The Impact of Relationship Quality Between Social Media Influencers and Their Followers on Brand Purchase Intention

Social media influencers (SMI) have been shown to influence the purchase decision-making of their followers. In this respect, it is necessary to explore the interactive relationship between them, i.e., relationship quality. Researchers will not know the extent to which social media influencers affect their followers and their persistence of persuasion until they understand the SMI-follower relationship quality. This study focused on whether the SMI-follower relationship quality enables followers to transfer their trust in SMIs and affects their attitude toward the advertised or recommended product and brand. Furthermore, it explored whether followers’ skeptical attitude toward advertising moderates the relationship between the SMI-follower relationship quality and trust transference, and whether followers’ self-monitoring moderates the relationship between their advertising skepticism and trust transference. This study distributed survey questionnaires online, and obtained a total of 235 valid copies. The data were used to conduct path analysis through the software PROCESS. The empirical data showed that the higher the SMI-follower relationship quality, the more likely are the followers to transfer their trust in SMIs to the advertised or recommended product or brand. In addition, it was found that trust transference positively mediates the impact of relationship quality on brand attitude. Further, the results indicated that the followers’ high level of advertising skepticism negatively moderates the impact of SMI-follower relationship quality on trust transference, and that a high level of self-monitoring inhibits the negative moderating effect of advertising skepticism on the relationship between SMI-follower relationship quality and trust transference to a larger extent.


Introduction
Social media influencers (SMIs), usually celebrities or opinion leaders (Enke & Borchers, 2019;Camplee & Farrell, 2020;Boerman, 2020), have the capacity to influence consumers' purchase decision-making (Lou & Yuan, 2019) via social media. McGuire (1989) proposed the input-output persuasion matrix and suggested the five elements that influence persuasion, namely: (1) source, (2) message, (3) channel, (4) receiver, and (5) destination. However, only the source, message, and receiver have been the focus of previous research. The element "source" explores how credibility and attraction influence the purchase behavior of followers (Ki et al., 2020). The element "message" examines which of the two types of value, namely informative value and entertainment value, generate the purchase intention of the brand that SMIs recommend; it was found that informative value has stronger effects (Lou & Yuan, 2019). Lastly, the element "receiver" investigates the parasocial interaction relationship between SMIs and their followers, which could increase the intention of the latter to purchase a particular brand or product. traditional advertising; the community and cultural capital of SMIs enable them to be more efficient in obtaining brand benefits (Chu, Kamal, & Kim, 2019). For SMIs, one way to cash in on their influence is to cooperate with brand manufacturers, making followers have a strong desire to use the same goods and services as them (Stoldt, Wellman, & Ekdale, 2019). McGuire (1989) proposed the input-output persuasion matrix and suggested the five elements that influence persuasion, namely: (1) source, (2) message, (3) channel, (4) receiver, and (5) destination. However, only the source, message, and receiver have been the focus of previous research. The element "source" explores how credibility and attraction influence the purchase behavior of followers (Ki et al., 2020). The element "message" examines which of the two types of value, namely informative value and entertainment value, generate the purchase intention of the brand that SMIs recommend; it was found that informative value has stronger effects (Lou & Yuan, 2019). Lastly, the element "receiver" investigates the parasocial interaction relationship between SMIs and their followers, which could increase the intention of the latter to purchase a particular brand or product.
To keep their influence lasting and effective, it is necessary for SMIs to establish and maintain good relationships with their followers. As described by Thibau and Kelley (1959), relationships are dynamic, and they should be mutually beneficial to both parties to persist. Thus, the SMI-follower relationship quality plays an important role in maintaining the relationship.

