Privacy, Trust, and Disclosure: Exploring Barriers to Electronic Commerce


  • Miriam J. Metzger

    Corresponding author
    1. (Ph.D. University of Southern California, 1997) is an assistant professor in the Department of Communication at the University of California, Santa Barbara.
    Search for more papers by this author

Address: Ellison 5814, Department of Communication, University of California, Santa Barbara, Santa Barbara, CA 93106. Tel: 805-893-8237.


This study proposes and tests a model of online information disclosure to commercial Web sites, which is an important component of e-commerce. Based on social exchange theory and research on self-disclosure in interpersonal contexts, the model emphasizes the role of trust and past online behavior in the disclosure of personal information to a commercial Web site created for this study. Data collected to test the model confirmed all predicted paths. A second aim of this exploratory research was to examine the nature and quality of online disclosure. Data show differences in online information disclosure depending on the characteristics of Internet users and the type of information requested from commercial Web sites.


The Internet and World Wide Web have greatly expanded opportunities for companies to communicate with their customers. Since the 1990s, the number of commercial Web sites has skyrocketed and continues to climb each year (Albarran, 2000; Kaye & Medoff, 2001). These Web sites offer several benefits to the companies that sponsor them; for example, they provide a cost-effective way to advertise and sell products to a global market. Internet users also benefit from commercial Web sites that offer convenient shopping from home 24 hours a day, increase their access to competing firms, and facilitate brand and price comparisons.Perhaps most important, the Internet's capacity for interactive, two-way information flows offer increased opportunity for buyers and sellers to learn about each other (Kaye & Medoff, 2001). For example, Web sites can provide customers with information about online vendors by listing contact information or answers to FAQs. At the same time, interactive technologies allow companies to collect information about their customers and to personalize communication with consumers in new ways (Albarran, 2000). The use of data-gathering techniques such as “cookies” or site registration are examples.1

For e-commerce to work, however, both online companies and Internet consumers must be willing to disclose information about themselves (Hatlestad, 2001). Personal information such as the customer's name, address, and credit card number are necessary for almost all online commercial transactions, and more detailed information about users' Internet behavior is required for Web site customization and personalization. However, not all customers are willing to give up this information, and many are shying away from electronic commerce as a result. In fact, several studies have found that the biggest barrier to the growth of e-commerce is the public's fears about privacy and security online (Albarran, 2000; Hoffman, Novak, & Peralta, 1999; Kaye & Medoff, 2001; UCLA Center for Communication Policy, 2003).

Fears about online privacy stem from the technology's ability to monitor and record almost every aspect of Internet users' behavior, and are further fueled by media reports of Web sites that have violated their own privacy agreements by distributing customer information without permission (Stout, 2000; Tweney, 1998). Recent opinion polls of Internet users reflect the magnitude of the public's concern over online privacy. A meta-analysis of major polls on the topic of online privacy found that people are very concerned about threats to their privacy when using electronic networks (Metzger & Docter, 2003). Some of the most frequently cited concerns include unfamiliar parties obtaining personal information and hackers stealing credit card information during business transactions taking place on the Web.

The public's fears appear to be well-founded. In 2002, the FBI's Internet Fraud Complaint Center referred 48,252 complaints to law enforcement or regulatory agencies and estimates the total dollar loss from online fraud to be $54 million (National White Collar Crime Center, 2003). And the problem appears to be worsening each year. The National Consumer League's annual Internet fraud statistics show consistent increases in the number of complaints annually since 1998 (National Fraud Information Center, 2003). At the same time, many commercial Web sites do not entitle Internet users to much privacy. Data from the Federal Trade Commission (FTC) find that between 85-97% of commercial Web sites collect at least one type of personal identifying information from site visitors, such as names and email addresses (FTC, 2000). Exacerbating the problem, many Web sites do not alert users that they collect these data, nor do they offer a purpose for collecting it (Culnan, 1999; FTC, 2000). For example, in their 2000 report, the FTC found that fewer than 20% of Web sites posted a comprehensive privacy policy.

Consequently, it is not surprising that some Internet users are reluctant to disclose personal information online. Hoffman et al. (1999), for example, found that 95% of the users in their sample indicated that they had at some point declined to provide personal information when it was requested by a Web site. The TRUSTe Corporation determined that 41% of their survey respondents would rather quit a Web page than reveal personal information (Quittner, 1997). And other researchers have estimated that consumers would spend roughly $6 billion more per year through e-commerce if privacy was not such a concern (Palmer, Bailey, & Faraj, 2000).

These data raise important questions regarding how businesses engaged in e-commerce can improve communication with customers. The factors that influence consumers to disclose information knowingly in the online environment is an open issue. Several variables are suggested by research in communication, marketing, and psychology. For example, trust, regard for the company, perceived privacy protection, general privacy concern, and past Internet experience have been shown to be important factors in both commercial and online exchange (e.g., Culnan & Armstrong, 1999; Fombrun, 1996; Jarvenpaa & Tractinsky, 1999; Jarvenpaa, Tractinsky, & Vitale, 2000). Using prior theory and research, this study proposes and tests a model of how these variables affect information disclosure on the Internet (see Figure 1). This model focuses on the role of trust and past online behavior in information disclosure, while exploring the concurrent influences of users' privacy perceptions and attitudes toward the Web site sponsor. In addition, this study also explores the quality of information disclosure on the Internet.

Figure 1.

Model of online information disclosure.


Trust is perhaps the most important influence on information disclosure (Hoffman et al., 1999; Jarvenpaa & Tractinsky, 1999; Jarvenpaa et al., 2000; Swaminathan, Lepkowska-White, & Rao, 1999). Trust is a function of the amount and type of control one has in a relationship (Heath & Bryant, 1992), and is a central concept of social exchange theory in interpersonal communication research (Roloff, 1981). Social exchange theory asserts that individuals weigh the costs and rewards in deciding whether to engage in social transactions. If the rewards are determined to outweigh the costs, then the individual is likely to enter into an exchange relationship. Trust is critical to this process because it is believed to reduce the perceived costs of such transactions. Indeed, several studies of interpersonal exchange situations have confirmed that trust is a precondition for self-disclosure because it reduces the perceived risks involved in revealing private information (e.g., Jourard, 1971; Rubin, 1975; Steel, 1991; Wheeless & Grotz, 1977).

