Does a firm’s participation on social networking sites contribute to
customer trust in a B2B setting?
Does a firm’s participation on social networking sites contribute to
customer trust in a B2B setting?
Master thesis
Author: Jing Liu Student ID: s2072645
Email: j.liu.6@student.rug.nl •
University of Groningen
Faculty of Economics and Businesses, Department of Marketing
Master Thesis MSc Business Administration - Marketing - Marketing Research January 2012
1st supervisor: S. Beckers
Management Summary
The explosion in the popularity of social networking sites (SNSs) all over the world is ongoing for a decade. These platforms offer various opportunities for a firm to join discussions with its customers and thus, develop ongoing relationships with them. Nevertheless, the application of SNSs in a business environment, especially in a business to business (B2B) setting is still in its nascent stage. This is mainly because of insufficient knowledge of the advantages of managing customer relationship via these communication channels.
The primary goal of this study is to examine the impact of SNSs in a B2B setting on one dimension of customer relationship: Trust. It is proposed in this study that the level of a customer‘s awareness of a firm‘s presence on SNSs and the usage of these platforms represent the influence of firm‘s SNSs on the customer. To achieve a clear understanding of how a firm‘s participation on SNSs can influence trust, a research model has been developed by introducing two perception variables: perceived social presence and perceived interactivity. These two perceptual variables help to link the awareness of a firm‘s presence on SNSs to trust and link usage of firm‘s SNSs platforms to trust. Due to the low variation in the usage variable this study only managed to examine the effect of awareness. By means of path analysis and mediation analysis, the direct and indirect effects in the research model have been quantified.
Content
1. Introduction ...1
1.1 Background ...1
1.2 Relevance ...2
1.3 Research Question and Structure of the Thesis ...3
2 Research Model and Hypotheses ...6
2.1. Research Model ...6
2.2. Hypotheses ...12
3 Research Design ...18
3.1 Sample and Data Collection Procedure ...18
3.2 Measurement ...21
4 Results ...26
4.1 Measurement Analysis ...26
4.2 Hypotheses Testing ...29
5 Conclusion ...40
5.1 Summary and Discussion of the Findings ...40
5.2 Recommendations ...43
6 Limitations ...46
Reference ...48
Appendix ...65
Appendix A: KMO and Bartlett's Test on factor analysis ...65
Appendix B: Output of seemingly unrelated regressions in conducting path analysis ...66
1. Introduction
1.1 Background
Social networking sites (SNSs) are applications that allow communication among multiple users. They enable people to connect online by creating personal information profiles and inviting friends (Kaplan & Haenlein, 2010). In the past decade, usage of SNSs has been growing exponentially. According to the report of comScore Media Metrix (2006), every second internet user in the United States has visited at least one of the top 15 SNSs. Moreover, an increasing amount of time is spent on SNSs according to the reports of Nielsen Company (2010). Meanwhile, traditional forms of communication such as radio, television, magazines, and newspapers are losing effectiveness (Hennig-Thurau et. al, 2010; Nail, 2005). Marketers have allocated their resources to SNSs and started to participate in conversations via these communication channels (Barnes & Mattson, 2009) predominantly because of the low cost, extensive reach (Boyd & Ellison, 2008) and the necessity to adjust marketing strategy to the emerging new patterns in customer behavior (Dutta, 2010).
SNSs create the opportunities to build online communities where people can share their interests, discuss various topics and consequently develop ongoing relationships (Kim et al, 2010). In the business environment firms are trying to develop relationships with their customers on SNSs. They create Facebook pages and LinkedIn pages and try to engage customers in the discussions on Twitter (Barnes & Mattson, 2009). However, research examining whether SNSs can truly help firms to manage relationships with their customers is still in its infancy. On the other hand, the capability of SNSs to constitute a new technological development in managing customer relationship attracted a lot of attention of both practitioners and scholars (MSI research priority, 2010; Hennig-Thurau et. al, 2010). In this paper, the capability of SNSs in managing firm-to-customera relationships in a business to business (B2B) setting will be discussed. More specifically, the focus of this research is on deriving how SNSs can help firms foster customer trust in B2B setting. This study strives to use perceived social presence and perceived interactivity to explain the relationship between firm employing SNSs and customer trust.
a
1.2 Relevance
The relevance of studying trust in a SNSs setting
In comparison to satisfaction, loyalty and some other intensely discussed dimensions of customer relationships, customer trust is most likely to be influenced by SNSs. This is a consequence of the fact that the core paradigm change caused by SNSs is communication (Coyle & Vaughn, 2008). With the help of SNSs, real time communications between a customer and the vendor and between a customer and other customers can be enabled. Communication, especially timely communication, is the most widely recognized antecedent of trust (Morgan & Hunt, 1994; Anderson & Weitz, 1989; Anderson & Narus, 1990; Gefen & Straub, 2004; Michele, 2001). Therefore, it seems to be highly probable that trust is affected positively by SNSs. On the other hand, antecedents of satisfaction and loyalty are unlikely to be affected by firm‘s activity on SNSs directly. Szymanski and Henard (2001) found that researches on antecedents of customer satisfaction have focused predominantly on modeling effects of performance, equity, expectation, disconfirmation of expectation and affect. An examination on antecedents of customer loyalty revealed that two groups of predictors were found by researchers: customer-related factors (customer satisfaction, trust, psychological commitment and loyalty program membership) and product-related factors (perceived value, perceived quality, perceived fairness, switching cost and brand reputation) (Pan et al., 2012). In general, these aspects are unlikely to be influenced by SNSs directly. Hence, it is argued that compared to trust, satisfaction and loyalty are less obvious outcomes that could be achieved by employing SNSs. As a consequence, the main research focus of this study is the effect of SNSs on customers‘ trust.
The relevance of studying online trust in a B2B setting
setting but increased to levels comparable to those associated with a face-to-face setting over time. Their finding could also be applied to the SNSs setting.
The relevance of studying SNSs in a B2B setting
Kho (2008) argued that applying SNSs in corporate strategy results in a greater payoff for B2B compared to B2C market. She asserted that real-time communication and more personalized interactions, which can be achieved by using SNSs, are more important in deepening customer-vendor relationships and enhancing corporate credibility in a B2B market than in a B2C setting. Although SNSs have a great potential in managing customer relationships in a B2B market, the application of these platforms is still in its nascent stage. Barnes & Mattson (2009) posited that although a large number of firms are present on at least one of the SNSs, the usage of these platforms is still very basic and limited resources are allocated for managing these sites. According to the White Horse Survey Report (2010), the main obstacle for SNSs to be extensively employed in a B2B setting is the difficulty of proving the ROI of SNSs strategies. Therefore, it is of focal importance to investigate whether the new communication channels (SNSs) indeed benefit buyer-seller relationships, so that tangible evidence could be provided to marketing practitioners. Practitioners by means of this information would be more confident while making decisions about the adoption of SNSs in a B2B setting and would be able to allocate more adequately resources for it.
