Master Thesis
MSc Business Administration – Digital Marketing 2021/2022
Selecting the Right Influencers in the Digital Era
Author: Samanta Vaivade Student Number:11691980 Supervisor: Hsin-Hsuan Meg Lee Date of Submission: January 27, 2022
University of Amsterdam
Statement of Originality
This document is written by Samanta Vaivade 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.
UvA Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
The rise and diversity of social media influencers have created a significant transformation in the marketing landscape. Brand use of social media influencers has shown to be an effective tool not only to promote purchase intentions but also to increase brand awareness through highly engaged influencer posts. Due to the proven marketing effectiveness of users’
willingness to engage with influencer posts, identifying and choosing influencers that users are more willing to engage with is imperative.
One of the most recent and relatively new shifts among social media influencers is the differentiation between human and virtual, also called computer-generated imagery (CGI), influencers. Thus, the objective of the current study is to fill the research gap on the varying effects that human and virtual influencers elicit on Instagram users' willingness to engage.
Moreover, determining the primary moderating role of influencer size based on the follower count and higher-order moderating effect of identification on the moderated relationship.
Eventually, to enrich the understanding of influencer marketing and for marketers to select the most suitable social media influencer.
Findings of the current study show that human influencers and virtual influencers do not significantly differ on users’ willingness to engage, and influencer size does not
significantly moderate this relationship. Exemplifying that these characteristics are not enough to determine users’ engagement and other factors should be taken into consideration too. Meanwhile, findings show that there is a significant positive secondary moderating effect of identification on this moderated relationship.
Table of Contents
1.0 Introduction ... 6
2.0 Literature Review and Hypothesis Development ... 9
2.1 Social Media Influencers ... 9
2.2 Users' Willingness to Engage ... 10
2.3 Human versus Virtual Influencers... 11
2.4 Micro versus Macro-Influencers ... 14
2.5 Identification ... 17
2.6 Study Overview ... 18
3. Method ... 19
3.1 Design ... 19
3.2 Sample ... 19
3.3 Procedure ... 20
3.4 Measures... 21
3.4.1 Willingness to Engage... 21
3.4.2 Identification ... 22
3.4.3 Attitude Towards the Influencers (ATTI) ... 22
3.4.4 Familiarity with the Influencer (FWTI) ... 22
3.4.5 Familiarity with Influencers in General (FWIIG) ... 23
3.5 Hypothesis Testing ... 23
4.0 Results... 23
4.1. Correlations ... 23
4.2 Normality Check ... 24
4.3 Hypothesis Results ... 25
4.3.1 Influencer Type and Willingness to Engage ... 25
4.3.2 Primary Moderation of Influencer Size ... 25
4.3.3 Secondary Moderation Effect of Identification ... 26
5.0 Discussion... 26
5.1 Findings ... 26
5.2 Practical and Theoretical Implications ... 29
5.3 Limitations and Future Research... 30
6.0 Conclusion ... 32
Appendices ... 46
Appendix 1: Survey... 46
List of Tables and Figures
Figure 1: Conceptual Framework
Figure 2: Summary of Hypotheses Results Figure 3: Summary Hypotheses
Table 1: Sociodemographic Characteristics Table 2: Manipulation Distribution
Table 3: Scale Reliability
Table 4: Means, SD and Correlations
1.0 Introduction
Fashion, lifestyle, and other brands have increasingly adapted their marketing activities using influential online users, known as social media influencers as their core source to build brand awareness and acquire more customers (Khamis, Ang, & Welling, 2017). While often sharing their lives in ordinary settings and being more relatable to the other platform users than traditional celebrities, influencers share endorsed thoughts about products and services on social media platforms, such as Instagram (De Veirman, Cauberghe,
& Hudders, 2017). According to the estimates, in the US influencer marketing is poised to surpass the $3 billion mark by the end of 2021 (eMarketer, 2021). Moreover, platforms like Instagram allow these brands to leverage the increasing marketing power of influencers to entice new as well as existing customers and co-create relationships through the engagement between an influencer and the other users (Michaelidou, Siamagka, & Christodoulides, 2011). According to Pick (2020), influencers are deemed to have a bigger influence on new generation consumers than conventional advertising activities. Therefore, the emergence and growth of social media influencers as brand image co-creators has made a significant
transformation in the marketing landscape and altered the way users communicate with the brands.
The literature on social media influencers varies from numerous aspects of their diversity and the effects they have on consumer decision-making (Power & Phillips-Wren, 2011). It has been shown that people interact and engage with influencers very differently based on the level of credibility they have towards an influencer (Lou & Yuan, 2019) as well as the level of relatability and perceived reciprocity when it comes to relationships that are parasocial by their nature (Nouri, 2018; Reinikainen, Munnukka, Maity, & Luoma-aho, 2020). All these aspects vary depending on more than the influencer’s follower count (Taylor, 2020). Therefore, since the global spending on influencer marketing in this creator economy keeps growing, it becomes more crucial for the brands to understand the
effectiveness of the content created by online platform users, not traditional companies (Bhargava, 2021).
Brands can have access to a large audience that comes from more than merely their own earned media, but also followers of other personal brands, namely, social media influencers (Pick, 2020). The diversity and number of social media influencers are growing involving users from different niches (Taillon, Mueller, Kowalczyk, & Jones, 2020).
Although there is a high variety of different influencers that brands can choose from, the
reach these goals should be selected. Knowing that user’s engagement with influencers plays a crucial role not only to influencers personal brand success but also to collaborative brands’
benefits to create the desired brand image and associations (Haenlein, Anadol, Farnsworth, Hugo, Hunichen, & Welte, 2020), it is important to understand what kind of influencers based on various aspects are the most appropriate and mutually beneficial for a brand.
One of the most recent distinctions of social media influencers evolves from their realness, namely, being real humans or virtual influencers (Silva, Farias, Grigg, & Barbosa, 2020). Many well-known brands in the fashion, technology and lifestyle industries have already made use of this relatively new form of non-human influencers to endorse their products (Drenten & Brooks; 2020; Callahan, 2021). Some of the most prominent virtual influencers, like Lil Miquela and Lu do Magalu, have gained more than 3 million followers in the time this study is taken. Not only virtual influencers have gained a high number of
followers, but they also have reached high engagement both from high numbers of likes and comments on their posts, as well as outstanding collaborations with luxury brands like Prada, Chanel and Balenciaga (Marwick, 2018). Furthermore, what makes these virtual influencers even more interesting and difficult to distinguish from the other real human influencers is their “realness” that comes from the fact that their represented behaviour and bodies are highly anthropomorphized (Arsenyan & Mirowska, 2021). As a result, this novel type of influencers has gained attention from various brands and Instagram users and certainly has become a necessary field of research in marketing. Since this is a yet new trend and main adapters of social media marketing have focused on human influencers, information about the subject of virtual influencers and their effect on users’ engagement is limited (da Silva
Oliveira & Chimenti, 2021). In fact, majority of even more recent studies of influencer marketing have completely disregarded this unit of analysis: non-human influencers (da Silva et al., 2021). Therefore, this study aims to develop a better understanding of this subject.
