SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References

This study applies Petronio’s Communication Privacy Management theory (CPM) to understand the tension between information disclosure and privacy within e-commerce relationships. It proposes that consumers manage their privacy concerns through decisions to reveal or conceal information about themselves in interactions with online retailers. The study investigates the degree to which privacy management strategies identified by CPM theory to regulate privacy and disclosure within interpersonal relationships, including withholding and falsifying information, as well as seeking information seeking from a relational partner, operate in the computer-mediated context of e-commerce relational transactions. Findings suggest that online consumers do erect boundaries around personal information and form rules to decide when to reveal information that are consistent with CPM theory. Overall, this study provides knowledge about privacy in online commercial transactions, serves as a basis for more directed theory construction in this arena, and has important practical and policy implications.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References

Despite increasing use of the Internet for electronic commerce, online privacy remains an important issue for consumers. Several polls find that privacy is the primary concern inhibiting people from engaging in e-commerce (Digital Future Report, 2005; Miyazaki & Fernandez, 2001). Privacy is implicated in e-commerce because of the risk involved in disclosing personally-identifying information, such as email addresses or credit card information, which is required for most e-commerce transactions. Specific privacy concerns in this realm include companies’ use of customers’ information for electronic surveillance (e.g., ‘cookies’), email solicitation (e.g., ‘spam’), or data transfer (e.g., when customer database information is sold to third parties or stolen) resulting in identity or credit card theft (Digital Future Report, 2005).

An important question for scholars examining online information disclosure, persuasion, and privacy, then, is how do people manage or cope with privacy concerns in e-commerce transactions? Under what circumstances do consumers decide to disclose or withhold information? This study invokes communication privacy management theory (Petronio, 2002) to address these questions. As such, this research considers privacy and e-commerce as a communicative process. The results are used to build an understanding of privacy regulation and disclosure during online commercial transactions, an area in which research is plentiful but theory is lacking (White, 2004). The findings of this study are then leveraged to help e-commerce practitioners craft messages that respond effectively to online consumers’ privacy concerns.

Privacy, privacy management, and communication privacy management theory

Privacy may be defined as an individual’s ability to determine when, how, and to what extent personal information is disseminated to others (Westin, 1967). Accordingly, consumers’ privacy fears in e-commerce transactions stem from their potential loss of control over personal information. At the same time, the convenience of e-commerce is attractive to many consumers. Communication privacy management theory (Petronio, 2002) addresses the tension between disclosure and privacy by examining how and why people decide to reveal or conceal private information across various relational contexts.

Undergirding CPM theory is the idea that disclosure has both benefits and risks, and thus that people must balance their competing needs for privacy and for disclosure. The benefits of disclosure range from self-expression to relationship development to social control. The risks include loss of face, status, or control. When people disclose, they give over something they feel belongs to them (e.g., private information), and therefore they feel they should retain the right to control it, even after disclosure. Disclosure renders people vulnerable to exploitation by others because information changes from being privately owned to being co-owned (Petronio, 2002). As such, disclosure always involves some degree of risk. This risk, according to CPM theory, leads us to erect boundaries around what information we consider public and private. These boundaries allow us to control who has access to the information and motivate us to set expectations for co-ownership of information once it is disclosed to others (Petronio, 2002).

CPM is a rule-based theory that proposes that individuals develop rules to aid decisions about whether to reveal or conceal private information and thus how to best protect personal privacy. The theory states that individuals develop rules to help them maximize the benefits while minimizing the risks of disclosure. The rules that are developed can stabilize over time through repeated use but are also highly situational and may be changed to fit new or evolving circumstances. Furthermore, many different rules are used throughout the boundary management process to decide what, when, and to whom to disclose.

The theory proposes that there are three processes of boundary management. First, ‘boundary rule formation’ stipulates that people develop rules to regulate when and under what circumstances they will reveal rather than withhold information. Second, ‘boundary coordination’ refers to the process of negotiating privacy rules between partners, for example, whether disclosed information can be revealed to others outside the relationship. As part of the coordination process, individuals enact rules to moderate boundary linkages (whether to link to others), boundary ownership rights (who should be included or excluded in the boundary), and boundary permeability (what information may be revealed to whom). Third, ‘boundary turbulence’ may result from differences in privacy rules used by individuals, privacy rule violations, or deficient boundary coordination, for example, when one partner shares information outside the relationship that violates the other partner’s expectations.

Applying CPM theory to privacy management online

CPM was developed to understand how people decide to disclose information within interpersonal relationships. However, the theory has expanded to explain disclosure within other settings, including group, organizational, and institutional relationships. Moreover, Petronio (2002) and others have discussed the applicability of CPM to privacy issues generated by new technologies, including the Internet (Altman, cited in Petronio, 2002; Stanton & Stam, 2003; West & Turner, 2004). Although there are significant differences between privacy issues in face-to-face versus computer-mediated communication (CMC) contexts, many of the basic premises of CPM theory likely endure in online privacy management.1 First, both benefits and risks to disclosure within e-commerce relationships exist, just as in other types of relationships. Benefits may include convenience, faster service, and lower prices. Risks include vulnerability to spam, theft, and electronic surveillance. Second, as within interpersonal relationships, studies find that people feel ownership over the personal information they provide to e-commerce retailers (‘etailers’), and believe they have a right to control access to information they give about themselves online (Federal Trade Commission, 1998; Fox, 2000). Finally, the main elements of boundary management—boundary rule formation, coordination, and turbulence—are evident in online privacy management.

With regard to boundary rule formation, preliminary evidence suggests that online consumers may construct rules to determine if and when they will disclose personal information to etailers, and that they will do this using similar criteria as in CPM, including culture, motivation, the specific situation, and risk-benefit analyses. For example, studies find that online disclosure is negatively related to an individual’s level of privacy concern, which is affected by larger cultural values surrounding privacy (Milberg, Burke, Smith, & Kallman, 1995).2 Motivations such as attraction/liking and expectations of costs or rewards that are known to affect disclosure rule formation in interpersonal contexts are also found in research on disclosure in e-commerce contexts. For example, people are more willing to disclose information to reputable etailers (Swaminathan, Lepowska-White, & Rao, 1999) and when etailers offer rewards for disclosure (White, 2004). Situational demands, such as whether a desired product is only available online, are likely to contribute to the rules people form to manage their privacy online. Evidence also suggests that risk-benefit calculations drive online consumers’ decisions regarding electronic information exchange, and that consumers balance their privacy concerns against the convenience of online shopping when deciding whether to engage in e-commerce (e.g., Bhatnagar, Misra, & Rao, 2000; Hann, Hui, Lee, & Png, 2002; Miyazaki & Fernandex, 2001; White, 2004).

Boundary coordination and turbulence are also evident in consumers’ efforts to manage their privacy online. As in face-to-face contexts, for many, part of the decision to disclose personal information to a website involves coordinating expectations about how the disclosed information will be treated and who will have access to the information outside the boundary. In other words, a set of privacy access and protection rules must be negotiated between parties. Furthermore, the nature of boundary negotiations is different in CPM versus in online privacy management. Whereas face-to-face boundary coordination may involve back and forth negotiation and agreement between relational partners, online consumers must negotiate boundaries in other ways, for example, through self-protective behaviors such as setting privacy preferences, rejecting cookies, opting-out of mailing lists, and only providing information to etailers who promise not to reveal information to third parties.

Finally, instances of boundary turbulence in e-commerce are available. The public uproar that ensued after America Online changed its privacy policy in 1997 to allow direct marketers to purchase subscribers’ information and after Amazon changed its policy in 2000 to reserve the right to sell customer database information to 3rd parties illustrate the potentially turbulent relations that can erupt over shifts in boundary conditions.

