Master thesis-Marketing track
“The different effect of general social media influencers and employees’
reviews on customer user engagement”
Student name: Xin Xu Student number: 11392444 Due date: 2017/6/23 Version: Final version
Qualification: MSc. in Business Administration – Marketing track Institution: ABS, UvA
Thesis supervisor: Meg Lee Word count: 13491
Statement of originality
This document is written by student Xin Xu who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
1. Abstract... 4
2. Introduction... 5
3. Literature review...11
3.1 Social media development and influencer marketing...11
3.1.1 General influencer...12
3.1.2 User Engagement... 16
3.1.3 Employee as influencer...18
3.2 Creative message strategy...22
3.2.1 Abstract/ concrete message style...23
3.2.2 Product-related/brand-related message content... 26
3.3 Conceptual framework... 32
4. Methodology...32
5. Results and analysis... 45
5.1 Preliminary analysis... 46
5.2 Main analysis... 48
6. Discussion and conclusion...59
6.1 Overall conclusion...59
6.2 Theoretical implications...64
6.3 Managerial implications... 64
7. Limitations and future research...66
7.1 Limitation... 66
7.2 Future research... 68
8. References...69
1. Abstract
Although much has been researched about influencer marketing in recent years, and social media influencer has been investigated for advertising effectiveness on social media, whereas influencer marketing is still in a narrow perspective in general. There’s employee that could also be a new influencer group when thinking about influencer marketing. Harris and de Chernatony (2001) said in their research that employees are perceived to act as the brand’s ambassadors by communicating brand’s values and delivering product information in the way they behave and have interaction with customers. Interestingly, if using the employees from the brand as marketing influencers, on the platforms of social media websites, the influence employees make on customers’ user engagement would be meaningful to study on compared to general social media influencers. If there’s such impact, it would contribute to the academics of influencer marketing field.
As what has been researched before, there are many factors interacting with customer user engagement and influencer groups. Message placement strategies would be critical to focus on as well in the research. Some scholars have demonstrated that message style – abstract/concrete style and the message content – product-related content and brand-related content have interaction with influencer groups (Liberman et al., 2007, Kim, et al. 2016). This study would include these factors as moderators to the interaction over reviewer groups.
This study conducted a 2 (reviewer groups – employee and general influencer) × 2 (message style – abstract/ concrete) × 2 (message content – product-related/brand-related) online experiment (n=241). The manipulation has been checked and yielded different results. A key finding is that using employees as social media influencers would have the effect as general social media influencers on customers’ user engagement. And message content has moderating effect with the reviewer groups in that employees are suitable to deliver product-related message while general influencers are suitable to deliver brand-related message. While there’s no interaction effect between message style (abstract/concrete style) and reviewer groups. Some researcher have already found support for these results. For future research, there should be scale on employee’s authenticity and actual impact of well-known social media influencers (similar to celebrities).
Key words: influencer marketing, employee reviewer, user engagement, social media use.
2. Introduction
With the fast development of social media technology and the daily use of the social media applications, people’s life has been changed by the ways they communicate with each other. Social media networking sites have become increasingly popular across the world. And it has changed the landscape of marketing as well. The comments and reviews play an important role in influencing others’
decisions. Everyone on the social media can be the spokesman of the brand and say something about the brand, especially the “influencers”—who have the power of changing people’s perceived value of the product and then influencing their decision making process (Bakshy et al. 2011). In Lu et al. (2014)’ research, they found that when products are recommended in blog posts, it is likely that consumers would have positive attitudes toward sponsored recommendation posts, which improves purchase intention. We can see that people’s comments and reviews are of great importance in customers’ purchase intention and behavior.
Another group included in this thesis is the employee. Employees could also be regarded as influencers of the brand. Aula and Mantere (2008) said that every employee is a reputation ambassador or someone who can destroy the reputation of the company. In light of the special position of employee for the brand, what about setting employees as brand influencers? Will it be different from the general influencers? At present, there’s little literature introducing employee as influencers and thus no comparison with the general influencers. Here in this thesis, the comparison between general influencers and employee influencers will be made to see the difference on customers’ user engagement.
When studying the effect of vividness of a message when persuading consumers, Royes, Thompson et al. (1980) found that a concrete message induced greater influence than an abstract message. The message style—whether concrete or abstract, can be assumed that it would make an effect on customers’ user engagement.
Concrete information facilitates access to script. Information that calls to mind a particular script will be assimilated to the script and subsequent judgments and actions will be guided by the contents of the script (Borgida and Nisbett, 1977). And information that is concrete and vivid would have much easier access to these scripts than by information that is pallid and remote (Nisbett et al. 1976). Abstract and concrete message would have different result in message communicating process.
On the other hand, there’s also difference between brand-related message and product-related message when communicating with customers. Smith et al. (2012) have compared the difference of brand-related user-generated content among a few most frequently-used social media platforms – Youtube, Facebook and Twitter. And they found that UGC was fostered by different social media websites with different characteristics. Product-related message differs from brand-related message in various contexts. Product-related message contains information that is easy to recall when seeing or using the product while brand-related message is too broad and difficult to feel (Kim, et al. 2009). From this, it can be assumed that product-related message would have different effect on customers’ user engagement compared to brand-related message. Message placement control helps to ensure that the intended audience is exposed to the message with sufficient frequency to exceed some threshold for effectiveness. And it also allows for the optimal timing and placement of those messages. Message placement strategy can be a tool used for marketers to attract target audience’s attention.
Despite the important managerial implications of the research above on customers, however, there’s few research about the comparison between employee and general influencer on customers’ user engagement at present since employees are not used as influencers in marketing. Research gap in this study is that although there are already many researches on employees’ performance and management field, most of them are in human resource management field, seldom of researchers would take employees into account in marketing field. With the widespread use of social media, and employee as user of social media as well, it is reasonable to expect that what if use employees as influencers for the company. And there’s no such hypothesis and research before, which could fill the gap of research in this area. In social media field, there have already been many studies about social media influencers’ impact on opinion leading of customers, and many of the studies are in the field of empirical research. It makes sense to see the difference between using employee as influencer and general social media influencer in marketing field. There are three reasons to compare these two types of people on marketing: first, they are both paid by the company and speak for the company, but the difference is that employees are “natural” speakers of company while extra money is used for general influencers. They work for the company while they have different identity. Second, the two types of people would have different impact on customers’ user engagement, which would have important managerial implications for marketers. Third, the result of the comparison would make company running effectively especially on cost-saving. Current
knowledge on the field is still unable to answer the different impact between employee and general influencer on customers, with the moderator effect under different context. From this research, marketers would know which group works better in delivering message to customers as well as the message style and message content. This thesis aims to add knowledge to social media advertising effectiveness literature by researching the following question:
“What’s the different effect of using employees as influencers compared to general social media influencers in delivering message on customer user engagement?”