When it comes to a more widely adapted distinction, namely, influencer size, previous studies have shown that there is a different level of trust and purchase intentions that come from micro-influencers versus macro-influencers (Kay, Mulcahy, & Parkinson, 2020;
Alassani & Göretz, 2019). These studies demonstrated that when a micro-influencer promotes a brand, there is higher reliance and trust amongst Instagram users and therefore higher purchase intentions of a promoted brand's products. Nevertheless, quite often the main goal of a brand that uses Instagram influencers as its marketing approach is to increase salience and brand awareness (Ferina, Sri, & Putu, 2021). This, however, is mainly achieved when an influencer gets high engagement of their posts and subsequently gets noticed by
more than their own followers through the explore page, individual shares on their stories or direct messages (Ferina et al., 2021). Thus, different sizes of influencers can be selected depending on whether the overall goal is brand exposure or the development of connection and fostering relationships with potential customers is more important. Although influencers size and popularity determine both the costs that the brand needs to devote and the actual effectiveness of it, there are no previous studies that show how this aspect interacts with different types of influencers, namely, human influencers versus virtual influencers.
Considering these two factors this study will test how users’ willingness to engage differs based on influencer type, namely, human versus virtual, and how this relationship is further moderated by influencer’s size.
Moreover, besides influencer follower count, it is also considered that users' online interaction with personal brands is affected by more personal factors that arise from users’
feelings and emotional connections with them (Leite & de Paula Baptista, 2021). While spending significantly more time on social media platforms, users develop more personal connections with influencers and the content that they choose to follow (Fan & Yost, 2019).
While creating their digital identities many individuals online life is blending into their offline world, making the distinction between offline and online interactions blurred (Blanch, 2016). Users create their own online personas and get influenced by the ones who they can identify with both in terms of perceived similarly as well as desirability to be alike.
Considering the feeling of identification, previous studies have shown that humans generally hesitate interactions with AI and robots (Darling, Nandy, & Breazeal, 2015). However, the increasing use of human-AI online interactions requires further research on how it is
moderated by human identification. In this era when human avatars are the potential future of the platforms like Instagram, it is crucial to understand how more personal human feelings like identification can play a role in social media engagement. Hence, the overarching aim of this study is to uncover the influence of social media influencer type, namely, virtual versus human on users’ willingness to engage with an influencer, and how this relationship is first moderated by influencer size, namely, micro versus macro and secondary moderated by users' level of identification. Therefore, the present study poses the following question: How does users' willingness to engage varies with different types of influencers and how
influencers size moderates this relationship, and further how is this moderated relationship moderated by the user's level of identification? By conducting a between-subjects survey design this research question is answered.
The theoretical contributions of this thesis are as follows. First, this research
investigates the different effects human and virtual influencers elicit on users’ willingness to engage. Second, how influencer size interacts in this main relationship between influencer type and users’ willingness to engage. Third, how identification feeling with an influencer moderates this moderated relationship. The practical insights will offer marketers more tangible tools when selecting each type and size of influencer for a successful influencer- brand collaborations.
This study is divided into sections. In the next section, the current literature, and the hypothesis development are discussed. Second, the methods and analysis part is covered, in which the design, data collection, and measures will be described. Third, the experiment results are discussed. Finally, the study results and implications together with the limitations and future research are described in the discussion.
2.0 Literature Review and Hypothesis Development
As a basis for answering the research question, earlier studies on the different theoretical concepts requires some explanation. Hence, in this section previous research and theories about these concepts will be discussed. Theories regarding social media influencers and users’ willingness to engage will be discussed. Moreover, human versus virtual and micro versus macro-influencers will be introduced. Lastly, the moderating effect of the feeling of identification will be summed up and linked to the moderated relationship between different types and sizes of influencers.
2.1 Social Media Influencers
By sharing their experiences and thoughts on social media, influencers are becoming a very powerful tool for marketers and advertisers. According to Kirwan (2018), social media influencers are “people who have large audiences of followers on their social media accounts, and they leverage this to influence or persuade this following to buy certain products or services”. These individuals have unmistakable influence over other platform users, and they are shown to be a great marketing tool for many brands by enabling an ongoing consumer- brand interaction (Evans, Phua, Lim, & Jun, 2017; Murdough, 2009). Marketing with
influencers is a trend on the rise, with an increasing number of brands who are investing their financial resources in influencer marketing (Linqia, 2017) According to Linqia (2020) 89%
of marketers claim that social media influencers can have a significant lead if their content about a brand is authentic and guides a further interaction with users and brand's product.
Past research has studied the effects and outcomes of different influencers on consumers behaviour, and according to Tengblad-Kreft, Hagman and Hessels (2017),
effective endorsements and partnerships with social media influencers will result in a positive perception of an influencer transfer to the brand. Moreover, influencer marketing has shown to be an effective tool to increase brand awareness, customer trust in a brand as well as purchase intentions (Fernando, Syahbani, Darmawan, Amalina, & Ikhsan, 2021). Therefore, with their assembled web of followers on social media platforms like Instagram, social media influencers are perceived as trustworthy experts within their niche (Arenas-Márquez,
Martínez-Torres, & Toral, 2021).
In the context of social media marketing, there are many reasons why brands choose to use influencers as their marketing tool. During the rise of new technologies that allow users to opt-out of ads and make it more difficult for brands to buy fame, the use of social media influencers helps the brands to find their new way to stand out (Holt, 2016). Moreover, besides influencers being known to be a beneficial tool for other brands to promote their products and services, many influencers who get monetary compensation for their collaborations, i.e., paid influencers, are using this strategy to build their own personal brands. Therefore, this exponential growth of influencers in social media platforms also opens new opportunities for people to develop their personal brands online (Hennessy, 2018).
2.2 Users' Willingness to Engage
The conceptual framework of this study introduces users' willingness to engage as the main outcome variable, which measures followers’ involvement in and responses to different influencers’ posts on Instagram (Arora, Bansal, Kandpal, Aswani, & Dwivedi, 2019). This users' response is measured by such behaviours as liking and commenting on influencers’
Instagram posts (Belanche, Flavián, & Ibáñez-Sánchez, 2020). Moreover, another measure of engagement is sharing of influencers posts either via stories or direct personal messages (Barger, Peltier, & Schultz, 2016; Coelho, de Oliveira, & de Almeida, 2016). As such, users' willingness to engage is the leading proof and numeric measure of influencers perceived favourability and their level of influence towards other Instagram users (Tafesse & Wood, 2021). Moreover, because follower engagement is known to play an important role in brand salience and users' purchase intentions, it has been widely applied in the brand engagement
literature (Pansari & Kumar, 2017; Van Doorn, Lemon, Mittal, Nass, Pick, Pirner, &
Verhoef, 2010; Argyris, Wang, Kim, & Yin, 2020; Harmeling, Moffett, Arnold, & Carlson, 2017). In line with the above, users’ willingness to engage has been shown to contribute to favourable user perception leading brands to increased customer acquisition, spending on promoted products as well as generating electronic word of mouth (eWOM) (De Vries, Gensler, & Leeflang, 2017; Harmeling et al., 2017; Erkan, 2015).