Relationship Quality
Previous studies pointed out that a transaction is likely to be completed if the two parties involved have a long-term relationship (Lovelock, 1983). During service transaction, service quality is necessary and relationship quality is indispensable. Relationship quality is a construct that measures the strength of the relationship between the buyer and the seller (Crosby, Evans, & Cowles, 1990); it is the overall evaluation of the strength of the relationship, assessing whether the needs and expectations of both parties are met. Therefore, a high relationship quality between the seller and the buyer is helpful in improving the performance of the seller.
When SMIs create posts to promote a certain brand, they are working as a salesman, marketing the brand and persuading their followers to purchase its products. In this respect, the SMI-follower relationship quality will affect their followers' purchase intention. Relevant literature pointed out that relationship quality has three constructs, namely: (1) trust, (2) commitment, and (3) satisfaction (Moorman, Deshpande, & Zalman, 1993;Morgan & Hunt, 1994;Oliver, 1997;Kumar & Pansari, 2016;Itani, Kassar, & Loureiro, 2019). Among them, trust is the key factor that enables a relationship to last (Santouridis & Veraki, 2017). Morgan and Hunt (1994) found that a high relationship quality is produced when both parties believe that their relationship is worth maintaining, which means that it is based on trust. Therefore, SMIs must be able to transfer their trust in the brand to their followers so that a relationship between the brand and the followers can be established. In this way, they can make their followers generate brand purchase intentions.

Trust Transference
Trust transference theory holds that when individuals trust certain objects, they tend to place their trust in unknown objects related to the latter (Doney & Cannon, 1997). Similarly, when we trust a person or an entity, we also trust relevant persons or entities through trust transference (Zhao, Huang, & Su, 2019). Doney and Cannon (1997) provided the five processes that establish trust between the seller and buyer, which are (1) calculative-based, (2) prediction-based, (3) capability-based, (4) intentionality-based, and (5) transference-based. This research applied transference-based trust to explore the SMI-follower relationship quality.
Transference-based trust means that people transfer their trust in individuals (or websites) towards certain objects that the latter recommend despite the former not having direct contact or being unfamiliar with the objects (Milliman & Fugate, 1988). For example, when a trusted blog recommends a link to a website (or product), the readers are likely to transfer their trust in the blog to the recommended website (or product).
Trust is an important construct for relationship quality (Ganesan & Hess, 1997;Johnson & Grayson, 2005). This study aims to explore whether the trust of followers in SMIs will be transferred to the product recommended by the latter, thus influencing the attitude of the former toward the product. It also approaches whether trust transference has mediating effects on the relationship between SMI-follower relationship quality and followers' attitude toward the brand recommended by SMIs. ijms.ccsenet.org International Journal of Marketing Studies Vol. 13, No. 4;2021 2.5 Brand Attitude Howard (1989) defined attitude as the extent to which consumers expect that a brand can meet their special needs. Consumers' attitude toward brands (i.e., their preference) can be evaluated through their subjective cognition. Brand attitude is the overall evaluation of the brand based on its various attributes, such as price, appearance, function, and quality (Lee, Lee, & Yang, 2017). Fishbein and Ajzen (1975) referred to brand attitude as the overall evaluation of consumers after experiencing a specific brand. This study adopted the definition of Aaker and Fournier (1995) who referred to brand attitude as consumers' overall emotional and affective evaluation of a brand. Assael (2004) divided brand attitude into (1) brand belief, which is the consumers' perceived characteristic of the brand; (2) brand evaluation, which is the extent to which the brand is liked by the consumers; and (3) purchase intention, which refers to the consumers' tendency to purchase the products of the brand.