1.3 Research Question and Structure of the Thesis
1.3.1. Research Question
The potential of employing SNSs in managing customer relationship is in the interest of both scholars and practitioners. The current study seeks to explore whether a firm‘s participation on SNSs can contribute to one aspect of customer relationship: trust. Thus the primary goal of this study is to answer the question:
“Can a B2B firm foster customers’ trust by participating on SNSs?”
A firm that attempts to participate on SNSs should create a business profile(s) on SNSs. It can then manage its profile(s) by constantly posting information and interacting with others on SNSs (Dutta, 2010). By means of participating on SNSs a firm provides extra communication channels: firm‘s SNSsb. Customers can get involved in this new communication channels in two ways: be aware of
b
the firm‘s SNSs and use firm‘s SNSs. Drawing on expectation-disconfirmation theory (Oliver, 1980), customers form a pre-usage cognition when they become aware of a firm‘s presence on SNSs. During the course of the usage of firm‘s SNSs, customers‘ beliefs and attitudes will change as customers learn from such use (Oliver, 1980). Therefore, it is proposed that the new communication channels - firm‘s SNSs - exert an impact on customers‘ attitudes and beliefs in two ways: they affect customers when customers become more aware of firm‘s SNSs and they affect customers during the course of a customer usage of firm‘s SNSs. Obviously, for those customers who are not aware of the firm‘s presence on SNSs platforms, firm‘s SNSs are irrelevant. Since it is the customer‘s level of engagement that decides how and to what extent the customer is being affected by firm‘s SNSs, the level of awareness and the level of usage of firm‘s SNSs are used to represent the influence of firm‘s SNSs. The rest of this study strives to validate the impact of a firm‘s participation on SNSs on trust by examining whether the awareness of firm‘s SNSs and the usage of the firm‘s SNSs are effective in influencing trust. As a consequence, to address the problem stated in the beginning of this section, two questions need to be answered. First, is there a difference in the level of trust among customers
with different levels of awareness of the firm’s SNSs and/or with different levels of usage of the firm’s SNSs? The second question is: does the difference result from the customer becoming more aware of firm’s SNSs and/or using firm’s SNSs more intensively? If the answer to both questions is
yes, it can be concluded that becoming more aware of and/or using firm‘s SNSs more intensively should be considered as an effective way(s) of enhancing customers‘ trust.
basic presence on SNSs at the moment (Barnes & Mattson, 2009), thus no actual change in the quality of interactions is expected. Therefore, changes in perception are examined to capture the impact of firm‘s SNSs. In this paper, two perceptual constructs - social presence and interactivity - are selected and tested as mediators in order to elucidate the nature of the relationship between awareness and trust and the relationship between usage and trust. These two perceptual constructs are used to explain why firm‘s SNSs might be able to influence trust. As a consequence, the research question is formulated as follow:
“Does the perceived social presence and perceived interactivity mediate the relationship between awareness of firm’s SNSs and trust and the relationship between usage of firm’s SNSs and trust?” To address this research question, the following sub-questions will be answered at the end of this paper:
1) Do the level of a customer’s awareness of firm’s SNSs and the level of usage of firm’s SNSs influence the customer’s perceptions of the firm’s social presence and interactivity?
2) Do the impacts of the level of awareness of firm’s SNSs and the level of usage of firm’s SNSs on perceived social presence and perceived interactivity influence the customer’s trust?
1.3.2. Structure of the Thesis
2 Research Model and Hypotheses
This chapter begins with developing a research model that can link the awareness of firm‘s SNSs to trust and link the usage of firm‘s SNSs to trust. Definitions of focal constructs will be given afterwards. Finally, the expected relationships between the constructs from the research model will be elaborated and specific research hypotheses will be derived.
2.1. Research Model
even when the customers rarely or never uses the firm‘s SNSs. Consequently, perceptual constructs are employed to capture the impact of firm‘s SNSs. It is proposed in this paper that customer‘s awareness and usage of firm‘s SNSs will affect his/her perception of the firm, which in turn exerts direct effect on the customer‘s trust towards the firm.
The capabilities of awareness and usage to influence perception were studied in earlier researches. The impact of awareness on perception was investigated by D‘Arcy et al. (2009) in the field of information systems (IS) in an organization. The authors argued that user awareness of security countermeasures directly influences perceived certainty and severity of organizational sanctions, which leads to reduced IS misuse intention. Their study validated that awareness is able to result in change in perception, which laid a foundation for examining the impact of awareness on perception in a SNSs setting. Besides awareness, this study will also examine the impact of usage on perception. Hartwick and Barki (1994) provided empirical evidence that usage of a system led to well-differentiated beliefs and feelings regarding usage of the system. The theory of cognitive dissonance basically posits that attitude changes occur because of a person's behavior (Festinger, 1957). Bhattacherjee and Premkumar (2004) argued that the change over time in believes and attitudes is a result of disconfirmation and satisfaction, consequences of actual usage experience. The study of Bhattacherjee and Premkumar (2004) could be generalized across various contexts with respect to technology and usage. Therefore, it is arguable that usage of a firm‘s SNSs is also capable of influencing customers‘ perception.
2.1.1. Awareness and usage of firm’s SNSs
Many studies have defined SNSs based on its applications (Boyd & Ellison, 2008; Kaplan & Haenlein, 2010; Trusov et al., 2009). Basically, these platforms constitute online applications allowing people or organizations to maintain and manage the network of people relating to them by creating personal profiles and communicate with relevant people. There are two levels at which customers could be involved with a firm‘s SNSs: they can be aware of a firm‘s presence on SNSs and they can use a firm‘s SNSs platform. Awareness of firm‘s SNSs is simply defined as customer‘s awareness that a firm is presenting on SNSs platforms. The level of awareness is conceptualized by inquiring the number of a firm‘s business profiles on SNSs that a customer is aware of. Usage of firm‘s SNSs includes a series of physical activities, such as demonstrating affinity to a firm on SNSs (e.g. following a firm on Twitter), acquiring information from firm‘s SNSs consciously or unconsciously (e.g. read a firm‘s post on Facebook wall) and interacting with a firm or other customers via firm‘s SNSs and so forth. In this study, the effect of awareness and usage will be examined separately.
2.1.2. Trust
2.1.3. Mediating Variables
The perception construct is proposed to mediate the impact of awareness of firm‘s SNSs on trust and the impact of usage of firm‘s SNSs on trust. Hence, it is perceived as an outcome of firm‘s SNSs and a contributor of trust. Therefore, the criteria for selecting a perception construct that would function as a mediator in this paper are twofold: first, it needs to be supported as an antecedent of trust by scientific research and second, it should be influenced by the level of customer‘s awareness and usage with firm‘s SNSs.
