The exponential growth of the online marketplace and the enhanced capability of online companies to collect, store, transfer, and analyze consumer data have raised concerns about online privacy (FTC 2000). Although search and transaction data are easily collectible, acquiring visitor, shopper, and customer information is deemed by many marketers to be key to creating a competitive advantage in online environments (Andrade, Kaltcheva and Weitz 2002; Denise and Geoffrey 2002; Sijun, Beatty and Foxx 2004). Using information about consumers and their shopping habits and preferences, marketers seek to customize and personalize customer benefits to build stronger relationships (Bush, Venable and Bush 2000; Gauzente and Ranchhod 2001; Stead and Gilbert 2001). To gain this information, marketers need consumers to respond willingly to requests for personal information with honest and accurate answers. In a cluttered marketplace, consumers would be expected to seek efficient, customized solutions to their needs. Yet some evidence indicates that large numbers of consumers deliberately falsify information in Web site exchanges (Fox 2005). Thus, it becomes vitally important to understand how consumers perceive and respond to such requests, particularly how they counterbalance the firms' desire for information with their own desire for privacy in the relationship (Carroll 2002; Charters 2002).

From a firm's perspective, the cost of not addressing this issue is quite high. According to Meister (2006), inaccurate data cost[s] the U.S. economy six hundred billion dollars annually - 5% of the American GDP. "This is in real tangible costs such as unnecessary postage, printing and staff overhead" (p.l). As a result of bad information - deliberate or otherwise - Krill (2000) estimated that 75% of companies that use Consumer Relationship Management (CRM) systems are unable to create an accurate and comprehensive profile of customers, which limits their ability to provide promised personalized services. It is estimated that "more than 25 percent of critical data in Fortune 1000 companies will continue to be flawed, that is, the information will be inaccurate, incomplete or duplicated [and] three-quarters of large enterprises will make little to no progress towards improving data quality until 2010" (Gartner, Inc. 2007).

Krill (2000) estimated the average cost of data cleanup for a firm at about $1 million. Data cleanup involves making the data collected usable in form and nature and includes correcting or deleting data which is in an incorrect format and removing duplicate postings or data which is blatantly false (e.g., names like Mickey Mouse). This figure does not even cover the costs that companies may incur as a result of managing deliberately falsified data, which are harder to identify and correct. From a public policy perspective, the Federal Trade Commission (FTC), among other entities, is concerned with the collection and protection of consumer personal information. Their focus includes the implementation of privacy protection practices required of financial institutions under the GrammLeach-Bliley Financial Modernization Act of 1999, as well as pursuing actions against companies that do not adequately protect customer data (FTC 2007a; 2008).

Understanding how customers choose to negotiate the online environment with marketers (i.e., how they decide when and what information to share) is crucial to improving the delivered products, services, and experiences with online marketers, as well as informing public policy. The purpose of this research is to discover rules, if any, that consumers use in online exchanges. To do this, unstructured interviews were used rather than relying on the documentation of practices, as previous research has done.

The rest of the paper is organized along the following lines. We first briefly discuss the existing literature on consumer privacy concerns and consumer responses to those concerns. Then we introduce the overarching model of consumer response and discuss the method and the results. We conclude the paper with a discussion of the findings and their relevance to academic literature, public policy makers and consumers.

DIMENSIONS OF CONSUMER PRIVACY CONCERNS

The public's concern with the way businesses handle the personal information they collect about their customers has increased since 1999 and two-thirds believe they have lost control over how it is collected and used (Taylor 2003). Lawmakers cognizant of consumer concerns have passed laws like the Gramm-Leach-Bliley Financial Modernization Act of 1999, The Identity Theft and Assumption Deterrence Act (1998), and the Fair and Accurate Credit Transactions Act of 2003 (FACTA) (FTC 2007a, Linnhoff and Langenderfer 2004). Legislation such as this has arisen, in part, because of greater consciousness about the value and uses of personal data, concerns about identity theft, and the ease with which identity thieves have used the online environment to their advantage (Milne, Rohm, and Bahl 2004).

Online exchanges, by their very nature, are remote exchanges over which consumers have limited physical control and no physical access to those providing the service at the time. Privacy concerns for consumers have been categorized in two dimensions: environmental control and secondary use of information control (Goodwin 1991; Hoffman, Novak, and Peralta 1999). Environmental control is defined as "the consumer's ability to control the actions of other people during a market transaction," whereas secondary use of information control is "the consumer's ability to control the dissemination of information related to or provided during such transaction or behaviors to those who were not present" (Hoffman, Novak, and Peralta 1999).

Environmental control, for example, includes the use of cookies. These remote programs, often placed on a person's computer hard drive without the user's knowledge or consent, may track his/her online behavior in addition to customizing his/her experience on specific Web sites (Marx 1999). This tracking of behavior without the explicit knowledge of the consumer is considered by many researchers to be a breach of an implied social contract (Miyazaki 2008). Recent revelations that these embedded programs in Web sites make it easier for hackers to capture personal data without the customer ever knowing about it have drawn significant attention (Acohido and Swartz 2008). To the extent that consumers are aware of and have concerns about their ability to control the transmission of their behavior online, they may be less willing to openly and honestly share their personal information (Zwick and Dholakia 2004).