Since marketers no longer have full control over the discussions that are disseminated about their brands, the knowledge of engagement through interactions with social media influencers as brand ambassadors becomes important (Smith, Kendall, Knighton, & Wright, 2018). Engagement with the brand through social media influencers is one of the key disrupters of traditional marketing communication channels, creating a new kind of brand- consumer interaction (Jin & Ryu, 2020). For instance, many brand ads that have been posted on their own social media platforms may go viral after someone with big audience shares it on their platform. Furthermore, consumers can add their comments to other social media influencers posts without the brand's full control over where and what opinions have been spread online (Delbaere, Michael & Phillips, 2021). Thus, by engaging with influencers who promote a certain brand, consumers have created a more active role in authoring brand stories (Delbaere et al., 2021).
Through their engagement with social media influencers, users' show their personal involvement with their content (Dhanesh, 2017). Previous research suggests that users
develop their willingness to engage with social media influencers by co-constructing personal interactions (Abidin & Ots, 2015). Moreover, according to Jiménez-Castillo and Sánchez- Fernández (2019) influential power exerted by interactions not only generates users’
engagement with influencers but also with the influencer promoted brands. Thus, since engagement creates a bond between consumers and brands, many businesses have started to use influencers to effectively engage with their targets. In sum, users’ engagement is
imperative for making them involved receivers of information about a brand and developing deeper user connections with the message sender, i.e., influencer and hence the brand itself (Ashley & Tuten, 2015).
2.3 Human versus Virtual Influencers
Although influencer size has been the most adapted level of classification of different kinds of influencers, one of recent trends and topics that has not been given enough research
is the distinction between human and virtual influencers. Whilst human influencers are real humans who share their content that has been made on a real-life basis, virtual influencers are non-human influencers also known as computer-generated imagery (CGIs) who can be both explicitly as well as not explicitly nonhuman in their physical appearance and personality (da Silva Oliveira et al., 2021; Moustakas, Lamba, Mahmoud and Ranganathan, 2020). These digital stars are making waves in the world of social media, revolutionizing the concept of influencer marketing. Virtual influencers have been designed realistically using fictive computer-generated images (CGI) that are built on artificial intelligence (AI) or other resources (da Silva Oliveira et al., 2021), and are carefully construed by their creators (Richters, 2020). With their content in fashion, shared experiences, emotions and opinions, virtual influencers inspire and influence their followers on Instagram like other human influencers (MediaKix, 2019). These online personalities have gained millions of followers on Instagram (Molin & Nordgren, 2019) and just like humans they have both their
professional as well as social lives (Maughan, 2018).
Despite their broad influence and identification with other users, human influencers have raised attention to some practical and ethical issues like their regular fashion
consumption and promotion of their environmentally unsustainable lifestyle (Thornton, 2021;
Denisova, 2021). Conversely, due to their nonexistence offline, virtual influencers can remove many of these concerns by being able to maintain a consistent image, the risk of indiscretions can be lowered because their words and behaviour can be calibrated in the background and their fashion consumption is not physically real (Tan & Liew, 2020; Molin et al., 2019; Wills, 2019). Furthermore, during the current pandemic-induced economic crisis which affected many brands and consumers (Mahmoud, Ball, Rubin, Fuxman, Mohr, Hack- Polay, & Wakibi, 2021), virtual influencers had a unique opportunity to “live” their life outside the lockdown circumstances, keeping their vibrant lifestyle and promoting brands in more “social” environments (Tan et al., 2020). This shows that virtual influencers can maintain their influential status and persuasive message through being human-like with their audio and visual features as well as their stereotypical standard of perceived attractiveness (Faddoul & Chatterjee, 2020; Khan & Sutcliffe, 2014). Moreover, despite potential
challenges with human-robot interactions and their potential risk for perceived brand- influencer fit, virtual influencers have already made collaborations with many prestigious brands, reaching high salience and engagement rates (Xie-Carson, Benckendorff & Hughes, 2021). Intriguingly, according to Baklanov (2019), the average engagement rate of virtual
the above mentioned, it seems apparent that many digital marketers have raised the belief that virtual influencers are a viable substitute for human influencers (Khamis et al., 2017).
Nonetheless, considering that during the year when the data of engagement was collected virtual influencers was a yet emerging trend, the above-mentioned engagement rates could have been due to virtual influencers being a novel phenomenon (Xie-Carson, et al., 2021). Generally, research suggests that virtual influencers have gained their exposure because of the novelty but there is still an ongoing discussion of their perceived authenticity as well as their reliability compared to the human influencers. Above that, in the past human- robot interactions have been proven challenging, and one of the suggested reasons for this is that robots do not have a high-level context-based cognition (El Hafi, Isobe, Tabuchi, Katsumata, Nakamura, Fukui, & Taniguchi, 2020). In addition, humans are not able to intuit the perception state of robots as they can for other humans (El Hafi et al., 2020).
Human influencers, however, have been grounded in their tendency to be virtually accessible and intimate by regularly sharing their personal life updates and showing more relatable parts of themselves (Abidin et al., 2015; Djafarova & Rushworth, 2017; Kim &
Song, 2016). Human influencers not only show how the products, clothes and experiences look but also share how they are applied and experienced in real life (Zak & Hasprova, 2020;
Sudha & Sheena, 2017; Feng, Chen, & Kong, 2021). By having a real-life experience, they openly share their thoughts and opinions as well as their personal feelings towards a brand making their opinions more reliable than those of virtual influencers (Chen et al., 2021). On top of that, some influencers also help brands to build a community of specific focus groups (Armenteras, Paech & Politiek, 2017).
Furthermore, research suggests that personal feelings of connection, perceived authenticity and true transparency are what make the use of influencer marketing a powerful tool for brands as their communication with modern consumers (Moustakas et al., 2020;
Arriagada & Bishop, 2021). Therefore, it is crucial that influencers are able to generate content that truly resonates with their followers and the rest of the Instagram users. However, because of their known to be not "real" content, which is created and managed by other humans and agencies, virtual influencers might lack this feeling of identification and personal connection with their followers (Moustakas et al., 2020). Although the reasons why users are willing to engage with influencers are quite ambiguous, previous studies have shown that perceived authenticity plays a significant role in this (Martikainen & Pitkänen, 2019;
Eigenraam, Eelen, & Verlegh, 2021; Pittman, Oeldorf-Hirsch & Brannan, 2021). In general, social media users are more interested in influencers that are perceived as authentic, resulting
in more willingness to engage with them (Pittman et al., 2021). According to Kowalczyk and Pounders (2016) communicating with the users without having authenticity is not effective in creating an enticing relationship with the users. Therefore, since users’ perceived authenticity and willingness to engage with virtual influencers are still in doubt, it is questionable whether brands utilizing virtual influencers as their marketing tool gain real marketing benefits from users’ engagement.