A literature review related to trust shows most antecedents of trust can be categorized into three groups. The first one consists of customer-oriented trust antecedents including individual‘s propensity to trust (Bhattacherjee, 2002; Kim et al., 2003), satisfaction with previous outcome (Ganesan, 1994) and relationship age (Anderson & Weitz, 1989). The second group is composed of firm-oriented trust antecedents including perceived organizational reputation (Casalo et al., 2007), perceived social presence (Gefen & Straub, 2004; Cyr et al., 2007; Hassanein & Head, 2004), perceived interactivity (Kim et al., 2011), perceived size of the organization, expertise, relational WOM, ease of use of the website, information quality, customization and personalization capacity, privacy assurances, security features and third-party guarantees (Geykens et al, 1998). Two-sided trust antecedents include communication, shared value (Morgan & Hunt, 1994), rapport (Macintosh, 2009), cooperation, cultural similarity (Geykens et al, 1998). Among these empirically supported antecedents, perceived social presence and perceived interactivity are selected as mediating variables. Because these two constructs are the most relevant in media studies (Short et al., 1976; Rafaeli, 1988) and because they are likely to be influenced by firm‘s SNSs.
relationships to the public (Trusov et al., 2009). SNSs push the presence of firms to a new level. Therefore, perceived social presence might be influenced if a customer sees a firm on SNSs and/or uses these platforms.
Interactivity is also highly relevant in media studies. Rafaeli (1988) proposed interactivity theory and he argued that interactivity is a variable characteristic for various communication settings: from the unmediated, face-to-face and intimate to the relatively anonymous and mass mediated. Apart from the fact that the interactivity is an important topic in media studies, it is also argued to be a distinctive feature of SNSs. Interactivity has been extensively studied as a unique feature of a website in comparison to other mass media (Duncan & Moriarty, 1998; Hoffman & Novak, 1996; Payne & Frow, 2005). SNSs powered by Web 2.0, which distinguishes itself by allowing end users to utilize the World Wide Web, provide an even more interactive environment than websites (Kaplan & Haenlein, 2010). SNSs create an opportunity of two-way communication that is open, transparent, non-hierarchical and real time (Dutta, 2010). In the SNSs environment, customers have more control over the information they receive and customers have more control over their communication with a firm and other related parties. Therefore, perceived interactivity might be influenced if customers are involving with firm‘s SNSs.
In conclusion, perceived social presence and perceived interactivity are antecedents of trust. They are highly relevant constructs in media research and they are likely to be influenced by SNSs. Therefore, these two constructs are selected as mediating variables in the relationship between the awareness of firm‘s SNSs and trust and in the relationship between usage of firm‘s SNSs and trust. Definitions of these two constructs will be given in the following two sections.
2.1.4. Perceived Social Presence
videoconferencing) (Yoo & Alavi, 2001), personal characteristics (e.g. gender) (Gefen & Straub, 1997) and the context in which the communication occurs (within zero-history group or established group) (Francescato et al., 2006). In this paper, customer's perception of a firm‘s social presence is measured and the influence of firm‘s SNSs on this construct will be examined.
2.1.5. Perceived interactivity
Interactivity is defined as the extent to which a user can participate in modifying the form and content of the mediated environment in real time (Steuer, 1992). Literature tries to measure interactivity in three approaches: functions-based, process-based and perception-based. As a result, the number of interactivity functions, actual interactivity and perceived interactivity are measured respectively (McMillan & Hwang, 2002). In this study, customer‘s perception is of focal interest. Therefore, the perception-based approach to measure interactivity is adopted in this paper. McMillan and Hwang (2002) grouped articles defining interactivity with focus on perception. Although there are different phrases used in various papers for denoting the same concept, it is generally agreed that interactivity has three dimensions: perceived two-way communication, perceived user control and perceived responsiveness (Mollen & Wilson, 2010; McMillan, 2002; Heeter, 1989).
With the two newly selected perceptual constructs (perceived social presence and perceived interactivity), a research model is finally established (Figure 1). In conclusion, this study proposes that the awareness of firm‘s SNSs and usage of firm‘s SNSs will exert impacts on perceived social presence and perceived interactivity respectively. These two perceptual constructs will then influence trust. In addition, studies have found that interactivity is a cause of social presence (Wang et al., 2007; Fortin & Dholakia, 2005). In this study, this relationship will also be considered. The rationale of these proposed relationships will be elaborated in section 2.2 and will be tested in section 4.2.
2.2. Hypotheses
In this section, the theoretical rationale for hypothesized relationships between constructs depicted in the research model will be reviewed. As a result, specific hypotheses will be derived accordingly.
2.2.1. Awareness and usage of firm’s SNSs positively influence perceived social presence Usage of firm‘s SNSs can reinforce perception of social presence in multiple ways. Firstly, it is found that social presence is enhanced through peer interactions in an online learning environment (Garrison & Cleveland-Innes, 1995). Thus, it is argued that customer‘s interactions with other customers via firm‘s SNSs can enhance his/her perception of social presence. Secondly, it was suggested by Steinbrück et al (2002) that using photographs in an online vendor‘s website is effective in creating social presence because it makes electronic exchange closer to face-to-face exchange. Thus, it can be expected that the disclosure of photos on firm‘s SNSs will also increase user‘s perception of social presence. Thirdly, electronic word-of-mouth, which usually spread in a SNSs environment (Trusov et al., 2009), is proved to be able to foster social presence (Kumar & Benbasat, 2006). Customers involving with firm‘s SNSs are likely to be exposed to word of mouth about the firm and thus should perceive that the social presence of the firm is enhanced. Finally, the customer takes advantage of the instant and two-way conversations - which is typical human interpersonal interaction - on firm‘s SNSs will also have higher perception of social presence (Wang et al., 2007). Recapitulating, customers using firm‘s SNSs should have higher perception of social presence. Among those who use firm‘s SNSs, it is proposed that customer characterized by higher level of usage has higher degree of social presence perception. Because the customers with higher level of usage interact with firm and other customers more frequently, have more knowledge about firm‘s activities on SNSs, and so forth. These customers are more likely to benefit from firm‘s SNSs and thus tend to have higher perception of firm‘s social presence. In conclusion, the impact of firm‘s SNSs on social presence is higher for customers that often use these platforms in comparison with those who rarely use it. Therefore, it is hypothesized that:
H1a: The level of a customer’s usage of a firm’s SNSs platform(s) has a positive impact on the
customer’s perception of the firm’s social presence.