Advertising Skepticism
Advertising skepticism refers to consumers' negative attitude (or disbelief) toward advertising claims. Skepticism means that consumers realize advertisers have some motives behind their advertising, such as guidance and persuasion, which may lead to biased communication (Boush, Friestad, & Rose, 1994;Brucks, Armstrong, & Goldberg, 1988). It is also described as the way in which consumers evaluate and process advertising messages. When faced with advertising, consumers with a skeptical attitude tend to recall relevant negative experiences (Campbell & Kirmani, 2000), resulting in a negative response. Thus, they are unable to be persuaded by the advertiser (Obermiller, Spangenberg, & Maclachlan, 2005). Obermiller, Spangenberg, and MacLachlan (2005) developed the SKEP scale to measure the consumers' tendency of disbelief toward advertising messages. They found that consumers' doubts on the authenticity of advertising reduce the value of information transmission. Consequently, Darby and Karni (1973) suggested that consumers are skeptical about the advertisement content (quality, durability and performance) of experience and credence goods. Darby and Karni (1973) classified products into three categories based on people's degree of trust in advertising, namely (1) search products, which are those with information that can be easily searched prior to purchase; (2) experience products, which are those that need to be experienced or used to obtain information; and (3) credence products, which are products with relevant information that can neither be searched nor known via short-term use. In this respect, Feick and Gierl (1996) pointed out that consumers are more skeptical about credence products than experience ones, and have a higher level of skepticism for high-priced experience products than low-priced ones.

Self-Monitoring
The concept of self-monitoring is based on classical pragmatism, and is regarded as a personality trait. Snyder (1979) proposed that individuals alter their own behavior according to changes in the external environments, which has stable and measurable individual differences. Individuals can maintain their positive self-image through self-monitoring (Gangestad & Snyder, 2000). Nantel and Strahle (1986) referred to self-monitoring as the individual's way of paying attention to the surroundings; the information they get is utilized as a clue about how they should present themselves, and how they should alter their behavior. Therefore, this personality trait makes people become potentially affected by others' remarks and attitudes, which could lead to a change in their behavior (Fine & Schumann, 1992).
Self-monitoring can be divided into high and low levels. High self-monitors are highly sensitive and good at capturing others' emotions during an interaction (Synder, 1979). They are able to alter their behavior according to what they have observed, for which they show high social adaptation (Lippa, 1976). High self-monitors are seen as social chameleons, who change their behavior to present a self-image that conforms to a given situation (Bedeian & Day, 2004). By contrast, low self-monitors often follow their feelings and show their actual attitude when presenting themselves. They do not care much about social contexts or situations, or examine whether social contexts are helpful to their behavior; they exhibit their behaviors with their inner attitude, values, and belief as their guidance. Low self-monitors are regarded as those who are loyal to themselves, for their performance is stable and does not change according to external situations (Bedeian & Day, 2004). Moreover, Jenkins (1993) believed that self-monitoring may change due to external influences, despite being considered as a stable personality trait. For instance, the elderly are more likely to act based on their own experience and belief, and are less likely to change their thinking compared with the younger individuals who tend to have a higher level of self-monitoring. Lavine and Snyder (1996) confirmed that low self-monitors are influenced by "messages making appeals to values", whereas high self-monitors are impacted by "appeals to status". In addition, it has been proven empirically that high self-monitors prefer prestigious brands that show their social status (Becherer & Richard, 1978). DeBono (2006) found that high self-monitors make judgments about the quality of a product based on its appearance, whereas low self-monitors make judgments via its performance.

