celebrity-influencers, but when this relationship is mediated by trust, micro-influencers do have an indirect larger positive effect on brand loyalty than celebrity-influencers.
References
Anderson, E. W. (1998). Customer satisfaction and word of mouth. Journal of service
research, 1(1), 5-17.
Anderson, E. W., & Sullivan, M. W. (1993). The antecedents and consequences of customer satisfaction for firms. Marketing science, 12(2), 125-143.
Bakshy, E., Hofman, J. M., Mason, W. A., & Watts, D. J. (2011, February). Everyone's an influencer: quantifying influence on twitter. Proceedings of the fourth ACM
international conference on Web search and data mining (pp. 65-74). ACM.
Bickart, B., & Schindler, R. M. (2001). Internet forums as influential sources of consumer information. Journal of interactive marketing, 15(3), 31-40.
Boerman, S.C., Willemsen, L.M., & Van der Aa, E.P. (2017). “This post is sponsored” Effects of sponsorship disclosure on persuasion knowledge and electronic word of mouth in the context of Facebook. Journal of Interactive Marketing, 38, 82-92. Booth, N., & Matic, J. A. (2011). Mapping and leveraging influencers in social media to
shape corporate brand perceptions. Corporate Communications: An International
Journal, 16(3), 184-191.
Carbone, L. (2018). This is how the instagram algorithm works in 2018. Blogi. Päivitetty, 19, 2018.
Chatard, A., Bocage-Barthélémy, Y., Selimbegović, L., & Guimond, S. (2017). The woman who wasn't there: Converging evidence that subliminal social comparison affects self-evaluation. Journal of Experimental Social Psychology, 73, 1-13.
Chaudhuri, A., & Holbrook, M. B. (2001). The chain of effects from brand trust and brand affect to brand performance: the role of brand loyalty. Journal of marketing, 65(2), 81-93.
Chaffey, D. (2017, April 27). Global social media research summary. Retrieved from https://www.smartinsights.com/social-media-marketing/social-mediastrategy/new- global-social-media-research/ on May 6th, 2018
Cheung, M. Y., Luo, C., Sia, C. L., & Chen, H. (2009). Credibility of electronic word-of- mouth: informational and normative determinants of on-line consumer
recommendations. International Journal of Electronic Commerce, 13(4), 9e38. http://doi.org/10.2753/JEC1086-4415130402.
Chu, S. C., & Choi, S. M. (2011). Electronic word-of-mouth in social networking sites: a cross-cultural study of the United States and China. Journal of Global Marketing, 24(3), 263e281. http://doi.org/10.1080/08911762.2011.592461
Chu, S.C. & Kamal, S. (2008). The Effect of Perceived Blogger Credibility and Argument Quality on Message Elaboration and Brand Attitudes. Journal of Interactive
Advertising, 8(2), 26-37. doi: 10.1080/15252019.2008.10722140
Cruz, F. (2018, May 16). Influencers: Micro & Macro. Retrieved from:
https://www.tapinfluence.com/influencers-micro-macro/ on May 4th, 2018 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.
Dick, A. S., & Basu, K. (1994). Customer loyalty: toward an integrated conceptual framework. Journal of the academy of marketing science, 22(2), 99-113.
Dinesh, D. (2017). Why Micro-Influencers Are a Social Media Marketing Imperative for 2017. EContent, 40(3), 14-15.
Eiser, J. R., Stafford, T., Henneberry, J., & Catney, P. (2009). “Trust me, I'm a scientist (not a developer)”: Perceived expertise and motives as predictors of trust in assessment of risk from contaminated land. Risk Analysis, 29(2), 288-297.