1973), customers rely on the observable signals they receive from SNSs through past experience to form a judgment (the level of social presence) about unobservable attributes on firm‘s SNSs. Nowadays, SNSs are widely used in everyday life. It is very likely that even if customers do not use firm‘s SNSs, they use SNSs in their private life (comScore Media Metrix report, 2006). Therefore, it is not difficult for customers to get a rough idea what is being offered on firm‘s SNSs, even if they are not really using these platforms. For example, customers should know that they have a chance to socialize with a firm and other customers if the firm is present on SNSs. By doing so, customers form a judgment about unobservable attributes on firm‘s SNSs without actually using these platforms. In addition, when customers are more aware of a firm‘s presence on SNSs, they perceive the possibility to socialize with the firm and other customers is higher. Consequently, customers with a higher level of awareness of the firm‘s presence on SNSs tend to have higher perception of social presence. Therefore, it is hypothesized that:
H1b: The level of a customer’s awareness of a firm’s presence on SNSs has a positive impact on the
customer’s perception of the firm’s social presence.
2.2.2. Social presence positively influences trust
validated in an avatar-mediated networking (Bente et al., 2008) website setting (Hassanein & Head, 2004), e-commerce context (Gefen & Straub, 2003; Gefen & Straub, 2004) and social virtual setting (Mantymaki & Salo, 2010). In this paper, the influence of social presence on trust in a SNSs setting will be examined. Therefore, it is hypothesized that:
H2: A customer’s perception of a firm’s social presence has a positive impact on the customer’s trust
towards the firm.
Figure 2: Comparison of popular online marketing tools on interactivity
Source: Liu and Shrum 2002
In addition, it is proposed that customers characterized by a higher level of usage of firm‘s SNSs have a higher perception of the firm‘s interactivity. Customers with a high level of usage are exploring more intensively the possibilities given by the communication tools on SNSs. Or at least they observe more conversation occurring between the firm and others. These customers see more opportunities for being involved in a two-way communication with the firm. Heavy users of firm‘s SNSs also take initiative while searching for information. Thus they have more control over the volume and source of information. All in all, the more the customers are using a firm‘s SNSs platform, the higher their perception of the firm‘s interactivity is. Therefore, it is hypothesized:
H3a: The level of a customer’s usage of a firm’s SNSs platform(s) has a positive impact on the
customer’s perception of the firm’s interactivity.
when customers are more aware of a firm‘s presence on SNSs, they perceive that they have more possibilities to engage in discussions about the firm. They also perceive that there are more information sources available to them. In conclusion, customers with a higher level of awareness are likely to have a higher perception of interactivity. Therefore, it is hypothesized that:
H3b: The level of a customer’s awareness of a firm’s presence on SNSs has a positive impact on the
customer’s perception of the firm’s interactivity.
2.2.4. Perceived interactivity positively influences trust
Perceived interactivity has three dimensions: user control, two-way communication and responsiveness. All dimensions exert a positive impact on trust and will be discussed successively. First, Dayal et al. (1999) found that those who perceive they have more control over the purchase process tend to be characterized by a higher level of online trust. Cyr et al. (2009) also argued that if users are allowed to control and access the information, they are more likely to trust the information. According to these findings, customers that perceive that their control is high tend to trust the firm. Second, two-way communication is also recognized as a critical factor for developing trust (Anderson & Weitz, 1989; Parasuraman et al., 1991). Third, employee‘s responsiveness is argued to generate customer trust towards a firm (Balasubramanian et al, 2003). Thus, it is argued that customers who perceive a firm to be responsive tend to have a high level of trust towards the firm. Moreover, with the possibility to talk directly to a firm (two-way communication) and receive response effectively (responsiveness), uncertainty can be reduced and thus trust can be increased (Anderson and Narus, 1990; Morgan & Hunt, 1994). According to the discussion above, a higher degree of perceived interactivity brings a higher level of trust. Therefore, it is hypothesized that:
H4: A customer’s perception of a firm’s interactivity has a positive impact on the customer’s trust
towards the firm.
2.2.5. Perceived interactivity positively influences perceived social presence
perception of social presence. Fortin and Dholakia (2005) suggested that interactivity generates feelings of social presence through open channels allowing for two-way communication. It was also found that an active interactive environment, in which a computer provides messages based on earlier input, contributes to children‘s perception of social presence (Tung & Deng, 2006). Animesh et al. (2011) also observed a high correlation between interactivity and social presence. The relationship between these two constructs will be considered in the research model. Therefore, it is hypothesized that:
H5: The level of perceived interactivity has a positive impact on the level of perceived social
3 Research Design
3.1 Sample and Data Collection Procedure
The data used to test the research model proposed in chapter two were collected through a mail survey sent out to customers of a B2B firm in the information industry. As firms nowadays are increasingly international in character and SNSs are not constrained by national borders, the necessity of establishing cross-national validity of the theoretical concept and model becomes relevant (Kumar et al., 1995a; Frazier et al., 1989). Based on this consideration, customers from six countries (Netherlands, UK, US, Australia, Belgium and Spanish) were targeted through the survey. These six countries were chosen because the firm participating in the research has established local profiles on SNSs in these six countries. The focal firm appeared for the first time on SNSs (Twitter, Facebook and LinkedIn) in 2009. However, like most of other B2B firms, the focal firm was used to develop hand-shaking relationships with its customers (Issa, Flood, Caglasin, 2003; Segev; Beam & Gebauer, 1997). These firms were doubtful about the value of using SNSs (White Horse Survey Report, 2010). The other firms in the B2B sector were accustomed to a similar strategic approach and, as a consequence limited resources have been invested in SNSs (Barnes & Mattson, 2009). So far, these firms did not manage to generate active customers‘ engagement through SNSs (White Horse Survey Report, 2010). All in all, these aspects made the focal firm a typical example, as it represents the majority of B2B organizations. Another reason why the focal firm constituted a suitable business case was because it was possible to obtain a natural control group. In case of firms that are actively engaging in SNSs activities, customers have full knowledge about which SNSs platform a firm is presenting on. For these businesses the level of awareness is consistent among all customers. Therefore, impact of awareness can hardly be tested. However, in the case of the focal firm, it was possible to distinguish customers with different levels of awareness. Thus a comparison can be made among these customers. The downside of observing the focal firm was that customers did not use the firm‘s SNSs heavily yet. This was partly caused by the fact that the focal firm was barely marketing this new channel. Therefore, the level of the customer‘s usage was very limited. As a consequence, the possibility of testing the impact of usage was a big concern in this study.
of the receivers completed the survey. There was a reminder email scheduled to be sent to those who did not open the first email one week after sending. Also making the survey available on the firm‘s profiles on SNSs was planned. However, these follow-ups were called off because three complaints were directed to the focal firm regarding the privacy issues. These changes in plan had some negative impact on the sample size. Especially, it is likely that these changes have decreased the amount of responses from those who were aware of the firm‘s presence on SNSs. Because there was a good chance that distribution of the survey on SNSs would have attracted more customers who were aware of the firm‘s presence and/or used the firm‘s SNSs.