This study applies CPM to online consumer relationships to understand information disclosure in e-commerce, focusing specifically on the ways in which consumers strive to protect their privacy in e-commerce relationships by enacting rules to regulate boundary permeability, boundary ownership rights, and boundary linkages. It is argued that online consumers use privacy-protection rules that guide the choices they make regarding whether to withhold information about themselves online, limit exposure by falsifying information online, and coordinate privacy rules with etailers by seeking information prior to disclosure to assess the danger of disclosure and to avoid turbulence. In each case, it is theorized that privacy protection rules are used by e-commerce participants to minimize the risk inherent in online disclosure. In this way, this research moves CPM theory to the CMC e-commerce environment and adds to our understanding of private disclosures beyond interpersonal settings.

Withholding information

According to CPM theory, boundaries around private information can be more or less permeable, and people regulate permeability by enacting various boundary access and protection rules. Under conditions of minimal risk, boundary access rules likely predominate, which results in high levels of disclosure. When risk of disclosure is perceived to be great, boundary protection rules prevail and more withholding results (Petronio, 2002). Although withholding information as a protection rule has been well documented in interpersonal contexts, it has never been investigated online. Nonetheless, there is some evidence to suggest that withholding information is a strategy used for protecting privacy in e-commerce situations. Polls find that people refuse to register for websites and fail to supply information requested by online marketers (Sheehan & Hoy, 1999). Some users cite threats to privacy as a reason for withdrawing entirely from e-commerce (Digital Future Report, 2005). Thus, the first hypothesis of this study proposes that, as in interpersonal scenarios, withholding information is a rule employed by Internet users to limit disclosure and protect personal privacy in CMC contexts:

H1: E-commerce participants will withhold information from a commercial website as a privacy protection rule.

However, maintaining impermeable (closed) boundaries around information may itself create risk. CPM predicts that relationship quality suffers when no information is allowed to pass through boundaries (Petronio, 2002). In other words, disclosure leads to relationship development, and without disclosure, relationships are often terminated. This explains why there is usually some disclosure and some withholding in most relationships (Petronio, 2002), even those between etailers and their customers who must disclose information to reap the benefits of online shopping despite their fears about privacy (Sheehan & Hoy, 1999). CPM theory thus stipulates that privacy protection and access rules guide not only whether people reveal or conceal information but also what information is withheld or disclosed within a relationship (Petronio, 2002).

Not all personal information carries with it the same degree of risk. As Petronio (2002) states, “private information changes in degrees of risk based on perceived repercussions for revealing and concealing” (p. 67). This suggests that some information is more readily disclosed or withheld according to the perceived consequentiality of disclosure. It further suggests that individuals develop privacy rules such that as information is perceived to be more risky to reveal, it is more likely to be withheld. Indeed, studies on interpersonal communication show this to be true. For example, some researchers have found that people use topic avoidance protection rulesto help them decide whether to avoid discussing particularly risky topics in order to protect themselves (Afifi & Guerrero, 2000).

Applying this notion of perceived repercussions to online privacy management, consumers may refuse to disclose certain types of information to etailers as a way to protect themselves, while disclosing other information in order to obtain desired products or services. CPM logic suggests that e-commerce consumers develop rules about withholding and disclosing information that carry different levels of risk. So, people may reveal personal information when trust of the etailer and benefit of disclosure are high, and/or may withhold more sensitive personal information when possible. The second hypothesis of this study is predicated on the idea that e-commerce participants may develop a rule about withholding certain types of information (e.g., more sensitive types) as a means of protecting their privacy. If true, this rule predicts that:

H2: More sensitive information will be withheld to a greater extent by e-commerce participants than will less sensitive information.

Falsifying information

CPM theory predicts that another strategy to protect privacy is to falsify information. Falsifying information allows an individual to experience the benefit of disclosure while maintaining his or her privacy (Petronio, 2002). In this way, deception may be conceptualized as a privacy protection rule that limits disclosure and thereby protects privacy boundaries. Deception can happen during boundary coordination or as a result of boundary turbulence. For example, in situations where one relational partner may prefer to keep certain information to him/herself but the other partner expects that information to be shared within the dyad, the discloser may choose to falsify some information rather than withholding as a way to maintain the relationship while retaining ownership of private information.

Analogous processes likely occur in e-commerce relationships, for example, when people deliberately falsify information to etailers as an online privacy protection rule that works by limiting exposure and thus the risk associated with disclosure of personal information. Online deception may also result from boundary turbulence, such as when a person wishes to enjoy the benefits of online shopping but has had (or has heard about) past negative experiences with disclosing information to etailers (e.g., receiving spam). Also, when customers are forced to provide personal information to complete a transaction online, they may falsify information, which allows the customer to keep his/her information private while fulfilling the etailer’s expectation for co-ownership of the information. Interestingly, in these situations, falsifying information is motivated by self-defense and so may not be felt as an act of deliberate deception by the consumer, since the consumer feels s/he owns the information (Petronio, 2002). In other words, deception is defined from the perspective of the receiver of the information in this context.

Evidence that e-commerce participants develop rules about falsifying information as a means to protect privacy exists, although studies show wide variation in terms of the rate of falsification. Estimates are that between 13% and 40% of online consumers have provided false information at one time or another (Kehoe, Pitkow, Sutton, Aggarwal, & Rogers, 1999; Pew Research Center, 2000; Sheehan & Hoy, 1999). These studies rely on self-report data, which may be subject to social desirability biases (Sheehan & Hoy, 1999). The present research explores falsifying information as a privacy protection strategy by examining actual online behavior in conjunction with self-report data in order to provide a more accurate understanding of the use of deception in e-commerce exchanges than is possible in survey research. Thus, although it is known that deception is used in both interpersonal relationships and online, the first research question seeks to determine the degree to which online consumers falsify information to etailers as a means of privacy protection:

RQ1: What percentage of e-commerce participants falsify information on a commercial website as a privacy protection rule?

Research on interpersonal relationships further suggests that whether a person lies depends on the type of information requested. Both CPM and social penetration theory (Altman & Taylor, 1973) predict that revealing sensitive information makes a person feel more vulnerable than revealing other types of information because the perceived risk of disclosure varies with information type. Jourard and Lasakow (1958), for example, found that people are more willing to disclose information about their interests, opinions, and work than they are to reveal their deep emotions, insecurities, or financial and health information. A direct analogy to the online context can be made, since revealing some kinds of information (e.g., credit card) is more risky than revealing other kinds of information online (e.g., general preferences). The third hypothesis is founded on the notion that e-commerce participants develop rules about what types of information to falsify based on their perceived repercussions of disclosure and, therefore, predicts that:

H3: E-commerce participants are likely to falsify more rather than less sensitive types of information as a means to protect their privacy.

Information seeking

According to CPM theory, seeking information from a relational partner prior to disclosing information is an important part of boundary formation and coordination processes. Information seeking may be viewed as a rule that regulates boundary linkages by helping a person decide whether to initiate a boundary linkage with a new partner. Information seeking is part of the cost-benefit analysis that the discloser goes through to determine, and thus control, the risk of disclosing information to a partner (Petronio, 2002). Assessing the trustworthiness of a partner prior to disclosure by determining whether the recipient will use the information responsibly is thus a key component of boundary formation (Petronio, 2002).

CPM theory further explains that after linkages are established, the decision to disclose private information to others involves negotiating expectations about how the information will be collectively managed once divulged. Boundary coordination, then, involves negotiating rules about boundary permeability and co-ownership, for example, determining who outside the collective boundary will have access to the information once it is revealed. For the discloser, the purpose of coordination processes is to assess the risk of revealing, and, consequently, having information about the confidant is crucial to the disclosure decision (Petronio, 2002). When there is a high degree of uncertainty about the partner, CPM theory predicts that there will be more information seeking prior to the decision to disclose private information (Berger & Calabrese, 1975; Petronio, 2002).