Similarly, consumers may not be aware of actual secondary use of their data and any threat it could pose. However, media coverage of instances of data and identity theft, as well as legal uses, is likely to create a general awareness among consumers that data collected on one site may be used by others. As such, it would be expected to influence behavioral responses to online requests for personal information.

In previous research, consumers reported using a variety of tactics, including so-called "guerilla tactics" such as creating all sorts of online identities (Fox 2005). Such tactics are a response to increased feelings of a loss of control with respect to control over their own data in online exchanges (Hoffman, Novak and Peralta 1999). Equity theory suggests that when people feel a loss of control in a transaction, they use strategies that help them regain that lost balance in a relationship (Adams 1963; Douglas, Cronan and Behel 2007). Although previous research has documented some of the practices, the focus here is on revealing the strategies used by consumers to establish and maintain that balance. Of interest is whether consumers may have simple response rules or more elaborate rules of engagement, and how these may vary across contexts and what motives determine the responses.

DETERMINING RESPONSES TO ONLINE REQUESTS FOR PERSONAL INFORMATION

Online requests for personal information are expected to vary based upon the context of the request, the relationship with the firm, the specific information sought, and individual traits. The context, such as whether the consumer is engaged in a transaction or in a search, may determine the willingness or extent of consumer sharing. Prior experience with and level of trust in either the particular business or type of transaction should affect the interpretation of the request. If a consumer is transacting with a well-known company or believes that the business will truly deliver a better service based on personal information (e.g., Amazon.com), one would expect consumers to provide more truthful information than with an unknown business.

Individual traits, such as a higher need for privacy, may systematically affect the likelihood to provide complete and accurate information (Sheehan and Hoy 1999). Online behavioral responses have also been attributed to personal state traits, such as emotions experienced while engaged in online exchanges with a firm (Eroglu, Machleit and Davis 2003; Menon and Kahn 2002). Particular responses, such as information fabrication, have also been tied to perceived anonymity and moral obligation (Lwin and Williams 2003). The goal of this study is to understand how consumers interpret online informational requests and whether their responses are simple or elaborate. Zwick and Dholakia (2004) suggest that consumers devise specific external informational strategies to maintain control over their digital representation. A consumer's response to online requests for personal information may be consistent across instances or may vary. The range of responses could include providing (1) complete and truthful information, (2) inaccurate and/or incomplete information, or (3) no information (Sheehan and Hoy 1999).

To understand how responses may vary and why, a more comprehensive model is used to explain these responses and their likely antecedents and moderators. We propose the stimulus-organism- response (S-O-R) model as the guiding framework for understanding the behavior of the consumers' response to online informational requests.

Framework: S-O-R Model

Environmental psychology researchers suggest that the S-O-R model provides a strong framework for understanding behavioral response to a physical environment (see Turley and Milliman 2000 for a review). Marketing-related studies applying the S-O-R framework in an online environment have used consumers' responses to variations in Web page stimuli as a proxy for approach and avoidance behaviors (Eroglu, Machleit and Davis 2003; Huang 2000; Menon and Kahn 2002). Within our study's context, a person providing complete and accurate information in response to an online request is an approach (positive) behavior. Alternatively, a person exiting a Web site without providing the requested information or providing false information is an avoidance (negative) response. Thus, the S-O-R model may serve as a guiding sequential framework of the factors that may influence a person's range of approach-avoidance behaviors based on their assessment of the stimulus and other context factors. This research attempts to understand the contexts and nuances that influence the way that consumers perceive online information requests that lead to different consumer behavioral responses.

METHOD

While prior research has used surveys to determine the degree to which people engaged in various types of known behaviors in Internet exchanges (Lwin and Williams 2003; Sheehan and Hoy 1999), this research relies on consumers' own descriptions of how they interpret and respond to online information requests. Using interviews allowed us to explore the precipitating cues and motivations for various behaviors rather than the extent of response (Lincoln and Guba 1985; Thompson, Locander, and Pollio 1989).

Consumers were recruited through personal and professional contacts to identify and solicit a variety of ages, genders, and Web experiences. Through purposive sampling in multiple cities, each was selected to represent a range of these characteristics. Initially, much younger and much older Internet users were interviewed because of the expected generational differences in experience with the advancement of technology and the degree of marketplace change. Specifically, older consumers have experienced firsthand the impact of technology on the marketplace and probably have evolved strategies from where and how they previously conducted an exchange or transaction to an online forum. Thus, we expected to elicit more varied responses by sampling older and younger Internet users. Similarly, differences in Web experience (i.e., the number of different activities for which they used the Web) were also sought. The described activities were categorized as information search, shopping, e-mail, online banking or financial transactions, social networking, blogging, gaming, or work-related use.

A sample of 21 Internet users allowed us to explore their experiences in-depth and develop a broad picture of how consumers interpret and respond to requests for information exchange (McCracken 1988). Repetitive themes emerged across multiple informants with later respondents providing little new information to expand on these themes. By standards suggested by McCracken (1988) and Lincoln and Guba (1985), the sample is considered adequate for the stated purpose.

