Women’s Issues in Transportation: Summary of the 4th International Conference, Volume 2: Technical Papers (2011)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

117 Privacy and Gender Reviewing Women’s Attitudes Toward Privacy in the Context of Intelligent Transportation Systems and Location-Based Services Caitlin D. Cottrill and Piyushimita (vonu) Thakuriah, University of Illinois at Chicago Limited previous research has shown that women value online privacy more than men, potentially influencing their online behavior or willingness to reveal personal data online. New generations of intelligent transportation systems (ITS) and location-based services (LBS) technol- ogies depend on the input of personalized and localized information to give, potentially, information that may uniquely address women’s complex travel patterns, but which may raise locational privacy concerns for women and cause them to hesitate to share the needed infor- mation. This paper examines gender differences in the propensity to reveal the potentially sensitive information necessary to make ITS and LBS highly personalized to individual travelers. The authors develop privacy indica- tors based on refusals to answer sociodemographic and location questions in a household travel survey to evalu- ate whether women have a significantly different atti- tude toward willingness to share data related to position and personal identifiers compared with men. The results show that gender differences regarding privacy prefer- ences are not statistically significant. However, this result is inconclusive because the survey overall achieved low response rates and participating households may already be self-selected into being open about divulging sensitive travel and locational information. Personal data are increasingly accessed and used in today’s technologies. websites deposit cookies on user’s computers, and the resulting data are used to enhance the user’s experience. Credit card compa- nies collect data on usage patterns to determine risk and maximize profits. electronic toll collection systems use personal data to enhance the ease of travel by allowing vehicles to move seamlessly through toll collection points, paying the fare from an electronically debited account. while arguments can be made for the benefits that accrue from each of these data collection systems (personaliza- tion, convenience, and efficiency), the amount of data being shared may have overwhelming ramifications for the privacy of the population. while awareness of some data collection technologies is reasonably high, the level of understanding of the potential implications of use and sharing of data may be substantially lower, leading some to share data that they might prefer to keep private if all implications were known, and some to refrain from sharing any data. with an increasing turn toward intelligent transporta- tion systems (ITS) and location-based services (LBS) and their reliance on user data for effectiveness, questions of privacy grow increasingly relevant. If users are reluctant to lessen their privacy by allowing collection of their per- sonal data for such uses as electronic toll collection, route navigation systems, or vehicle-to-vehicle safety applica- tions, it is possible that they will elect not to participate in the system, which may hamper the overall effectiveness of “crowdsourcing” as a method of collecting real-time traffic information. Over time, such technologies will have the potential to provide information that addresses women’s complex and unique travel patterns (such as real-time itinerary planning or dynamic ridesharing with members of a woman’s social network); however, such technologies will depend on the input of information

118 wOMeN’S ISSUeS IN TRANSPORTATION, vOLUMe 2 that is even more personalized and localized, potentially making individuals hesitant to share such information and raising greater locational privacy concerns. One subset of the general question of the relationship between privacy and willingness to share data in ITS and LBS technologies is that of gender influences. Prior research in online behavior has indicated that women tend to have stronger privacy preferences than men, and this may influence their online behavior or willingness to reveal personal data online (Sheehan 1999; Garbarinoa and Strahilevitz 2004). The similarities between methods of data collection and types of data collected online and in the context of ITS and LBS are strong enough that it is possible that such preferences will also translate to the realms of transportation; however, differences in the complexities of women’s travel behaviors compared to those of men will necessitate ensuring that appropriate (and detailed, personalized, and localized) data are col- lected in order to accurately reflect the travel needs and desires of the female traveler. Little research is available that examines the gendered relationship between loca- tional privacy preferences, willingness to share data, and potential impacts on adoption and use of ITS and LBS. Such a study is difficult, in part because the persons whose privacy preferences are highest are unlikely to partici- pate in surveys that would provide an indication of their trade-off preferences between data provision and benefit accrual. As a result, this study is limited to analysis of the responses of those persons who have already con- sented to survey participation—in this case, the House- hold Travel Tracker Survey of the Chicago Metropolitan Agency for Planning (CMAP). Unwillingness to partici- pate in surveys is an early indication of weaker privacy concerns (Singer et al. 1993; Couper et al. 1998; DeMaio 1980). Proxy measures of privacy based on refusals to answer potentially invasive sociodemographic and loca- tional information were developed that were asked in the CMAP survey, partly motivated by prior research that suggests that item nonresponse in surveys may also be linked to privacy concerns (Mayer 2002; Zmud and Arce 2000; Riphahan and Serfling 2005; Dixon 2005). This paper explores whether there may be cause for con- cern related to the potential for not gathering accurate or useful data on women’s travel behavior due to gender- influenced privacy preferences and then evaluates the context of privacy–utility trade-offs within ITS and LBS to determine why differences may not be found. BackgRound The issue of privacy has gained increasing interest as more and more of our lives are recorded, placed online, or take place in the public or semi-public sphere. The ubiquity of such technologies as surveillance cameras, online cookies, and data (“reality”) mining software has significantly impacted the expectations there may be for privacy, and current research indicates that the issue may be of particular concern for women. women consistently indicate a higher level of concern for privacy than men (Kate 1998; O’Neil 2001), which may be a result of such concerns as safety and security, discomfort with loss of control over personal information, and uncertainties regarding use of data, and may impact willingness to share personal data. Privacy is of particular interest