Culnan and Armstrong (1999) point out that although most research has focused on trust and self-disclosure in interpersonal contexts, similar balancing dynamics are used in electronic environments. That is, the risks of disclosing personal information are weighed against the benefits when deciding to provide information to a Web site, and so trust is the key to disclosure in both interpersonal and online relationships. Research reported by Hoffman et al. (1999) supports this notion; 63% of Internet users who have declined to give information to Web sites reported doing so because of lack of trust.

Furthermore, both the Internet Consumer Trust Model developed by Jarvenpaa and Tractinsky (1999) and the Electronic Exchange Model by Swaminathan et al. (1999) find trust to be a key antecedent to engaging in consumer transactions online because it reduces the risks associated with purchasing goods and services over the Internet. These risks include theft of financial or identity information. For example, Jarvenpaa and Tractinsky found that trust increases confidence in a company, which lowers the perceived risk of electronic exchange with that company and, therefore, increases the likelihood of consumers engaging in electronic transactions. Similarly, Swaminathan et al. established that consumers prefer to do business with Web sites that they perceive to be reliable, honest, consistent, competent, fair, responsible, helpful, and altruistic—key components of trust.

The role of trust in facilitating disclosure may be particularly important in online exchanges where computer-mediated communication replaces physical contact. In an environment of reduced social cues, trust may be more difficult, yet more important, to establish than in interpersonal contexts (Boyd, 2003). Added to this is the fact that the Internet shopping experience is relatively novel, may have technical difficulties, lacks the ability for consumers to physically inspect goods prior to purchase, abounds with new companies that have not yet established good reputations, and other problems (Culnan & Armstrong, 1999; Javenpaa & Tractinski, 1999; Jarvenpaa et al., 2000).

In sum, it is reasonable to suggest that trust of a commercial Web site will influence disclosure of personal information to that site. If trust is particularly significant to the process of online exchange, then it is important to ask what fosters trust of a commercial Web site, especially during initial encounters with the site. Some factors may be user attitudes toward or evaluations of the Web site sponsor, perceptions of Web site privacy protection or security, and users' overall online privacy concerns (Boyd, 2003; Culnan & Armstrong, 1999; Jarvenpaa & Tractinsky, 1999; Jarvenpaa et al., 2000; Sultan, Urban, Shankar, & Bart, 2002).

Regard for the Company

Consumers' overall regard for a company, defined as their affective evaluation of that company, is likely to influence perceptions of trust. Affective evaluations are made on the basis of many things, for example, firm reputation or personal experience with a company, both of which have been linked to trust in past research (see Jarvenpaa et al., 2000). However, in the case of new companies where reputation and personal experience are lacking, affective evaluations are made on the basis of a company's observable attributes or “signals” (Fombrun & Shanley, 1990; Kirmani & Rao, 2000). These signals include presentation of self in public communications, for example, through advertising, customer service interactions, or stated policies (Fombrun & Shanley, 1990; Kilbourne & Mowen, 1986). In turn, these things influence trust perceptions (Fombrun, 1996; McKnight, Cummings, & Chervany, 1998; Sultan et al., 2002).

Support for the notion that regard for the company impacts trust comes from research on social influence, as well as studies of trust in ecommerce environments specifically. For example, evidence that liking for a communicator positively influences judgments of trustworthiness is found across a number of studies examining communicator credibility (see O'Keefe, 2002 for a review of this literature). Extending the credibility literature to the Web, Fogg (2003) found that Web sites of respected organizations that were visually pleasing were rated high in trustworthiness and expertise (see also Sultan et al., 2002). Also along these lines, Jarvenpaa and colleagues found a positive relationship between attitudes toward an Internet store and store trustworthiness (Jarvenpaa & Tractinsky, 1999; Jarvenpaa et al., 2000).

Perceived Web Site Privacy Protection

The degree to which Internet users feel that a Web site protects their privacy may also have an impact on their trust of the site. Although perceptions of privacy protection likely differ depending on the user, culture, and other factors, Web site design and content elements can communicate information about privacy protection (Palmer et al., 2000). With regard to design, site complexity and layout (e.g., presence and number of hyperlinks, obtrusiveness of advertising, ease of navigation, and general professionalism) have been shown to affect perceptions of the authority of the site (Alexander & Tate, 1999; Flanders & Willis, 1998; Tseng & Fogg, 1999), which may impact privacy perceptions. Web site content also influences perceptions of a site's privacy protection through such elements as privacy statements and seals, for example, TRUSTe or BBBonline (Culnan, 1999; Hansen, 2000; Palmer et al., 2000).

Research suggests that enhancing perceptions of Web site privacy protection via features such as privacy statements and seals may increase regard for the company and trust. For example, Fombrun (1996) shows that demonstrating goodwill and respect for customer needs is an important component of improving customer relations and firm reputation, which he defines as the consumer's overall affective evaluation of a company. With regard to trust specifically, Culnan and Armstrong (1999) find that revealing information collection procedures increases consumers' feelings of security and organizational trust. Similarly, Sultan et al. (2002) found that having an easy-to-understand privacy policy which explains how the company will use customer information predicts trust of a Web site. Finally, surveys have found that people are more willing to provide personal information to Web sites if they post privacy statements or display privacy seals (Kehoe, Pitkow, Sutton, Aggarwal, & Rogers, 1999; Hoffman et al., 1999). Together, these findings suggest that the degree to which Internet users believe a commercial Web site protects their privacy will positively influence their overall regard for the company and trust of the company's Web site.

Concern for Privacy

Another factor that may predict trust of commercial Web sites is a person's general concern for privacy (Culnan & Armstrong, 1999). Berscheid (1977) argues that individuals differ in the degree to which they desire and value personal control over information about themselves. With regard to information privacy values, studies have found a negative relationship between the value people place on privacy and perceptions of control over personal information, which is the very foundation of trust as defined earlier (Stone, Gueutal, Gardner, & McClure, 1983).

Research on these issues in the online environment is scarce, although Sheehan and Hoy (1999) found that as individuals’ concern for privacy increases, users report registering for Web sites less frequently and providing incomplete information when they do register, presumably because they have less trust in the Web site. Further, Jarvenpaa and Tractinsky (1999) argue that one's general trusting stance can generalize to trust of Internet-based companies. This suggests that Internet users’ general concern for online privacy will negatively predict their trust of commercial Web sites.