The focus of this research will be the key difference between the two variables – employee and general influencer on customers’ user engagement and consequently have managerial implications for marketers. And the key difference from prior research will be the comparison between these two variables: prior researches mainly focus on employees and general influencers separately, yet no comparison of these two. The research will collect data in both the general influencers and employees in order to (1) compare the potential difference and (2) to provide insights that enable generalization.
Given the increasing WoM and frequently use of social media, the main research objectives are:
1. To identify the difference of employee and general influencer on customers’ user engagement based on different moderator effect
(abstract/concrete message style, brand-related/product-related message content).
2. Explain the managerial implications based on the results of the research. The result of the comparison will have important managerial implications for marketers in the way they develop their communication strategies. From the reviewer group, if employees work better than general social media influencers, it makes sense to not hire general social media influencers in order to reduce cost. The message style (abstract/concrete) and message content (product-related/brand-related) have different effect in different contexts, which would also provide some implications for marketers to decide what strategy to use when delivering information to customers.
In order to attend the research objectives, the research will be structured in 6 main sections. The methodology will be introduced, and analysis and results, conclusion, implications and limitations. The literature review covers the theoretical background of key concepts, leading towards the research gap which is addressed in the conceptual framework chapter, as well an overview of the hypotheses. Methodology and data analysis will explain how the research is set up, conducted and analyzed. Conclusion chapter will draw conclusions from this research and have some implications.
3. Literature review
3.1 Social media development and influencer marketing
With the fast social media technology development, social networking sites and social media have become increasingly popular across the world in recent 10 years. The use of Web 2.0 technologies makes the Internet applications easy to reach and use. Social media has already become an essential part in people’s daily life. From the survey by Madden and Zickuhr (2011), it is reported that 65% of the adult Internet users in USA use social media websites and mobile apps, and the percentage was more than double to the result that was reported in 2008 (29%). And Facebook is reported to have more than 2 billions active users (as of 2012) since its beginning in 2004 (www.facebook.com). On average, consumers spend almost one third of their spare time on online social media websites and apps (Lang, 2010). Most of them are everyday users and they log on the social media websites more than once a day. With the number of social media users growing at a geometric rate, marketing managers should realize and understand the condition, and then response properly.
Social media has also changed the traditional way of people’s communication with each other. People now can share photos and communicate with any other people around the world without seeing or even knowing each other. It is proposed that social media have been explicitly designed for relationship building purpose (Fournier and Avery, 2011). People are connected as a network within social media and
communication patterns have changed (Naaman, Boase and Lai, 2010). Because of the use of social media, social interaction ways are now mainly on the Internet. People use WhatsApp, Wechat, etc. to contact with others, use Instagram, Facebook and Twitter to share recent story of them. Given this current condition, marketers have realized the power of social media on marketing. Social media marketing is a kind of marketing track that use social media platforms and websites as promotion tools to generalize a product or service to customers (Felix, Rauschnabel and Hinsch, 2017). In order to use social media more effectively, most firms encourage customers and Internet users to post user-generated content (UGC). UGC is different from marketer-prepared advertising copy (Drell, 2011), it is any forms and kinds of information generated by users of a system and this kind of information is made available to the public on that system (Beal and Vangie, 2017). Influencer marketing is part of social media marketing, and the content influencers post is included in UGC as well. In this research, general influencers and employee influencers will be introduced and comparison will be made between these two groups.
3.1.1 General influencer
Usually, before people buy something or make decisions, they would like to listen to others’ opinions, experience and suggestions. What other people say about the product or brand often affects the former in people’s decision making, and are aptly termed as the influential (Agarwal et al, 2008). This kind of peer effect is empirically exclusive in the social science. With the fast development of technology
and easy use of social media applications, people feel much easier to express their opinions on social websites. Then, the contents shared on social media websites are connecting to a virtual community and the people who share the information, and lead the opinion direction become “influencers”. Freberg et al. (2011) said in the research that social media influencers are those who develop and change the attitudes of users through blogs, tweets, and other social media Internet websites. Bakshy et al. (2006) proposed that everyone can become an influencer. Social media influencers are those who exhibit some combination of desirable attributes—whether personal attributes like credibility, expertise, or enthusiasm, or network attributes such as connectivity or centrality—that allows them to influence a disproportionately large number of others (Gladwell, 2006). These people are often market-movers. They can have the influence over buying decisions of their followers. In Agarwal, Liu and Tang (2011)’s research, they also defined that influential bloggers are those who have at least one influential blog post and the followers are influenced by the secondhand information they post. Social media influencers are now representing a new category of independent third party endorsers. They play the role of potential buyers themselves but they also provide secondhand information for customers. They may exist either in the so-called value-added people or in the supply chain (Brown and Hayes, 2008). The social media influencers (SMIs) have the power to promote a brand or organization. They are seen by people as trustworthy and non-purposive because they are users, too. Freberg et al. (2011) also proposed “SMI capital” content in their paper, and they
found that individual SMIs have more influence on people’s opinions than social media advertisements because customers feel that they are not being tricked into thinking the ways the selling companies want them to. Agarwal (2008) said in the research that these influencers can have great help in customer support and troubleshooting since the solutions from these influencers are usually regarded as more trustworthy because of the sense of authority they naturally own. Also, they found that social media influencers were highly correlated with each other. They share the information mutually and usually they have a network themselves.
And how to identify these social media influencers? Most prior researches to identify SMIs depend on factors such as the number of times a post is shared, the quantity of daily hits on a blog or twitter, or number of followers the blogger/Instagrammer owns. Basille (2009) and Straley (2010) said that regarding the cognition that online influence is mainly about quality, not quantity, these standards can only be viewed as a starting point. In Elihu Katz (1957)’ s article, he pointed out three factors to identify opinion leaders: expression of values, professional competence and nature of their social network. In modern society, social media websites provide them with more intimate social network, the communication becomes much more easier. Freberg et al. (2011) added another new dimension to identify social media influencers – subjective audience impressions and perceptions. Participants in the research viewed SMIs as often verbal, smart, ambitious, productive, and poised.
Now, there have been approximately 64% advertising companies have realized this trend on marketing and as a result, influencer marketing exists. Influencer marketing is a kind of marketing in which the focus is mostly gathered on specific key individuals and customers rather than the whole target market. Its goal is to influence this kind of opinion leaders first, usually pay for them, and then promote and expand the products to more customers with the recommendation of these influencers.