With regard to e-commerce transactions, online consumers may be motivated to seek information about how a website with which they are considering doing business will handle their personal information. Etailers’ privacy policies contain information about how the company will treat and disseminate customer information, which may help consumers make the risk-benefit calculations that CPM theory predicts regulate the decision to disclose information. Thus, people may develop rules about reading privacy statements to assess the risk of disclosure as a way to coordinate boundaries around co-owned information. However, not everyone takes the time to read etailers’ privacy policy statements (Culnan, 2001). This leads to the following research question concerning the extent to which e-commerce participants develop rules about information seeking for deciding whether to disclose personal information:

RQ2: To what degree do people seek information from a commercial website’s privacy policy as a strategy for privacy management during e-commerce transactions?

As discussed, seeking information may be used by relational partners to guide disclosure decisions in interpersonal contexts. Information-seeking may work similarly in computer-mediated relationships as well. Privacy policies are used by etailers to increase consumers’ disclosure to commercial websites by lowering the perceived risk of engaging in electronic exchanges (Jarvenpaa & Tractinsky, 1999; Swaminathan, et al., 1999). By providing information about how personal data will be treated and to whom it will be revealed, privacy policies can decrease risk by giving customers a greater sense of control over the exchange (Federal Trade Commission, 1998). In CPM terms, privacy policies communicate explicit rules for whether, how, and when personal information will be allowed to permeate the collective boundary after disclosure, and individuals can use that information to decide whether the website’s rules align with their own privacy goals. Privacy policies may also lower risk via enhancing the perceived security of online transactions by providing information about the security measures taken by the etailer to protect customer information (Culnan & Armstrong, 1999; Palmer, Bailey, & Faraj, 2000; Swaninathan, et al., 1999).

Of course, not all policies provide the same level of privacy assurances. Indeed, studies reveal a great deal of variation in the content of etailers’ policies, with some promising greater or lesser degrees of privacy protection (Federal Trade Commission, 2000). Theoretically, perceived risk will be decreased only to the extent that the policy shows that the company actually protects customers’ privacy interests. Policies that have limited information about their privacy protections should not have the same effect of reducing perceived risk as more comprehensive policies. Therefore, seeking information from an etailer’s privacy policy should influence disclosure only to the degree that the policy lowers consumers’ perceived risk. Interestingly, the level of assurances used in websites’ privacy policies has never been investigated in any prior studies of online privacy. This study fills that void to predict that seeking information from an etailer’s privacy policy guides disclosure decisions as a part of privacy boundary formation and coordination processes. The fourth hypothesis thus states:

H4: Greater privacy protections offered by a commercial website’s privacy policy will result in more disclosure and less withholding on the part of e-commerce participants.

Individual differences in online privacy management

CPM theory stipulates that privacy rules may vary situationally depending upon an individual’s motivation or cost-benefit analysis but may also coalesce into consistent patterns of disclosure based on more stable criteria, such as one’s gender, culture, or certain life experiences (Petronio, 2002). The third research question explores this issue with regard to privacy management in e-commerce.

Gender

Studies of dyadic relationships find some consistent patterns to revealing and concealing information based on gender. For example, women appear to disclose more information than men overall (Petronio, 2002). Differences may be due to variations in how men and women are socialized, sex-role expectations, or in how men and women use different criteria in defining and controlling private information (Petronio, 2002). There is some evidence for gender differences in online privacy management. Studies find significant differences in how men and women respond to online privacy issues. Women express greater concern about their privacy online and report providing incomplete information to commercial websites more than do men (Pitkow & Kehoe, 1997; Sheehan, 1999).

Studies have yielded conflicting results with regard to gender differences in falsifying information online. For example, while Sheehan (1999) discovered no gender differences in the amount of lying to websites, a survey by the Pew Research Center found that males reported lying slightly more often than females (Fox, 2000). Small gender differences have also been observed in studies of lying in interpersonal contexts (Burgoon, Buller, Grandpre, & Kalbfleisch, 1998; DePaulo, Epstein, & Wyer, 1993).

Privacy concern

Just as individuals vary in their perceived risk of disclosure within interpersonal relationships (Petronio, 2002), studies find that Internet users vary in the degree to which they feel e-commerce is risky. This is manifested in substantial variance in consumers’ stated level of online privacy concern (Digital Future Report, 2005; Federal Trade Commission, 2000) and in their behavior of providing information to online marketers (Sheehan & Hoy, 1999). With regard to concern for privacy, CPM theory predicts that people with high privacy concern would be motivated to protect their privacy, perhaps by forming privacy protection rules that might involve withholding, deception, and information seeking. Indeed, Sheehan (1999) found some differences in lying based on Internet users’ level of concern for privacy.

Experience

Disclosure in e-commerce settings may also be contingent upon an individual’s past experience with the Internet and/or with electronic commerce specifically. Chelune (1987) found that past experience with information exchange affects the value placed on such exchanges and influences both expectations about and actual disclosure in future interpersonal exchanges. Extending this logic, past experiences with online disclosure may affect future e-commerce behavior by mitigating or amplifying perceived risk. Indeed, research confirms that willingness to reveal personal information to marketers is determined in part by whether people have done so in the past (Culnan & Armstrong, 1999; Metzger, 2006). Others have found that Internet experience is negatively associated with perceived risk of e-commerce (Miyazaki & Fernandez, 2001) and positively associated with willingness to provide personal information online (Kehoe, et al., 1999).

Experience with the Internet or with e-commerce may also affect online deception. Past online experiences may motivate an individual to be more or less forthright in future online interactions. Certainly, boundary turbulence experienced during past online disclosure should have a negative impact on consumers’ willingness to disclose information in subsequent transactions. Although this issue has not been tested empirically, Fox (2000) found that experienced Internet users reported providing false personal information to websites more often than less experienced users.

Finally, as mentioned earlier, not everyone reads privacy policies. Those with greater concern about online privacy, or those who have suffered privacy problems as a result of past online disclosure, may be more motivated to read them. Other factors may influence motivation as well. For example, as discussed earlier, higher concern for privacy has been linked to gender, where women have been shown to be more concerned than men about privacy. Also, past Internet and e-commerce experience may impact one’s motivation to read privacy statements as a means to protect privacy, although it is difficult to predict in what direction: Greater experience could lead people to read privacy policies less frequently, as trust in the security of online commerce is developed through positive experiences (see Miyazaki & Fernandez, 2001). Alternatively, negative online or e-commerce experiences could have the opposite effect, leading consumers to scrutinize etailers’ policies more thoroughly. As a first step toward examining these issues, the last research question asks whether factors identified by CPM theory, including gender, motivation, and experience, impact disclosure in e-commerce settings:

RQ3: Do individual differences in gender, level of concern about online privacy, and Internet or e-commerce experience influence online consumers’ rates of information seeking, falsification, or withholding on a commercial website?

Method

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References

Participants

Students enrolled in introductory communication courses at the University of California, Santa Barbara served as participants in this study. Two hundred and thirteen students who were members of an IRB-approved subject pool established at the researcher’s university received course credit for their voluntary participation. Participants’ average age was 19.54 years (SD = 1.42), and average Internet experience was 5.26 years (SD = 1.98). Most participants indicated that they spent 1-7 hours per week online, not including using email. Most also reported making between 1-6 online purchases “last year” and spending an average of $159. The sample reflected the university population in terms of race (63.7% Caucasian) and family income ($60-$120K) and also mirrored a gender skew within the communication major (75.0% female).

Procedure

Participants came to an on-campus computer lab where they were asked to view a website that sold music CDs as part of a study of e-commerce. After providing informed consent, subjects browsed the site for at least 10 minutes and were instructed as follows: “Even though you are looking at this website here in the lab, this is a real site—so for example, any orders you place while on the site or any offers made on the site are real. It is important for our research that you respond to the site just like you would if you had come across this site on your own, as if you were at home.” When participants were finished browsing, they completed a questionnaire about their experiences and were debriefed afterward.