As the mailing list contained a few contacts that are suppliers or partners of the firm, a screening question was used asking the receiver‘s relationship with the focal firm to ensure that only customers‘ responses were obtained. After filtering out non-customers responses, responses with constant score or logically inconsistent answers (e.g. indicated he/she posted something about the focal firm but the frequency of posting is ―never‖) were deleted from the sample as well. In addition, inconsistent responses to reverse-coded items were examined and if this occurred to several pairs these responses were eliminated. To detect outliers, three factor analyses were run to calculate respondents‘ scores on: trust, social presence and interactivity respectively. Factor analyses were executed with items validated in early studies for the three constructs. One factor was set to be extracted for each construct. The factor scores were standardized within each construct. In five cases the standardized scores of at least one of the three constructs excessed the critical value of 2.58 (p=0.01). These five responses were regarded as outliers and hence eliminated. Nine respondents who completed the majority of questions but left a few items incomplete were included in the final sample. The reason to include these incomplete responses is caused by the fact that each respondent is important due to the relatively small sample size. The incomplete answers were labeled as missing values for these nine responses.
(trust, social presence, interactivity, awareness, usage & relationship age) among all five groups. Cross-tabulation was used to test variables with nominal scales (gender and nationality). Only variables presenting nationality and gender were significantly different between the five groups. Nationality difference can be explained by the time difference. The survey was sent out at 10:00 am Amsterdam time. In the first two hours only European respondents could fill in the survey while the respondents from the US and Australia started the survey after four hours. Gender difference is also partially a result of nationality. In European countries (UK, The Netherlands, Spain, and Belgium) the ratio of female respondents to male respondents is 1.9, in Australia and the US together this ratio is 5.6. It can be concluded that non-response bias was not a problem for this study.
Since this research obtained a self-reported survey relying on responses in a cross-sectional single setting, it is likely to suffer from the common method bias (Animesh et al., 2011). Common method bias is believed to be a major flaw when implementing subjective measures (Wall et al., 2004). Common method bias occurs when systematic error variance - resulting from the method used to collect the data - has a serious confounding influence on the empirical results and generates potentially misleading conclusions (Podsakoff et al., 2003). The extent of common method bias was evaluated by employing Harman‘s single-factor test according to Podsakoff et al. (2003). All subjective items (all items designed to measure trust, social presence and interactivity) were loaded into a principal component factor analysis and an unrotated factor solution was derived. Although one factor accounted for 44.58% of the total variance, no single factor emerged from the factor analysis or no single general factor accounted for the majority of the covariance between the measures (Podsakoff et al., 2003). Consequently, Harman‘s single-factor test indicated that the study was robust against the common method bias. The relatively high variance accounted by the one general factor appeared partially due to the causal relationships among the three constructs. Not only social presence and interactivity are the causes of trust, interactivity is also recognized as a cause of social presence (Fortin & Dholakia, 2005; Animesh etc., 2011).
regular mail (7.04%), visits (2.55%), video meeting (0.36%) and SNSs (0.27%). The distribution of responses from the six countries were corresponding with the contact lists of the focal firm with 33.33% respondents coming from the Netherlands, 27.27% from the UK, 14.39% from the US, 10.61% from Australia, 10.61% from Belgium and 3.79% of the respondents coming from Spain.
3.2 Measurement
3.2.1 Focal constructs
The main purpose of the survey was to ask the customers to report on: trust towards the focal firm, awareness of the firm‘s presence on SNSs, usage of the firm‘s SNSs, and perception of the focal firm‘s social presence and interactivity. Whenever possible, established multi-item measures that previously had been validated were used. Trust, social presence and interactivity were measured with well-established seven-point scales. Awareness of firm‘s SNSs and usage of firm‘s SNSs have not been found measured by any researcher yet. The development of scales in order to measure these two constructs was guided by construct definitions and practice in reality. The items measuring these focal constructs can be found in table 1 and 2.
Trust was measured by ten items developed and validated by Kumar et al. (1995a, b). As discussed
in section 2.1.2, only affective trust will be examined in this research. The ten items measured two aspects of affective trust: honesty and benevolence.
Perceived social presence was measured by five items that were adapted from Gefen and Straub
(1997). The five items have been validated by many researchers (Qiu & Benbasat, 2009; Fortin & Dholakia, 2005; Animesh etc., 2011).
Awareness of a firm‘s presence on SNSs has not yet been studied. A straightforward item was
created: Are you aware that Firm Xc is present on the following social networking sites? Twitter, Facebook, LinkedIn, None (you can choose multiple options). In this way different levels of awareness can be derived. There are in total four levels of awareness: the focal firm was present on none, one, two or three SNSs.
Usage of firm‘s SNSs has not been conceptualized in earlier studies either. According to the
observation and monitoring on SNSs, it was discovered that SNSs were mainly used for acquiring information, sharing information and communication (Hanna & Donnelley, 2009). These can be measured by asking whether a customer has ever read information on firm‘s SNSs, shared information about the firm on SNSs and corresponded with the firm on SNSs. Additionally, another item inquiring whether the customer ―follows‖ a firm on Twitter or LinkedIn or ―likes‖ it on Facebook was incorporated in the questionnaire. In this way four groups of items measuring usage were developed. The frequency of the different types of usage was also added in the survey so that the intensity of usage could be reflected.
3.2.2 Control variables Gender
Gender should be used as a control variable because Gefen and Straub (1997) found that women perceive social presence to be higher in comparison with men. Gefen and Straub (1997) asserted that in oral discourse women‘s inclinations are associated with a stronger sense of social presence and human contact than men.
Relationship age
The age of relationship (the duration of being a customer) has long been supported as an indicator of trust. This is because a long relationship generates experience which breeds trust (Anderson & Weitz, 1989). In addition, the older the relationship the greater the likelihood it has passed through a critical ―shakeout‖ period of conflict (Anderson & Weitz, 1989). Furthermore, partners learn from each other and deepen mutual understanding over time which improves the affective quality of the relationship (Sullivan & Peterson, 1982). In this study affective trust is argued to be affected by the duration of the relationship between customer and firm.
c
Nationality
The reason for controlling for nationality in this study is twofold: firstly, there might be a difference in the propensity to trust across different countries (Huff & Kelley, 2003) and secondly, the local offices of the focal firm might perform differently across countries which might lead to a different level of trust in a specific region.
To validate these items in real practice in such a way that it could be adapted to the focal firm, an expert group interview was set up. The expert group consisted of one global marketing manager and two marketing specialists. They helped to improve the wording of the items in order to make it easier for customers to understand them. As a consequence, the items were made more realistic in this particular setting. A pretest with customers of another B2B firm was conducted in order to examine the validity of the survey to improve the layout and so forth.
Table 1: Operationalization, mean, variance and source of the scales measuring the construct of trust, social presence and interactivity
Variable Variable
label
Operationalization
Seven point scale from strongly disagree (1) to strongly agree (7)
Mean (Var)
Trust (Kumar, Scheer and Steenkamp (1995a,b ))
Honesty T1 Even when Firm X gives me a rather unlikely explanation or
information, I am confident that Firm X is telling the truth.