Research Framework
Based on the relationships between variables stated above, this study explored the impact of the SMI-follower relationship quality on brand attitude via trust transference. Moreover, advertising skepticism and self-monitoring were set as moderators. The research framework is shown as follows. SMI/celebrity endorsement and product marketing are common marketing strategies that brand manufacturers adopt. Previous studies have already pointed out that endorsers must have professionalism, credibility, reliability, persuasiveness, and preference. In addition, consumers' familiarity with endorsers affects their perception of the recommended products, influencing their purchase intention (Stoldt, Wellman, & Ekdale, 2019).
Previous researches on relationship quality mainly focused on the interaction between enterprises/brands and customers. Crosby, Evans and Cowles (1990) first introduced the blogger-reader relationship quality. Relationship quality is the foundation of good cooperative relationships. Through long-term operation and maintenance, a blogger is able to establish professionalism to his/her readers. By sharing reliable information, the blogger is able to convince the readers that he/she is trustworthy and cares about their interests. Wang et al. (2015) pointed out that readers tend to compare the differences between themselves and the bloggers. The closer the readers get to the actual or ideal image of the blogger, the better the blogger-reader relationship quality. When the satisfaction of readers is sufficient to make them commit to maintaining a long-term relationship with the blogger, they become willing to adopt an inclusive and trusting attitude towards the blogger (Fournier, 1998). The study by Wang et al. (2015) found that the higher the blogger-reader relationship quality, the more positive is its impact on readers' word-of-mouth product transmission. This confirmed that relationship quality truly influences the relationship between the SMIs and the followers.
Trust constitutes the basis of relationship quality. Stewart (2003) explored consumers' trust transference between shopping websites, and found that consumers believed that shopping websites have similarities and interactivity. If consumers trust one website, they would trust an unknown website to some extent. Bergami and Bagozzi (2000) believed that when customers identify with the company, they will have a sense of belonging and identity, making them show positive attitudes and behaviors like taking pride in their work.
Moey and Kincade (2003) pointed out that attitude is one's tendency to react to certain people, matters, and issues, giving a persistent and consistent evaluation which could be either positive or negative. It is also an antecedent of consumers' behavior, which can directly predict their purchase behavior. When customers identify with a company, they will support it and continuously purchase its products, making it more competitive than its ijms.ccsenet.org International Journal of Marketing Studies Vol. 13, No. 4;2021 6 rivals in the long run (Armstrong & Kolter, 2000). Moreover, Smith (1998) observed that readers were willing to establish a long-term relationship with their trusted bloggers, believing that the latter are capable of meeting their needs and act in their interests.
This research holds that SMIs share some similarities with bloggers. Thus, based on the above statements, the following hypotheses are proposed: H1: The higher the SMI-follower relationship quality, the more likely it is for the followers to transfer their trust in SMIs towards products that the latter recommend.
H2: The higher the extent of trust transference, the more positive followers' attitude toward the brand endorsed by SMIs.
3.1.2 SMI-Follower Relationship Quality, Trust Transference, and Advertising Skepticism Mangleburg and Bristol (1998) noted that consumers have a tendency toward disbelief of advertising claims because they realize that advertisers have certain motives behind the advertising. Furthermore, Obermiller and Spangenberg (1998) found that consumers with a higher level of advertising skepticism take a more negative attitude toward advertising than those with a lower level of skepticism. In addition, it was confirmed that consumers with a high level of skepticism do not rely on advertising much, so they tend to avoid it because they do not believe in persuasive stimuli like suggestions, proposals, or changed messages. Therefore, they are considered as cautious and careful (Obermiller & Spangenberg, 1998). These explain why consumers with different levels of skepticism make different responses.
Hwang and Jeong (2016) revealed that consumers' advertising skepticism affect their purchase intention; more specifically, consumers with a high level of skepticism often take a negative attitude toward advertising, decreasing their purchase intention.
Also, Lee and Hong (2019) found that a high SMI-follower relationship quality is more likely to generate trust transference. Meanwhile, a high level of skepticism results in a negative attitude toward advertising. Therefore, this research infers that skeptical character has moderating effects; that is, the higher the level of consumers' advertising skepticism, the less likely the followers will transfer their trust in SMIs towards brands endorsed by the latter. Based on the above, the following hypothesis is proposed: H3: Followers' high level of advertising skepticism negatively moderates the relationship between SMI-follower relationship quality and trust transference.
3.1.3 SMI-Follower Relationship Quality, Trust Transference, Advertising Skepticism, and Self-Monitoring As previously mentioned, the higher the level of advertising skepticism, the more likely it is to negatively moderate the relationship between SMI-follower relationship quality and trust transference. Self-monitoring is a personality trait; individuals with high self-monitoring are susceptible to external information flow and have the tendency to exhibit herd mentality (Bikhchandani et al., 2011).
SMIs represent an ideal state that followers expect to get into, so followers with high self-monitoring are likely to change their behavior or appearance to get close to the "ideal state", or to conform to the viewpoint of the public. Enterprises can use this to attract the attention of followers and recommend products to them, which could lead to consumer purchase behaviors. By contrast, it is difficult to influence followers with low self-monitoring via the recommendation of SMIs. Therefore, consumers with high self-monitoring are likely to inhibit the negative moderating effect of skepticism on SMI-follower relationship quality and trust transference (Ferguson, Lau, & Phau, 2016;Kauppinen-Räisänen et al., 2018;Pornsakulvanich, 2018). Accordingly, this research proposes the following hypothesis: H4: Followers' high self-monitoring inhibits the negative moderating effect of advertising skepticism on the relationship between SMI-follower relationship quality and trust transference.

