Social Presence to Online Trust:
The Mediation Role of Information Quality
Independent research supported by FDUROP
(Fudan Undergraduate Research Opportunity Program)
Design Research / Psychological Research / Web Design
The Challenge

Social presence, the extent to which people consider the technology conveys human contact, has a positive impact on online trust. However, less is known about the mechanism of such an impact. This research explores the mediation effect of information quality and information-task fit.

My Role

Literature review

Experimental design

Web design and development (HTML/CSS/JavaScript) 

Subject recruitment and testing​

Data analysis (R/SmartPLS)

Writing reports

Abstract

Social presence is the extent to which people feel the web conveys human contact. Previous research has demonstrated that social presence enhances online trust. However, less is known about the mechanism of such impact. The study investigates two potential mediators for the effect: information quality and information-task fit. Structural equation modeling (SEM) indicated that information quality is a plausible mediator; information-task fit does not have a direct impact on online trust. Research suggests that when implementing social presence elements in the aim of increasing online trust, it is important to examine whether the implementation enhances information quality.

 
Introduction

Social presence was first introduced as the capacity of a media to transmit rich information in the Media Richness Theory (Short et al., 1976). The theory claims that various medias have different levels of social presence, thus making some media superior over others on certain tasks. The concept later developed into two accounts. The Computer-mediated Communication (CMC) perspective argues that social presence is the inherited quality of a media (Gefen and Straub, 2003), while another account considers social presence as a psychological connection a person has with the technology (Gunawardena, 1995). In this research, we conceptualize social presence as a perception people hold toward technology. It is the extent to which people feel the web is sensitive, personal, sociable, warm and conveys human contact (Cyr, Hassanein, Head, & Ivanov, 2007).

The Computer as Social Agent (CASA, Reeves, & Nass, 1996) theory claims that people can response socially toward computers. Nass and colleagues found that certain computer personality can be manipulated by adding cues and that people interacted with these agents as they interact with other human beings. In this sense, social presence shortens the perception of distance between the user and the vendor, thus increasing trust. Previous studies have also confirmed the social presence increased trust in e-commerce applications (Cyr et al., 2007; Hassanein and Head, 2006). Therefore, we hypothesize that:

H1. Higher perceived social presence will result in higher online trust.

Few studies have not found any relationship between social presence elements and online trust. For instance, Riegelsberger and Sasse (2002) found that presenting employees’ photos cannot increase users’ trust toward the web. Participants in the study felt that the media was a deliberate manipulation, which caused mistrust toward the web. Another research found that an increase in social presence did not increase the response rate for the survey (Tourangeau, Couper, & Steiger, 2003). These results called for a clearer understanding of the mechanism behind social presence’s impact on online trust. Specifically, it is important to examine possible mediation relationships because it helps us understand in what contexts can social presence increase online trust.

 

 

Specifically, information quality is expected to mediate the relationship between social presence and online trust. Social presence may increase information quality as it provides additional visual and audio cues of the product. Meanwhile, detailed information is thought to increase trust toward vendor’s web as users cannot feel or touch the item in an online setting (Liao, Palvia, & Lin, 2006). Meanwhile, In the e-commerce (McKnight, Lankton, Nicolaou, & Price, 2017), e-health (Mun, Yoon, Davis, & Lee, 2013) and information exchange (Nicolaou and McKnight, 2006) domain, information quality is examined to have a positive influence on online trust. Hence, we hypothesize that:

H2. Perceived information quality mediates the relationship between perceived social presence and online trust.

H2a. Higher perceived social presence will result in higher perceived information quality.

H2b. Higher perceived information quality will result in higher online trust.

Information-task fit is also expected to mediate the relationship between social presence and online trust. Information-task fit refers to the extent to which information on a web satisfies the needs for users to complete their tasks (Loiacono, Watson, & Goodhue, 2002). In the above cases when social presence fails to increase online trust, the social presence elements do not fit the tasks. For instance, while the employee’s picture on the web does not increase online trust, it also does not have a direct relationship with customer’s aim to know the product more. Therefore, information-task fit is also expected to have a positive effect on online trust as it helps to effectively complete the task. Hence, we hypothesize that:

H3. Information-task fit mediated the relationship between perceived social presence and online trust

H3a. Higher perceived social presence will result in higher information-task fit.
 