4,75(2.05) T2 Firm X usually keeps the promises that it makes to our organization. 5,37(1.57)
T3 Whenever Firm X gives me advice on my operation, I know that Firm
X shares its best judgment.
5,26(0.99) T4 Our organization can count on Firm X to be sincere. 5,45(1.08)
T5 Firm X often provided me with information that has later proven to be
inaccurate.a
3,04(2.42) Benevolence T6 Though circumstances change, I believe that Firm X will be ready and
willing to offer me assistance and support.
4.53(0.90) T7 When Firm X is making important decisions, Firm X is concerned
about welfare of our organization.
5,24(0.92) T8 When I share our problems with Firm X, I know that Firm X will
respond with understanding.
5.24(0.93) T9 In the future, I can count on Firm X to consider how its decisions and
actions will affect our organization.
4,66(1.19) T10 When it comes to things that are important to our organization, we can
depend on support from Firm X
Perceived Social presence
(Gefen and Straub (1997))
To what extent does Firm X gives you following senses?
S1 1. Human contact 4,93(1.68)
S2 2. Personalness 4,88(1.76)
S3 3. Sociability 4,78(1.38)
S4 4. Human warmth 4,74(1.44)
S5 5. Human sensitivity 4.75(1.57)
Interactivity (Song and Zinkhan (2008); Voorveld et al. (2011))
User control I1 I feel that I have a great deal of control over my interaction with Firm
X
5,13(1.42) I2 While I am interacting with Firm X, I can choose freely what I want to
hear/read and say/contribute.
5,30(1.15) I3 While I am interacting with Firm X, I have absolutely no control over
my communication. a
2,52(1.09) I4 While interacting with Firm X, my action decides the kind of
experience I get.
4,54(1.33) Two-way
communication I5
Firm X facilitates two-way communication.
5,20(1.31)
I6 Firm X gives me the opportunity to talk back. 5,50(0.92)
I7 Firm X facilitates concurrent communication 5,06(1.14)
I8 Firm X enables conversation. 5,01(1.69)
I9 Firm X does not encourage customers to talk back. a 2,85(1.77)
I10 Firm X is effective in gathering customers‘ feedback. 4,58(1.28)
Responsiveness I11 Firm X responds to my requests very quickly. 5,00(2.03)
I12 I am able to obtain the information I want from Firm X without any
delay.
4,70(1.93) I13 When I communicate with Firm X I feel I received instantaneous
information.
4,46(2.08) a:
Table 2: Operationalization, means and variance of awareness and usage.
Variable Variable
label
Operationalization Mean (Var)
Awareness A1
Are you aware that Firm X is present on the following social networking sites?
- Twitter - LinkedIn - Facebook
0.35(0.63)
Usage U1 Are you following Firm X on the social networking sites?
- Twitter - LinkedIn - Facebook
0.03(0.03) U2 Do you ever read information on Firm X's social networking site
profile(s) below?
- Twitter - LinkedIn - Facebook
0.04(0.03)
U3 Have you ever had any correspondence with Firm X on following
social networking sites?
- Twitter - LinkedIn - Facebook
0.01(0.01)
U4 Do you ever tweet, share or post any information about Firm X
on social networking sites below?
- Twitter - LinkedIn - Facebook
4 Results
The five focal constructs in this study are: usage of firm‘s SNSs, awareness of a firm‘s presence on SNSs, perceived social presence, perceived interactivity and trust. As can be seen in table 2 the usage of a focal firm‘s SNSs was extremely low. There was not enough variation in this construct in order to provide any significant results. Hence, the impact of usage of firm‘s SNSs cannot be examined with the available data. As for awareness, 27 out of 142 respondents were aware that the focal firm was present on at least one of the three SNSs platforms. The level of awareness is measured by inquiring the number of a firm‘s business profiles on SNSs that a customer is aware of. For trust, social presence and perceived interactivity, exploratory factor analysis was employed to evaluate the consistency of the measures and to compute scores of these measures for further analysis.
4.1 Measurement Analysis
messages but also senders of messages, customers should perceive themselves as having more control. Factor four was named as responsiveness because all three variables designed to measure responsiveness loaded highly on this factor. There were two variables loading highly on factor four unexpectedly: variable I1 and I10. These two variables both reflect a firm‘s responsiveness to some extent, therefore this finding is reasonable. It is understandable that variable I1 loaded highly on factor four because customers perceive to have more control over their interaction with a firm when the period of waiting for response is short. It is also reasonable for variable I10 to load high on factor four because variable I10 measures the effectiveness of a focal firm gathering feedback and the word ―effective‖ reflects whether customers perceive that their feedback was heard and timely processed. UCTC and responsiveness represent two aspects of interactivity.
Table 3: Pattern Matrix table of the four-factor solution
Variable label
Factor
1 2 3 4
T1 Even when Firm X gives me a rather unlikely explanation or information, I am confident that Firm X is telling the truth.
.530
T2 Firm X usually keeps the promises that it makes to our organization. .525 -.427 T3 Whenever Firm X gives me advice on my operation, I know Firm X
shares its best judgment.
.547
T4 Our organization can count on Firm X to be sincere. .695
T5 Reversed Firm X often provided me with information that has later proven to be inaccurate.
.462
T6 Though circumstances change, I believe that Firm X will be ready and willing to offer me assistance and support.
.659
T7 When Firm X is making important decisions, Firm X is concerned about welfare of our organization.
.637
T8 When I share our problems with Firm X, I know that Firm X will respond with understanding.
.650
T9 In the future, I can count on Firm X to consider how its decisions and actions will affect our organization.
.715
T10 When it comes to things that are important to our organization, we can depend on support from Firm X.
.792
S1 Human contact -.816
S2 Personalness -.789
S4 Human warmth -.854
S5 Human sensitivity -.838
I1 I feel that I have a great deal of control over my interaction with Firm X. .415 -.426
I2 While I am interacting with Firm X, I can choose freely what I want to hear/read and say/contribute.
.506
I3 Reversed While I am interacting with Firm X, I have absolutely no control over my communication.
.639
I5 Firm X facilitates two-way communication. .661
I6 Firm X gives me the opportunity to talk back. .696
I7 Firm X facilitates concurrent communication. .488
I8 Firm X enables conversation. .493
I9 Reversed Firm X does not encourage customers to talk back. .692
I10 Firm X is effective in gathering customers’ feedback. -.385
I11 Firm X responds to my requests very quickly. -.875
I12 I am able to obtain the information I want from Firm X without any delay. -.878
I13 When I communicate with Firm X, I feel I received instantaneous information.
-.810
Cronbach‘s alpha coefficients were calculated to measure the reliability of the four factors. As represented in table 4, the Cronbach‘s alpha coefficient values for all four factors greatly exceeded the minimum acceptable value (0.7) (Hair et al., 2006). It indicates that measures yield consistent results.