Although the subject of virtual influencers has a lot of implications that raise further questions, the currently existing literature has mainly focused on parasocial interaction between users and virtual influencers through their perceived source credibility (Molin et al., 2019), users’ response to virtual influencer based on their perceived visual humanness (Arsenyan et al., 2021). In accordance with da Silva Oliveira and Chimenti (2021), there are only 32 peer-reviewed English articles that are related to the keywords virtual influencer (VI) or digital influencer (DI). This suggests that there is still a strong need for studies on virtual influencers and their comparison with human influencers and how this plays a role in users’
willingness to engage with them. Thus, the increasing popularity of virtual influencers and the emerging use of influencer marketing, in general, makes this a worthwhile domain for further research. Based on the above discussed, human influencers are characterized as having higher users’ willingness to engage rates in contrast to virtual influencers. Hence, the following will be examined:
H1: Users’ willingness to engage will be higher with human influencers than with virtual influencers.
2.4 Micro versus Macro-Influencers
In practice, there are different classifications of social media influencers based on their size. Some practitioners are suggesting three levels, namely micro, macro and celebrity (Porteous, 2018; Childers, Lemon, & Hoy, 2019; De Veirman, Hudders, & Nelson, 2019), meanwhile others divide them into two levels, namely micro and macro (Dhanik, 2016;
Barker, 2017; Pavlika, 2017; Pierucci, 2017). However, as it appears from both academia and practice the most widely adapted classification of the size among these influencers is a two- level classification between macro and micro-influencers, and it is based on influencers' number of followers. The number of influencer followers determines influencer size and therefore the popularity and recognition among other Instagram users (De Veirman et al.,
2017). Although this distinction between micro and macro-influencers is also sometimes based on more than follower count, this paper will use the number of followers as the basis to measure the size of the influencer. Therefore, to make this distinction clear, according to Yılmaz, Sezerel and Uzuner (2020) micro-influencers are Instagram users whose number of followers is between 1,000 and 100,000, whilst macro-influencers are users who have between 100,000 and up to 1 million followers.
Although follower count is a crucial element to enhance influencers’ post reach and increase influencers' popularity and their financial worth (Arora et al., 2019; Belanche et al., 2020), a high number of followers does not necessarily translate into true influence and positive brand attitude (De Veirman et al., 2017). Furthermore, an influencer's number of followers has the power to change users' perceptions of a product that is being promoted in the sense that divergent design loses its perception of its uniqueness when it is being promoted by an influencer with a high number of followers (De Veriman et al., 2017). This finding corresponds to the other study that shows that highly popular influencers make the products to be perceived as less unique since there are many others who are interested in them (Machleit, Eroglue, & Mantel, 2000). Consequently, an increasing number of brands prefer to partner with micro-influencers more often since they have more capabilities to portray and align their lifestyle and visuals to a specific niche (Chen, 2016; Barker, 2017).
Social media practitioners have suggested that micro-influencers have a
proportionally higher engagement both in comments as well as direct messages with their followers, and therefore can be seen as more effective for collaborations with different brands and have better persuasion than other less reachable influencers (Kay et al., 2020). For
instance, Dhanik (2016) suggested that influencers who have fewer followers can be more effective since they can establish a more personal connection with their followers. Other practitioner-based research suggests that as the influencers obtain a higher number of
followers their engagement with followers drops and so do their followers' perception of their credibility and relatability, suggesting that micro-influencers are in the “sweet spot” having the right number of followers (Chen, 2017).
When it comes to users’ behaviour, studies have mainly examined followers’
responses to influencers’ product endorsements in terms of attitude towards the product that is being promoted (Kim & Kim, 2021; Schouten, Janssen, & Verspaget, 2020) and users’
willingness to purchase (Lou et al., 2019; Ladhari, Massa, & Skandrani, 2020). For instance, previous studies suggested that micro-influencers are perceived as more credible and thus lead to users' higher willingness to purchase their promoted brands (Gupta & Mahajan, 2019).
Nevertheless, one of the few studies that focused on how engagement varies depending on influencer's content is the study done by Tafesse and Wood (2021). By scraping an online influencer database, the study found that an influencer with a lower follower count is positively related to the number of comments and likes. In addition, previous findings have shown that influencers with a higher follower base began to resemble traditional celebrities and hereby followers’ ability to feel related and identify with them became weaker and consequently, their participation decreased (Belanche et al., 2020).
Although these studies support the proposition that micro-influencers have a higher engagement, they do not examine the willingness to engage from the users' willingness perspective and how influencer size interacts with different types, i.e., human versus virtual influencers. Based on the previous studies on human-robot interaction, this study expects human influencers to have a higher users’ willingness to engage than virtual influencers. As indicated above, micro human influencers are shown to be more relatable and approachable and hence users are expected to be more willing to engage with them than with macro human influencers. Nonetheless, knowing that virtual influencers are not as widely known and are still perceived as a novelty in influencer marketing, the engagement with micro virtual influencers might have a different effect on users’ willingness to engage than that with micro human influencers. According to Shin and Lee (2020) when it comes to engagement with virtual influencers, the high engagement that is only present with the most popular virtual influencers such as Lil Miquela. Other less know virtual influencers who do not have such a high media exposure and follower count, however, are shown to have a relatively low engagement rate (Shin et al., 2020). Considering this and previous knowledge about human- robot interaction, it is expected that influencer size, i.e., micro versus macro, interacts in relationship between human vs. virtual influencers and users ‘willingness to engage in different ways for human and virtual influencers.
In sum, although human micro-influencers have far fewer followers than macro- influencers, the interconnected relationships they foster with other Instagram users is more favourable for brands in terms of engagement behaviours and brand attitude. However, the lack of perceived authenticity and curiosity caused by the novelty effect of the virtual influencers, might play a different role of influencer size when the influencer is virtual. The following will be examined regarding the primary moderating effect of influencer size on the relationship between different influencer types and users’ willingness to engage:
H2: The effect of different types of influencers and willingness to engage is moderated by the size of the influencer.
H2a: Users' willingness to engage is higher with micro-human influencers than with macro-human influencers.
H2b: Users' willingness to engage is lower with micro-virtual influencers than with macro-virtual influencers.