for ITS and LBS, for which user-generated or crowdsourced information will play an increasingly important role in generating col- laborative travel opportunities, traffic management, and multimodal transportation use. Many emerging ITS and LBS technologies depend upon the ability to collect and share large amounts of data from locational sensors in handheld devices such as cell phone, personal digital assistants, and vehicles with a variety of infrastructure- based sensors. while many technologies will be embedded in vehicles or the roadway, thus requiring no conscious participation choices from the traveler, other forms of ITS and LBS will require the traveler to consciously opt in to services, which will require ensuring that potential users are aware of and comfortable with the associated consequences of the sharing and use of this data. The novelty of many proposed ITS and LBS technologies has made willingness to adopt a concern for developers, particularly for those applications that will require fairly extensive deployment for maximum effectiveness (e.g., the greater the number of cell phones or mobile devices generating information on congestion or safety concerns, the better the quality of information). Given the amount of data needed for many applications, one critical aspect of this willingness to adopt will likely be the willingness to share, or privacy preference, of travelers, a factor that may have gendered attributes. If it is true that women will more carefully guard their private information, the combined concerns of willing- ness to share information and potential willingness to adopt and use ITS technologies may create an imbalance between the amount of information available on travel patterns and habits of men and women. This may, in turn, affect planning and implementation of applications that have the potential to address women’s travel needs, such as personalized route planning and trip itinerary development for trip chains; dynamic ridesharing with friends; real-time walking-group formation from transit stations and stops to overcome personal security con- cerns; integrated weather, construction, and travel infor- mation; and other applications that would allow specific travel preferences of women to be addressed by informa- tion technology. The paper first reviews the general findings in the literature related to the influence of gender on privacy

119PRIvACY AND GeNDeR and technology use. Next, an analysis of the willingness of persons to share identification and position informa- tion in the CMAP household travel survey is undertaken in order to determine if there are gender influences in willingness to share the types of information that may affect the level and usefulness of ITS and LBS. while the empirical evidence explored in this paper indicates that little, if any, difference exists, questions are raised that should be explored future, including those related to basic privacy–utility trade-offs. Finally, gendered recom- mendations pertaining to potential privacy protections for ITS and LBS are discussed. ReseaRch oBjectives The aim of this research is to study gender in the context of willingness to share information supporting personal- ization and localization of information available from ITS and LBS. Pertinent research related to privacy preferences and willingness to adopt is first reviewed. The authors then empirically evaluate the revealed privacy preferences of women in relation to a CMAP travel survey by using ques- tion response refusals (those cases in which respondents refuse to answer questions in contrast to choosing “don’t know” as a response) as a proxy indicator. The validity of using question refusals in this study of privacy stems from previous survey research that has linked refusals to ques- tion sensitivity, in contrast to “don’t know,” which has been found to more closely reflect cognitive difficulties in answering the questions. For example, Shoemaker et al. (2002) note that questions regarding income generally produce refusal rates of 20% or higher, presumably due to question sensitivity, and observe that refusals in this context may be used to avoid embarrassment, reduce the risk of giving responses that violate social norms, or to protect privacy. The authors argue that a “don’t know” response may be linked more to a lack of motivation to interpret the question and determine the answer than to a lack of opinion or knowledge. Under this interpretation, the use of question refusals may be regarded as a valid proxy for general privacy preferences. Question refusals are separated into demographic questions (associated with personal identification) and questions related to travel behavior (associated with revealing position or location information). Statistical models of propensity to refuse to answer will allow the determination of the role, if any, that privacy preferences may potentially play in women’s willingness to share the types of personalized and localized information that may be necessary to develop travel assistance from informa- tion systems. The study is intended to identify general trends in women’s willingness to share data in the trans- portation context in order to determine if this is an area that requires additional study. gendeR influences on willingness to adopt and use Fairly limited research has been conducted on women’s willingness to adopt new transportation technologies in relationship to their privacy concerns. Friedman et al. (2006) reported that in a systematic search of journal articles and conference papers released from 2000 to 2004 related to human–computer interaction, 32 papers recorded empirical findings on privacy, but none included an analysis of gender impacts and effects. This gap in the literature, combined with the complexity and number of trips reported by women (Hu and Reuscher 2004), indi- cates that this is an area that requires additional study. For example, McGuckin and Murakami (1999) found that women tend to take more responsibility for trips related to household sustainability (such as shopping, errands, and children-oriented trips) and are more likely than men to chain these trips with work trips. The poten- tial for personalized ITS and LBS information would be particularly beneficial for women in this context; how- ever, it will first be necessary for women to adopt these technologies and be willing to share the data that will make them effective. Consumer adoption of technology has been noted to occur in five stages—innovators, early adopters, early majority, late majority and laggards—which are generally differentiated on the basis of innovativeness, or “the degree to which an individual or other unit of adoption is relatively earlier in adopting new ideas than other members of a social system” (Rogers 1995). Influ- encing factors on this innovativeness may also include demographic characteristics (including age and educa- tion) and prior experience with technology (Munnukka 2007). while males have traditionally dominated the categories of innovators and early adopters (Caruso and Salaway 2007), there is evidence that women are becom- ing more willing to participate earlier in technological innovations. For example, in a study of online social net- working sites, a majority of users of such sites as Face- book (63%), MySpace (63%), and Friendster (58%) were found to be female (Rapleaf 2008). According to the Business Week analysis of the study, males tend to gravitate toward more transactional-based sites such as those targeted toward news, sports, and financial infor- mation, while women’s online behavior is more geared toward relationship-driven sites (Hoffman 2008). The nature of shared information on these sites is interest- ing given findings in an earlier study of Internet users that indicated that males believe “censorship” to be the greatest threat to the Internet, while females cited their greatest concern as “privacy” (Herring 2001). Internet privacy has been noted to be more of concern to women than men by westin (1998), Ackerman et al. (1999), Kate (1998), and O’Neil (2001). These characteristics of