Stimuli

Participants viewed one of three versions of a stimilus website created for this research but closely patterned after real music etailer sites. The three versions varied only in terms of the presence and/or wording of the privacy policy.3 Specifically, each participant viewed one site that had either no privacy policy, a “weak” privacy policy, or a “strong” privacy policy. The weak and strong privacy policies were based on existing policies of real etailer websites. The strong privacy policy provided all of the information that the Federal Trade Commission (2000) identifies as necessary for a comprehensive policy, including notice (notifying users that the website 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 contact information. Site visitors’ privacy and data security were ensured on each of these dimensions in this condition of the study. By contrast, the weak privacy policy did not provide access, choice, or security, and it notified site visitors that their data not only would be collected but could be passed to third parties.

A pretest of the privacy policies revealed significant differences in the degree to which site visitors’ felt the policies protected their privacy. One hundred and twenty-eight student participants in a separate sample read either the strong or weak privacy policy and were asked to what degree they would consider the website to be safe, secure, and trustworthy, based on the privacy policy statement. Pretest participants also indicated the extent to which the company issuing the privacy statement they read protects the rights and interests of customers, ensures that personal information is kept private, treats customer information properly, and is trustworthy. Differences between the strong and weak privacy statement versions were in the expected direction and all were significant at the p < .001 level.

Sample and procedure appropriateness

The use of a student sample in a laboratory setting raises two issues with regard to the quality of data. First, college students may differ from older e-commerce populations in ways that might impact the results of this research. Second, the laboratory setting may reduce perceived risk of disclosure. Several measures were taken to test for and mitigate these potential problems.

First, evidence suggests that college students are fairly similar to the overall adult Internet user population on several key dimensions. Like most active e-commerce adult participants in the United States, college students tend to come from middle or upper-middle class families and have access to and skill using Internet technology (Pastore, 2000a). In terms of online purchases specifically, a 2002 WSL Strategic Retail poll using a national representative sample found that an almost equal percentage of consumers ages 18-34 (29%) shop online as those in the 35-54 age range (27%).Other studies using large samples of adult Internet users provide further points of comparison (Wharton Forum on Electronic Commerce, 2000; UCLA Center for Communication Policy, 2003). The percentage of students in this study who indicated that they had made at least one online purchase “last year” (70%) fell between estimates from UCLA (50%) and the Wharton School (81%) for the same time period. Similarly, the amount of money the student participants spent online also fell between estimates from the UCLA and Wharton samples. This indicates that the participants in this study do not differ dramatically from the larger adult population in terms of general e-commerce behavior. Also, to ensure sample appropriateness and external validity, the product category selected for this study (i.e., music CDs) was among the most common items purchased online by both college students and adult Internet consumers at the time of data collection (Pastore, 2000a, 2000b).

Second, the data indicate remarkable similarity between the students who participated in this study and the more general Internet population in terms of perceived risk. For example, the UCLA (2003) study found that 94.6% of Internet users were “very or extremely concerned” about the privacy of their personal information when buying online, and 5.5% were “not at all concerned.” In the present study, 94.9% of participants indicated that they were extremely to somewhat concernedabout privacy issues related to e-commerce, whereas 5.1% indicated they were not concerned about these issues. Again, the college sample is strikingly similar to the larger population of Internet users in terms of perceived risk of engaging in e-commerce.

The question of whether the participants felt any risk while in the laboratory was also explored. Indeed, the majority of participants who did not disclose any information during the session indicated it was due to privacy and security concerns, as discussed later. Also, data show that disclosure in the lab environment was similar to what it would be in the home, although a few participants said they would have disclosed more information at home than they did in the lab, indicating that they actually felt more rather than less risk in the lab. Thus, disclosure in this study may be slightly underestimated. Finally, participants were queried during debriefing about their perception of the website they viewed in the lab, and all indicated that they believed that the site was real.

Measures

Information disclosure and withholding

As a way of getting at participants’ privacy access and protection rules, participants’ actual disclosure and withholding of information to the website were measured. All three versions offered two opportunities for disclosure. First, every page of the website included a flashing “Get a Free CD” graphic that linked to a page offering a free CD in exchange for completing a “marketing survey to help improve customer service.” The survey asked for participants’ first and last name, postal address (street address, city, state, and zip code), home telephone number, email address, age, sex, education level, income, marital status, number of people living in their household, race, political party affiliation, time spent online per week, hobbies/interests, last online purchase, favorite kind of music, favorite website, and social security and credit card numbers. Although all information was provided to the site voluntarily, participants were informed that in order to obtain a free CD, their name and address were required. Second, participants could order CDs by completing an online order form that asked for their first and last name, address, telephone number, credit card number and expiration date, and CD order. The information that participants provided to the website was captured automatically in a data file.4

The post-exposure questionnaire asked respondents to indicate whether they provided any untruthful information to the site. To reduce social desirability problems, the wording of the question stem read, “Many people give false or inaccurate information to websites. For example, they might give a fake name or email address. We are interested in what kinds of false information you might have provided while on the website today. We will not ask you to correct the information you gave falsely, nor will you be penalized in any way for having provided false information to the website, so please be honest.” Participants indicated any type of information that they falsified to the stimulus site on a checklist.

A score of 0 was assigned to any piece of information that was either not provided or was false, and a score of 1 was assigned to any piece of truthful information provided while online. A “truthful disclosure” scale score was obtained by summing across all of the items, whereas information falsification was computed by counting the number of items respondents indicated they provided falsely. Information withholding was operationalized by calculating the difference between the maximum amount of information that could have been disclosed and the amount of truthful information each participant actually disclosed. In addition, an open-ended question probed participants’ reasons for withholding information from the website.

Information sensitivity

To further explore the boundary access rules used by participants, data from the pretest were used to determine Internet users’ sensitivity to revealing various kinds of personal information while online. Pretest participants (N = 128) were asked how comfortable they are in giving each of the 23 types of personal information that the stimulus websites used in the main study requested (first name, last name, credit card number, etc.) on scales ranging from 1 “Very Comfortable” to 4 “Not at all Comfortable.” Mean sensitivity scores were calculated for each information type and are used in the analyses of Hypotheses 2 and 3.5

Privacy policy information seeking

Rules governing boundary coordination by seeking information from the website’s privacy policy were explored by unobtrusively capturing each participant’s ‘clickstream’ data during browsing, and by having participants report whether they recalled seeing the stimulus site’s privacy policy on the post-exposure questionnaire. Analyses were run on both self-reported and actual attention to the privacy policy.

Concern for privacy and security

Six items were combined into a scale of participants’ concern for online privacy and security, including their extent of agreement with statements such as: “There should be new laws to protect people’s privacy on the Internet,”“I am concerned about people I don’t know getting my personal information over the Internet,”“The amount and type of personal information collected by websites should be limited,” and “A user ought to have complete control over which websites get what personal information about them.” Items were derived from prior studies of this concept (Fox, 2000; Jarvenpaa & Tractinsky, 1999; Miyazaki & Fernandez, 2001; Swaminathan, et al., 1999). Analyses showed the scale to be unidimensionaland acceptably reliable for exploratory research (Cronbach’s alpha = .66; Nunnally, 1967).

Internet and e-commerce experience

Internet experience was assessed by asking participants to indicate how much time they spend per week on the Internet (not including using email) in hourly increments and how many years they have been surfing the Web. Two items were used to measure participants’ e-commerce experience: the number of online purchases made in the past year and the total amount of money they spent online last year.

Demographics

Finally, basic demographic information including gender, age, race/ethnicity, and approximate parents’ income was measured.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References

The first hypothesis proposed that online consumers may withhold information from commercial websites as a privacy protection rule. This was tested by investigating how many people withheld information from the commercial website, how much information participants withheld, and the reasons given for withholding information. Of the 213 participants, 89 (41.8%) disclosed at least one piece of information to the website, whereas 124 (58.2%) withheld all information about themselves. The mean amount of withholding across all participants was 16.61 items.