Internet Experience and Past Disclosure

Besides users’ concern for privacy, individuals’ Internet experience and past online information disclosure are also likely to influence their disclosure to commercial Web sites. Based on the truism that the best predictor of future behavior is past behavior, prior disclosure of personal information is likely to affect later disclosure. Chelune (1987) explains that past experiences with interpersonal exchanges affect the value placed on such exchanges and influence expectations regarding future interactions, which then guide future exchange behaviors. Accordingly, studies of disclosure in interpersonal contexts find that past disclosure positively affects future willingness to disclose when the initial disclosure is rewarded, because the perceived risks associated with self-disclosure are reduced (Gilbert, 1976).

This logic applies to other contexts as well. For example, Culnan and Armstrong (1999) found that people's willingness to be profiled, that is, to give up personal information for marketing purposes, was a function of their past experience with revealing information to marketers. They suggest that people who have had prior experience with direct or targeted marketing are more likely to understand the benefits of consumer profiling. As a result, they should be more willing to have their personal information used for this purpose because they are likely to perceive profiling as “compatible with their existing values and past experiences” (pp. 108-109). Furthermore, due to efficient data storage and transfer capacities of computer-based technologies, consumers may believe that after they disclose personal information online once, the damage has already been done and so they may feel less inhibited about revealing that information online again. In this way, future disclosure is not necessarily associated with higher perceived risk in the Internet environment. For these reasons, past disclosure of personal information online is likely to predict future disclosure of personal information to a commercial Web site.

But what leads a person to disclose information online initially? Unfortunately, there is little data to answer this question. Two interesting possibilities are Internet experience and concern for privacy. A survey of several thousand Internet users in 1998 found that more experienced and “expert” Internet users were more willing to give up demographic information (Kehoe et al., 1999). Similarly, Cheskin (2000) found that Internet users identifying themselves as “experts” were more willing to sacrifice their privacy for better prices than were beginning Internet users. Finally, the Pew Research Center (2000) found greater online experience was associated with less concern about online privacy (but see Javenpaa & Tractinsky, 1999 and Hoffman et al., 1999 for conflicting results).

Concern for online privacy may also predict whether people have ever disclosed personal information on the Internet. According to communication boundary management theory, withholding information in interpersonal relationships is a way to protect felt privacy needs and reduce feelings of vulnerability (Derlega & Chaikin, 1977; Petronio, 2000). The research described earlier in which Internet users cite privacy fears as a barrier to engaging in Web-based business transactions supports this idea in the online environment (e.g., Sheehan & Hoy, 1999). Further, other studies have found that concern about Internet privacy has a negative influence on people's likelihood of engaging in electronic exchange and willingness to disclose personal information on the Web (Culnan & Armstrong, 1999; Swaminathan et al., 1999).


The preceding theory and research on trust and disclosure in both interpersonal and computer-mediated communication contexts suggest the following hypotheses, which are illustrated by the paths depicted in the model shown in Figure 1.

H1: The degree to which Internet users' believe a company's Web site protects their privacy positively influences their regard for the company.

H2a&b: Internet users' (a) regard for the company and (b) degree to which they believe the company's Web site protects their privacy positively influence their trust of the Web site.

H3: Internet users' concern for privacy online negatively influences their trust of a commercial Web site.

H4: Internet users' concern for privacy online negatively influences their past online information disclosure.

H5: The amount of time users spend on the Internet positively influences their pattern of past information disclosure.

H6a&b: Internet users' (a) trust of a company's Web site and (b) past online disclosure positively influence their current information disclosure to the company's Web site.

Quality of Online Disclosure

In addition to exploring the antecedents to online information disclosure, a second aim of this study is to examine the characteristics of that disclosure. For example, the lack of social cues in the online environment might promote dishonesty on the part of Internet users when asked to provide personal information to Web sites. Knowing the type and amount of false information that consumers supply is important for businesses that are trying to accurately gauge the online market and improve customer service. Other questions relevant to e-commerce and online marketing are: What kinds of information will people give to commercial Web sites and what kinds of information are they reluctant to disclose? And, do different kinds of people disclose differently?

Unfortunately, there is little research on these questions and extant findings are inconsistent. For example, studies vary markedly in their estimates of the amount of lying online. Some find that as few as 15% of Internet users have provided inaccurate information when registering for a Web site (Sheehan & Hoy, 1999), while others find as many as 24% (Pew Research Center, 2000), or even 47% have falsified information that they were asked to provide online (Kehoe et al., 1999). Sheehan and Hoy suggest that the inconsistency of these estimates may be due in part to methodological differences in the studies, such that respondents who were not anonymous might have given socially desirable responses. This indicates that studies of online behavior that rely entirely on self-report data may lack external validity.

Although the issue of the truthfulness of online disclosure has received some scholarly attention, questions about who provides information and what types of information is given online have been overlooked. The literature on self-disclosure in interpersonal settings may provide a starting point for answering these questions. For example, in a study of the conditions and dimensions of self-disclosure with known others, Jourard and Lasakow (1958) found that not all information is divulged with equal frequency. People are more willing to disclose information about their interests, attitudes and opinions, and work than they are to reveal their deep emotions, insecurities, or financial and health information. It is likely that certain topics will be more readily divulged than others in the online environment just as in face-to-face contexts, but the topics may differ on the Internet due to the unique risks involved in most e-commerce transactions and the reduced social cues inherent in the technology.

Differences in both the amount and topics of self-disclosure between males and females have also been observed in the literature on interpersonal communication (Cozby, 1973; Derlega, Metts, Petronio, & Margulis, 1993). Overall, women have been found to disclose both more and more intimate information than men, although this finding depends on the specific social context of the disclosure (Derlega et al., 1993). According to Derlega and colleagues, sex differences in disclosure within interpersonal contexts may be attributable to “subcultural” differences between males and females. That is, due to socialization, the norms and expectations for self-disclosure may be different for males and females raised in a particular culture. It would be interesting to see if sex differences in information disclosure exist in the culture of the Internet, which until recently has been dominated by males (Morahan-Martin, 1998; Pew Research Center, 2000).

Given the dearth of research on self-disclosure in the online environment, the following research questions are proposed:

RQ1: What kind of personal information is voluntarily disclosed to commercial Web sites?