John Hall (2016) divided the social media influencers into two types: earned influencers and paid influencers. Earned influencers are those who share the secondhand information to the public in order to further their own personal social growth. Paid influencers are those who get sponsorship from companies and brands and act endorsement for them. Budgets usually vary widely and are often based on audience reach. In this research, the focus group is the latter one since it is more suitable with another comparison group – they are both paid by the company or brand. In this research, general social media influencers would specifically refer to the paid influencers, which is comparable with employee group in that they are both paid by the company.
In general, influencers would be those whose posts are shared many times, who are active on social media, usually with a large number of followers, and those who are general opinion leaders that can influence people’s opinions, and those who are specially paid by the brand or company.
3.1.2 User Engagement
In the context of social media world, user engagement is usually used to evaluate the influence of influencer marketing (Kim, et al. 2013). Online user engagement means that social media users are active participants on the Internet rather than passive viewers. It is conceptually defined as a series of behavior which has no any intentional mindset like control, attention, focus, curiosity and/or intrinsic interest (Kim, et al. 2013). It is a kind of state characterized by energy and involvement. In the contexts of organizational and individual levels, there’s a strong relationship between customers’ user engagement and company’s financial profitability through customer satisfaction, increased sales, employee retention and productivity (Harter, et al. 2002). The use of social media allows all the users, especially the customers in business to express and share comments and reviews of a product or service. Each customer participating in the process becomes part of the social media system, as the viewers and other customers interested in the product or service read the positive or negative comments or reviews. It is a fundamental step to get existing customers and potential customers engaged online in order to succeed in social media marketing (Evans, 2010). When firms increase customers’ user engagement and build a possible environment that would contribute to foster user engagement, the chances of success in their business will increase significantly (Harter, et al. 2002). Customers’ user engagement and customer satisfaction can lead to future loyal users (Oliver, 1980). It becomes increasingly important to get more
customers become interested in the product or service, since customers’ user engagement would eventually be transformed into purchase intention and purchase behavior. Social media user engagement has emerged into customer loyalty in the new context of marketing, which both aim to build customer participation and brand loyalty.
Various definitions of user engagement have translated different aspects of what user engagement is. O’Brien and Toms (2008) proposed in their research that user engagement is a new kind of user experience featured in attitudes of challenges, positive effects, attention, endurability, aesthetic and sensory appeal, variety/novelty, interactivity, feedback, and perceived user control. Fundamentally, prior research indicates that user engagement consists of users’ activities, attitudes, goals and mental models, and motor skills, and that it is manifested in the form of intrinsic interest, attention, curiosity and motivation (O’Brien and Toms, 2008). And some of previous studies also indicate that user engagement consists of system feedback, user control, attention, motivation, and the ability of the system to challenge users at extent appropriate to their knowledge and skills. Yu, Aoki and Woodruff (2004) said in their research that user engagement measures the level of commitment with customer interaction. It describes the degree of a participant’s interest in the conversation and the intrinsic attention to the conversation. In this research, we mainly focus on online user engagement. According to Forrester (2008), online user engagement is characterized as constructing close connections and relationships with customers that
drive customers’ purchase intention and decisions, interaction, and participation over time. It is a kind of rich associations formed between the brand and customers, and eventually would translated into purchase behavior and brand loyalty. It is a kind of proactive behavior social media users usually has which in co-creating the individual experiences and perceived value of themselves with companies through active, ongoing, explicit, and continuous interactions.
From the definition of user engagement, it is obvious that user engagement is critical for the success of brands and companies in the social media connected society. Having direct communication with customers enable them to be involved in the system and marketing process, giving them control in exchanging their information, and to increase brand loyalty. It increases the success rate of promoting products or services, thus enhancing the awareness and loyalty of the brand and company.
3.1.3 Employee as influencer
As an important part of the company, employees act the role of brand ambassador of the company. In marketing, it is critical to associate customers with the company since the more intimate the relationship between the brand and customers, the more likely customers would recall and recognize the brand when purchase behavior happens. Then the brand ambassador is the key to relate customers with the brand. Brand ambassadors represent the brand and help to increase brand awareness and promote sales (Brand Ambassador World, 2015). Then, employees are important roles in communicating the brand message to customers as brand ambassador. Harris
and de Chernatony (2001) said in the research that employees are perceived to act as brand ambassadors through communicating the corporate values and brand’s products in the way they behave and have interaction with its customers. Employees are regarded by customers that they are literally “living the brand” (Gotsi and Wilson, 2001). The recent literature has advocated that the employee as a brand ambassador can have great help in building stronger brands. And employees can influence the brand image a customer makes with the corporate brand. Employee is the symbol of the company that would have an effect on corporate brand associations (Gary and Rosa, 2010). Employees are the public face of the company or brand, and their performance and behavior is another communication path with the customers for the brand (Wallace and Chernatony, 2009). Furthermore, Zeithaml et al. supported that employees “are” the brand. Employee’s performance can bring the light for the customers, and thereby customer loyalty and customer retention are created. Then over time, customers can build close and strong relationships with employees, which would enhance trust and loyalty to the brand for future development (Zeithaml et al. 2006).
Fast development of social media technology has caused user-generated content (UGC) blow up on the Internet. Social media platforms like Facebook, Twitter, etc. allow everyone to post any information of their daily life freely. Employees, as part of the users on social media, also gain the power to express themselves on the Internet, especially exposing the content of the company customers are interested in. Usually,
employees have their own personal social media accounts, it is more likely that public would perceive the information post on their personal accounts to be more trustworthy and authentic compared with the information provided by the brand (Snyder and Honig, 2016). Yan, Param and Anindya (2015) have analyzed social-media content creation and consumption behavior of employees within an enterprise. And they found that different blog content shared by employees have different impact for readers. Work-related content, which is often brand-related or product-related content, the utility consumers derive from is higher than what they derive from leisure-related content blogging. And the information customers get from this work-related content have a positive effect on customers’ perceived value of the products since employees’ level of trust is assumed to be higher than the source of the brand (Snyder and Honig, 2016). What also needs to be considered is that trust affects the level of commitment in a positive way (Morgan and Hunt, 1994). Therefore, it can be assumed that employees’ use of social media is positively related to the brand’s social connection with the customers and the level of trust would enhance in this way. Employees use social media to increase the exposure of the brand to public, and it has been shown that these exposures on social media by the brand’s employees can have a huge impact on company’s avenue. These socially engaged companies are 57% more likely to get more sales leads according to the National Business Research Institute.