Responses to the open-ended question asking why participants chose to withhold information from the site were analyzed to determine the extent to which privacy-related concerns were mentioned. Each participant’s response was coded as reflecting privacy-related reasons, ambiguous reasons, or privacy-unrelated reasons for withholding. Of the 146 respondents who provided reasons, 85 (58.2%) cited a privacy-related reason for withholding information from the website (e.g., afraid of spam, didn’t trust the site, never give personal information online because of risk), 8 (3.8%) cited an ambiguous reason (e.g., wasn’t required to get the CD), and 53 (36.3%) gave a reason that was not related to online privacy concerns (e.g., did not want the CD, too lazy, already belong to similar site).6 The differences in these percentages are significant, χ2[2]= 61.49, p<.001. These data suggest that withholding information is indeed a strategy for privacy management in the e-commerce setting, thus supporting H1.

Hypothesis 2 predicted that online consumers develop rules about withholding specific types of information, in particular more sensitive information, as a way of protecting privacy. As a first step, H2 was investigated by testing for differences in the percentages of participants who withheld each of the 23 types of information requested by the site. The omnibus test showed significant differences (Cochran’s Q[22]= 808.16, p < .001). Table 1 shows that participants were most likely to withhold some types of financial information or information that could link their identity to financial records, specifically their credit card and social security numbers (Wilcoxon z=−4.96, p < .001). Personal contact and consumer preference information were the next-most withheld items, and included email address, telephone number, favorite website, hobbies/interests, and last purchase made online, although income and political party affiliation fell in this range as well.7

Table 1.  Mean sensitivity scores and percentage of disclosure by information type
Type of InformationInformation Sensitivity% Disclosed
  1. Notes: Standard deviations are in parentheses. Information sensitivity means with matching superscripts do not differ significantly. Participants were required to provide their name and address to receive the free CD.

First name1.48 (.64)f39.9
State1.84 (.87)f39.0
Zip code2.43 (1.02)d39.0
Last name2.07 (.89)e38.5
City2.15 (.97)e38.5
Street address3.35 (.86)c38.0
Sex1.62 (.80)f37.6
Education level1.39 (.54)f37.1
Favorite kind of music1.86 (.93)f36.6
Race1.85 (.92)f36.2
Marital status1.74 (.82)f35.7
Time online1.57 (.71)f35.2
Number of people in household2.07 (.97)e33.3
Age2.15 (.89)e31.9
Email address2.63 (.99)d31.5
Income2.62 (1.04)d30.0
Telephone number3.54 (.69)b29.6
Political party affiliation2.06 (.96)e29.1
Favorite website1.73 (.85)f27.2
Hobbies/interests1.70 (.79)f23.0
Last online purchase2.19 (1.02)e22.1
Social security number3.87 (.38)a7.5
Credit card number3.57 (.61)b0.0

At the other end of the spectrum, participants were least likely to withhold their name and address, which were required for purchasing or obtaining the free CD. Aside from the required information, participants were least likely to withhold general demographic information about themselves, for example, their sex, race, education, marital status, time spent online, number of people in their household, and age. As a more direct test of H2, the correlation between information sensitivity and information withholding was computed and found to be both positive and significant, r= .61, p < .001. These findings support H2.

Taken together, the results of the first two hypotheses indicate that withholding information within e-commerce relationships is a common privacy-protection strategy and depends on the sensitivity of the information requested by the site. These findings are consistent with theoretical predictions, as delineated earlier. They also parallel research on self-disclosure that finds greater withholding of intimate or more sensitive information during the initial stages of interpersonal relationships (Altman & Taylor, 1973; Jourard & Lasakow, 1958).

Research Question 1 addressed the extent to which online consumers falsify information as a privacy protection rule. Of the participants who disclosed information to the website, 39.6% falsified some of the information they disclosed. Most people, however, falsified only a few items; out of the 23 items requested by the website, participants falsified only 2-3 items on average.

Delving further into online deception, the third hypothesis examined which types of information participants were most likely to falsify. Table 2 shows the percentages of lying for each type of information requested. Participants who revealed information generally falsified more sensitive information, especially information that could be linked to financial and identity records. Nearly everyone (93.8%) who provided their social security number to the website falsified it. Participants were next most likely to falsify their name, direct contact, and basic demographic information. Participants were most truthful about their mailing address (recall this was needed for receiving the CD), time spent online, marital status, and hobbies/preferences.

Table 2.  Lying by information type
Type of Informationn disclosedn falsified% falsifying
Social security number161593.75
First name851517.65
Last name821417.07
Email address671014.93
Last online purchase47714.89
Income64914.06
Age68913.24
Political party affiliation62812.90
Telephone number63812.70
Number of people in household71912.68
Race77911.69
Education level79911.39
Favorite website58610.34
Street address8189.88
City8289.76
State8389.64
Zip code8389.64
Sex8078.75
Time online7568.00
Marital status7656.58
Hobbies/interests4936.12
Favorite kind of music7845.13

To test H3 directly, the correlation between information sensitivity and information falsification was computed and found to be positive and significant (r= .59, p < .01), as hypothesized. Considered jointly, the results of H2 and H3 show that participants are least likely to provide sensitive information online and most likely to falsify it when they do, compared to other kinds of information, which supports both hypotheses as well as prior research findings on disclosure in other contexts.

Research Question 2 examined information seeking as a strategy for online privacy management. Only those participants in the two conditions that had a privacy policy on the website were used in the analyses (n = 145). The research question probed the extent of participants’ privacy policy information seeking as a means to manage their privacy during e-commerce transactions. Analyses revealed an interesting discrepancy between participants’ self-report and clickstream data. While 33% of participants reported that they saw the privacy policy, only 18.6% actually clicked on the link taking them to the page with the text of the policy. One interpretation of the discrepancy is a social desirability bias, such that participants lied about seeing the privacy policy because they know they “should” read privacy policies, and yet they did not here.

One piece of data indirectly supports this interpretation: Those who lied about seeing the privacy policy had more Internet experience than those who did not lie about seeing the policy (t[143]=−1.93, p < .05), which suggests that those who are relatively more experienced with the Internet may be motivated to lie about their privacy-protection behavior so that their behavior appears consistent with social prescriptions for privacy protection on the Web. In any case, the data show that a minority of participants sought information from the site’s privacy policy.

The fourth hypothesis proposed that greater privacy protections offered by a website result in less withholding and greater disclosure to the site. Analysis of H4 was approached in two ways. First, participants’ withholding and disclosure were compared across all three privacy conditions. Second, withholding and disclosure were compared among only those who indicated that they actually saw the privacy policy.

Compared to the overall rate of withholding, more people withheld information in the weak policy condition (71.2%) and fewer withheld information in the no policy condition (47.1%). The proportion of participants who withheld information in the strong privacy policy condition (55.5%) was very similar to the overall rate of withholding. These differences are significant, χ2[2]= 8.77, p < .05, which suggests that the content of a website’s privacy policy may in fact matter. However, post hoc tests showed no differences in disclosure between the strong policy condition and the condition with no privacy policy (χ2[1]= 1.01, p= .35), although a significant difference was observed between the weak privacy policy condition and the no policy condition (χ2[1]= 8.54, p < .01). Thus, the data only partially support H4.

Next, data were analyzed looking only at those who self-reported that they saw (n= 48) and those whose clickstream data show they actually clicked on (n= 27) the site’s privacy policy. In terms of the likelihood of disclosure, the self-report data indicate that those who said they saw the strong privacy policy were more likely to disclose information (60.7%) than those who reported seeing the weak policy (20.0%), χ2[1]= 7.86, p < .005. The clickstream data showed that 62.5% of the participants who clicked on the strong version of the privacy policy disclosed some information to the website, whereas only 37.5% on the weak version did so, but this difference was not statistically significant (χ2[1]= .94, p= .33). Given the substantial difference in percentages, it is possible that the chi-square test failed to reach statistical significance because of the low statistical power derived from the 27 cases available for analysis. Indeed, a power analysis showed the power to detect differences to be only .25.