Operational Definition of Research Variables and Question Items
This section discusses the operational definitions of the five variables being explored, namely: (1) SMI-follower relationship quality, (2) trust transference, (3) brand attitude, (4) advertising skepticism, and (5) self-monitoring. Also, the scales that measure these variables are presented. The questionnaire used for this research consisted of short-answer and multiple-choice questions. The multiple-choice questions were scored using a seven-point Likert scale (1 for strongly disagree and 7 strongly agree).

Relationship Quality
This research evaluated the relationship between SMIs and followers using the items from the brand relationship  (2019) provided the three constructs of relationship quality, which are trust, commitment, and satisfaction. Therefore, with the aid of experts, this research selected three questions, one for each construct, from the brand relationship quality scale. The questions are as follows: 1) I think this SMI should be like a good friend of mine.
2) I believe in the professional knowledge of this SMI.
3) I feel disappointed and lost if I do not watch the videos or browse the articles posted by this SMI for a period of time.

Trust Transference
This study referred to the trust transference scales by Doney and Cannon (1997), Zhao, Huang and Su (2019), and Liu et al. (2018), and selected three questions through experts' discussion. The questions included are as follows: 1) I think that this SMI is sincere and reliable.
2) This SMI recommend products sincerely, which makes me think that the products can be trusted.
3) The products recommended by this SMI are reliable.

Brand Attitude
This research divided brand attitude into three constructs, namely cognitive, emotional, and behavioral, as proposed by MacKenzie, Lutz and Belch (1986). Based on these constructs, three items for measuring the brand attitude were selected from the scale by Rosenberg et al. (1960). These are as follows: 1) I think the brand (or products) recommended by this SMI is trustworthy.
2) I think the brand (or products) recommended by this SMI can meet my expectations.
3) I think I have a strong sense of identity with the brand (or products) recommended by this SMI.

Advertising Skepticism
This study adopted the skeptical character scale developed by Zhang, Ko and Carpenter (2016), which incorporated the word-of-mouth skepticism scale modified by Obermiller and Spangenberg (1998) and Tan (2002). The skeptical character scale includes three dimensions, namely (1) trustfulness of the message sender, (2) motives of the message sender, and (3) identity of the message sender. The present study selected three questions and revised them. The revised questions are as follows: 1) I can hardly get the truth from the articles or videos posted by SMIs.
2) Generally, the articles or videos posted by SMIs do not truthfully present the totality of the products.
3) Most SMIs intentionally lead me in forming my views on certain products.

Self-Monitoring
This study adopted the self-monitoring scale compiled by Lennox and Wolf (1984) based on the scale developed by Synder (1974). It consists of two dimensions which are the ability to modify self-presentation and the sensitivity to the expressive behavior of others. For the current research, three out of 13 items from the study by O'cass (2000), which conducted factor analysis of the self-monitoring scale, were selected: 1) I am able to change myself in order to leave a good first impression on others.
2) I am able to alter my behavior to adapt to the environment.
3) I can easily change my behavior when the situation requires it.

Research Participants
The study explored whether SMIs influence their followers' impression and purchase intention of the products they recommend by establishing relationships with their followers. There are various kinds of SMIs, and respondents have different preferences for SMIs. To ensure that research participants were not limited to a particular type, the respondents were asked to write down the SMIs who they like and the category of the recommended products.

Survey Method
The participants of this study were Internet users who had been following a certain SMI for a long time. The SMI being followed must have posted commercial advertising. Surveycake was utilized as a tool to conduct an online survey; the questionnaires were distributed through various Internet platforms.