120 wOMeN’S ISSUeS IN TRANSPORTATION, vOLUMe 2 behavior online may have implications for understand- ing how women, compared to men, will respond to new technologies in the mobile environment. Privacy concerns become even more critical in the public and mobile environment—called locational privacy—in which considerations of direct personal safety and security may arise. For example, Marchau et al. (2008), evaluating the potential for users to adopt a personal intelligent travel assistant (PITA), state that “Another certain vulnerability would be that travelers resist the willingness to buy PITA because it affects their privacy—i.e., it seems like ‘Big Brother’ watching their travel behavior.” In “The watcher and the watched,” Friedman et al. (2006) examined whether gender influ- ences people’s judgments about privacy in public. After an extensive literature review, the authors found that men and women view privacy similarly on some dimen- sions, but that on others women tend to be more con- cerned with privacy. In order to test these findings, the authors conducted an empirical study that asked males and females to respond to a series of questions regard- ing feelings of intrusiveness from being videotaped in a public place with the images shown in a variety of set- tings and to a variety of people, and whether their feel- ings changed based on whether they were the watcher or the watched. The authors found that women were more likely to express concerns about the display of images sent from a camera in a public place, and that these con- cerns did not change on the basis of the condition of watcher or watched. Friedman et al. (2006) concluded that “compared to men, women feel more vulnerable, especially in terms of physical safety and psychologi- cal well-being.” while the study was limited to physical surveillance via cameras, some of these findings may be translated to general reactions in locational privacy. pRivacy conceRns in the moBile enviRonment Because it is concerned primarily with privacy in the pub- lic sphere, and due to the variety of commercial and gov- ernmental interests in tracking travel patterns and habits, issues of control and context are particularly relevant to the understanding of privacy in the mobile environment. From a general standpoint, traditional concepts of pri- vacy occur at a static point of reference, while locational privacy exists in the context of a rapidly changing route in the transportation network. elements of change over time and space heighten the degree of awareness and information needed by the individual to make informed decisions regarding the collection and use of her data and increase the potential that aspects of control may be lost. For example, when an individual uses the Internet she is bound by the privacy policy of the associated Internet service provider (ISP). If the individual wishes to switch ISPs, she is able to review and accept or decline that ISP’s policy. In a dynamic environment, the rapidity of change related to context and authority may make such consid- erations impossible in terms of efficiency. For example, if when using a cellular phone one enters a roaming envi- ronment, this indicates that the cellular provider does not have coverage at that location but has entered into a contract with another provider in order to ensure ser- vice coverage. In this case, the consumer is subject to the policies and terms of the secondary cellular provider, and may not be aware of potential changes in the guiding privacy policy. This situation demonstrates the impor- tance of where and when decision making takes place. Additionally, collected data that include both spatial and temporal identifiers may increase the likelihood that per- sonally identifying information may be gleaned, as it is possible to determine not only where a traveler has been, but also at what times, thus introducing the potential to identify activities that may have taken place. The ability of data from the mobile environment to be used as both spatial and temporal identifiers may also heighten the privacy concerns specific to women. As noted by Friedman et al. (2006), a sense of vulner- ability may be more evident in women than men, and the potential for ITS and LBS to pinpoint both location and identification information for individual travelers may be regarded as a benefit or a detriment. The benefi- cial aspects of Global Positioning System (GPS) tracking have been revealed in several recent cases in which GPS- enabled cell phones were used to track and recover miss- ing or abducted persons (Barnes and williamson 2009; Bolduan 2008). Additionally, such services as GM’s OnStar, a subscription-based service offering emergency communications, in-vehicle security, and turn-by-turn navigation, have been increasingly targeted toward women. “‘we’ve tailored the media so it impacts more of the women’s market,’ (Young 2003). ‘we’re not trying to walk away from the men. Men look at it as a way to protect their wife and family. women look at it as some- thing that could protect them’” (Fahey 2003). while the feeling of security engendered by these services is criti- cal for their increased deployment, issues of privacy and insecurity are also increasing in visibility. In addition to the traditional concerns relating to sec- ondary uses of data, issues of unauthorized marketing, surreptitious monitoring, and malicious use by unauthor- ized agents have also been raised. One particular concern for women is that, without well-integrated security mea- sures, it may be possible for GPS stalking to occur, as happened in December 2002 when a wisconsin man was arrested and charged with stalking a former girlfriend after placing a locating device under the hood of her car (Mayer 2003). while this case involved an outside actor placing a device on a person’s vehicle, the potential