So, what about using employees as influencers of the brand? Empowering employees to “live the brand” through social media platforms not only helps extend
the reach of the core message, but it also allows employees to deliver superior service that results in WoM endorsements, customer loyalty and customer retention (Paula, 2008). At present, there are rare researches and applications of using employees as brand’s social media influencers, this research is going to figure out the influence on customers’ user engagement when using employee as influencer and then compare employee influencer with general influencers.
Much of the recent researches have centered on the effectiveness of source credibility when communicating messages to customers. When involvement is low, peripheral cues such as source credibility may become a major determinant of how customers perceive the message or advertisement (Buda and Zhang, 2000). Employees and general social media influencers stand for two different groups of information sources for customers. General social media influencers are usually from ordinary social media users and employees are from the brand or company itself. As companies and brands are increasingly concerning about online marketing, they also are more often facing a certain of hostility on social media as in customers’ eyes, most brands seem to be inauthentic, and their presence is regarded as intrusive and out of place (Fournier and Avery, 2011). Comparing general influencer and employee influencers, employee influencers are seen more trustworthy and authentic because it is assumable that they know more about the brand and the information they provide seems more authentic and trustworthy (Fullerton, 2003). And they use their personal social media accounts, which in result that customers would perceive this behavior
more as interpersonal rather than organizational by the brand. While general influencers are usually paid by the brand and their social media accounts seem to be more “official” than employee influencers. From the above, the hypothesis is that:
Hypothesis 1: Employee as an influencer of the brand would have a positive
influence on customers’ user engagement.
H1a: Employee as influencer would have a more positive influence on
customers’ user engagement than general social media influencers. 3.2 Creative message strategy
At the same time, some scholars proposed using message strategies to appeal to different audience groups besides using the specific person groups. Product messages are usually copious in today’s advertising market. Customers are encountering tens of thousands of advertisements everyday, information explodes cause their inaccurate judgment and consequent improper purchase decisions. Many marketing and advertising researches have studied the ways to enhance the effectiveness of advertisements and messages communicated to customers. Some researches have agreed that the form of information presented to customers has an effect on customers’ processing of such messages. In other words, the way how information is structured or framed may influence consumers’ judgment, perceived value and purchase intention about the products or service (Ganzach and Karsahi, 1995). Lee, Kim and Yoo (2015) proposed in their research that message framing types—abstract and concrete message have effect on message effectiveness. Other scholars, like Hastak
and Olson (1989), have proposed that customers’ cognitive responses to brand-related and product-related message are different, and the research they did suggested that brand-related message would lead to brand cognitive response, and brand-related message is the primary moderator of ad effects on customers’ cognition. Here in this research, it will mainly focus on these two types of message framing types as moderators: abstract/concrete message style and product-related/brand-related message content.
3.2.1 Abstract/ concrete message style
When communicating message, abstract and concrete forms are usually used in the information delivering process. Abstract message is message describes an object with no physical referents, and concrete message is message describes an object with physical referents. According to Construal Level Theory (CLT), the greater a person’s psychological distance from an object or event, the greater possibility that the person conceptualizes the object or event in an abstract rather than a concrete manner (Liberman et al., 2007). So when customers are involved in low user engagement, message presented in abstract form would be more effective than concrete form. And vise versa. In previous research, some scholars had demonstrated that vivid and concrete information can have a greater impact than abstract information on individual’s memory, initial impressions, and social acknowledge (Hamill, et al. 1980; Reyes et al. 1980). And concrete information is generally easier to recall or more
relative ease of generating general causal explanations or scenarios for each – the causing process. For example, concrete data, such as case histories, contain rich details that may make it easier to create explanations or more likely to generate explanations. When it is based on concrete information even though it’s weak, people may preserve more in their initial perception than when based on abstract information even it is strong (Anderson, 1983). In this case, concrete message is more effective in communicating to recipients than abstract message.
Also, the information these two types of message presenting is different and the inducement of mindset construal level is different. Abstract and concrete words have different reliance on associative and similarity-based information. Abstract words always more rely on association than similarity, and concrete words vice versa (Crutch, 2006). The more abstract a word, the greater it would depend on associative information and the smaller possibility it would rely upon similarity-base information (Crutch and Warrington, 2010). So information with abstract form will rely more on the previous knowledge of customers and will not be easily disturbed by peripheral information. If customers have no such previous knowledge of the product, it would be difficult for them to recognize the message. And because of the independent thinking way of customers, it is hard to control the ideas customers thinking about. While concrete message usually elaborates objects in detailed manners, which would induce recall and realization in depth. Concrete message usually doesn’t rely on customers’ own understanding of the product or brand, it leads customers to thinking
in the way the brand wants. Concrete message have a quantitative advantage over abstract message in that concrete message is always supported by more semantic units, more semantic predicates, and greater availability of contextual information. A possible explanation of the difference between abstract and concrete message is the dual-coding theory (Paivio,1986), which proposed that concrete and abstract words are supported by different verbal systems. Concrete words usually receive more support from a second system based on sensory representations compared to abstract words. From the above, the hypothesis could be:
Hypothesis 2: Concrete message style is more effective in communicating
message to customers than abstract message style in general.
Successful firm performance is derived from employee performance. As employees of the company, they are on their way of delivering product message to customers. Employees’ activities are related to the internal operation of company, and they provide timely and accurate feedback on the efficiency and effectiveness of operations to company managers (Dangelico, 2015). Employees are usually believed to be professional about their company’s product. Concrete information they provide tends to be more believable and reliable in that concrete words involve more contextual information than abstract words, and concrete information could revive some parts of the potential memories of the brains(Jacobson and Tufts, 2013). Employees are in the center of the dealing process of company’s production and operation, they have the competence to state concrete words about the product or
brand (Jacobson, 2013). Employees who use their personal social media accounts for communicating with customers are perceived in some levels of persuasive ability to customers partly because of their mixed use of personal life and profession, which is assumed to be authentic in that they get no other paid by doing advertising on personal social media accounts (Levin et al. 2012). Employees’ use of personal social media accounts could be a bond to remain the relationship with customers, the more concrete and specific the information they reveal to customers, the more trustworthy and believable they are in customers’ eyes (Kietzmann et al. 2011). They seem to tell the truth even it could be a lie. And then the trust could transform into customer loyalty for the company. While influencers are seen as “outsider” of the brand, they know less knowledge about the product and therefore work in a harder way to deliver concrete message to customers. From the above, the hypothesis could be:
Hypothesis 3: Message style (abstract/concrete style) would have moderating
effect between reviewer group (employee influencer/general influencer) and customers’ user engagement.