RQ2: How truthful is the information that is disclosed to commercial Web sites? What kind of information do people lie about to these sites?

RQ3: Are there sex differences in the quantity or type of online disclosure to commercial Web sites?



A pretest was conducted to determine Internet users' sensitivity to revealing various kinds of personal information while online. Students enrolled in communication courses at a west coast university participated in the pretest for extra course credit (N = .47). Participants were instructed to think about how comfortable they would be in giving each of 23 pieces of personal information that commercial Web sites often request from visitors (e.g., first name, last name, credit card number, etc.) and then to rate the “privateness” of each piece of information on scales ranging from 1 (“Not Really Private”) to 7 (“Very Private”).

The information from the pretest was used to develop a weighting scale to measure the depth of online information disclosure, which is discussed in detail later. The results of the pretest are presented in Appendix A. Not surprisingly, financial information, including credit card number, social security number, and banking information were considered to be the most private types of information, whereas basic identifying information, such as sex and first name were rated by participants as the least private types of information.


Participants in the main study were recruited from introductory communication courses at the same university as the pretest. A total of 189 students, mostly freshmen and sophomores, participated in the study in exchange for course credit. Of the respondents, 73.5% were female and 25.9% were male (one participant failed to provide his or her sex).2 All participants said that they use the Internet regularly, and spend about 1 hour per week online (excluding email) on average. With respect to online commerce, 73.4% of the sample indicated that they had engaged in e-commerce during the previous year. A majority of those participants (53.7%) made between 1-3 online purchases and 26.8% of them made more than four purchases during the year.

Although the use of a college student sample is not ideal, there is some evidence that college students are, in many ways, similar to online consumers more generally. For example, college students have both access to the Internet and skill in using it, and they typically come from middle to upper-middle class families. According to Pastore (2000), these characteristics mirror those of the most avid e-commerce participant groups in the U.S. Also, because the emphasis in this exploratory study is on theoretical model testing, rather than on determining average levels of adult's e-commerce usage, the use of a college sample is justifiable here.

Procedures and Measures

A fictitious but realistic commercial Web site that sold posters was created for this study. The site was called “” and was modeled after a real online poster vendor. The site was created by downloading the real online poster vendor's site and changing the name of the company.3 Thus, the site was extremely realistic. The use of a fictitious site offers some key advantages over using real Web sites in this study. First, no real Web sites possessing the features needed to test all of the hypotheses and research questions adequately (e.g., ample opportunity to disclose a wide variety of information) could be located. Second, it allows for much greater control than would be possible had real sites had been used. Third, and more important, it allows for an examination of actual versus self-reported online behavior, since customer information from real Web sites is proprietary, and thus is unavailable to the public, even for educational or research purposes.

Students were instructed to come to an on-campus computer lab where they were told that they would be evaluating a real Web site, and that they should interact with the site as naturally as possible, “as if they had come across it while surfing the Web on their own.” After reviewing the Web site for at least 10 minutes, participants completed a pencil-and-paper questionnaire assessing key variables in the study.

Perceived Web Site Privacy Protection

To measure this variable, participants indicated the degree to which they agreed that the Web site “seems to protect my privacy.” Response categories ranged from 1 (“strongly disagree”) to 5 (“strongly agree”). As discussed earlier, individuals' perceptions of privacy protection come from many sources. However, to ensure that there would be enough variance in the perceived privacy protection variable, the privacy statement on the Web site was varied slightly to convey strong, weak, or no privacy protection, and approximately one-third of the participants saw each version.

The “strong” privacy statement provided all of the information that Culnan (1999) identified as necessary for a comprehensive privacy statement, including notice (notifying users that the Web site collects information), choice (giving users a choice for whether to provide the information), access (explaining who has access to the information and for what purposes), security (explaining how the information is securely handled) and whether the Web site provides host contact information. The “weak” privacy statement offered only notice and contact information for the company. The text of both statements appears in Appendix B.

Concern for Privacy Online

Participants’ concern for online privacy was assessed using items adapted from prior studies of privacy in both Internet and organizational contexts (Jarvenpaa & Tractinsky, 1999; Jarvenpaa et al., 2000; Pew Research Center, 2000; Stone et al., 1983; Swaminathan et al., 1999). Based on their face validity, seven items were selected for the measure of concern for online privacy. Respondents were asked to indicate their agreement with each of the following statements on a 5-point Likert scale: “I am concerned about businesses and people I don't know getting my personal information on the Internet,”“Web sites that collect and store users’ information should not have the right to release it to others,”“I should be able to visit Web sites anonymously,”“When I want a product or service, I am comfortable giving my personal information on the Internet” (this item was reverse coded), “The amount and type of personal information collected by Web sites should be limited,”“There should be new laws to protect people's privacy on the Internet,” and “A user ought to have complete control over which Web sites get what personal information.” A factor analysis revealed that the scale is unidimensional and Cronbach's alpha for the scale was .61. Although the reliability of this scale is low, it is considered acceptable for exploratory research (Nunnally, 1967).4

Regard for Company

Following the definition of regard for the company as a consumer's overall affective evaluation of a firm, participants were asked to indicate the degree to which they liked the company. Liking of the company was rated on a 5-point scale anchored at 1 (“I did not like the company at all”) to 5 (“I liked the company very much”).

Trust of a Web Site

A scale measuring trust of the Web site was developed from prior research on trust in organizational, marketing, and ecommerce contexts (Doney & Cannon, 1997; Jarvenpaa & Tractinsky, 1999; Jarvenpaa et al., 2000; Mayer, Davis & Shoorman, 1995; McKnight & Chervany, 2001; Morgan & Hunt, 1994; Sultan et al., 2002; Swaminathan et al., 1999). Four overlapping dimensions of trust emerge consistently across these literatures: trust is the degree to which an organization is perceived to be reliable, competent, benevolent, and to have integrity. Furthermore, assessments of these dimensions may happen over time or during initial encounters (McKnight et al., 1998).