H3b. Higher information-task fit will result in higher online trust.

Trust disposition is proposed to have a positive influence on online trust. To be specific, individuals with a higher trust disposition are more likely to trust a vendor in e-commerce settings than those with a lower trust disposition when given the same amount of information (Salam, Iyer, Palvia, & Singh, 2005). Empirical studies found that trust disposition has a positive effect on online trust (Gefen & Straub, 2004; Lu, Fan, & Zhou, 2016). Thus, it is needed to control trust disposition when examining the mediation effects. Hence, we hypothesize that:

H4. Higher trust disposition will result in higher online trust. 
 
Methods
Experimental Design

The current study adopted a one-factorial between-subject design. I manipulated the levels of social presence (high, low) with two independent groups. Participants were randomly assigned to each condition. The experiment was conducted online. The github.com was used to host both experimental websites, as it offers a cost-free, stable and fast hosting service. Participant each received e-mail instructions on how to conduct the experiment and a link that directed them to one condition only.

Web Design

The two conditions are shown for an example. The conditions were adapted from Hassanein and Head (2007). The experimental websites feature product title, product image, product information, purchase information, and customer evaluation. The layout of the webs was adapted from online shopping sites like www.Gap.com. The webs have top navigation and side navigation bars. The websites were coded using HTML5, CSS3, and JavaScript. Cascading Style Sheets (CSS3) was used to provide layouts that adhere to current web design protocols, thus increasing task realism. JavaScript was used to allow for interaction.

 

Condition 1 was the basic treatment that had less social presence elements. The images on the web included cloth and excluded human presence. It also does not have any comment in the “comment” section. Condition 2 had more social presence elements. The images on the web included both the cloth and the human image. It also had four comments for each product in the “comment” section.

Click here to view the high social presence version

The sites were written in Chinese

Results

I used a structural equation modeling (SEM) approach in the current study to uncover causal relationship among the variables. Specifically, I used the Partial Least Square (PLS) method since it is suitable for exploratory study and can be applied to small sample size. Bootstrapping was applied (500 subsamples) to uncover significance levels for path coefficients. Approximately 64.8% of the variance in online trust was accounted for by all variables in the model.

 

In the original model shown below, both path coefficients from social presence to online trust, and from trust disposition to online trust were significant. In the overall model, path coefficients from social presence to information quality, and from information quality to online trust were significant, while the path coefficient from social presence to online trust is not, indicating a full-mediation effect of information quality. On the other hand, though path coefficient from social presence to information-task fit is significant, path coefficient from information-task fit to online trust is not. Meanwhile, path coefficient from trust disposition to online trust is significant. Thus, hypotheses H1, H2, H2a, H2b, H3a and H4 are supported; while hypotheses H3 and H3b were not. Table x provides the t-value for hypotheses testing results.

 
 
Discussion

The primary purpose of the study was to explore possible mediation effects from social presence to online trust. First, we found a significant mediation effect of information quality. It indicates that an increased perception of information quality may be the underlying cause of social presence’s effect on online trust. Surprisingly, social presence does not yield a statistically significant effect on online trust, which indicates that the effect of social presence may be fully mediated by information quality. Meanwhile, no significant mediation effect was found of information-task fit.

The results imply one main recommendation: the role of social presence should not be overestimated. When implementing social presence elements, it is needed to evaluate whether the implementation enhances the information quality. Social presence elements that do not convey useful information may not lead to an increase in online trust.

 
 
Relfections
  • Turn to your participants when you failed the manipulation. The websites were not in their current way. When my first manipulation failed, I asked several participants about how and why they felt certain things about the websites. I made many modifications to my first websites. These recommendations include but not limited to adding more reviews, making the web look less appealing, providing the most important piece of information and so on.

  • Always have a pretest. Pretest can tell you whether your experiment design works. It also preps you for the real experiment. Same applies to user research as well.​

©2018 by Yu Zhao.