121PRIvACY AND GeNDeR for vehicle- or device-based ITS and LBS to be used for malicious purposes, particularly if personally identifying information is not adequately protected, is great. The level of trust that an individual has about an information-sharing situation is also an important con- sideration. A trusted entity is one with which an indi- vidual may feel a greater degree of comfort in revealing otherwise private information. There are two aspects to the concept of trust in the current context. First, trust can be built or enhanced with different strategies; for example, Title 13, U.S. Code Protection of Confidential Information of the U.S. Bureau of Census, guarantees the confidentiality of census information and establishes pen- alties for disclosing this information. In the case of Inter- net transactions, secure sockets layer (SSL) technology enables encryption of sensitive information during online transactions and communications to the customer and guarantees the vendor’s need to comply with her privacy and security requirements. Second, as noted by Sheehan and Hoy (2000), a person’s familiarity with the informa- tion-seeking entity (Gronroos 1994; Pitkow and Kehoe 1996; Rogers 1996; wang and Petrison 1993) can affect the degree of privacy concern. It may be the case that public agencies carry a degree of familiarity and aware- ness of lack of trust violations in the minds of individuals when they are faced with situations in which personal or locational information are to be revealed. Therefore, the organization that manages and operates the ITS or LBS technology may have a bearing on the level of trust that an individual has about the technology. oveRall ReseaRch design Constructing an empirical model for determining pri- vacy preferences of women is difficult due to data and response limitations; however, defining a basis from which to conduct future research will depend on estab- lishing a baseline for evaluating women’s privacy prefer- ences by using existing data. This study uses an existing household travel survey (the 2007 CMAP Travel Tracker Survey) to study the relative willingness of women to respond to demographic and travel-related questions when compared to men. Item nonresponse is used as a proxy for privacy preferences in the absence of specific questions related to privacy. Question nonresponse is modeled against the gender of the respondent and other demographic variables to determine if significant rela- tionships exist between gender and willingness to pro- vide data. Results of the model are used to draw general conclusions regarding the relative privacy preferences of women, and suggestions are made for future studies on how these preferences may impact privacy–utility trade- offs and the related willingness to adopt ITS and LBS technologies. empiRical model using cmap tRavel tRackeR data According to information from the analysis presented above, two primary considerations in determining the willingness of persons to participate in ITS and LBS applications relate to information concerning position and identifiers. Identifying information can include names, vehicle license plates, demographic information, and other data that allow one to be personally identified. Position information is the spatiotemporal information that allows a person’s whereabouts to be known at a given location at a given time. The following analysis of CMAP Travel Tracker Sur- vey data is intended to provide an exploratory analysis of the willingness of women to share such information. while perhaps the most obvious method of determin- ing willingness to share would be to conduct an analy- sis of the sociodemographics of households that refused to participate in the survey, available data do not lend themselves to such interpretation. Therefore, item non- response is used as a proxy for determining the general trend of women’s privacy preferences regarding the shar- ing of data. A number of researchers (Zmud and Arce 2000; Riphahan and Serfling 2005; Dixon 2005; Shoe- maker et al. 2002; Biemer and Lyberg 2003) have indi- cated that item nonresponse is related, in part, to privacy concerns. Two types of factors may be analyzed in order to study privacy preferences: those relating to personal identification and sociodemographics and those relat- ing to the individual’s use of transportation and location information. while many empirical situations give the opportunity to examine refusals and sensitivity or pri- vacy regarding personal identification and sociodemo- graphics, a household travel survey provides a unique opportunity to analyze the types of privacy concerns that may arise in the context of ITS or LBS. A total of 14,315 households consisting of 23,808 persons in the greater Chicago metropolitan area (con- sisting of Cook, DuPage, Grundy, Kane, Kendall, Lake, McHenry, and will counties in Illinois, and Lake, Por- ter, and LaPorte counties in Indiana) took part in the Travel Tracker Survey (NuStats 2008). A subsample of the respondents were queried as to whether they would be willing to participate in further survey activities, including using vehicle-based or portable GPS devices. The purpose of the dual computer-assisted telephone interviews (CATIs) on trip making and GPS recording of trips is to reduce trips that are unreported, which is a problem with recall in survey-based efforts. In an earlier paper, the authors modeled the propensities of house- holds to refuse to participate in the GPS portion of the survey (Cottrill and Thakuriah 2008). Refusal to use the GPS devices potentially represents individuals and

122 wOMeN’S ISSUeS IN TRANSPORTATION, vOLUMe 2 households who perceive ubiquitous locational tracking to be problematic, and thus is partly indicative of loca- tional privacy preferences and concerns about security, although the refusal to use the GPS may also be indica- tive of difficulties with technology in general, concerns about potential damage to the vehicle, distractions, and so on. The results indicated that while households that made a greater number of trips and with more working members were more likely to agree to participate in the GPS survey, english-speaking, higher-income households with children less than 16 years of age tended to have a weaker propensity to take part in the GPS survey. Of the participants, 12,625 (or roughly 53%) were female. According to the NuStats (2008) documen- tation from the survey, the following list comprises question nonresponses and the magnitude of those non- responses: • Recruitment instrument: – Number of persons depending on you for trans- portation (0.7%), – Number of bicycles (0.6%), – Residence type (0.3%), – Home ownership status (1.3%), – Tenure at current residence (0.1%), – Number of cell phone lines (1.3%), – Number of landlines (0.7%), – Household income (10.4%), – vehicle year (2.1%), – vehicle make (0.5%), – vehicle body type (1.1%), – where vehicle is parked at home (3.7%), – Gender (<0.01%), – Age (1.8%), – Hispanic origin (0.7%), – Race (1.6%), – Disability status (0.8%), – Licensed driver status (0.2%), – employment status (0.1%), – volunteer status (0.1%), – work status if not employed (0.7%), – Typical mode to work (0.3%), – Type of work schedule (1.2%), and – educational attainment (0.7%). • Retrieval instrument: – Parking details (if auto trip) (10.7%), – whether the trip was made on expressway or tollway (if auto trip) (18.8%), and – Car available (if transit trip) (<0.01%). These item nonresponses were coded to allow for analy- sis