Using this definition of trust as a basis, an 8-item index of Web site trust that taps into the four dimensions was created, drawing from questions developed by other researchers (e.g., Jarvenpaa & Tractinsky, 1999; Sultan, Urban, Shankar, & Bart, 2002). To assess reliability, respondents rated the degree to which they felt the My Web site was dependable and responsible; competence was measured by how professional respondents felt the site was; benevolence included the degree to which the site was rated as fair, secure, and non-exploitative; and integrity was assessed by how trustworthy and honest respondents believed the site to be. All items were measured on 5-point scales ranging from “strongly agree” to “strongly disagree.” Responses were subjected to factor and reliability analyses, which found the trust scale to be internally consistent (Cronbach's alpha = .90).

Web Site Disclosure

This study measured disclosure of personal information to the Web site more directly than has been done in past research because, as mentioned earlier, results from studies relying exclusively on self-report data are susceptible to social desirability biases. The Web site used in this study offered participants two opportunities to disclose personal information. First, participants could order a poster online by completing an order form that asked for their first and last name, address, telephone number, credit card number and expiration date, and poster order.5 Second, each page of the Web site included a link that offered participants a free poster in exchange for completing “a marketing survey to help improve customer service.” The survey asked for the participant's first and last name, email address, postal address, telephone number, age, education level, sex, race/ethnicity, marital status, occupation, income, religion, political party identification, time spent online per week, number of people living in the household, hobbies/interests, most recent e-commerce purchase, favorite genre of posters, and social security number. Although all information was provided to the site voluntarily, participants were told that in order to obtain the free poster, their name and address were required. The information that participants provided online was captured automatically into a data file and was compiled by the researchers.6

Next, respondents were asked on the post-exposure questionnaire to check off any false information they provided to the site on a list of all the information types requested by the site. To reduce social desirability in answering, wording of the question stem read, “Sometimes people falsify the information they provide while online for a number of reasons; put a check mark next to any information that you provided today that was false. You will not be required to provide ‘corrected’ information, nor will you be penalized for having provided false information to the Web site, so please answer honestly.”

The measure of the amount of truthful online disclosure was developed by assigning a score of 0 to any piece of information that was either falsified or not provided. A score of 1 was then assigned to any piece of truthful information provided by participants. A truthful disclosure scale score was computed for each respondent by summing across all of the disclosure items.

A second measure, depth of truthful disclosure, was computed by weighting each piece of truthful information provided according to how private that information had been rated on the pretest. The weights were calculated from the pretest data as follows: First, the mean privacy level of each item was divided by the grand mean of all items, resulting in a ratio reflecting the relative privacy level of each piece of information. Next, the responses to each disclosure item (0 = did not provide information or provided false information, 1 = provided accurate information) was multiplied by this relative privacy ratio. Thus, information that was more private and provided truthfully by participants was given greater weight. Finally, all of the weighted disclosure items were then summed to create a measure of depth of truthful disclosure.

Past Online Information Disclosure

The questionnaire instructed participants to indicate what information they had disclosed to any Web site in the past using a check-off list. Each piece of information that they said they had provided in the past received a score of 1, and answers across all the items were summed to form the measure of past online information disclosure.

Demographics and Time Online

Participants were asked to provide their sex, year in school (freshman, sophomore, junior, senior), and number of online purchases in the past year (0, 1-3, 4-6, 7-9, 10 or more). Time spent online per week was measured in 30-minute increments, and answer categories included 0-30 minutes, 30-60 minutes, 60-90 minutes, 90-120 minutes, 120-240 minutes, and more than 240 minutes.


The hypothesized model of Web site trust and online disclosure was tested via path analysis using hierarchical regression analysis. Due to the exploratory nature of this research, hierarchical regression analysis was preferable to other structural equation modeling techniques because it tests the model equation-by-equation, using least squares rather than maximum likelihood estimates. The advantage of this analytic approach is that, unlike other structural equation modeling techniques, specification errors in any one equation are not diffused throughout the entire model (Cohen & Cohen, 1983). The path analysis proceeded by testing each endogenous variable to determine, first, if the hypothesized paths were confirmed by the data (as indicated by significant standardized regression coefficients) and, second, whether any unspecified paths were significant after controlling for the hypothesized paths. It should be noted that the model was tested using both breadth and depth of Web site disclosure measures, but because the relationships and coefficients were nearly identical in both models, only the results for breadth of disclosure are presented here.7 Sex was included as a control variable in all analyses. A correlation matrix showing relationships between the variables in the model is presented in Table 1.

Table 1.  Correlation matrix of variables used in the path analysis (N = .189).
 Concern for online privacyWeb site disclosureRegard for companyPast online disclosurePerceived Web site privacy protectionTime online
  1. Note: ***Correlation is significant at the .001 level (2-tailed). ** Correlation is significant at the .01 level (2-tailed).

Concern for online privacy      
Web site disclosure-.06     
Regard for company-.03.43***    
Past online disclosure-.17**.48***.11   
Perceived Web site privacy protection-.12.08.31***-.06  
Time online-.10.06-.04.29***-.12 
Trust of Web site-.22**.24**.58***.01.67***-.14

As shown in Figure 2, all of the hypotheses were supported. The first hypothesis concerned the effect of perceived privacy protection on regard for the company. This path was significant and in the hypothesized direction (β= .28, p < .001). Hypotheses 2a and 2b specified that trust would be positively predicted by regard for the company and perceived Web site privacy protection. The data confirm both paths (β= .41, p < .001 and β= .53, p < .001, respectively). As predicted in Hypotheses 3 and 4, concern for online privacy negatively predicted trust of the Web site (β= -.15, p < .01) and past online information disclosure (β= -.15, p < .05). The amount of time spent online positively predicted past online disclosure (β= .26, p < .01), as proposed by Hypothesis 5. And finally, the data confirmed Hypotheses 6a and 6b, which stated that disclosure to the Web site would be predicted by trust (β= .22, p < .01) and past online information disclosure (β= .47, p < .001).

Figure 2.

Path coefficients for the model of online information disclosure.

Note. ***Path coefficient is significant at the .001 level. **Path coefficient is significant at the .01 level. *Path coefficient is significant at the .05 level.

The overall fit of the model was assessed using Land's Simultaneous test, which examines the degree to which the observed variance/covariance matrix matches the predicted variance/covariance matrix. The goodness-of-fit statistic produced by this test has a chi square distribution, and a nonsignificant result indicates that the data support the model. In this case, the data fit the model proposed in this study (X2[28]= 9.91, p > .05). It is important to note, however, that this finding does not rule out the possibility that other patterns of causality would not also fit the data, or perhaps produce an even better fit. In fact, the model analyses indicated that one unspecified path was significant: Disclosure to the Web site was predicted by regard for the company (β= .36, p < .001). This result suggests that the model should be revised to include a direct path between these two variables. The suggestion that a person's regard for the company directly influences their willingness to disclose information to the company's Web site is intuitively reasonable, and will be discussed further later. None of the other unspecified paths was significant.