As can be seen in Table 1, the final sample consisted of 10 men and 11 women, ranging from 20 to 74 years of age, with an average age of 43.4. Professions ranged from full-time students to retired persons and education averaged 15.5 years. The estimated number of years of Internet experience ranged from 3 to 15, with all respondents reporting using it for e-mail or information searches. Ten percent reported conducting no transactions online (i.e., shopping or financial management). Just over half reported doing banking, investing, or other financial management online and 80% reported shopping or making purchases online.

Interviews averaged around 45 minutes in length, ranging from 20 minutes to more than an hour. Interviews began with asking informants about the various ways in which they use the Internet and then continued in a free-form manner. Informants were encouraged to recall and elaborate on actual experiences online where they were asked to provide personal information. Prompts and neutral probes were used to clarify and expand on what they thought and felt about requests across various types of information exchanges, as well as toward the Web sites and Web site providers. A general outline guided the interviews; however, actual discussions evolved based upon the informant's description and recall of personal experiences. A priori expectations were that consumers may not always be comfortable sharing personal information through Web sites and that, as reported in prior research and anecdotally, may adopt certain strategies to limit their exposure to potential known and unknown risks. It was expected that such strategies might vary from offline strategies, given that information shared online is always given to a virtual rather than an actual entity. Our focus was not just on strategies that might be used but on what precipitated those strategies.

The interviews were tape-recorded and transcribed, producing over 163 pages of single-spaced informant data. Using ATLAS .ti software, the transcripts were independently coded by two of the researchers using the S-O-R framework as a guide for initial stages of analysis and code categorization. Additional codes were identified and used as both researchers worked inductively and deductively through the data (Spiggle 1994) with stimuli cues, organism (consumer) reactions and response behaviors coded into the S-O-R framework, respectively.

An example of the deductive coding process was the initial classification of behavioral responses that depicted informants as providing "truthful information," "false information," or "no information." Through inductive interpretation, informant interviews revealed an overarching construct referred to as "relational norms." Relational norms characterize the interplay between the informant and Web site in terms of information requests. This interplay is positioned within the organism section because these emergent themes describe how the informants perceive the requests from a psychological perspective. Stated differently, the three codes "criticality of exchange," "felt invasion," and "fair play" - describe how the informant felt and/or what he/she thought when presented with the information request. These three themes represent different properties that vary in dimensional ranges. The identification of properties and dimensions permits the researcher to explore and define relationships across categories and constructs (Spiggle 1994).

Once coding was complete, each researcher alone used the codes to determine the compatibility of the responses with the S-O-R model. The researchers analyzed the framework compatibility individually, and then collectively, providing dependability to the framework from a humanistic perspective (Hirschman 1986). Once the overall findings and constructs were finalized, the results, along with the code sheets, were provided to the third member of the research team for triangulation and independent verification of the findings.

RESULTS

The results will be discussed in terms of the S-O-R structure, but starting with the behavioral responses (R), (the focal variable), followed by the stimuli (S), and finally the organism (O) factors, as outlined in Figure 1, with the emergent themes discussed in the organism section. Thus, the goal is first to explain how consumers respond, and then explain what makes the consumer respond in that manner and finally touch on why the consumer seems to respond in a certain manner.

Responses

Respondents were asked to think of situations where they were actively searching for information or engaging in some transaction and then recall their responses when those Web sites requested some personal information. Behavioral responses described by consumers reflected experiences from a range of contexts and activities, such as online searches, shopping, banking, and online chats.

Responses were broadly categorized into three forms: disclosure, falsification, and exit. Consumers may respond truthfully, disclosing the information requested (Zwick and Dholakia 2004) in partial or complete fulfillment of the request. Alternatively, they may falsify some or all information, including creating false identities aimed at establishing a sense of distance between them and the company (Fox 2005; Lwin and Williams 2003). Finally, they may choose to simply end or exit the interaction, rather than engaging in other behaviors. Additional discussion of each type of behavior follows.

Disclosure

Disclosure often occurred when consumers recognized that Web sites sometimes ask for information so as to better serve them. As James (68, M) said, "They [marketers] have a job to do and they're trying to be conscientious about it and if they want a straight answer to a straight question, we'll give it to them." For online purchases, consumers understand that providing information is a necessary part of the exchange, so under those circumstances, they are more likely to provide truthful information. Lance (73, M): "When you do business with somebody ... it may require that [providing personal information]. In order to do business with them, you have to do that." For example, when a Web site requires consumers to submit specific information in order to be able to provide them a quote for a mortgage or an insurance policy, informants were more willing to provide accurate information. James (68, M): "I may ask them what's behind the questions, but that's about it." If the consumer expected that the information would be used to educate the firm and enhance future communication between the consumer and the firm, he/she was more likely to provide truthful information. Deb (39, F) exemplifies this sentiment: "I typically am honest with them. . . . If I take the time to fill up the response ... I will honestly answer the questions. Because clearly people are going to use that to market, so if I am going to spend the time, I am not going to lie. I am not going to up my income or lower my income, because I don't want to be marketed [inappropriately], . . .