Hypothesis 3a: Employees would work better in using concrete message style to
deliver message to customers compared to general influencers. 3.2.2 Product-related/brand-related message content
Today, companies are facing more and more challenges, especially that they find it hard to create sufficient differential advantages over other competitors (Clancy and Trout, 2002). Managers are trying every way searching strategies to establish
appropriate brand associations in customers’ minds in order to stand out from other competitors. Many of them do this by feature positioning – product feature and brand feature strategies. Brand-related strategy is “the act of designing the company’s offering and image to occupy a distinctive place in the mind of the target market” (Kotler, 2003). Keller (1993) had proposed that making a differential advantage/value proposition, a well-positioned brand can draw the attention of a target customer segment with satisfying their specific needs. Product-related attributes are defined as the necessary elements for promoting and performing the product or service sought by consumers, which is often related to a product’s physical function or a service’s practical requirements (Keller, 1993). And Keller also elaborated product-related attributes and brand-related attributes in benefits level. Functional benefits are the more intrinsic advantages in product or service consumption process. These functions usually correspond to the product-related characteristics, and these functions often are linked to relatively basic motivations and intentions, such as physiological and safety needs, including a desire for problem removal or avoidance. Experiential benefits are characterized as what it feels like to use the product or service and are also usually related to the product-related functions. These benefits satisfy individual’s experiential needs such as sensory pleasure, entertainment, variety, and cognitive stimulation. Then, symbolic benefits relate to brand-related attributes, which are the more extrinsic needs and desire of using the product or service. Symbolic benefits relate to underlying functions for social approval or personal expression and
outer-directed self-esteem. Product-related message and brand-related message satisfy different needs of customers. The message content delivered closely relates to the mind route of customers when they process the information they receive from advertisements.
Product-related message is the information that is related to the attributes of a product’s physical or functional aspects, while brand-related message is the information that reveals the overall features of the brand, including the brand name, brand reputation, and brand emotion, and brand history, etc. From the intrinsic perspective, brand-related message helps customers to build initial impression of a brand when firstly know the brand. And brand-related message builds customers perception in an abstract construal route of the mindset while product-related message guides customers in another more concrete construal mindset route. Often product-related message is used when the brand has already got some awareness. In previous research, Kim, et al. (2009) proposed that product-related ads have different advantages and impact of effectiveness from corporate (brand) ads in terms of customers’ involvement effects on advertising attitude/behavioral motivations. In corporate/brand advertising, only when mediated by customers’ involvement with the ads, pre-dispositional involvement would have an impact on customers’ attitudes towards the ads and behavioral motivations, whereas pre-dispositional involvement would have both a direct and an indirect impact on attitude and motivations in product ads. It can be assumed that product-focused ads work in a broader situation.
In general, product-related message content and brand-related message content are different from two perspectives: functional and emotional. Product-related message is functional in that it provides information in details to customers, which would help customers to step further about the product itself. It’s useful when customers are in low involvement with the brand. Brand-related message usually reveals the brand’s feature to customers, which is often more abstract and emotional for customers who already have some involvement with this brand (Kim, et al. 2016). Usually, when individuals are with higher levels of involvement with a certain product or brand, the decision-making process is more complicated. They are usually more involved in elaborate information processing and have more product-related thoughts and makes more inferences (Steenkamp, 1990). And they often infer more qualities and traits attributes about the brand from their previous consumption experience (Ballester and Aleman, 1999). Then the consumption experiences are usually from the brand’s specific products. It can be assumed that when customers are in high involvement with a specific product, product-related message would help more when they are making decisions. From the above, the hypothesis could be:
Hypothesis 4: Product-related message works better than brand-related
message in general.
As part of the company, employees are likely to be perceived as important components of the brand when they have interaction with customers. Besides the role of job description, they can represent the brand to customers to some extent. They are
believed to have more experiences with the product or the company (Gelb and Deva, 2014). Usually, before telling employees how to communicate the word on social media about a specific product or service of the company, one of the obvious needs is to identify whether they have already understood and figured out the product or service to be boasted about or to be complained about—and in either case, how the employees perceive the brand’s strength and weakness, advantages and disadvantages (Gelb and Deva, 2014). Once the company consults them, they are concerned to be the “inside-knower” of the company. The information they know and provide to customers is perceived to be more reliable and authentic in general. Sirianni et al. (2013) proposed that product-related information delivered by employees would have a greater positive effect on customer alignment with the brand, and the overall brand evaluation would be more positive as well as customer-based equity would be stronger. When customers initially encounter a product that is unfamiliar to them, it is critical important for employees to deliver product-related message that supports the brand’s espoused image. And if this occurs during a customer’s initial experience, this product-related information would lead to potential better overall brand evaluation. And based on the highly concern of product information relative to sustainability messages, and the desire for individuals to reduce stress by information explosion, employees are supposed to use product-related information in the persuasion process and therefore increase sustainability and trust of the brand (Peloza, 2012). And the use of product-related sustainability will lead to a greater halo effect and additional credit.
Employees are on the way to push the brand’s reputation using product-related message delivery.
As for influencers, Rappaport (2007) defined social media influencers as “the high relevance between brands and customers, and the development of an emotional connection between customers and brands”. Influencers are connecting bridges between customers and brands. Usually it is difficult for a brand manager to directly have conversations with customers due to customers to customers dialog, however, they have the opportunity to shape the brand by using influencers as intermediaries. Social media influencers are often connected with brand and are now regarded as the online opinion leaders due to their knowledge, expertise, experience and influential power among followers. They are “potential powerful makers and breakers of brands” (Smith et al. 2009), and they have the ability to provide current and advanced product information to customers (Hsu and Tsou, 2011). Social media influencers are essentially “outsider” of the company compared to the company’s employees. Most of the case is that it’s easy to acquire brand knowledge from them (Lee et al. 2012). From the above, the hypothesis could be:
Hypothesis 5: Message content (product-related message content/brand-related
message content) would have moderating effect between reviewer group (employee influencer/general influencer) and customers’ user engagement.
Hypothesis 5a: Employees would work better in delivering product-related
would be more suitable to deliver brand-related message to customers. 3.3 Conceptual framework
Given the overview of this research, the relationship among different variables and moderators are visualized in the following conceptual model, and it can be helpful to understand the research subject better.