Research Question 1 asked about the type of personal information that is disclosed to commercial Web sites. Overall, participants in this study were willing to give information about themselves online: 129 (68.3%) of the participants chose to give some information to the Web site, either through the order form or through the free poster marketing survey. On average, these participants gave 14.06 of the 21 pieces of information requested by the Web site. Moreover, only 3 (1.6%) indicated that they have never given personal information to any Web site in the past.

In terms of the type of information given voluntarily by participants to the MyPosters site, they were most willing to provide basic demographic information (e.g., sex, age, education level, marital status), and slightly less willing to provide information about their actual online behavior (past purchases and time spent online), religion, political party identification, race, hobbies/interests, and occupation. Respondents were by far most protective of their personal contact information (telephone number and email address) and financial information (credit card number, social security number, and income), which makes sense given that disclosing these types of information carries the risk of personal intrusion or monetary loss.

Research Question 2 questioned the accuracy of information disclosure to commercial Web sites. Of the participants in this study who disclosed to the MyPosters site, 25 (19.4%) indicated that they falsified some of that information. About half of those participants falsified only one piece of information, whereas the other half gave inaccurate information for between 2 and 7 pieces of the information they provided. Participants tended to give inaccurate information for the items that were rated as more private on the pretest. For example, social security number, income, and postal address were among the most-often falsified items and were rated as some of the most private types of information in the pretest (see Appendix A). This was not true in all cases, however, as 7% of the respondents falsified their first name on the Web site, whereas this information was rated to be the second least private type of information in the pretest.

The last research question was concerned with sex differences in the quantity and type of disclosure to commercial Web sites. Analyses revealed no differences between males and females in the amount of inaccurate information provided to the MyPosters Web site (t[127]= .45, p > .05), and in either the breadth (t[185] = -.77, p > .05) or depth (t[186]= -.85, p > .05) of truthful disclosure to the site. However, one difference did emerge when the data for each type of information were examined. Fewer females and more males gave their telephone number to the Web site than was expected by chance (X2[1]= 4.48, p < .05). Also, when asked whether they had ever provided various kinds of information to any Web site in the past, females indicated that they had revealed their sex online significantly less often than males did (X2[1]= 4.35, p < .05).


A tension exists between commercial Web site sponsors and online consumers. On one hand, the ability of online companies to collect and use customers' personal information is important for their survival. On the other hand, concerns about privacy make many consumers reluctant to provide information while online. To explore this tension, this study proposed and tested a model of Web-based disclosure that emphasizes trust. As Hoffman et al. (1999) suggest, “ultimately, the most effective way for commercial Web providers to develop profitable exchange relationships with online customers is to earn their trust” (p. 85). Results of this study using a student sample support earlier research that found trust to be a key antecedent to online information exchange (e.g., Hoffman et al., 1999; Jarvenpaa & Tractinsky, 1999; Jarvenpaa et al., 2000; Swaminathan et al., 1999). The data also confirm that past online behavior and attitudes toward the Web site sponsor are also important in the decision to disclose personal information to a commercial Web site, as has been suggested by previous theory and research (Culnan & Armstrong, 1999; Swaminathan et al., 1999).

Moreover, the results indicate that regard for the company may have a particularly strong effect on disclosure, both mediated by, and independent of, trust. Although unexpected, the finding that regard for the company directly influences online disclosure is consistent with the theory of reasoned action and its extension, the theory of planned behavior (Azjen, 1991; Fishbein & Azjen, 1975). These theories find that beliefs and attitudes, such as favorable evaluations of a target, influence behavioral intentions toward that target, which then affect behavior. Applications of this theoretical model to consumer psychology find that it is a good predictor of actual purchasing behavior (e.g., East, 1993; Sahni, 1995). Jarvenpaa and Tractinsky (1999), for example, used these theories to predict that users' attitudes toward an Internet-based store would affect their willingness to buy from that store. They found that attitudes directly affected online purchase decisions, which is similar to this study's results showing a direct influence of attitudes toward a company on disclosure to the company's Web site.

The data from this study also demonstrate the importance of privacy as an antecedent to Web site disclosure. Participants' general concern for privacy and the degree to which they believe Web sites protect their privacy were found to influence online information disclosure through the trust these variables engendered. These results are consistent with other models of online exchange, including the Internet Consumer Trust Model (Jarvenpaa & Tractinsky, 1999) and the Electronic Exchange Model (Swaminathan et al., 1999). The significant positive relationship between perceived privacy protection and trust offers some support for Boyd's (2003) hypothesis that trust in the online environment is built through such rhetorical devices as providing descriptions of the Web site's credentials and competence, as well as by offering structural assurances and assessments of risk, all of which were present in the privacy statements.

Together, these findings suggest that companies should be sensitive to consumers' privacy concerns and to find ways to communicate this sensitivity on their Web sites. Boyd (2003) offers some rhetorical strategies for building trust in the online environment that are relevant to this point. For example, he recommends that sites offer assessments of risk in clear and simple terms in their online communications, provide evidence or a record of their past performance in terms of security and predictability, include indicators of credentials or competence, and display customer feedback and/or testimonials. Indeed, there are several ways for an e-tailer to demonstrate that it cares for customer privacy and security concerns, and Web site operators might consider using multiple strategies in order to suit various audiences and levels of privacy concern.

Internet experience, or time online, was also a factor in the decision to disclose personal information to the commercial Web site used in this study. This result supports prior findings that Internet users with more experience are less concerned with privacy and are more willing to provide information online than less experienced users (Cheskin, 2000; Pew Research Center, 2000), and contradicts other research that finds Internet experience is associated with greater mistrust and desire to control personal information (Hoffman et al., 1999; Jarvenpaa & Tractinsky, 1999). However, Jarvenpaa and Tractinsky's findings about mistrust were admittedly weak and, in contrast to the present study, Hoffman et al. only looked at a special group of users who refused to engage in e-commerce.