An identical pattern of findings was observed for amount of disclosure. The self-report data showed a significant difference in the amount of information that participants disclosed, depending on whether they saw the strong (M= 12.07) or weak version (M= 4.00) of the privacy policy (t[45.65]= 2.85, p < .01). Again, however, the clickstream data (n= 27) failed to reach statistical significance (t[25]= .64, p= .53) likely because of low power (power = .08), despite the fact that those in the strong policy condition disclosed 7.23 pieces of personal information on average compared to those in the weak version condition who only disclosed 4.71 pieces of information on average. So, while the self-report data support H4, results from the clickstream data are inconclusive, although all percentages were in the anticipated direction.

The results of Research Question 3 show few patterns of information withholding, falsification, or seeking based on participants’ gender, experience, or level of concern about online privacy. Although a slightly greater percentage of men (64.2%) than women (56.6%) withheld all personal information, this difference was not significant (χ2[1]= .93, p= .33). To assess individual differences in deception, participants’ likelihood of lying (whether participants falsified information or not) and their amount of lying (total number of items falsified per participant) were examined among only those who provided some information to the site (n= 89). Gender was not related to likelihood or amount of lying (χ2[1]= .68, p= .41; t[86]=−.28, p= .78). Finally, analyses revealed no gender differences in seeking information from the website’s privacy policy (χ2[1]= .08, p = .78).

Similar results were found for level of privacy concern. No difference was observed for level of privacy concern between those who did and did not withhold information from the website (t[210]= .37, p = .72). No relationship was found between concern about privacy issues for either the likelihood (rs= .02, p = .88) or amount (r=−.07, p = .50) of falsifying information. Surprisingly, concern for online privacy also showed no relationship with privacy policy information seeking (rs= .01, p = .90), although participants who did seek privacy information showed slightly greater privacy concern (M= 1.72) than those who did not (M= 1.64), but this difference was not significant (t[143]=−.70, p= .49).

Neither amount of experience with the Internet (rs = .05, p = .48) or with e-commerce (rs = .02, p = .83) was related to withholding information, in contrast to past e-commerce research (Culnan & Armstrong, 1999). Participants’ likelihood of lying was also not related to Internet experience (rs = −.02, p = .83) or to e-commerce experience (rs = −.15, p = .16). However, participants’ overall amount of lying was positively related to e-commerce experience (r = .23, p < .05) and marginally related to Internet experience (r = .19, p = .07). Finally, results show no relationship between either Internet experience (rs = −.04, p = .60) or e-commerce experience (rs = −.03, p = .77) and seeking information from the website’s privacy policy.

In sum, results generally support the four hypotheses. With regard to the research questions, there was some evidence that a sizable percentage of participants falsified their personal information and that some participants sought information from privacy policies as a means to protect their privacy, although not to a great extent. On the other hand, there was very little evidence that gender, level of privacy concern, or online experience have much impact on disclosure and information seeking within the e-commerce context.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References

This study shows that consumers manage their privacy online through their decisions to reveal or conceal information about themselves to online retailers. In particular, this study examines information withholding, deception, and seeking as online privacy management strategies. The research also provides insight into factors including gender, past online and e-commerce experience, concern about online privacy issues, type of information requested, and the specific language used in etailers’ privacy policies that may or may not influence decisions to disclose or withhold information.

Results inform past work on disclosure and interpersonal relationships that served as the basis for this study. Findings demonstrate that similar kinds of balancing dynamics appear to operate in the Web environment as they do in face-to-face situations, thus extending CPM into the domain of CMC, and e-commerce relationships specifically. Evidence that online consumers use strategies predicted by CPM theory—including information withholding, deception, and, to a lesser extent, information-seeking—is confirmed in this research. The study suggests that online consumers erect boundaries around personal information and form rules to decide when to reveal information to etailers, as predicted by CPM theory.

For example, the results confirm that, as in interpersonal communication, people are more likely to engage in deception and are less likely to disclose more sensitive information within e-commerce contexts. In CPM terms, online consumers may regulate access to personal information by making the boundaries around different types of information more or less permeable (i.e., “thicker” or “thinner”), depending on the degree of perceived risk involved in revealing more or less sensitive information. The boundary coordination processes observed in this study similarly illuminate how online consumers may use risk-benefit calculations to decide what to disclose within e-commerce relationships. Specifically, the fact that disclosure was higher in the strong privacy policy condition compared to the weak condition suggests that participants may have used policy information to “coordinate” boundaries to see if there was a match between their own and the etailer’s privacy expectations. Given that participants’ privacy concerns were generally quite high (M= 1.66 on a 5-point scale, where 1 indicated highest concern), a match was more likely to be found by those in the strong policy condition compared to the weak conditions. The strong privacy assurances may have lowered participants’ perceived risk of disclosure and encouraged greater boundary permeability than those in the weak policy condition in which boundary expectations may have been felt to be uncoordinated between the participant and the etailer. Of course, future research is needed to test this interpretation directly.

The data also encourage future research to extend CPM theory’s predictions of how boundary turbulence may impact disclosure decisions to e-commerce contexts. This is most evident in the open-ended questionnaire responses, such as “Whenever I give my information I get a lot of spam” or “I’ve had bad luck with giving out information about myself.” These responses indicate that prior negative experiences with online disclosure in e-commerce contexts may be a primary reason for withholding personal information. Results showing greater deception for those with more e-commerce experience also imply that boundary turbulence in past e-commerce relationships may play a role in people’s future online disclosure decisions. This is consistent with research showing that experienced Internet users were nearly two times more likely to provide false information to websites compared to less experienced users (Fox, 2000).

Together, the results of this study provide a basis to begin conceptualizing online consumers’ disclosure decisions in more systematic ways than have been attempted before. CPM theory predicts many aspects of the online behavior observed in this research and, thus offers a first step toward building a theory of online privacy management. At the same time, however, this study suggests that CPM must not be applied without accommodation for fundamental differences in the online context compared to face-to-face settings. For example, there was no evidence that online boundary rules are formulated on the basis of gender or general online privacy concerns, which is somewhat surprising given past findings in the interpersonal and e-commerce research literatures.

One reason why gender was less important in this study than in interpersonal relationships might be that the nature of disclosure in e-commerce contexts is quite different from that in most interpersonal relationships. Research on disclosure within interpersonal relationships finds that females tend to disclose more intimate and emotional information than do men (Derlega, Metts, Petronio, & Margulis, 1993). The lack of gender differences for withholding information found in this study might be explained by the fact that e-commerce transactions require disclosure of factual and largely non-emotional information. The results of this study agree with some prior e-commerce research that failed to find a relationship between gender and online deception (e.g., Sheehan, 1999) in spite of studies showing that women are more concerned about their privacy online than are men.

An explanation for the lack of findings with regard to concern for privacy in this study is the existence of a “privacy paradox” when it comes to online disclosure. Recent research finds that, despite expressing high levels of concern about privacy and security online, consumers are still willing to provide personal information to commercial websites (LaRose, 2004; Spiekermann, Grosslags, & Berendt, 2001). E-commerce incentives such as giveaways, lower prices, greater selection, the convenience of online shopping, and consumers’ feelings of powerlessness to protect their personal data on the Web have all been advanced as explanations for this paradox, and may have been operating in this study. Furthermore, this is not dissimilar to studies using CPM that have observed that people are sometimes willing to give up privacy when they seek security, in other words, that dialectical tensions sometimes shift from privacy-disclosure to privacy-security.

A second possible explanation is variable measurement. Privacy concern had low variance and acceptable, yet less-than-ideal reliability, making for an extremely conservative test of this concept statistically. Despite efforts to base the measure on prior research and the face validity of the items comprising the scale, future studies must develop better operationalizations of this concept.