of patterns related to women’s willingness to reveal identification and position identifiers. The following identification and sociodemographic variables and position variables relating to individuals in the CMAP Travel Tracker CATI Survey were used to construct the six composite variables shown in Table 1: • Identification and sociodemographic (ID) variables: – Residence type, – Home ownership status, – Tenure at current residence, – Household income, – Gender, – Age, – Hispanic origin, – Race, – Disability status, – Licensed driver status, – employment status, – volunteer status, – work status if not employed, – Typical mode to work, and – educational attainment. • Position and travel behavior (TB) variables: – Parking details (if auto trip), – whether the trip was made on expressway or tollway (if auto trip), and – where vehicle is parked at home. The composite variables are either indicator or count variables. The objective of these composite variables is to identify respondents who refused to answer at least one question in the CATI survey or otherwise generated an item nonresponse by not answering a question. Table 1 also provides gender-based statistics on the six composite variables. The first and second columns of the data set give the composite variable names and descriptions. The third column gives the percentage of all respondents with the attribute for the three compos- ite dummy variables (D_IDRefs, D_TBRefs, and D_All- Refs) and the mean value of the three composite count variables (C_IDRefs, C_TBRefs, and C_AllRefs) for respondents with a positive number of refusals. Overall, only a small percentage of respondents had refusals of any kind. For example, only 3.83% of all respondents had at least one identification or sociode- mographic information variable refusal, 13.9% had at least one position and travel behavior variable refusal, and 16.31% of all respondents had at least one refusal of either type. The percentage of females with at least one refusal in each category is only slightly higher than the percentage of males, and a t-test of significance indicated that the male–female differences in refusal percentages for each composite variable was not significant at any reasonable level of significance. Further, the statistics on the count variables also indicate virtually no difference in the mean number of refusals between males and females for all three types of composite indicators. The objec- tive then becomes to examine if these lack of differences

123PRIvACY AND GeNDeR between males and females persist after controlling for factors other than gender that might potentially affect the composite variables giving refusal propensity. model Results In order to analyze differences in propensity to give infor- mation, three logit regressions (for D_IDRefs, D_TBRefs, and D_AllRefs) are estimated. The control variables and results obtained from each of the three models are shown in Table 2. The public use data set from the CMAP Travel Tracker Survey was joined to data sets containing a number of environmental factors based on the Census tract of the home location of the persons surveyed. envi- ronmental factors developed for the Chicago area spatial decision support system (SDSS) (Thakuriah et al. 2003) were used to hold elements that might impact feelings of security versus privacy (such as crime levels and traf- fic crashes) constant and to reflect the urban ecological context of the surveyed population (such as transit avail- ability). A number of models, including ordinary least squares regression, zero-inflated Poisson, and negative binomial, were tested by using various combinations of gender, age, and presence or absence of children at the person and household levels. None of the variables asso- ciated with gender were found to be significant, includ- ing a dummy variable indicating gender, gender by age, or gender by presence of children in the household. The general logistic model results shown in Table 2 indicate the influence of those factors that may affect identifica- tion refusals, position refusals, or total refusals. while the degree of response refusal or uncertainty was mod- eled, it was not significantly different from simply mod- eling the presence or absence of a refusal to one or more relevant questions. As shown in Table 2, the general gen- der attribute of FemDum had no significant influence on the propensity of a person to refuse to answer questions related to specific factors. Those factors that are shown to have a significant degree of influence, such as living in a Census tract with a high minority or high low-income population (eJ_tract), age (Age), and income (Incom), among others, tend to be relevant to person-based, as opposed to group-based, characteristics. various factors may be considered in understanding the result of lack of significant differences among men and women. The first is that the agency for which the survey was conducted, CMAP, is a generally trusted agency, and well-known in the Chicago region. The issue of trust, as noted above, is one of the most influential in the deter- mination of whether a person will respond to a survey or respond to all questions asked. The lack of major trust vio- lations of information collected by government agencies in recent history may have contributed partially to the low rates of participant refusals. The fact that this survey was conducted by a trusted agency may have influenced the willingness of the respondents to answer all questions. A second issue to address is that of nonresponse. For the pilot test, an overall response rate of 9% was achieved (NuStats 2006), while a 10% response rate for the full study was obtained (NuStats 2008). The authors did not have access to information on nonresponders, making it difficult to evaluate fully the influence of gender on this set. while various factors influence nonresponse, including time and availability, it is likely that privacy concerns will impact the willingness of some persons to participate. Con- sidering westin’s privacy index, it is possible that many of those persons who fall within the privacy fundamental- ist or privacy pragmatist groups chose not to participate in the survey at all, and consequently, their information TABLE 1 Composite Variables and Explanation and Summary Statistics Composite Respondents (%) variable Description All Male Female D_IDRefs = 1 D_IDRefsa Dummy (D_IDRefs = 1) indicating at least 1 ID refusal 3.83 3.65 3.98 D_TBRefs = 1 D_TBRefsb Dummy (D_TBRefs = 1) indicating at least 1 TB refusal 13.90 13.92 13.96 D_AllRefs = 1 D_AllRefsc Dummy (D_AllRefs = 1) indicating at least 1 ID or TB refusal 16.31 16.00 16.58 Mean of C_IDRefs for (C_IDRefs > 0) C_IDRefs Count of ID refusals 1.20 1.20 1.22 Mean of C_TBRefs for (C_TBRefs > 0) C_TBRefs Count of TB refusals 0.27 0.28 0.27 Mean of C_AllRefs for (C_AllRefs > 0) C_AllRefs Count of all (ID plus TB) refusals 1.47 1.47 1.48 a IDRefs refer to those question refusals associated with identification or sociodemographic information. b TBRefs refer to those question refusals associated with position or travel behavior. c AllRefs is the total of IDRefs plus TBRefs.