In addition to confirming the proposed model, the data from this study extend theoretical models of social exchange, boundary management, and self-disclosure from interpersonal contexts to the online environment. However, due to differences in interpersonal and computer-mediated communication, this extension is not perfect. For example, computer-based commercial communication in particular may be more anonymous, less intimate, and more impersonal because of greater social distance between communicators than in most interpersonal communication (see Culnan & Armstrong, 1999). These and other differences suggest that researchers should continue to develop new models for communication that takes place online.

Beyond understanding why respondents choose to disclose information to commercial Web sites, this study was also concerned with the quality of the information that is disclosed. Overall, about one-fifth of the students falsified at least some of the information that they provided to the Web site. This number is somewhat lower than other recent estimates, which may be reassuring to e-tailers. However, Web site operators should still be wary about the accuracy of the information customers provide online because, clearly, many people will lie to protect their privacy, especially about more sensitive information such as financial and personal contact information. Further, inaccurate customer information undermines companies' ability to personalize sites, one of the main advantages of e-commerce compared to traditional marketing techniques (Sheehan & Hoy, 1999). Interestingly, the type of information either withheld or falsified by participants in this sample parallels findings about self-disclosure in interpersonal contexts (Jourard & Lasahow, 1958).

This research also provides insight into sex differences in online information disclosure, a topic that has received no scholarly attention until now. For the most part, the male and female respondents behaved similarly when interacting with a commercial Web site. The women were, however, slightly less willing than the men to provide certain information about themselves to the site, specifically their telephone number and their sex. These differences may be due to women's greater susceptibility to victimization in general, and particularly on the Internet. Recent studies have found that women are significantly more likely than men to have experienced cyberstalking or sexual harassment online (Goodson, McCormick, & Evans, 2001; Griffiths, 2000). Not surprisingly, the Pew Research Center (2000) found that women are more concerned about their privacy online than are men.

Findings regarding the quality of online disclosure raise questions about how more accurate information may be obtained from Internet consumers. It was suggested earlier that privacy statements might help to increase accuracy by imparting a sense that the Web site protects users' privacy. Interestingly, the data in this study do not support this prediction. Despite the fact that most participants (67%) noticed the privacy statement when it was present, a post-hoc analysis found that the level of privacy protection offered in the statements did not directly affect their information disclosure to the Web site (F[2, 187] = 1.65, p > .05). This seems to suggest that privacy statements may be less important for commercial Web sites than previously assumed. However, privacy statements are not unimportant, since they may have an indirect influence on disclosure through trust and regard for the company. Regardless of their effectiveness, the results of this study suggest that companies should consider alternative means of encouraging Internet users to disclose accurate information about themselves.

For example, future research should examine ways for companies to communicate the security of their Web sites to potential customers. Such Web site features as the use of secure servers or trusted third party privacy seals could be explored for their effectiveness in promoting feelings of security and truthful information disclosure among online consumers. Site navigation features and aesthetics could also be examined for their impact on perceived trustworthiness of and disclosure to commercial Web sites (Boyd, 2003; Jarvenpaa & Tractinsky, 1999; Sultan et al., 2002; Tseng & Fogg, 1999).

The results of this study also suggest that more research is needed on the effects of company reputation on online trust and information disclosure. The finding that regard for the company had a stronger relationship with disclosure than did trust is interesting because most theories assume that company regard (or, by extension, reputation) affects consumer behavior by fostering trust (see Jarvenpaa & Tractinsky, 1999). It could be that, in the case of new or unknown Web retailers, trust may be somewhat difficult for users to gauge and so is relied upon to a lesser extent than are other factors when deciding whether to disclose personal information. In any case, future research should explore ways in which new businesses can build a good reputation in the online environment since it appears to be so important to electronic exchange.

Future research on issues of online trust and information disclosure is also needed because this study had some limitations. First, the use of undergraduates in this research means that the sample is not representative of all Internet users, and readers are cautioned against overgeneralizing the results of this exploratory study. Some studies have found that younger adults are less concerned about privacy online and are more willing to give up personal information (Cheskin, 2000; Pew Research Center, 2000), even though others find college students to be fairly representative of Internet consumers in general (Pastore, 2000). Future research using more representative samples of all Internet users is necessary to confirm the estimates of the antecedents to and the quality of online information disclosure obtained in this study. That said, a strength of this study is that it is the first to measure actual rather than self-reported online disclosure, which was feasible only in a controlled environment such as a campus computer laboratory. Also, because today's college students are likely the Internet shoppers of tomorrow indicates that they may be a particularly useful population to study when seeking to understand e-commerce behavior in the near future.

Second, this research was limited to only one type of e-commerce business and to user behavior with unfamiliar organizations. It is possible that disclosure of personal information to other types of commercial Web sites could differ from the type studied here, which was a retail poster vendor. For example, users might be more or less willing to disclose to sites selling other kinds of merchandise or to sites that are more familiar to them. Prior research has found differences in perceptions of credibility and trust of Web sites depending on name recognition and amount of informational content on the sites (Metzger & Flanagin, 2003). Also, the type and amount of information that consumers are willing to disclose to commercial Web sites will likely change over time, both as consumers spend more time online and as they develop ongoing relationships with particular e-tailers (Hoffman et al., 1999; Sheehan & Hoy, 1999). All of this points to a need to conduct this type of research on trust and disclosure on a greater variety of online companies, as well as on familiar versus unfamiliar companies.

Finally, as mentioned earlier, this study overcomes criticisms of past research on Internet privacy by measuring disclosure more directly, rather than relying on self-reports exclusively. Some might argue, however, that conducting this study in a campus computer lab could have impacted participants' online behavior. Two pieces of evidence argue against a Hawthorne effect operating in this study: First, none of the participants indicated that they believed the company was not real when asked during the debriefing. Second, participants were asked on the questionnaire how likely they would be to disclose the same kinds of information to the Web site if they had found the site at home on their own. A strong, positive correlation was found between actual disclosure and what participants said they would have disclosed at home (r= .54, p < .001). Furthermore, although the laboratory setting of this study is certainly a limitation of the research, the direction of potential influence is unbiased, meaning that some participants may have felt compelled to disclose less information than usual during the study (e.g., because they knew they were being examined), while others might have provided more information (e.g., because of social desirability or reactivity effects). In any case, future studies of online privacy and disclosure should be conducted in truly naturalistic environments to the fullest extent possible.