4. Methodology Research design
In order to examine the different relationship between general influencer and employee’ reviews on user engagement, the method used in this research is a
combination of experiment and survey. The use of experiment is in that this is the only research method that can demonstrate the inner difference between these two influencer groups because there’s no such comparison before. And the use of employee as influencer is also innovative in research. The experiment was processed randomly in a group of people, everyone in the experiment group would only meet one situation of the 8 experiment scenes. After the experiment, there’s analysis to see the difference and inner relationship of the independent variables and dependent variables in these 8 different experiment scenes. The independent variables in this study are employee reviews and influencer reviews on social media, the dependent variable is user engagement, and the moderators in this research are abstract/concrete message style and product-related/brand-related message content.
This study follows a 2 (general influencer review, employee review) × 2 (abstract style, concrete style) × 2 (product-related content, brand-related content) factorial design. So there are 8 different experiment scenes in this research in total. They are:
General influencer/ Employee influencer
Product-related message Brand-related message
Abstract style Abstract-style/product-related message
Abstract-style/brand-related message
Concrete style Concrete-style/product-related message
Concrete-style/brand-related message
All the respondents can only see just one scene of these 8 scenes at a time. They are distributed randomly and the probability they meet every scene is the same. We have collected 241 responses in total finally, and almost 30 responses averagely for each scene.
Procedure
At the start of the questionnaire, all of the respondents were informed that they were taking part in an experiment concerning general influencer and employee reviews on social media. This information highlighted that this research was conducted under the supervision of the University of Amsterdam and guaranteed the respondents’ confidentiality. Subsequently, respondents were asked some general questions about their use of social media. From the frequency question, it shows that everyone of the respondents has social media accounts and most of them use their social media accounts everyday. Before the experiment, general user engagement questions about the reviews of employee or influencer were asked, and the question either is about employee’s review or general influencer’s review is linked to the sequent experiment scene (either about employee or general influencer) followed by the first survey question. Then the questionnaire goes to the experiment page, 8
different experiment scenes were shown to different participants, and they were asked to complete evaluation questions in the scene they encountered. In this evaluation process, control questions were asked firstly to clarify if the message post is abstract/concrete or product-related/brand-related. Then 4 questions were asked to scale user engagement. After this evaluation under the experiment scene, the online experiment was completed. Gender and age were asked after the experiment. Following this, the whole questionnaire was finished. The online experiment questionnaire is in the Appendices part.
Stimulus material
In this research, two Instagram accounts were created and used as stimulus material in order to have complete control over two review groups – general influencer reviewer and employee reviewer. And the product message post on the accounts in every experiment scene is classifies into 8 different content. These two Instagram accounts both use the same pictures as the account’s profile pictures and the product picture is all the same in both accounts in order to have control over the image exposure to the participants’ evaluation. The chosen product in this research is earphone, which would have less bias evaluation between male and female participants. The brand is Sennheiser, which is a famous audio company especially good at earphones. A famous brand would help participants quickly catch the product and brand, and famous brand usually have millions of followers on social media websites, which would make the accounts more believable and acceptable. Then the
two accounts respectively post message about Sennheiser’s new earphone, using abstract/concrete style and posting product-related/brand-related content on the post area. Under different review groups and moderation effect, participants make evaluation of their user engagement with the earphone.
Manipulation in the experiment
In the experiment part, in order to control other variables in this research, there’s manipulation in these 8 different experiment scenes.
From the above that have been showed, the same experiment scene has the same picture of the product, same picture of the reviewer’s profile, and the same post content in the Instagram post area in order to control all the other variables that may have an impact on the participants’ evaluation of the product from the post. The only difference is in the introduction part of the post which informs the participants that the post is whether about the product or about the brand, and whether the post is from the general influencer (the Instagrammer) or the employee. And following the post picture, there is a manipulation question that has been asked for participants to figure whether the message style is abstract or concrete.
Sample and participants
online experiment questionnaire. Through this tool, participants were able to do the online experiment conveniently in their natural surroundings. And this also reduces the cost and time to gather groups of participants together to do the experiment offline. And it reduced the likelihood of participants respond desirable expected evaluation that would disturb the result of the experiment thus it enhanced the reliability of this study. However, there’s also some limitation. In the process, the researcher did not have full control over the participants’ answering circumstances whether they took the online experiment seriously. Fortunately is that with the enough amount of the passionate participants had put into this research, it can be assumed that most of the participants have participated with serious intentions.
There are 241 responses in total. From the 241 respondents who had answered the questionnaire, everyone of them completed the questionnaire entirely. The vast majority of the participants were from different universities and most of them are in their master’s. Additionally, the questionnaire was distributed through Facebook, Wechat link and other social media apps.
The gender of the responses showed that the questionnaire was distributed relatively averagely in the participants group. From the 241 participants, 53.1% were female (N=128) and 46.9% were male (N=113). And the average age was 23.5 (M=23.5, SD=1.654). The overview of these results is given in table 1 and table 2. Measurement
engagement. There are four items in the experiment to measure customers’ user engagement, including participants’ attitude towards the product (Raimondo and Farace, 2013), judgment of the product (Reid, MacDonald and Du, 2013), interaction with the online reviewers on social media websites (Jussila et al. 2012) and perceived value of the product change (Chang and Tseng, 2013). All four of these items are crucial to measure the dependent variable – customers’ user engagement in this research and from these four items, the final measurement of the dependent variable would be more accurate and comprehensive.
2. Control variables: in this research, control variables are age, gender, frequency of social media use and general change of customers’ user engagement. Participants were asked to respond their age and gender at the final part of the experiment. Social media use is very popular among young people (Kietzmann, et al. 2011). In other words, older people do not use social media as much as youngsters. Age is an interesting factor to include in this research as it may influence the dependent variable directly. Age was measured on an interval scale and gender was measured on a nominal scale in this research. Another factor included in this research is gender. A small quantity of scholars found that gender is a factor that influence social media use (Jelenchick et al. 2013), in order to have overall control of this research, gender is also collected for the participants in the questionnaire. Gender was coded as “0” for “female”, “1” for “male” for further research. Frequency of social media use was measured on an interval scale. General change means before the experiment, what is
the participant’s response to the reviewer’s comments of a product in general. This was asked before the experiment in the online questionnaire. General change is measured on a scale ranging from 1 to 5 (1 means strongly disagree and 5 means strongly agree).