124 wOMeN’S ISSUeS IN TRANSPORTATION, vOLUMe 2 was not included in the analysis. Additional information on the nonrespondents would allow for a more nuanced determination on the influence of gender. On the basis of the obtained results, particularly those associated with age, it is possible that the persons who elected to participate in the study have a lower degree of privacy sensitivity. If this is the case, it would be useful to conduct a different study that would stratify persons according to westin’s index to determine if the lack of gender influence holds true across all categories. In general, far greater variation is needed in the preferences of subjects studied for these insights to be gleaned from empirical studies. Finally, the finding of no significance is, of itself, a find- ing. Little research has been conducted on gender influ- ences on privacy preferences, beyond basic analyses such as those conducted by westin (1998) and Ackerman (1999) as part of larger studies. while this study has a number of limitations, outlined more fully below, the general findings in relationship to privacy preferences of women related to identification and position variables may provide a basis upon which to base future studies, particularly if more work is conducted to tease out relationships regarding the balance between privacy, security, and utility that the respondent believes will accrue by providing information (including better traffic management and congestion allevi- ation, time savings, and so on). One potential would be to conduct specific research on gendered privacy preferences by looking at both stated and revealed preferences based on participation and response. conclusions The above review identifies a number of issues for concern in the relationship between transportation, data needs, and gender-based privacy preferences. Transportation agencies and actors are increasingly reliant on data to effectively plan and implement transportation improvements. Dif- ferences in travel patterns between genders (Turner and Niemeier 1997; McGuckin and Murakami 1999; Matthies et al. 2002) highlight the need to make certain that rep- resentative data have been collected from both men and women. Two primary concerns related to privacy and data, that of general willingness to share data and the poten- tial for that willingness to affect willingness to adopt ITS and LBS, are emergent in the transportation realm. while the research reported above has indicated that men and women may have divergent privacy preferences, it is pos- sible that the presence of a trusted entity in data collection efforts may have encouraged those respondents who have already agreed to participate to provide the data requested, thus minimizing item nonresponse in general and question refusals in particular. The relatively low response rate of 10% achieved by CMAP, however, may indicate that the TABLE 2 Model Variables and Results estimate Model I: Position Model II: Identification Model III: All Refusals: Refusals: Refusals: Parameter Description p(D_TBRef = 1) (p(D_IDRef = 1) p(D_AllRef = 1) Intercept 28.1922 23.8629 24.1687 FemDum Person is female 20.0791 20.00111 0.0102 Crime_low Crime rate of the home Census tract is in the lower quartile 20.00744 0.0618 0.0531 Crime_med Crime rate of the home Census tract is in the second or third quartile 20.0003 0.1018 0.0865 Crime_high Crime rate of the home Census tract is in the upper quartile 0.0123 0.0963 0.1164 eJ_tract Home Census tract is characterized by a minority or low-income population above the region median 0.2449 0.1978 0.2358 Age Age of respondent 0.0624 0.0143 0.0243 HHSize Household size 0.3563 20.0817 20.0218 educa Highest level of educational attainment 0.0595 20.0386 20.0203 Incom Income variable 0.1598 0.3234 0.3046 CookCo Household resides in Cook County 28.94e-06 0.0617 0.0658 PTrips1 Number of person-trips reported during interview 0.0372 20.0112 20.00618 TAI Transit Availability Index (see SDSS website at http://www.utc.uic.edu) 0.0249 20.1085 20.0372 TeAI Origin employment Accessibility Index (see SDSS website) 0.2019 20.1504 20.1013 LicDum Survey respondent holds a valid driver’s license 21.2046 20.2536 20.4961 Homemkr Survey respondent is a homemaker 20.0953 0.3063 0.2501 Tot_crash Total number of automobile crashes reported in the home Census tract 0.00738 0.00455 0.00464