2.5 Identification
To have more insight in the underlying processes of the users' willingness to engage with influencers based on their various characteristics the secondary moderating role of users' identification with an influencer will be studied. According to Schouten, Janssen and
Verspaget (2020) identification is “the extent to which one identifies with an influencer”.
Through the identification process individuals either consciously or unconsciously conform to the perceived identity with others (Brown, 2015). However, this can be further measured with two different subcomponents. One of the subcomponents is users' own perception of their similarity with an influencer or perceived extent of the things they have in common (Schouten et al., 2020). This, however, can depend on perceived similarity in terms of their values, lifestyle, visual style etc. Further, another component is wishful identification, which can be best explained as someone's level of desire to be like someone else (Shoenberger &
Kim, 2017; Hoffner & Buchanan, 2005). It generally endures beyond the mere visual experience of the social media representation; therefore, it is considered to be a long-term component of identification (Shoenberger et al., 2017).
The level of identification is considered to change social media users’ psychological responses as well as actual behaviours (Hu, Min, Han & Liu, 2020). Previous studies have shown that after experiencing wishful identification individuals, especially the younger ones, imitate the accoutrements and the actual behaviours of fictional characters (Hoffner et al., 2005; Bond & Drogos, 2014). Through mirroring an influencer who they wish to be like by, for example, using the same product that was shared by the influencer, followers might see themselves as more of a part of their community and hence be more involved (Shoenberger et al., 2017). Moreover, users are in general more involved with the content posted by
influencers with who they can identify (Bond et al., 2014; Uzunoğlu & Kip 2014).
When it comes to influencer size, within the research of human influencers the most recent trends in influencer marketing have indicated that social media users are more likely to trust and relate to micro-influencers (Brewster & Lyu, 2020; Schouten et al., 2020). When
users relate to influencers, they perceive them as similar (Sokolova & Perez, 2021). As previously mentioned, users’ identification with influencers is highly determined by their perceived similarity (Gräve, 2017). Furthermore, human macro-influencers as opposite to human micro-influencers are often seen as a new kind of social media celebrity and are seen as unavailable whilst micro-influencers are perceived as more approachable, like having an online friend (Djafarova et al., 2017). In contrast to macro-influencers, micro-influencers frequently address their followers, by directly replying to the comments, replying to the direct messages as well as sharing their follower’s content, which in turn connotes a certain
familiarity making followers see them as their peers and strengthening the feeling of identification (Erz & Christensen, 2018; Gannon & Prothero, 2018). Moreover, because micro-influencers are perceived as more similar to “ordinary people” than macro-influencers and followers believe more that they can be like them, this perceived similarity might
strengthen both subcomponents of the identification, namely perceived and wishful
identification (Schouten et al., 2020). Therefore, it is expected that identification will have a positive secondary moderation on the primary moderated by influencer size relationship between influencer type and willingness to engage. In sum:
H3: The moderated relationship between the types and sizes of the moderators on the willingness to engage is moderated by the identification.
H3a: When the identification is high, the moderated effect will be stronger.
2.6 Study Overview
The information above is summarized in the conceptual framework shown in Figure 1. This framework presents an overview of the hypotheses discussed along with the
theoretical framework of the study.
3. Method 3.1 Design
The study used survey 2 (influencer size: micro versus macro) x 2 (influencer type:
human versus virtual) between-subjects design, with one dependent variable: users'
willingness to engage, and one moderator variable: identification. The size of the influencers was manipulated based on the number of the influencer’s followers (micro-influencers: <
100,000 followers, macro-influencer: >100,000 followers). The type of influencers was manipulated based on the influencer type that each participant was looking at depending on whether an influencer is a real human being or a computer-generated image, i.e., virtual being. The between-subject design of this study decreases the chances of respondents finding out the research goal, and consequently increases the research validity. Furthermore, the independent variables of different social media influencer profiles were manipulated into four conditions. Where the first two conditions were human macro-influencer and human micro- influencer profile. The second two conditions were a manipulation of the virtual macro- influencer and virtual micro-influencer profiles. After viewing the influencers profile of the assigned types and sizes, respondents were asked subsequent questions. Moreover, the equal distribution and random allocation of each manipulation control for extraneous variables in this study (Hart, Albarracn, Eagly, Brechan, Lindberg & Merrill, 2009). The study aimed to find the effect of a post made by Instagram influencers, based on the different attributes of human influencers and virtual influencers as well as the moderating roles of influencer sizes and identification level.
3.2 Sample
Data was collected through means of a Qualtrics survey (Appendix 1) that was
distributed on Facebook, Instagram, and Reddit during the November and December of 2021.
Within these online environments, certain groups and discussion channels were selected based on their overall community knowledge and awareness of social media influencers. The sample for this study was Instagram users between the ages of 18-44. The sample was selected using a non-probabilistic, self-selection, convenience sampling method. A total of 235 responses were collected. Once the participants who did not finish the whole survey were excluded, the survey resulted in a final sample of 167 respondents. Of these respondents 32%
were male, 65% were female, 3% were non-binary and 0% preferred not to say.
This survey was open to all genders since the topic and content were gender neutral. Table 1 shows the sociodemographic characteristics of the study participants.
Table 1: Sociodemographic Characteristics
Furthermore, all 167 respondents were assigned to one of the four manipulations. In order to ensure that the independent variable has effectively been manipulated and that participants did regard the independent variable in the various ways that it was intended, each participant was asked to indicate the username and the follower count of the influencer that they looked at. This was later checked within the data. The random distribution between the
manipulations and participants is presented in Table 2.
Table 2: Manipulation Distribution
3.3 Procedure
When opening the survey link, respondents were presented with a short introduction about the content, information that their responses will be kept anonymous and only for
research purposes, as well as contact information for any questions they may have. During the questionnaire, respondents were presented with a short text giving the description of a specific type and size of influencer they should look at on Instagram and an example of an influencer with these attributes (Appendix 1, Survey Q3). After seeing both the Instagram profile and Instagram post, control variables, including familiarity and attitude with specific as well as influencers in general were measured. Following, willingness to engage and the feeling of identification with the influencer was checked within the questionnaire. Lastly, general demographic information such as gender, age and education level were asked.
3.4 Measures
One dependent variable (users' willingness to engage), one secondary moderating variable (identification) and six control variables, namely, age, gender, education level, familiarity with the influencer, familiarity with the influencers in general and attitude towards the influencer were measured. All variables were recoded so the most favourable response
“Strongly Agree” was associated with 7 and the least favourable response “Strongly Disagree” was associated with a 1. Additionally, the moderator variable identification was measured in two ways: with a seven-point Likert scale as well as using the inclusion of other in the self (IOS) scale by Aron, Aron and Smollan (1992). Moreover, in this study a
Cronbach alpha of 0.7 is considered acceptable (Griethuijsen, Eijck, Haste, Brok, Skinner, Mansour, N., et al., 2014).