In conclusion, this exploratory study is an initial step toward understanding consumers' trust of and information disclosure to commercial Web sites. It offers some of the first data on why people disclose personal information to Web sites, under what conditions they are more or less likely to disclose, and it more accurately gauges the quality of online disclosure than previous studies that have relied exclusively on self-reports. Such data come at a crucial time in the evolution of e-commerce, as the recent economic downturn has caused consumers to be more cautious with their spending (UCLA Center for Communication Policy, 2003), and as Internet-based companies struggle to establish business models with long-term viability. This study gives some insight into the barriers to e-commerce brought about by the public's concerns over privacy, and thus benefits scholars who seek to understand online behavior and Web site sponsors looking for ways to capitalize on new marketing opportunities afforded by advanced communication technologies.


The author wishes to thank Richard Grant for his help in data collection and early conceptualizations of this project.


  • 1

    Data may be gathered from consumers online both with (e.g., site registration) and without (e.g., cookies) Internet users’ consent. Although this study is concerned with information that is disclosed knowingly by users, several new technologies to help online consumers stop the surreptitious collection of their personal data are currently in development and should be examined for their effectiveness at protecting users’ privacy while online (Cranor, 1999; Lievrouw, 2000).

  • 2

    Because of the gender imbalance in the sample, the effects of sex were studied on all variables in the model. No significant differences were found for any variables except Internet experience. Males spent slightly more time online per week than females (t[185] = -3.57, p < .001). On average, males indicated that they spend about 90 minutes online per week (excluding email) while women said they spend closer to 60 minutes. Consequently, the data analysis was run controlling for respondent sex.

  • 3

    University attorneys deemed this to be a “fair use” under copyright laws, given the Web site's use for legitimate educational and research purposes and because it could have no bearing on the market value of the real Web site. As will be explained later, in addition to changing the name of the original Web site, a “marketing survey” feature was added.

  • 4

    The problem with low reliability is that it increases the error term in statistical tests, thereby making it more difficult to achieve significance. As will be demonstrated later, achieving statistical significance with this variable was not a problem in this study. Future research should, however, develop a more reliable scale to measure concern for online privacy.

  • 5

    Interestingly, none of the participants ordered a poster during the study, although several (37%) indicated that they would likely buy from this Web site in the future, and a majority took advantage of the free poster offer.

  • 6

    To protect the privacy of study participants, credit card and social security data were not actually recorded. The experimental Web sites were programmed such that the numbers entered into the Web site by participants were displayed and recorded as asterisks. Also, the model was tested using, first, all of the information provided by participants and, then, using only the information that was not required by the Web site (truly voluntary information only). The results were the same using both methods of calculating online disclosure.

  • 7

    Rather than complicating the picture by adding nearly redundant tables and figures to this article, results for the model using depth of disclosure are available from the first author.



Appendix A. Pretest ratings of information privacy

First name2.281.651-6
Education level2.351.301-5
Time spent online per week2.451.401-6
Marital status2.491.431-5
Product preferences2.771.141-5
Email address3.491.641-7
Number of people in household3.491.721-7
Political party affiliation3.601.781-7
Recent online purchases3.721.721-7
Last name4.062.001-7
Postal address5.131.381-7
Telephone number5.981.311-7
Credit card number6.631.001-7
Banking information6.660.961-7
Social security number6.740.941-7

Appendix B. Texts of Strong Privacy Statement and Weak Privacy Statement

Text of Strong Privacy Statement understands that our customers are concerned about their privacy and security. We appreciate your business and we promise to handle your personal information carefully and sensibly. This notice describes our privacy policy. We believe that our privacy policy should give you confidence whenever you use hours a day, seven days a week, 365 days a year.

What Personal Information About Customers Does Gather?
  • Information You Give Us: We receive and store any information you enter on our Web site or give us in any other way. You can choose not to provide certain information, but then you might not be able to take advantage of many of our features such as our current FREE POSTER OFFER.
  • Automatic Information: We receive and store certain types of information whenever you interact with us. For example, like many Web sites, we use “cookies,” and we obtain certain types of information when your Web browser accesses
Why does collect information?
  • The information we collect about our customers helps us personalize and continually improve your shopping experience at
  • We use the information that you provide for such purposes as responding to your requests, customizing future shopping for you, improving our stores, communicating with you, and handling orders.
Does Share the Information It Receives?
  • will always contact you and ask for permission before releasing any of your personal data.
  • Information about our customers is an important part of our business, and we are not in the business of selling it to others.
How Secure Is Information About Me?
  • We work to protect the security of your information during transmission by using Secure Sockets Layer (SSL) software, which encrypts information you input.
  • Shopping at is 100% safe-guaranteed.
  • You should feel confident that the information you provide us will not be intercepted or stolen by a third party.
What Information Can I Access?
  • gives you access to your account/order information so you can make sure everything is accurate. In order to receive a copy of these records, please CONTACT US and make a request. We will be more than happy to oblige.
Contact us with questions and concerns!
  • Please feel free to CONTACT US and give us feedback on the Website and your shopping experience. Or if you have any questions or concerns about your account or an order, please do not hesitate to CONTACT US.
  • In the case of technical difficulties, our e-mail address is or call (805) 893-8237.
Text of Weak Privacy Statement understands that our customers are concerned about their privacy and security. This notice describes our privacy policy.

What Personal Information About Customers Does Gather?
  • Information You Give Us: We receive and store any information you enter on our Web site or give us in any other way. You can choose not to provide certain information, but then you might not be able to take advantage of many of our features such as our current FREE POSTER OFFER.
  • Automatic Information: We receive and store certain types of information whenever you interact with us. For example, like many Web sites, we use “cookies,” and we obtain certain types of information when your Web browser accesses
Why does collect information?
  • The information we collect about our customers helps us personalize and continually improve your shopping experience at
  • We use the information that you provide for such purposes as responding to your requests, customizing future shopping for you, improving our stores, communicating with you, and handling orders.
Contact us with questions and concerns!
  • Please feel free to CONTACT US and give us feedback on the Website and your shopping experience. Or if you have any questions or concerns about your account or an order, please do not hesitate to CONTACT US.
  • In the case of technical difficulties, our e-mail address is or call (805) 893-8237.