125PRIvACY AND GeNDeR presence of a trusted entity is a necessary, but not suffi- cient, condition for encouraging participation. However, far greater variation is needed in the preferences of sub- jects studied for these insights to be gleaned from empirical studies, an idea also noted by Hann et al. (2002). without additional information regarding refusal to participate it is difficult to draw larger conclusions on the revealed privacy preferences of women as a whole. On the basis of the results of the empirical study, it may tentatively be concluded that once the initial hurdle of participation is overcome, women do not differ signifi- cantly from men in their willingness to provide responses to questions of demographics and travel behavior—that is, women who choose to participate will quite possibly be no different in terms of privacy preferences compared to men. This finding indicates that the survey data that are currently used are likely adequate in this regard for rep- resenting both men’s and women’s travel behaviors. This finding, combined with other studies revealing attitudes toward privacy expressed by women in general, may indi- cate that while privacy concerns may present a barrier toward women’s adoption of ITS and LBS, for those per- sons who are likely to adopt, data will be fully reported. Such a conclusion, though somewhat speculative, would support the use of data collected from ITS and LBS appli- cations for transportation purposes, as it is likely that gen- dered differences would be accurately reflected. One potential method of overcoming the propensity toward nonresponse would be to more fully determine the privacy–utility trade-offs, such as those seen in the OnStar experience. For users of OnStar, the safety and security benefits attained outweigh the privacy cost of surrender- ing certain private data. Hann et al. (2002) argue that privacy concerns are not absolute, but rather that they are subject to trade for economic benefits. In the context of ITS and LBS adoption, such benefits may be accrued through savings of time or safety, and these might balance out concerns of improper access or other privacy-related issues. while the study found no significant gender-based preferences, it is possible that further evaluation of such trade-offs might identify specific issues or concerns that would better encourage women to participate in ITS and LBS adoption despite privacy concerns. A few research limitations should be noted. First, infor- mation on survey nonrespondents is not available, which limits the degree to which the conclusions may be claimed to represent the population as a whole. Second, the use of proxy variables to reflect privacy preferences is not ideal. Item nonresponse has been linked to privacy prefer- ences vis-à-vis reluctance to share information; however, it should not be concluded that privacy preferences are the sole reason for item nonresponse. Discomfort over income or age, an inability to recall information, or other facts may influence the ability or willingness of a person to respond to a given question. The use of item nonre- sponse as a privacy proxy in this study has been done in order to utilize existing relevant data to determine if a sys- tematic relationship may exist in patterns of nonresponse and gender and environmental factors. RefeRences Ackerman, M. S., L. F. Cranot, and J. Reagle. 1999. Privacy in e-Commerce: examining User Scenarios and Privacy Preferences. Proc., 1st ACM Conference on Electronic Commerce, Denver, Colo., pp. 1–8. Barnes, G., and D. williamson. Mass. Officers Use GPS to Find Kidnapped Girl. Telegram & Gazette, January 8, 2009. Biemer, P. P., and L. Lyberg. 2003. Introduction to Survey Quality. John wiley and Sons. Bolduan, K. Is GPS a High-Tech Crime-Fighting Tool or Big Brother? CNN, August 18, 2008. http://www.cnn .com/2008/CRIMe/08/18/gps.tracking/index.html. Caruso, J. B., and G. Salaway. 2007. The ECAR Study of Under- graduate Students and Technology, 2007. eDUCASe Center for Applied Research. Cottrill, C., and P. Thakuriah. 2008. GPS Use by Households: early Indicators of Privacy Preferences. Proceedings of Intel- ligent Transportation Systems World Congress, New York. Couper, M. P., e. Singer, and R. A. Kulka. 1998. Participation in the 1990 Decennial Census: Politics, Privacy, Pressures. American Politics Research, vol. 26, No. 1, pp. 59–80. DeMaio, T. J. 1980. Refusals: who, where, and why. Public Opinion Quarterly, vol. 44, No. 2, pp. 223–233. Dixon, J. 2005. Comparison of Item and Unit Nonresponse in Household Surveys. BLS Statistical Survey Paper. Fahey, J. OnStar’s Scare Tactics. November 26, 2003. http:// www.forbes.com/2003/11/26/cz_jf_1126feat.html. Friedman, B., P. H. Kahn, Jr., J. Hagman, R. L. Severson, and B. Gill. 2006. The watcher and the watched: Social Judg- ments About Privacy in a Public Place. Journal of Human– Computer Interaction, vol. 21, No. 2, pp. 235–272. Garbarinoa, e., and M. Strahilevitz. 2004. Gender Differences in the Perceived Risk of Buying Online and the effects of Receiving a Site Recommendation. Journal of Business Research, vol. 57, pp. 768–775. Gronroos, C. 1994. From Marketing Mix to Relationship Marketing: Towards a Paradigm Shift in Marketing. Man- agement Decisions, vol. 22, March, pp. 4–20. Hann, I. H., K. L. Hui, T. Lee, and I. Png, I. 2002. Online Information Privacy: Measuring the Cost–Benefit Trade- off. Proc., 23rd International Conference on Information Systems, Completed Research Track, Barcelona, Spain. Herring, S. C. 2001. Gender and Power in Online Communica- tion. CSI working Paper. No. wP- 01-05. University of Indiana, Bloomington. Hoffman, A. The Social Media Gender Gap. Business Week, May 19, 2008. http://www.businessweek.com/technol- ogy/content/may2008/tc20080516_580743.htm.