3.4.1 Willingness to Engage
Instagram engagement can be exemplified by liking, privately or publicly sharing as well as commenting on others' posts. Hence, these three actions made up the elements within this construct. In this research, engagement is considered when a user and personal brand have interactive experiences with one another (Hollebeek, Glynn & Brodie, 2014). To measure this variable participant were asked their willingness to engage with different influencers' posts based on a published scale (Lutfeali, Ward, Greene, Arshonsky, Seixas, Dalton, & Bragg, 2020) that consists of a 3-item scale. The items were measured with a seven-point Likert scale, with 1 for “Strongly Disagree” and 7 for “Strongly Agree”. The following items were included: “I would ‘like’ this post on social media”, “I would share this post on my story or to another person”, “I would comment on this post on social media”.
Cronbach’s alpha of the three items was found reliable (α=.836).
3.4.2 Identification
To measure the moderator variable, namely, identification two different measures were used. The first way to measure identification was with five seven-point Likert scale items by Hoffner and Buchanan (2005). Examples of items are as follows: “He/she is the sort of person I want to be like myself”, “Sometimes I wish I could be more like him/her”,
“Sometimes I wish I could be more like him/her”, “He/she is someone I would like to be similar to”. The original measure tests young adults' wishful identification with television characters, but in this study, it will be reworded to test participants’ feelings of identification with different social media influencers.
Additionally, the IOS scale by Aron, Aron, and Smollan (1992) was used to measure the same variable. In this measure, respondents see seven pairs of circles that range from just touching with 1 for "No overlap" to almost completely overlapping with 7 for "Most
overlap". One circle in each pair is labelled “You” and the second circle is labelled “X”.
Respondents are asked to refer to the "X" as the influencer that they looked at. Next,
participants choose one of the seven pairs to answer the question, “Which of the seven pairs of circles best describes your relationship with this influencer?". Overall, the Cronbach’s alpha reliability coefficient of the items of both scales combined was found to be very reliable (α=.945).
3.4.3 Attitude Towards the Influencers (ATTI)
Participants attitude towards the influencer they looked at was measured with four seven-point Likert scale items. The items stem from Lee and Eastin (2020) and are as follows: “I perceive this influencer as interesting”, “I have a positive perception about this influencer”, “I perceive this influencer as favourable” and one of the items was reverse coded and stated, “I perceive this influencer as unpleasant”. Cronbach’s alpha reliability coefficient of this scale was found to be very reliable (α=.926).
3.4.4 Familiarity with the Influencer (FWTI)
Participants familiarity with the influencer they looked at was measured with three seven-point Likert scale items. The items were as follows: “I know this influencer very well”,
“I have been following this influencer”, “I have previous experience with this influencer”.
Cronbach’s alpha of the three items was found to be (α=.764), therefore, this scale is reliable.
3.4.5 Familiarity with Influencers in General (FWIIG)
Participants’ familiarity with social media influencers, in general, was measured with three seven-point Likert scale items. The items are as follows: “I have previous experience with influencers in general”, “I have been following Instagram influencers in general”, “In general my purchase intentions are affected by influencer recommendations”. Cronbach’s alpha reliability coefficient of this scale was found to be reliable (α=.844).
Table 3: Scale Reliability
3.5 Hypothesis Testing
In this research, the first testing of the main hypothesis was done with one-way ANCOVA, and the primary moderating effect of influencer size with two-way ANCOVA.
ANCOVA allows one to understand if differences among means within the sample are statistically significant. Further, the secondary moderator of interaction was tested with the three-way interaction with PROCESS model 3 from Hayes (2018). Moreover, since the interactions can better explain the dependent variable's variance the results of the main hypothesis and the primary moderating effect were finalised based off the results of PROCESS model 3 from Hayes (2018) where the secondary moderation was present.
4.0 Results 4.1. Correlations
To see whether the variables used in the study correlate with each other, Pearson’s correlation was used. The correlation matrix is presented in Table 1. This study makes use of Hair Jr., Wolfinbarger, Ortinau, and Bush (2013) correlation effects, where effect below .4 is weak or too weak, from +/- .41 to +/- .6 moderate, from +/- .61 to +/- .80 strong and from +/-
.81 to +/- 1.00 very strong effect. From this correlation matrix is it evident that willingness to engage has a strong, positive correlation with the level of identification (r = .63, p < .001), moderate correlation with the attitude towards the influencer (ATTI) (r = .44, p < .001), and weak familiarity with the influencer (r = .38, p < .001). Moreover, the secondary moderating variable, i.e., identification has a significant correlation with the control variables, namely, strong positive correlation with the attitude towards the influencer (r = .64, p < .001),
moderate correlation with familiarity with the influencer (FWTI) (r = .58, p < .001), and very weak positive correlation with familiarity with the influencers in general (FWIIG) (r = .17, p
< .05).
Table 4: Means, SD and Correlations
*. Correlation is significant at the 0.05 level (2-tailed)
**. Correlation is significant at the 0.01 level (2 tailed)
4.2 Normality Check
To check if the data was normally distributed a normality check for all variables of this study was conducted. The Kolmogorov-Smirnov and Shapiro-Wilk test of normality indicated that the variables were not normally distributed since p <.05. Nonetheless, there were no outliers identified among the variables and the histograms showed the data to have the approximate shape of a normal curve. In addition, the Q-Q plots of the variables showed the nonnormally distributed data did not look dispersed along the line, meaning that all variables of this study either matched the line or were close by the line. Hence, based off the fact that the histograms showed approximately normal curves, the Q-Q plots did not show
significant deviation from the lines and the box plots were close to symmetrical, the decision to proceed with the use of parametric tests was made.
4.3 Hypothesis Results
4.3.1 Influencer Type and Willingness to Engage
Hypothesis 1 stated that users’ willingness to engage will be higher with human influencers than with virtual influencers. For the first test a one-way ANCOVA was used to the differences between the two types of influencers, namely, human, and virtual. The test indicated that human influencers (M = 3.64, SD = 1.51) and virtual influencers (M = 2.14, SD
= 1.09) have a significant difference on the users’ willingness to engage (F (1,158) = 20.264, p = .000). However, to see if the results remained significant after a three-way interaction of size and identification was present, the final effects were analysed using PROCESS model 3.
The results from three-way interaction, however, showed that the main effect of influencer type on users’ willingness to engage are not statistically significant (b = -.2661, se = .3013, t
=-.8834, p >.10, 95% CI = [-.8614, .3291]). Therefore, based on this hypothesis 1 is rejected.