There are other limitations of this study. Although the experimental design afforded many benefits, it forced participants to look at the stimulus website. It is possible that if respondents had come across the site naturally while shopping for CDs, more would have been interested in ordering a CD or taking advantage of the free CD offer. Consequently, they might have disclosed more information to the website and/or they might have been motivated to read the privacy policy. Indeed, some participants indicated that they probably would have disclosed more information if they had found the site on their own, as discussed earlier. This study, then, may provide a somewhat conservative view of online disclosure and information seeking. However, there is no indication that participants’ use of deception should be impacted in this study, since participants were not forced to provide their information to the website and because the motivation for lying to the site was no different than it would be to a similar e-commerce site.

It would be ideal to replicate this study using a greater variety of e-commerce sites and with a larger, more representative sample of Internet users. For example, brand name is likely a factor in online consumers’ privacy and disclosure behavior (see Metzger, 2006). Also, a larger sample would help to eliminate problems of statistical power seen in the analysis of H4. Indeed, the discrepancy in findings for the clickstream and self-report data demonstrates that results may differ depending on the method of observation, which highlights the need for researchers to measure actual online behavior in addition to self-reports. Similarly, evidence that many participants lied about reading the privacy policy is indicative of social desirability problems with survey-based research on this topic and shows that, to judge their effectiveness, researchers should go beyond self-report data to gauge how many people are reading privacy policies.

Implications of the research for policy and practice

Since the late 1990s, the federal government has struggled to respond to public outcry over online privacy issues, oscillating between forcing etailers to post privacy policies on their websites and encouraging etailers to protect online consumers’ privacy through self-regulation (Federal Trade Commission, 2000; Muris, 2001). The finding here that only a small percentage of site visitors bothered to read the privacy policy argues against the need for strict government mandates requiring etailers to post policies. On the other hand, results show that ensuring effective privacy protection may be important to those who do read privacy policies. An implication is that government efforts might be better spent—and certainly better appreciated by the public—by developing a comprehensive policy to standardize privacy and security protections for all e-commerce transactions (see the EU’s Privacy Directive), rather than forcing or relying on individual etailers to post policies that will inevitably vary in their assurances and practices.

This is among the first studies to examine how the content of privacy policies impacts consumers’ willingness to disclose personal information online, which has significant implications for practitioners of e-commerce. The finding that disclosure was higher in both the strong and no privacy policy conditions compared to the weak condition suggests that etailers should take an all-or-nothing approach to privacy protection. Indeed, having no policy at all may be preferable to offering weak privacy protections because weak “protections” may cue site visitors to privacy concerns while failing to address them adequately. The results showing that few people read privacy policies also may interest online marketers because it signals that etailers’ assurances of privacy and security, regardless of their content, may not be very effective in stimulating customers to disclose personal information. Together, the implications of these findings are that etailers striving to increase business by offering strong privacy protection should make their policies short, explicit, and clearly visible to users of their website.

The fact that willingness to disclose was significantly impacted by the sensitivity of information requested by the website implies that marketers need to be aware of and sensitive to consumers’ perceptions of risk when asking for users’ personal information (see also White, 2004). For example, results suggest that online marketers will likely have greater success eliciting less threatening information from consumers. Indeed, etailers may do well to wait until after a relationship has been established and trust proven—perhaps after a satisfactory transaction with a customer has been completed—to request more sensitive information. Although this would hinder efforts toward customized marketing in the short run, it may provide significant payoff in the long term with regard to eliciting disclosure from consumers.

Finally, the findings on deception offer both good and bad news for e-commerce practitioners. The good news is that the majority of participants who disclosed personal information in this study were truthful, and if they lied, it was typically about only a few items. The bad news is that a substantial number of consumers will use deception to protect their privacy. Prior survey estimates show an average of 20% of online consumers regularly falsify information (Kehoe, et al., 1999; Pew Research Center, 2000; Sheehan & Hoy, 1999; SurveyNet, 1997). This study shows that closer to 40% of participants falsified information, implying that previous estimates may underestimate online deception and perhaps reflect a social desirability response bias in self-reports. The implications for etailers are somewhat ominous, especially when viewed in light of the fact that consumers with greater e-commerce experience lied more than those with less experience. This suggests that efforts to elicit truthful disclosure will become more difficult for etailers in the future, as consumers gain experience with e-commerce.

Overall, the study further suggests that people may be cautious in giving their personal information online because they know they will have little opportunity to negotiate mutually-acceptable privacy rules. To cope with this, they may withhold the more private information and give the minimum to achieve their goal. Under circumstances where pre-set privacy rules disallow negotiation, then, people settle for providing only the minimum information to get what they want. Thus an interesting dilemma arises: By pre-setting privacy rules in their privacy policies, companies may be limiting the amount of information they can expect to receive from consumers.

Conclusion

By framing decisions to provide or withhold information, either truthfully or falsely, in terms of CPM theory, this study helps to understand the privacy decisions that consumers make during e-commerce transactions. Results suggest that specific elements of CPM migrate well to the e-commerce environment, and that the notion of boundary management has theoretical traction when applied to this context. This research also highlights similarities and differences between interpersonal relationships and online commercial transactions, suggesting that information disclosure and veracity in e-commerce are somewhat a function of the type of information requested, past e-commerce experience (with regard to the amount of lying), and the specific language used in privacy policies. Together, findings from this study serve as a basis for more directed theory construction in this arena.

Notes
  • 1

    One key difference of CPM theory as applied to online privacy management is that CPM focuses on “private” disclosures, or disclosing information that is not publicly available or has not been revealed to many people in the past. Information disclosure in online privacy management is conceptualized more broadly to include information that may be publicly available but may be accessed only with some effort (e.g., email address), and information that the discloser did not want revealed outside of a particular relationship or linked with other types of information about them stored in electronic databases.

  • 2

    The European Union Directive on Privacy and Electronic Communications is another example of how culture shapes privacy regulation. The EU views privacy as a fundamental human right, which stems from Europe’s unique historical and political experiences over the last century. This view was codified into its policy on online privacy, which affords greater protection of individuals’ privacy than does U.S. law. In the United States, privacy is seen as important, but not as a fundamental right. Thus, individual privacy concerns are often “balanced” against companies’ rights to collect information about consumers.

  • 3

    Familiar and unfamiliar CD vendor websites were balanced such that half of the sites were from a well-known music etailer and half were from an unknown etailer. The unknown etailer sites were created by simply changing the name of the familiar vendor’s site, and thus all sites were identical except in name. An examination of the differences between the familiar and unfamiliar sites is available (Metzger, 2006).

  • 4

    To protect the privacy of study participants, the website was programmed such that credit card and social security numbers entered into the website by participants were displayed and recorded as asterisks. Seventy-nine participants (37.1%) opted to receive a free CD, which was mailed to them after the study was complete.

  • 5

    Differences in information sensitivity among the information types were analyzed using contrast analysis using SPSS’s repeated-measures GLM procedure. See Table 1 for these results as well as the mean sensitivity scores for each of the information types.

  • 6

    Both the author and an independent coder coded the data. Scott’s Pi was .91, indicating very good intercoder reliability. Disagreements were resolved via discussion.