Samples and Descriptive Statistics
The questionnaire survey was conducted between April 19, 2020 and May 2, 2020 through Surveycake. A total of 363 questionnaires were collected. If the respondents were long-time followers of an SMI who had not advertised in the past, their data were not included in the study. After removing the invalid questionnaires, 245 were recognized as valid, and the effective recovery rate was 67.49%. According to Bentler and Chou (1987), the number of valid samples should be 5−10 times that of the measurement items. There are 15 items in this research, so at least 150 samples are required.

Reliability Analysis
A variable with a Cronbach's α value of more than 0.7 is considered to have good reliability (Nunnally, 1978). In this study, the Cronbach's α values obtained for the SMIs-followers relationship quality, trust transference, brand attitude, skepticism character, and self-monitoring were 0.875, 0.766, 0.919, 0.710, and 0.805 respectively. All exceeded 0.7, indicating that they have good reliability.

Validity Analysis
Convergent validity refers to the degree to which questions or tests about similar potential traits truly fall onto the same construct. In this study, the average variance extracted (AVE) was used to examine the average variation explanatory power between the observed and the latent variables. Fornell and Larcker (1981) suggested that AVE should exceed 0.5 and its composite reliability (CR) should exceed 0.6.
Discriminant validity refers to the extent to which two constructs that are not related, are truly not correlated with each other. In this study, discriminant validity was measured using the square root of AVE. According to Bagozzi and Yi (1988) and Fornell and Larcker (1981) the construct being measured should be greater than the correlation between every construct to be considered as having good discriminant validity. As shown in Table 2, the square roots of AVE of the present study's variables were greater than the square of the correlation coefficients; thus, all have good discriminant validity.  Diagonal values are values of the square root of AVE; * indicates that p-value is less than 0.05 while ** indicates that p-value is less than 0.01.

Relationship Among SMI-Follower Relationship Quality, Trust Transference, and Brand Attitude
This study used the model 11 in software PROCESS 2.163 (Hayes, 2013) and the bootstrapping method (M=5000) to verify the mediating and moderating effects; relevant data are listed in Table 3 and Figure 2. For the relationship between SMI-follower relationship quality and trust transference, the values obtained for regression coefficient and t-value were 2.985 and 3.866 respectively, with a p-value of 0.000. Since the p-value is less than 0.05, this means that the explanatory power was significant. Therefore, H1 is supported; that is, the relationship quality has a positive impact on trust transference indicating that the higher the SMI-follower relationship quality, the more likely the followers will transfer their trust in SMIs to brands that the latter recommend.
As for the relationship between trust transference and brand attitude, the regression coefficient was 0.661, t-value was 12.581, and p-value was 0.000 which is less than 0.05. Therefore, the explanatory power of trust transference to brand attitude reached a significant level, supporting H2. This means that trust transference exerts a positive impact on brand attitude; that is, the higher the extent to which trust is transferred, the more positive the brand attitude of followers will be.

The Moderating Effect of Skeptical Character and Self-Monitoring
Regarding the interaction of skeptical character and relationship quality with trust transference, the regression coefficient was -0.475, t-value was -3.401, and p-value was 0.000, which is less than 0.05 indicating that advertising skepticism negatively moderates the impact of relationship quality on trust transference. This means that H3 is supported; that is, a high level of advertising skepticism lowers the relationship quality and the extent of trust transference.
In terms of the influence of self-monitoring on the moderating effect of advertising skepticism on the relationship between SMI-follower relationship quality and trust transference, the values obtained for regression coefficient and t-value were 0.074 and 2.840 respectively, and the p-value was 0.000. Since the p-value is less than 0.05, this means that H4 is supported. When followers' self-monitoring is high, it inhibits their advertising skepticism and strengthens the positive relationship between relationship quality and trust transference.