4.3.2 Primary Moderation of Influencer Size
Hypothesis 2 stated that users' willingness to engage will be higher with human influencers than virtual influencers and this will be moderated by the size such that users will be more willing to engage with human micro-influencers than with human macro-influencers but less willing to engage with virtual micro-influencers than virtual macro-influencers. First, a two-way ANCOVA was used to test this hypothesis. The ANCOVA test showed that the interaction between influencer type and size is statistically significant (F (1,1558) = 17.837, p
= .000) with micro human influencer (M = 4.56, SD = 1.29), macro human influencer (M=
2.89, SD = 1.21) and micro virtual influencer (M = 1.81, SD = .94) macro virtual influencer (M=2.46, SD = 2.46), providing initial support. However, to see if the results remained significant after three-way interaction was present, the interaction of influencer size was tested using PROCESS model 3 from Hayes (2018). The results with secondary interaction, however, did not support this hypothesis, as they showed that there was no significant interaction effect of size (b =1.1852, se = .6058, t=1.9562, p>.05, 95% CI = [-.0119, 2.3822]). Therefore, since the three way interaction results showed that the results are not statistically significant hypothesis 2 is rejected.
4.3.3 Secondary Moderation Effect of Identification
Hypothesis 3 stated that the moderated relationship between the influencer types and sizes of the moderators on the willingness to engage is moderated by the identification such that when the identification is high, the moderated effect will be stronger. PROCESS model 3 from Hayes (2018) was used to test this hypothesis. Results supported this hypothesis, as they showed that there is a significant interaction effect of identification (b = .9003, se = .3324, t
=2.7087, p <.05, 95% CI = [.2434, 1.5570]). Therefore, hypothesis 3 is supported. These results are shown in Figure 2.
Figure 2: Summary of Hypotheses Results
5.0 Discussion
In this section the overall results and findings of the study will be discussed.
Furthermore, theoretical, and managerial implications will be indicated. This will be followed up by the limitations of the current study and recommendations for the future research.
Lastly, general conclusion will be presented.
5.1 Findings
The aim of this research was to find the influence of social media influencer type, namely, virtual versus human, on users’ willingness to engage. While additionally testing if
positively moderates this moderated relationship. This was answered through an online survey, where three hypotheses were tested. A summary of the hypotheses is presented in Figure 3.
In the first hypothesis, this study expected that users ‘willingness to engage will be higher with human influencers compared to virtual influencers. Contrary to the expectations, when the three-way interaction including identification was present, there was no significant effect found between different types of an influencer and users’ willingness to engage. The expectations of this study were based off previous studies that suggested that Instagram users are generally more willing to engage with influencers that they perceive as authentic and relatable (Arriagada et al., 2021). Since previous literature has suggested that virtual influencers are generally not perceived as authentic and credible the measure of users’
willingness to engage with virtual influencers was expected to be significantly lower. In addition, the expectations of this study were based off the findings that users are generally more willing to engage with a content that resonates with them and gives enough
transparency about the personal brand (Moustakas et al., 2020). Despite the previous
literature on this matter, this was shown to not be the case within the current research. One of the possible reasons for this could be attributed to the fact that quite a lot of participants of this research sample were not familiar with the influencers that they looked at. According to Thomas and Fowler (2015) existing familiarity, long-term online relationships, and perceived closeness play important role towards the attitude and behaviour with brand endorser.
Thereby, it is possible that the participants of this sample did not display the same behaviour they would have when interacting with influencers that they have known for a longer time.
From this perspective it could be argued that users demonstrate different willingness to engage with human and virtual influencers only when they have established a longer-term online relationship with an influencer. This is in accordance with findings by Copeland, Lyu and Han (2021), which showed that general attitude and behaviour towards the brand was more beneficial in terms of online interactions when individuals were already familiar with the personal brand. In the current study, however, 40% of participants showed that they were not following or did not previously know the influencer they looked at, indicating that they have not established a longer-term relationship and potentially felt indifferent towards them.
Therefore, the findings indicate that the effects might differ in a real-life situation when participants willingness to engage is measured with influencers that they have already established a better knowledge about. With this line of reasoning, the influencer type might
be not such strong factor as previously expected than users’ actual familiarity and already established feeling towards an influencer.
Furthermore, this study sought to determine the primary moderating effect of influencer size, i.e., micro and macro. The findings, however, show that micro-influencers and macro-influencers did not have a significant moderating effect on the relationship between human and virtual influencers and users’ willingness to engage. This finding is contradictory to previous literature which showed that generally, when it comes to human influencers, Instagram users find micro-influencers more relatable and approachable and hence are more willing to interact with them than with human macro-influencers (Djafarova et al., 2017). Moreover, regarding virtual influencers, some studies have suggested that users may be engaging with them mostly because of the novelty effect and consequently the opposite effect of influencer size was expected in this study. In other words, it was expected that users’ willingness to engage will be lower with micro virtual influencers than with macro virtual influencers. Nonetheless, there was no significant difference found in this study in regard to micro and macro influencers on the relationship between influencer types and users’
willingness to engage. This might be due to the fact that the participants were not highly familiar with the influencers that they looked at and as previously mentioned close relationships with an influencer is what promotes their willingness to interact through commenting, liking, or sharing their posts (Tiba, 2020). Therefore, the measures of different types and sizes of influencers were inaccurate to measure in a survey suggesting that the effect may be different in a real-life setting where participants demonstrate their willingness to engage with an influencer they have been following or have known for a longer time.
Regarding the level of identification with an influencer, this study expected that when users have a high level of identification with an influencer the moderated relationship by size to the relationship between the influencer type and users’ willingness to engage will be strengthened. The findings show that secondary moderator, namely, identification
significantly moderates the relationship between influencer type and users’ willingness to engage and primary moderation of influencer size. This is in line with previous literature that shows that higher identification plays an important role in brand advertisement effectiveness (Schouten et al., 2020). While also this adds to literature that the level of identification has a significant effect interacting with such factors as different influencer types and sizes on the users’ willingness to engage.
Although the main effect and primary interaction of influencer size, showed
way interaction was present, showing that the interaction of identification better explained the variance of the willingness to engage. To sum up, the study found that there was no
significant effect of influencer type on users’ willingness to engage, and influencer size did not significantly moderate this relationship. Nonetheless, the study found that the level of identification has a significant moderating effect on this moderated relationship. Possible explanations for the insignificant results will be discussed in the limitations and future research section.
Figure 3: Summary Hypotheses
5.2 Practical and Theoretical Implications
As previously mentioned, the digital advancements and changing marketing tools like social media influencers provide many new opportunities for marketers in terms of different types and sizes of influencers to use as the sources to promote a brand. However, to make the best selection, it is imperative to learn more about what factors beyond previously researched characteristics like influencer size might have different effects on users’ online behaviour.