  • 7

    Some of the student participants may not have had an established political party affiliation, favorite website, or last online purchase, which could explain why these items were among the most withheld. Alternatively, favorite website and hobbies/interests may have struck some participants as less important to provide and were located at the bottom of the page, so there could also be a respondent fatigue issue as well with these items. Excluding these items, direct contact and information that could be linked to identity and financial records comprise the most-withheld items. Differences in withholding between the various information types were assessed via Wilcoxon signed rank tests, and test statistics are available from the author.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References
  • Afifi, W. A., & Guerrero, L. K. (2000). Motivations underlying topic avoidance in close relationships. In S.Petronio (Ed.), Balancing the Secrets of Private Disclosures (pp. 165179). Mahwah, NJ: Erlbuam.
  • Altman, I., & Taylor, D. A. (1973). Social Penetration: The Development of Interpersonal Relationships. New York: Holt, Rinehart, & Winston.
  • Berger, C., & Calabrese, R. (1975). Some explorations in initial interactions and beyond: Toward a developmental theory of interpersonal communication. Human Communication Research, 1(2), 99112.
  • Bhatnagar, A., Misra, S., & Rao, H. R. (2000). On risk, convenience, and Internet shopping behavior. Communications of the ACM, 43(11), 98105.
  • Burgoon, J. K., Buller, D. B., Grandpre, J. R., & Kalbfleisch, P. (1998). Sex differences in presenting and detecting deceptive messages. In D. J.Canary & K.Dindia (Eds.), Sex Differences and Similarities in Communication: Critical Essays and Empirical Investigations of Sex and Gender in Interaction (pp. 263280). Mahwah, NJ: Erlbaum.
  • Chelune, G. J. (1987). A neuropsychological perspective of interpersonal communication. In V. J.Derlega & J. H.Berg (Eds.), Self-Disclosure: Theory, Research, and Therapy (pp. 1034). New York: Plenum.
  • Culnan, M. J. (2001, December 4). Consumers & privacy notices. Paper presented at the Federal Trade Commission’s Interagency Public Workshop, Get Noticed: Effective Financial Privacy Notices, Washington, D.C. Retrieved December 12, 2006 from http://www.ftc.gov/bcp/workshops/glb/
  • Culnan, M. J., & Armstrong, P. K. (1999). Information privacy concerns, procedural fairness, and impersonal trust: An empirical investigation. Organization Science, 10(1), 104115.
  • DePaulo, B. M., Epstein, J. A., & Wyer, M. M. (1993). Sex differences in lying: How women and men deal with the dilemma of deceit. In M.Lewis & C.Saarni (Eds.), Lying and Deception in Everyday Life (pp. 126147). New York: Guilford.
  • Digital Future Report. (2005, December). USC Annenberg School Center for the Digital Future. Five Years of Exploring the Digital Domain. Retrieved December 15, 2005 from http://www.digitalcenter.org/pages/current_report.asp?intGlobalId=19
  • Federal Trade Commission. (1998, June). Privacy Online: A Report to Congress. Retrieved January 23, 2002 from http://www.ftc.gov/reports/privacy3/index.htm
  • Federal Trade Commission. (2000, May). Privacy Online: Fair Information Practices in the Electronic Marketplace. A Report to Congress. Retrieved January 23, 2002 from http://www.ftc.gov/reports/privacy2000/privacy2000.pdf
  • Fox, S. (2000). Trust and Privacy Online: Why Americans Want to Rewrite the Rules. Retreived November 2, 2000 from http://pewinternet.org/reports/toc.asp?Report=19
  • Hann, I.-H., Hui, K.-L., Lee, T. S., & Png, I. P. L. (2002, December 15-18). Online information privacy: Measuring the cost-benefit trade-off. Paper presented at the Twenty-Third International Conference on Information Systems, Barcelona, Spain. Retrieved December 12, 2006 from http://www.comp.nus.edu.sg/~ipng/research/privacy_icis.pdf
  • Jarvenpaa, S. L., & Tractinsky, N. (1999). Consumer trust in an Internet store: A cross-cultural validation. Journal of Computer-Mediated Communication, 5(2). Retrieved December 15, 2006 from http://jcmc.indiana.edu/vol5/issue2/jarvenpaa.html
  • Jourard, S. M., & Lasakow, P. (1958). Some factors in self-disclosure. Journal of Abnormal & Social Psychology, 56, 9198.
  • Kehoe, C., Pitkow, J., Sutton, K., Aggarwal, G., & Rogers, J. D. (1999, May). GVU’s 10th WWW User Survey. Retrieved December 12, 2006 from http://www.gvu.gatech.edu/user_surveys/survey-1998-10/
  • LaRose, R. (2004, May). Promoting I-safety: The effect of consumer information and privacy seals on risk assessment and online privacy behavior. Paper presented at the International Communication Association annual meeting, New Orleans, LA.
  • Metzger, M. J. (2006). Effects of site, vendor, and consumer characteristics on website trust and disclosure. Communication Research, 33(3), 155179.
  • Milberg, S. J., Burke, S. J., Smith, H. J., & Kallman, E. A. (1995). Values, personal information, privacy, and regulatory approaches. Communications of the ACM, 38(12), 6574.
  • Miyazaki, A. D., & Fernandez, A. (2001). Consumer perceptions of privacy and security risks for online shopping. The Journal of Consumer Affairs, 35(1), 2745.
  • Nunnally, J. (1967). Psychometric Theory. New York: McGraw-Hill.
  • Palmer, J. W., Bailey, J. P., & Faraj, S. (2000). The role of intermediaries in the development of trust on the WWW: The use and prominence of third-parties and privacy statements. Journal of Computer-Mediated Communication, 5 (3). Retrieved December 15, 2006 from http://jcmc.indiana.edu/vol5/issue3/palmer.htm
  • Pastore, M. (2000a, April 24). Demographics influence online spending. Retrieved December 12, 2006 from http://www.ecommerce-guide.com/news/trends/article.php/346411
  • Pastore, M. (2000b). U.S. college students use Net for shopping. Retrieved March 17 2003 from http://cyberatlas.internet.com/big_picture/demographics/article/0,1323,5901_432631,00.html
  • Petronio, S. (2002). Boundaries of Privacy: Dialectics of Disclosure. Albany, NY: State University of New York Press.
  • Pitkow, J. & Kehoe, C. (1997, June). GVU’s 7th WWW User Survey. Retrieved December 12, 2006 from http://www.gvu.gatech.edu/user_surveys/survey-1997-04/
  • Sheehan, K. B. (1999). An investigation of gender differences in on-line privacy concerns and resultant behaviors. Journal of Interactive Marketing, 13(4), 2438.
  • Sheehan, K. B., & Hoy, M. G. (1999). Flaming, complaining, abstaining: How online users respond to privacy concerns. Journal of Advertising, 28(3), 3751.
  • Spiekermann, S., Grossklags, J., & Berendt, B. (2001). Stated privacy preferences versus actual behaviour in EC environments: A reality check. Paper presented at the 5th International Conference Wirtschaftsinformatik, Augsburg, Germany. Retrieved March 24, 2004 from http://www.sims.berkeley.edu/~jensg/research/paper/final_privacy.pdf
  • Stanton, J. M., & Stam, K. R. (2003). Information technology, privacy, and power within organizations: A view from boundary theory and social exchange perspectives. Surveillance & Society, 1(2), 152-190. Retrieved December 12, 2006 from http://www.surveillance-and-society.org/articles1(2)/boundarytheory.pdf
  • Swaminathan, V., Lepkowska-White, E., & Rao, B. P. (1999). Browsers or buyers in cyberspace? An investigation of factors influencing electronic exchange. Journal of Computer- Mediated Communication, 5(2). Retrieved December 15, 2006 from http://jcmc.indiana.edu/vol5/issue2/swaminathan.htm
  • UCLA Center for Communication Policy. (2003, February). UCLA Internet Report: Surveying the Digital Future, Year Three. Retrieved August 25, 2002 from http://ccp.ucla.edu/
  • Westin, A. (1967). Privacy and Freedom. New York: Atheneum.
  • West, R., L., & Turner, L. H. (2004). Introducing Communication Theory: Analysis and Application (2nd ed.). Boston: McGraw-Hill.
  • Wharton Forum on Electronic Commerce. (2000, March 1). Is online consumer spending slowing down? Retrieved August 25, 2002 from http://knowledge.wharton.upenn.edu/index.cfm?fa=printArticle&ID=142
  • White, T. B. (2004). Consumer disclosure and disclosure avoidance: A motivational framework. Journal of Consumer Psychology, 14(1&2), 4151.
About the author
  1. Miriam J. Metzger is Associate Professor of communication and Associate Director of the Center for Film, Television, and News Media at the University of California, Santa Barbara. Her research interests include studies of the credibility of information in the new media environment, problems of online privacy and security, the impact of media on public opinion, and the theoretical and regulatory changes brought about by the development of new media technologies.

    Address: 5814 Ellison Hall, Santa Barbara, CA 93106, USA