Conclusion
This study explored the relationship among SMI-follower relationship quality, trust transference, advertising skepticism, self-monitoring, and brand attitude. Based on the results, this study concludes the following: (1) The higher the SMI-follower relationship quality, the easier it is to generate trust transference; (2) trust transference is a partial mediator, which has a positive impact on relationship quality and brand attitude; (3) followers' high level of advertising skepticism negatively affects the relationship between SMI-follower relationship quality and trust transference; and (4) consumers' high degree of self-monitoring inhibits the negative moderating effect of advertising skepticism on SMI-follower relationship quality and trust transference.
The empirical results showed that a strong relationship between SMIs and followers enables the latter to transfer their trust in the former to the recommended products. To build a good relationship, one should work on it from three aspects, namely trust, commitment, and satisfaction. Therefore, SMIs have to ensure that the content of their posts is authentic, and should not casually recommend products to their followers merely for business benefits. Once the followers sense that the SMI is only after the profit, they are likely to end their relationship with the SMI. SMIs are suggested to stick to their original intention and provide valuable contents to satisfy their followers. When endorsing brands, SMIs must confirm whether the product meets the needs of their followers. Therefore, to enable the SMI economy to last, SMIs should pay close attention to the quality of their relationship with their followers.
In addition, the results showed that advertising skepticism negatively moderates the impact of relationship quality on trust transference. The higher the advertising skepticism, the higher the relationship quality is and the lower the trust transference is. A high level of self-monitoring inhibits the negative effect of SMI-follower relationship quality on trust transference; that is, the high self-monitoring trait of the follower will positively influence the recommendation of the SMI; followers will like and purchase the recommended product, and even recommend them to others.

SMIs Advised to Improve SMI-Follower Relationship Quality
SMIs have advantages over celebrities in terms of having a diverse background and knowledge, closer affinity with followers, and varied platforms that increase their public reach. Therefore, SMIs have easier access to followers than celebrities. In addition to a specific audience, they are able to cultivate loyal followers through long-term efforts. This research confirmed that the higher the SMI-follower relationship quality, the more effective their recommendation will be. If SMIs mean to commercialize their relationship with followers via advertising, they would need to maintain their professionalism and have great interaction with followers to retain their popularity.
Moreover, SMIs are advised to get to know their followers. This research found that followers with high self-monitoring are more likely to be influenced by SMIs. If their followers have a high level of advertising skepticism, they should be careful in advertising via their posts. SMIs can get to know the personality and characteristics of their followers through daily interaction, questionnaire survey, lucky draws, or with the aid of organizations. In brief, the better SMIs know their followers, the more likely it is to increase their conversion rate.

The Way Brand Manufacturers Select SMIs
This empirical study revealed that relationship quality has a partial mediating effect on trust transference. In the context of SMI economy, how brand manufacturers select the right SMIs as their brand spokespeople and optimize the return on their investments in advertising is a key issue. To help brand manufacturers solve this problem, a growing number of websites provide social media influencer rankings, such as NoxInfluencer, socialblade.com, and KOL Radar, which serve as references for them in terms of SMI selection.
In addition to the number of followers and viewing rate, brand manufacturers can also select SMIs based on the quality of their relationship with followers, based on their recent posts, audience groups, and interaction rate. When looking for SMIs to cooperate with, brand manufacturers should take into account the extent to which SMIs are related to the positioning, product, and target market of the brand, the SMI-follower relationship quality, and whether the number of ads is acceptable for the followers of SMIs who they mean to cooperate with. Otherwise, the investment of brand manufacturers may not reach the target consumers accurately, and may even cause the disgust of potential consumers.

Research Limitations of Present Research and Suggestions for Future Research
This study employed two personality traits, i.e., self-monitoring and advertising skepticism, as moderators. However, humans are complex, and may have other traits that can better reflect their thoughts and behaviors.
It is suggested that subsequent research explore whether SMI-follower relationship quality shows an inverted U-shaped curve after SMIs endorse certain brands; in other words, they should investigate whether followers will be averse to SMIs because of excessive endorsement, and even destroy their relationship.