This study found that when the higher order moderator, namely identification, is involved, then willingness to engage with human micro influencers is significantly higher than when identification is not present. This showed that the strength of two-way interaction between influencer size on the relationship between influencer type and users’ willingness to engage depends on users’ level of identification. Despite the predicted separate effects of influencer types and moderated effect of influencer sizes on the willingness to engage, the
study found that the effect holds to be significant only when identification moderation is present within the moderated relationship. This implies that marketers should look further into the reasons and other characteristics beyond influencer types and sizes that play a crucial role in users’ attitude and online behaviour with social media influencers. Moreover, this indicates that more personal feelings towards influencers have a significant effect on relationships with different types and sizes of influencers. Because the feeling of
identification both in terms of perceived similarity as well as wishful identification play important role, it is crucial to understand the causes and other characteristics beyond influencer size and type that promote the feeling of identification towards an influencer.
Furthermore, it is essential for marketers to learn what are the influencers that users generally show to feel highly identified with.
Moreover, with the rise of virtual influencers and general development of digital identities on social media platforms, it is essential for marketers to learn how virtual influencers differ from human influencers and which influencer type is more suitable to deliver the desired brand message as well as succeed with their marketing strategy. Although this study did not show the significant difference between users’ willingness to engage between these different types, more research on this matter is necessary to understand human and virtual influencers differing effect on online users’ behaviour as well as interaction of different influencer sizes.
5.3 Limitations and Future Research
The current study gives some more guidance for future research within influencer marketing domain. Nonetheless, there are some limitations to consider. The first limitation of this study is the small sample size. Whereas the present sample size of 167 satisfies size guidelines and is considered adequate, the overall consequences of small sample size is a reduced power of statistical analyses and limited generalizability. Moreover, regarding the sample of this study, most of the participants were highly educated with 92% of the overall sample having obtained or enrolled in a bachelor’s degree or a higher-level degree. This means that the sample is not big and diverse enough to be representative of the population.
Thus, to increase the internal validity and generalize the results, further research with bigger sample and more diverse group of participants is needed. Nonetheless, despite the small sample size, the quality of the obtained data is notable.
Second limitation of the present study relates to the fact that the methods used to measure the relationships use of a single questionnaire to collect data from all sets of participants, and hence increase the possibility of common method variance (CMV) bias creating a false internal consistency (Fuller, Simmering, Atinc, Atinc, & Babin, 2016).
Therefore, to avoid or minimize the presence of CMV bias, it is suggested for future research to employ different data sources. In line with this, the current study has a low validity which is a general problem of online surveys. Hence, future studies are encouraged to investigate the effects using different study methods that do not distort the correlation between the variables.
Third, the current study used a cross-sectional design to collect data, implying that causal inferences from these results cannot be drawn. The primary limitation of cross- sectional study is that the measures of the study may have reciprocal relations meaning that there is no explanation of why certain event occurs within the sample. Therefore, in order to address causality, it is suggested that further studies employ experimental or longitudinal designs.
As previously mentioned, the current study showed that 40% of the sample were familiar with influencer that they looked at. Since high familiarity as well as lack of it might create different results, it would be interesting to look at how these separated groups of individuals, i.e., those who are already familiar versus those who are unfamiliar with an influencer, can differently affect the outcomes. According to Copeland et al. (2021) familiarity with influencers or any kind of brand ambassadors is a powerful predictor of consumers’ behavioural responses including consumer loyalty and parasocial interaction.
Further, taking the volatile and visual nature of Instagram posts, it would be great to study how one specific post of each type and size of influencer that is visually similar amongst them all, affects users’ behaviour and perception about these different types and sizes of influencers. Previous studies have shown that different content characteristics in social media platforms affect users’ perceptions and evaluations, which consequently
determine user’s behaviour like commenting and liking (Casaló, Flavián, & Ibáñez-Sánchez, 2017).
Lastly, even though Instagram is currently the most prominent and widely used platform among different social media influencers, it would be interesting to investigate how the effects on outcomes might vary on different social media platforms such as TikTok and YouTube. Previous study by Kapitan and Silveira (2016) showed that different social media platforms have different outcomes because they all have their own unique cultures. Hence,
the research on different platforms would be beneficial to understand how the effects differ among more visual and audio-based content as well as vary among different user cultures.
6.0 Conclusion
Influencer marketing has become a mainstream form of social media marketing and it is safe to say that it is not going away anytime soon. Therefore, it is essential for marketers to keep up to date on the newest advancements and to be strategic in selecting the most suitable influencers for a brand. This study examined whether virtual and human influencers elicit different users’ willingness to engage with their posts. On top of that, the primary moderating role of influencer size and secondary moderating role of user’s level of identification was investigated. The main contribution of this study is that effect of influencer types on users’
willingness to engage and moderating effect of influencer sizes do not have as strong effects as previously expected but the secondary moderating role of identification does, however, significantly affect this moderated relationship. Thus, the study presents that other
characteristic that are more related to users established feelings might be what make the significant difference in users’ behaviours. Therefore, this study opens a discussion on influencer qualities that evoke such users’ feelings as identification towards an influencer.
References
Abidin, C., & Ots, M. (2015). The Influencer’s dilemma: The shaping of new brand professions between credibility and commerce. In AEJMC 2015, Annual Conference, San Fransisco, CA, August 6-9. Retrieved from: http://urn.kb.se/resolve?urn=urn:nbn:se:hj:diva-27812
Alassani, R., & Göretz, J. (2019, July). Product placements by micro and macro influencers on Instagram. In International conference on human-computer interaction (pp. 251-267).
Springer, Cham. DOI: https://doi.org/10.1007/978-3-030-21905-5_20
Arenas-Márquez, F. J., Martínez-Torres, M. R., & Toral, S. L. (2021). How can trustworthy influencers be identified in electronic word-of-mouth communities?. Technological Forecasting and Social Change, 166, 120596. DOI: https://doi.org/10.1016/j.techfore.2021.120596
Argyris, Y. A., Wang, Z., Kim, Y., & Yin, Z. (2020). The effects of visual congruence on increasing consumers’ brand engagement: An empirical investigation of influencer marketing on Instagram using deep-learning algorithms for automatic image classification. Computers in Human Behavior, 112, 106443. DOI: https://doi.org/10.1016/j.chb.2020.106443
Armenteras, J., Paech, R., & Politiek, I. (2017). Influencer Marketing's Effects on B2C Brand Image: A Case Study of Nike. LBMG Strategic Brand Management-Masters Paper Series.
Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion of other in the self scale and the
structure of interpersonal closeness. Journal of personality and social psychology, 63(4), 596.
DOI: https://doi.org/10.1037/0022-3514.63.4.596
Arora, A., Bansal, S., Kandpal, C., Aswani, R., & Dwivedi, Y. (2019). Measuring social media influencer index-insights from Facebook, Twitter and Instagram. Journal of Retailing and Consumer Services, 49, 86-101. DOI: https://doi.org/10.1016/j.jretconser.2019.03.012
Arriagada, A., & Bishop, S. (2021). Between Commerciality and Authenticity: The Imaginary of Social Media Influencers in the Platform Economy. Communication, Culture and Critique, 14(4), 568-586. DOI: https://doi.org/10.1093/ccc/tcab050