Table 1
Gender
Frequency Percent Valid Percent
Cumulative Percent Valid female 128 53.1 53.1 53.1 male 113 46.9 46.9 100.0 Total 241 100.0 100.0 Table 2 Age
N Range Minimum Maximum Mean DeviationStd. Variance AGE 241 10 18 28 23.50 1.654 2.734
Valid N
(listwise) 241
Table 3
Frequency of social media use
Frequency
Frequency Percent Valid Percent
Cumulative Percent Valid Everyday 239 99.2 99.2 99.2 2-3days 1 .4 .4 99.6 Once a week 1 .4 .4 100.0 Total 241 100.0 100.0
General change
Change (general)
Frequency Percent PercentValid CumulativePercent
Valid Somewhat disagree 6 2.5 2.5 2.5
Neither agree nor disagree
75 31.1 31.1 33.6
Somewhat agree 141 58.5 58.5 92.1
Strongly agree 19 7.9 7.9 100.0
Total 241 100.0 100.0
3. Independent variables: in this research, the independent variable – reviewer groups were coded as “0” for “employee”, “1” for “general influencer” for further research. Moderators: in order to enable use of message style and message content as moderators, message style “concrete” was coded as “0”, “abstract” was coded as “1”. And message content “brand-related” was coded as “0”, “product-related” was coded as “1”. And also the control variable – gender, was coded as “0” for “female”, “1” for “male”.
Data analysis
Data were collected by means of an online experiment. To perform the statistical analyses, the Statistical software Package for Social Science (SPSS) was used in this research. After recoding the reverse coded items, scaling reliabilities of the variables, descriptive statistics, skewness, and normality tests were conducted. One-way ANOVA analysis was undertaken to test the difference between groups and
within groups in this research. The one-way ANOVA analysis was conducted for the independent variable – the two reviewer groups and the two moderators in this research – abstract/ concrete message style and product-related/brand-related message content to see the different effects on dependent variable – customers’ user engagement in these groups.
In order to test the moderating effect of the message style and message content, an SPSS macro by Andrew F. Hayes (2013) was used. The macro of Hayes (2013) computed the moderating analyses results of message style (abstract/concrete) and message content (product-related/brand-related) in order to test hypotheses 3 and 5.
ANCOVA analysis was also conducted to see the interactions between variables in this research. By including variables and moderators step by step, these interaction relationships were clearly computed in the result. Reviewer groups (employee/general influencer), message style (abstract/concrete) and message content (product-related/brand-related), their between-subjects effects were clearly demonstrated in the result part.
5. Results and analysis
The following chapter will address all the hypotheses previously mentioned, the results of the preliminary analysis from SPSS and where necessary a clarification of those results.
5.1 Preliminary analysis
This analysis was distributed and conducted in 241 participants in total. In order to deduct some inappropriate responses that would make the result unreliable and doubtful, outlier test and frequency test were conducted to see if there is any error, missing data or abnormal values in the responses. And lastly, only the responses from the participants who have social media accounts (social media users) were taken into account in this research. This means that those participants who don’t use social media to search information and have interaction with other users were excluded from the sample. At the start of the online questionnaire, participants were asked whether they have any social media accounts, the results show that everyone of them owns social media accounts. Then all of the 241 participants’ responses were recorded as sample in this research.
Missing value
All variables under investigation were checked for missing data. The outlier check was made for all the variables. There were 241 respondents in total and the result reveals that there’s no missing data and outliers in all the variables. So there’s no need to cope with the missing data in this research.
A preliminary frequency and descriptive analysis were conducted for all the variables and moderators to check on any error values or missing data. The results of frequency and descriptive analysis were further detailed in Appendices. The age of participants range from 18 to 28, and the average is 23.5 (Mean=23.5, SD=1.654).
And the gender composition of participants is that 53.1% were female (N=128) and 46.9% were male (N=113). The groups of reviewers are classified nominally, and as well as message style and message content. Only the user engagement was scaled and included 4 items to qualify.
Reliability analysis
The expected relationship between different reviewer groups’ reviews to user engagement and the moderation effect of message style and message content were firstly tested on the reliability of the scale items and the overall Cronbach Alpha for each of the attributes. Reliability analysis enables to examine the consistency of measurements. Reliability checks were run for the dependent variable – customers’ user engagement. There are 4 items in total to measure the dependent variable in this research. The Cronbach’s alpha, which represents the estimator of the internal consistency, had been tested to verify if all the items in one scale measure the same, or if some questions should not be used for analysis. As exhibited in table 4, all of the 4 items have a Cronbach’s alpha > 0.7 (Cronbach’s Alpha = 0.796), which indicates high level of internal consistency. The sample is reliable for following analysis.
Table 5: Reliability analysis
Reliability Statistics
Cronbach'sAlpha N of Items
In this research, manipulation check is necessary to check whether the
experiment manipulation has had some preliminary effect. The manipulation check in this research is to check whether the participants had understood the instructions and information exposed in the scene. In this case, those participants who failed to pass the manipulation check would be excluded in this research. ANOVA and ANCOVA were used in this research to check if the manipulation of the independent variable – reviewer groups made a statistically significant difference on the dependent variable – customers’ user engagement. ANOVA and ANCOVA analysis would be
demonstrated in the main analysis part. Computing scale means
As the final preliminary step, new variables as a function of existing variables were created for hypothesis testing. The means of all items that was used to describe a variable was calculated. Means and standard deviations of all variables are exhibited in table 5 (also the correlation matrix). The mean score of user engagement is 3.65 and the standard deviation is 0.69.
5.2 Main analysis ONE-WAY ANOVA
In this research, one-way ANOVA is used to test hypotheses related to the differences of means between groups. In this research, there was a statistically significant effect of message style on levels of user engagement, F (1,239) = 145.43, p
< 0.05, which indicates that concrete style is significantly different from abstract style. There was also a statistically significant effect of message content on levels of user engagement, F (1,239) = 3.47, p < 0.05, which indicates that there’s also significant difference between product-related content and brand-related content. However, there’s no statistically significant difference between employee and influencer group from the data (p = 0.37). Which indicates that there’s no significant difference between employee group and influencer group. These two groups’ reviews have the same effect on customers’ user engagement. Thus hypothesis 1 is totally rejected.
In the tests of between-subjects effects, message style has moderating effect over customers’ user engagement (Sig. = 0.000), which is statistically significant in this research. And the mean value of concrete message style is 4.1226, the mean value of abstract message is 3.2741, where there is a huge difference between these two styles. However, message content and the reviewer groups were not significant in Sig. coefficients (Sig. = 0.117 and Sig. = 0.971 separately), which indicates that there’s no significant between-subjects effect in message content and customers’ user engagement as well as in reviewer groups and customers’ user engagement. Also from the mean value table, the mean value of brand-related content is 3.5671, while the mean value of product-related content is 3.7309, which also indicates that there’s little difference. And the mean value of employee group is 3.6875 while the mean value of general influencer group is 3.6074, there’s no significant difference between these two reviewer groups in this research. However, there is interaction between message
content and reviewer groups, whose the Sig. is statistically significant in 0.000. Table 6
Tests of Between-Subjects Effects
Dependent Variable: EnTOT (User engagement)
Source
Type III Sum
of Squares df Mean Square F Sig.