126 wOMeN’S ISSUeS IN TRANSPORTATION, vOLUMe 2 Hu, P. S., and T. R. Reuscher. 2004. Summary of Travel Trends: 2001 National Household Travel Survey. FHwA, U.S. Department of Transportation. Kate, N. T. 1998. women want Privacy—women More Concerned About Privacy Issues. Louis Harris & Associ- ates Survey for Privacy & American Business. American Demographics, January. Marchau, v., w. walkera, and R. van Duin. 2008. An Adap- tive Approach to Implementing Innovative Urban Trans- port Solutions. Transport Policy, vol. 15, No. 6, pp. 405–412. Matthies, e., S. Kuhn, and C. A. Klöckner. 2002. Travel Mode Choice of women: The Result of Limitation, ecological Norm, or weak Habit? Environment and Behavior, vol. 34, No. 2, pp. 163–177. Mayer, R. N. 2003. Technology, Families, and Privacy: Can we Know Too Much About Our Loved Ones? Journal of Consumer Policy, vol. 26, No. 4, pp. 419–439. Mayer, T. S. 2002. Privacy and Confidentiality Research and the U.S. Census Bureau: Recommendations Based on a Review of the Literature. Research Report Series: Survey Methodology #2002-01. McGuckin, N., and e. Murakami. 1999. examining Trip- Chaining Behavior: Comparison of Travel by Men and women. In Transportation Research Record: Journal of the Transportation Research Board, No. 1693, TRB, National Research Council, washington, D.C. Munnukka, J. 2007. Characteristics of early Adopters in Mobile Communications Markets. Marketing Intelligence & Planning, vol. 25, No. 7, pp. 719–731. NuStats. 2006. Pilot Study. Travel Tracker Survey, Chicago Metropolitan Agency for Planning. NuStats. 2008. Travel Tracker Survey, Chicago Metropolitan Agency for Planning. O’Neil, D. 2001. Analysis of Internet Users’ Level of Online Privacy Concerns. Social Science Computer Review, vol. 19, No. 17. Pitkow, J. e., and C. Kehoe. 1996. Sixth www User Sur- vey. http://www.cc.gatech.edu/gvu/user_surveys/survey- 04-1996/. Accessed June 20, 2009. Rapleaf. Rapleaf Study Reveals Gender and Age Data of Social Network Users. San Francisco, Calif., July 29, 2008. http:// www.rapleaf.com/company_press_2008_07_29.html. Riphahn, R., and O. Serfling. 2005. Item Non-Response on Income and wealth Questions. Empirical Economics, Springer, vol. 30, No. 2, pp. 521–538. Rogers, e. 1995. Diffusion of Innovations, 4th ed. Free Press, New York. Rogers, J. L. 1996. Mail Advertising and Consumer Behavior. Psychology and Marketing, vol. 13, winter, pp. 211–33. Sheehan, K. B. 1999. An Investigation of Gender Differences in On-Line Privacy Concerns and Resultant Behaviors. Jour- nal of Interactive Marketing, vol. 13, No. 4. Sheehan, K. B., and M. G. Hoy. 2000. Dimensions of Privacy Concern Among Online Consumers. Journal of Public Policy & Marketing, vol. 19, No. 1, pp. 62–73. Singer, e., N. A. Mathiowetz, and M. P. Couper. 1993. The Impact of Privacy and Confidentiality Concerns on Survey Participation: The Case of the 1990 U.S. Census. Public Opinion Quarterly, vol. 57, No. 4, pp. 465–482. Shoemaker, P. J., M. eichholz, and e. A. Skewes. 2002. Item Nonresponse: Distinguishing Between Don’t Know and Refuse. International Journal of Public Opinion Research, vol. 14, pp. 193–201. Thakuriah, P., P. S. Sriraj, and J. F. Ortega. 2003. Spatial Data Integration for Low-Income worker Accessibility Assess- ment: A Case Study of the Chicago Metropolitan Area. Proc., Mid-Continent Transportation Symposium, Iowa State University, Ames. Turner, T., and D. Niemeier. 1997. Travel to work and House- hold Responsibility: New evidence. Transportation, vol. 24, No. 4. wang, P., and L. A. Petrison. 1993. Direct Marketing Activities and Personal Privacy. Journal of Direct Marketing, vol. 7, winter, pp. 7–19. westin, A. F. 1998. E-Commerce & Privacy: What Net Users Want. Privacy & American Business, Hackensack, N.J. Zmud, J., and C. Arce. 2000. Item Nonresponse in Travel Sur- veys: Causes and Solutions. In Transport Surveys: Raising the Standard. Proceedings of an International Conference on Transport Survey Quality and Innovation, Grainau, Germany, 1997. ResouRces Bricka, S. 2009. Chicago Regional Household Travel Inven- tory: Draft Final Report. NuStats. http://www.cmap. illinois.gov/uploadedFiles/regional_data/TravelTracker Survey/TravelTrackerSurveyMethodsReport.pdf. Clarke, R. Introduction to Dataveillance and Information Pri- vacy, and Definitions of Terms. August 15, 1997. http:// www.rogerclarke.com/Dv/Intro.html. Long, S. J. 1997. Regression Models for Categorical and Lim- ited Dependent Variables, 5th ed. Sage. McDonald, A., and L. F. Cranor. 2006. How Technology Drives vehicular Privacy. I/S: A Journal of Law and Policy for the Information Society, vol. 2, No. 3, pp. 981–1015. Palen, J. 1997. The Need for Surveillance in Intelligent Trans- portation Systems. University of California PATH, Berke- ley, Intellimotion, vol. 6, No. 1, pp. 1–3, 10. Salaway, G., and J. B. Caruso, with Mark R. Nelson. 2008. eCAR Study of Undergraduate Students and Information Technology. Research Study, vol. 8. eDUCAUSe Cen- ter for Applied Research, Boulder, Colo. http://www.edu cause.edu/ecar. Taylor, H. Most People Are “Privacy Pragmatists” who, while Concerned About Privacy, will Sometimes Trade It Off for Other Benefits. Harris Poll #17, March 19, 2003. http://www.harrisinteractive.com/harris_poll/index.asp? PID=365.