Table 4: Summary of Cronbach‘s alpha of four factors
Factor label Cronbach’s alpha Number of variables
Trust 0.902 10
Social presence 0.945 5
UCTC 0.883 7
Responsiveness 0.914 5
4.2 Hypotheses Testing
This section begins with employing seemingly unrelated regressions (SUR model) to conduct path analysis. By using path analysis, the direction and extent of direct effects in the research model are examined. Due to the complexity of mediated effects in the path analysis, each indirect effect suggested in the research model is analyzed separately using mediation analysis and a Sobel test. In the end, a summary of direct and indirect effects will be given.
Path Analysis
Path Analysis is a method used to examine causal relationships between two or more variables. It is used mainly in the attempt to understand comparative strengths of direct and indirect relationships among a set of variables (Mitchell, 1993). Because of its unique application, in this study path analysis was employed to acquire a clear understanding of how and to what extent the constructs are related to each other in the research model. To conduct path analysis, a SUR model was estimated. Zellner (1962) formulated the SUR model as p correlated regression equations, while regular standard ordinary least square (OLS) ignore any correlation among error across equations. Employing SUR model as a solution to path analysis is suggested by Beasley (2008). He argued that SUR model is of particular relevance while conducting path analysis, because unlike a standard linear OLS model, a SUR model allows a variable to be both dependent variable and independent variable, which fits exactly the situation in path analysis.
To conduct path analysis it is required that for each endogenous variable (Y) a regression predicting Y from all other variables which are hypothesized to have direct effect on Y should be estimated (Pedhazur, 1982). According to this requirement, a SUR model consisting of four equations was estimated (Figure 3) (The output of SUR model can be found in Appendix B). Since awareness does not influence trust directly, it was not included in the first equation to predict trust.
Figure 3: Seemingly unrelated regressions (SUR): estimated equations
a:
The relevant results produced by SUR are shown in table 5. Based on the results of SUR figure 4 was drawn. The path diagram shown in figure 4 depicts the path coefficients, path significances, and variances explained (R2) by the research model.
Table 5: Results of SUR model
Equationsa
Dependent variable
Independent variable P-value
of equation R2 awareness Social presence UCTC Responsiveness 1 Trust NA 0.114(0.045)b 0.174(0.033) 0.373(0.000) 0.000 55.5% 2 Social presence 0.214(0.017) NA 0.667(0.000) 0.195(0.014) 0.000 48.6% 3 UCTC 0.217(0.015) NA NA NA 0.013 8.2% 4 Responsiveness 0.266(0.023) NA NA NA 0.022 7.4% a
Relationship age, gender and nationality were controlled for in these equations.
b
Standardized estimate (p-value).
Figure 4: Path diagram of research model
a) Number 1(number 2) represent the standardized coefficient (p-value).
b) R2trust, R2SP, R2uctc and R2Re represent R-square of the equation where trust, social presence, UCTC and
As can be seen in figure 4, all hypothesized direct effects are significant. The coefficients of path ―awareness social presence‖, path ―awareness UCTC‖ and path ―awareness responsiveness‖ suggest that awareness has significant and positive impact on perceived social presence and perceived interactivity (UCTC and responsiveness), supporting H1b and H3b. The coefficients of path ―social presence trust‖, path ―UCTC trust‖ and path ―responsiveness trust‖ suggest that perceived social presence and perceived interactivity have significant and positive effects on trust, which support H2 and H4. The coefficients of paths ―UCTC social presence‖ and path ―responsiveness social presence‖ suggest a significant and positive impact of interactivity on perceived social presence, which supports H5.
The W Statistic proposed by Specht (1975) was employed to evaluate how good the research model fits the data. The fit of the research model was tested by estimating the model‘s omitted paths. The ―just-identified model‖ (Figure 5), which was obtained by including the path ―awareness trust‖ that was previously not analyzed by the research model, was estimated using the SUR model. Moreover, two reduced models based on the research model were also tested and benchmarked against the research model. Reduced model (1) was obtained by eliminating the path ―UCTCsocial presence‖ from the research model (Figure 6). Reduced model (2) was the result of eliminating the path ―responsivenesssocial presence‖ from the research model (Figure 7). The values of R-square for equations within each model estimated by means of SUR were used to calculate the W statistic (Table 6).
Figure 5: Path diagram of the just-identified model
Figure 6: Path diagram of the reduced model (1)
Figure 7: Path diagram of the reduced model (2)
Table 6: Summary of the W statistic to evaluate model fit
Comparison Original model New model W statistic p-value
1 Just-identified model Research model 0.083 0.773
2 Research model Reduced model (1) 4.1312 0.042
3 Research model Reduced model (2) 0.649 0.394
model (1) suggested that the elimination of path ―UCTCsocial presence‖ significantly decreased the fit of the model with the data (p= 0.042). Furthermore, the comparison between the research model and the reduced model (2) leads to the conclusion that the research model outperforms the reduced model (2) in explaining data (p= 0.394). Besides, the fact that the reduced model (2) surpasses the explanatory power of the reduced model (1) indicates that the path ―UCTCsocial presence‖ is more relevant than the path ―responsivenesssocial presence‖.
Mediation Relationships
An examination of the research model suggests four potential mediated relationships. As interactivity is represented by UCTC and responsiveness, seven indirect effects (also known as mediated effects (Bollen, 1987)) are suggested. Mediation analysis is employed in order to discover the significant relationships among the seven presumed mediation relationships. By doing this, a clear picture of the system of indirect effects in the path analysis can be achieved. In a multiple mediation model where multiple mediators account for the relationship between independent variable and dependent variable, it is recommended that mediators should be tested simultaneously (Preacher & Hayes, 2008). Although testing multiple mediators simultaneously is possible, it is only appropriate when these mediators are conceptually distinct and not too highly correlated (Tasa et al., 2011). In this study the three mediators are conceptually related with each other and correlation analysis also suggests significant correlations between them. Thus, each of the seven presumed mediation relationships in the research model will be tested separately using mediation analysis and the Sobel test.
mediation occurs. Whether there is an effect is decided on whether c’ is zero or nonzero, instead of its statistical significance. Baron and Kenny (1986) argued that trivially small coefficients can be statistically significant in large sample sizes and very large coefficients can be non-significant in small sample sizes.
Figure 8: Paths of a mediation effect
In this study three sets of mediation relationships are proposed in the research model: first, social presence and interactivity function as mediators in the relationship between awareness and trust; second, interactivity functions as a mediator in the relationship between awareness and social presence; third, social presence functions as a mediator in the relationship between interactivity and trust. As interactivity is represented by UCTC and responsiveness, seven mediation effects will be tested. To evaluate these mediation effects in the research model a series of regressions were estimated (Appendix C). Table 7 displays the four critical coefficients (a, b, c, and c’) of each mediation relationship. The Sobel test is employed to test the significance of each mediated effect (Sobel, 1982). The outcomes of the Sobel test are also displayed in table 7.