Corrected Model 58.265a 10 5.827 24.474 .000
Intercept 8.975 1 8.975 37.699 .000
AGE 3.139E-5 1 3.139E-5 .000 .991
CHANGE 2.292 1 2.292 9.629 .002 GENDER .006 1 .006 .027 .871 STYLE 42.132 1 42.132 176.974 .000 CONTENT .588 1 .588 2.471 .117 REVIEWER .000 1 .000 .001 .971 STYLE * CONTENT .773 1 .773 3.247 .073 STYLE * REVIEWER .283 1 .283 1.188 .277 CONTENT * REVIEWER 11.896 1 11.896 49.970 .000 STYLE * CONTENT * REVIEWER .130 1 .130 .548 .460 Error 54.755 230 .238 Total 3319.000 241 Corrected Total 113.021 240 a. R Squared = .516 (Adjusted R Squared = .494)
Table 7
Descriptive Statistics
Dependent Variable: EnTOT (User engagement)
STYLE CONTENT REVIEWER Mean Std. Deviation N Concrete Brand Employee 3.7600 .37832 25
Influencer 4.2400 .57500 25
Total 4.0000 .53927 50
Product Employee 4.4911 .42754 28
Influencer 3.9732 .31431 28
Total Employee 4.1462 .54472 53
Influencer 4.0991 .47114 53
Total 4.1226 .50738 106
Abstract Brand Employee 3.1486 .49812 37
Influencer 3.3958 .60467 36 Total 3.2705 .56328 73 Product Employee 3.5417 .60559 30 Influencer 3.0313 .42951 32 Total 3.2782 .57813 62 Total Employee 3.3246 .57901 67 Influencer 3.2243 .55674 68 Total 3.2741 .56802 135
Total Brand Employee 3.3952 .54236 62
Influencer 3.7418 .72164 61 Total 3.5671 .65839 123 Product Employee 4.0000 .70866 58 Influencer 3.4708 .60557 60 Total 3.7309 .70723 118 Total Employee 3.6875 .69531 120 Influencer 3.6074 .67764 121 Total 3.6473 .68624 241 ANCOVA analysis
ANCOVA test was also conducted in this research to see all the variables’ effects and interactions in this model. In order to see the main effect of different variables separately, the variables were put into the model step by step.
Firstly, the reviewer groups were put into this model. From the table, it is obvious that there’s no main effect between reviewer groups and customers’ user engagement in this research (Sig. = 0.773). And there’s little difference between the two groups’ mean value as well.
Descriptive Statistics
Dependent Variable: EnTOT (User engagement)
REVIEWER Mean Std. Deviation N
Employee 3.6875 .69531 120
Influencer 3.6074 .67764 121
Total 3.6473 .68624 241
Table 9
Tests of Between-Subjects Effects
Dependent Variable: EnTOT (User engagement)
Source Type III Sumof Squares df Mean Square F Sig. Corrected Model 2.221a 4 .555 1.183 .319 Intercept 14.789 1 14.789 31.500 .000 REVIEWER .039 1 .039 .084 .773 AGE .500 1 .500 1.065 .303 CHANGE 1.003 1 1.003 2.137 .145 GENDER .409 1 .409 .872 .351 Error 110.800 236 .469 Total 3319.000 241 Corrected Total 113.021 240 a. R Squared = .020 (Adjusted R Squared = .003)
Secondly, message style was put into the model. From the table, it is revealed that there’s indeed main effect in message style in this research (Sig. = 0.000). However, there’s no interaction between reviewer group and message style to customers’ user engagement (Sig. = 0.514), which indicates that there’s no moderation effect between reviewer group and message style in this research.
Descriptive Statistics
Dependent Variable: EnTOT (User engagement)
REVIEWER STYLE Mean Std. Deviation N
Employee Concrete 4.1462 .54472 53 Abstract 3.3246 .57901 67 Total 3.6875 .69531 120 Influencer Concrete 4.0991 .47114 53 Abstract 3.2243 .55674 68 Total 3.6074 .67764 121 Total Concrete 4.1226 .50738 106 Abstract 3.2741 .56802 135 Total 3.6473 .68624 241 Table 11
Tests of Between-Subjects Effects
Dependent Variable: EnTOT (User engagement)
Source Type III Sumof Squares df Mean Square F Sig. Corrected Model 45.226a 6 7.538 26.017 .000 Intercept 12.819 1 12.819 44.247 .000 AGE .249 1 .249 .858 .355 CHANGE 1.725 1 1.725 5.955 .015 GENDER .138 1 .138 .477 .491 REVIEWER .001 1 .001 .003 .957 STYLE 42.854 1 42.854 147.914 .000 REVIEWER * STYLE .124 1 .124 .427 .514 Error 67.794 234 .290 Total 3319.000 241 Corrected Total 113.021 240 a. R Squared = .400 (Adjusted R Squared = .385)
Thirdly, message content was put into the model. From the table, it shows that there’s no main effect between message content and reviewer group (Sig. = 0.085).
However, there’s significant interaction between reviewer group and message content to customers’ user engagement (Sig. = 0.000). From this, it indicates that message content has moderation effect with reviewer groups in this research.
Table 12
Descriptive Statistics
Dependent Variable: EnTOT (User engagement)
REVIEWER CONTENT Mean Std. Deviation N
Employee Brand 3.3952 .54236 62 Product 4.0000 .70866 58 Total 3.6875 .69531 120 Influencer Brand 3.7418 .72164 61 Product 3.4708 .60557 60 Total 3.6074 .67764 121 Total Brand 3.5671 .65839 123 Product 3.7309 .70723 118 Total 3.6473 .68624 241 Table 13
Tests of Between-Subjects Effects
Dependent Variable: EnTOT (User engagement)
Source
Type III Sum
of Squares df Mean Square F Sig.
Corrected Model 14.994a 6 2.499 5.965 .000 Intercept 10.314 1 10.314 24.621 .000 AGE .012 1 .012 .030 .863 CHANGE 1.323 1 1.323 3.159 .077 GENDER .094 1 .094 .225 .635 REVIEWER .054 1 .054 .128 .721 CONTENT 1.253 1 1.253 2.992 .085 REVIEWER * CONTENT 11.656 1 11.656 27.824 .000 Error 98.027 234 .419 Total 3319.000 241 Corrected Total 113.021 240