Table 7: Relevant information in testing for mediation effect
Mediation path
standardized coefficient (p-value) Sobel test
a b c c’ P-value Std. indirect effect
1 AWSPTR 0.285(0.001) 0.5(0.000) 0.216(0.010) 0.073(0.331) 0.002 0.143 2 AWUCTC—TR 0.197(0.019) 0.564(0.000) 0.216(0.010) 0.104(0.135) 0.023 0.111 3 AWRETR 0.185(0.028) 0.670(0/000) 0.216(0.010) 0.092(0.141) 0.030 0.124 4 AWUCTCSP 0.197(0.019) 0.630(0.000) 0.285(0.001) 0.161(0.015) 0.021 0.124 5 AWRESP 0.185(0.028) 0.506(0.000) 0.285(0.001) 0.191(0.009) 0.035 0.094 6 UCTCSPTR 0.531(0.000) 0.141(0.075) 0.254(0.001) 0.180(0.039) 0.084 0.075 7 RESPTR 0.210(0.012) 0.141(0.075) 0.528(0.000) 0.498(0.000) 0.142 0.030
a) AW represents awareness; SP represents social presence; RE represents responsiveness; TR represents trust;
b) The arrow shows the path from independent variable to mediator and then to dependent variable;
The test for three sets of mediation relationships will be discussed successively. Firstly, the mediation relationship between awareness and trust was evaluated. In the research model, social presence and interactivity are suggested to function as mediators in this relationship. In the evaluation of the mediating effect of social presence in the relationship between awareness and trust, coefficients a and b were significant, while c was reduced from 0.216 to 0.073 (c’) when social presence was controlled for. Therefore, the criteria proposed by Baron and Kenny (1986) were satisfied. The Sobel test confirmed the significance of the mediation effect (p=0.002). As interactivity is represented by UCTC and responsiveness, the relationship between awareness and trust mediated by UCTC and responsiveness were assessed separately. In the evaluation of the role of UCTC as a mediator, coefficient a and b were significant, while c was reduced from 0.216 to 0.104 (c’). The Sobel test proved a significant mediation effect (p=0.023). In the evaluation of the role of responsiveness as a mediator, coefficients a and b were significant, while c was reduced from 0.216 to 0.092(c’). The Sobel test proved a significant mediation relationship (p=0.030). In the evaluation of any of these three mediation effects (social presence, UCTC and responsiveness as mediators), the coefficient c’ is not zero, therefore partial mediations were suggested. It is notable that when any of the three constructs - social presence, UCTC or responsiveness - was controlled for, the effect of awareness on trust became not significant anymore. This could be a result of the small sample size (Baron & Kenny, 1986). Since the impact of awareness on trust was not significant when any of the three mediators was controlled for, including all three mediated effects in the research model will certainly over-estimate the effect of awareness. Therefore, it has been decided that only the mediated effect through social presence will be considered in the research model while the other two will be neglected. The reasons to retain the mediating effect through social presence, instead of through the other two mediators, is twofold: first, the indirect effect through social presence is higher so that established effect of awareness on trust can be retained to a larger extent and second, interactivity (UCTC and responsiveness) has a positive impact on social presence. It is possible that the indirect effect of awareness mediated by interactivity influences trust via social presence.
relationship between awareness and social presence, coefficient a and b were significant and in addition coefficient c reduced from 0.285 to 0.161 (c’) was also significant when UCTC was controlled for. Since c’ was not zero and significant, partial mediation occurs. The Sobel test also showed that the mediation effect was significant (p=0.021). When evaluating the mediating effect of responsiveness in the relationship between awareness and social presence, partial mediation was also observed and the significance of the mediation effect of responsiveness was supported (p=0.035). Furthermore, in order to better understand of how each aspect of interactivity mediates the relationship between awareness and social presence, responsiveness was controlled for when evaluating the mediation effect of UCTC and UCTC was controlled for when evaluating the mediation effect of responsiveness. Nonetheless, when responsiveness was controlled for, the effect of awareness on UCTC was no longer significant and when UCTC was controlled for, the effect of awareness on responsiveness was no longer significant. The interdependence of the three dimensions of interactivity (user control, two-way communication and responsiveness) makes it difficult to separate the effect of awareness through two aspects of interactivity. As a consequence, including the indirect effect of awareness through both UCTC and responsiveness in the research model will result in an overestimation. Therefore, the higher indirect effect of awareness on social presence, which was through UCTC (0.124), was taken into consideration and the indirect effect of awareness on social presence through responsiveness (0.094) was neglected.
social presence and trust were not correlating with UCTC. When social presence was controlled for in the regression predicting trust, the original coefficient of responsiveness (c) decreased from 0.528 to 0.498 (c’), which is a marginal decrease. Hence the mediation relationship is not suggested by mediation analysis proposed by Baron and Kenny (1986). In accordance with mediation analysis, the Sobel test also showed a non-significant mediation effect (p=0.142). Therefore, it can be concluded that social presence mediates the relationship between UCTC and trust and does not mediate the relationship between responsiveness and trust. This finding is in accordance with the results of the W statistic, which presents that path ―UCTCsocial presence‖ is more relevant than path ―responsivenesssocial presence‖.
Direct, indirect and total effects in the research model
Table 8: A summary of direct, indirect and total effects
Direct effect Indirect effect Total effect
Trust Social presence 0.114(0.045) 0 0.114 UCTC 0.174(0.033) 0.076 (0.084) 0.250 Responsiveness 0.373(0.000) 0 0.373 Awareness 0 0.041 (0.002) 0.041 Social presence UCTC 0.667(0.000) 0 0.667 Responsiveness 0.195(0.014) 0 0.195 Awareness 0.214(0.017) 0.145 (0.015) 0.359 UCTC Awareness 0.217(0.015) 0 0.217 Responsiveness Awareness 0.266(0.023) 0 0.266
a) Number 1(number 2) represent effect size (p-value).
b) Figures within brackets in column ―indirect effect‖ are the p-values of the Sobel test of related mediation.
According to table 8, awareness had positive and significant direct effects on social presence (0.214), UCTC (0.217) and responsiveness (0.266). Besides a direct effect, awareness also influenced social presence indirectly through UCTC (0.217*0.667=0.145). As discussed earlier, the effect of awareness on social presence through responsiveness was decided to be ignored. Therefore, the total effect of awareness on social presence was 0.214+0.145=0.359. Since only an indirect effect of awareness on trust through social presence is considered, the effect of awareness on trust can be calculated as 0.114*0.359=0.041. Social presence had only a direct impact on trust, which was 0.114. UCTC had an indirect impact on trust through social presence: 0.677*0.114=0.076. Together with its direct effect, the total effect of UCTC on trust was 0.076+0.174=0.250. As evaluated earlier in this section, the mediated effect of responsiveness on trust through social presence was not significant. Therefore, responsiveness turned out to have only a direct effect on trust, which was 0.373.
