The Effect of Formal and Informal Social Capital on Diffusion of Wireless Encryption Practices: A Case Study

Sorin Adam Matei

Purdue University

<sorin@matei.org>

Introduction

The emergence of wireless computer networks supported by 802.11 protocols has the potential to change the way information is distributed and used. This wireless technology, popularly known as WiFi, can be used to connect a large number of devices to each other and to the Internet through radio waves (Jordan & Abdallah, 2002; Rao & Parikh, 2003). Computing devices of all kinds can be freely moved within the coverage area of a wireless local area network (WLAN), revolutionizing the very idea of local area network. LAN structures can be reshaped by simply moving a computer from one desk, room or location to another.

The creation and implementation of this technology was necessitated by the proliferation of computer networking over the last decade, which caused a demand for new interfaces that would enable multiple computers to connect to each other and to the Internet more efficiently. The complexity of these new local computer networks posed a daunting dilemma in terms of infrastructure, as reconfiguring or wiring older buildings and production facilities for increasingly faster traffic and applications. In response to these issues, a wireless method of networking computers was developed to alleviate the costly management and updating of cable-based computing environments . Although 802.11 wireless networks were first proposed as a “patch” technology for the business environment, they were soon repurposed as a solution for residential users. Many households felt the need to link multiple computers through a home network and/or to share a single Internet access point.

This technological drift from a business to a residential product was not entirely foreseen by the developers of 802.11 wireless technology. In addition, the home use of the technology raised specific concerns. The reach of the radio waves that support WiFi technologies cannot be confined by walls and fences. The only spatial limitation is that imposed by how far the electromagnetic signal can travel. Although this can be dampened by wood, brick, mortar or other types of obstacles, the only real limitation is that imposed by the power of the emitting station. Although this power is mandated by FCC to be quite low (a few watts), a radio wave broadcast by a wireless antenna can easily travel under certain conditions a mile or even more. Shortly, radio signals can travel much farther than their immediate beneficiaries might want them to go. Thus, WiFi networks are intrinsically exposed to abuse, unauthorized use, and privacy infringement and home users seemed to be less aware of these problems than other users.

Despite these problems, the 802.11 technology was developed to promote unlimited access to networks by any compatible hardware. It is true that some safety measures to prevent unwanted parties from accessing the network itself or individual computers on the network were designed into the technology to keep data secure. All wireless protocols include methods of filtering or authorizing connections and signal encryption procedures. Encryption usually consists of an authentication protocol which allows individual computers to connect to each other and to the Internet, via the wireless network base station (access point) only after authenticating and implementing an encryption key. Yet, in most cases, these procedures have to be manually set by the user and require a degree of comfort with utilizing and customizing computer systems. Companies that produce the hardware for home wireless networks tout the “out of the box” simplicity of their products over the encryption and safety capabilities (that are not very easy to master). Because of this relative difficulty of use, encryption has become in the transition from business to residential use a secondary and, in many situations, ignored feature .

Thus, adoption of wireless networks in the home market requires a careful evaluation of their costs and benefits. However, current understanding of the extent to which users engage in this kind of analysis and how frequently they decide to adopt encryption practices is fragmentary. The present paper will try to explore the social diffusion of encryption-related social practices over time, from a diffusion of innovations and social capital perspective. In this context, our focus is not on the technology itself, but on the practice of choosing to use or not to use encryption, much like choosing whether to use other computer technologies such as the internet, e-mail, and the like.

Using as theoretical framework the “diffusion of innovations” and “social capital” research paradigms applied to communication technologies (Simpson, 2005), the current study hopes to provide a new insight into the role played by formal and informal local social bonds in adoption and diffusion of technological practices. Specifically, the aim of this study is to explain the diffusion patterns of wireless computer networks, especially encryption, and to examine the role that social capital plays in shaping these patterns. Currently, the relationship between social capital and wireless networks is virtually unexplored; the current study will break new ground in determining whether these two are linked and will advance our knowledge base on diffusion processes.

Review of Literature

Diffusion of Innovations

The use of encryption in home wireless computer networks is an innovation that people can choose to adopt or reject. Like many innovations that can affect the way people live their lives, encryption will be embraced by some and rejected by others. We know, however, that innovations that are not adopted are not always inferior or defective. There are a number of factors that determine whether a new innovation is diffused or not that do not deal with the inherent value of the innovation itself.

Rogers defined diffusion as a process by which an innovation is communicated through certain channels over time among the members of a social system. He identified a number of factors that play a part in the adoption or rejection of innovations, citing the importance of communication channels. The four-step process of communicating about an innovation consists of 1) the innovation itself; 2) a person that has knowledge of or experience with the innovation; 3) a person who does not have knowledge of or experience with the innovation; 4) a communication channel connecting the two units. While mass media channels can reach a large number of people rapidly, Rogers states that diffusion research shows that most people base their evaluation of an innovation on “near peers,” similar individuals who have already adopted an innovation and provide their own subjective evaluations of its value. According to Rogers, “dependence on the experience of near peers suggests that the heart of the diffusion process consists of the modeling and imitation by potential adopters of their network partners who have adopted previously. So diffusion is a very social process…” (Rogers, 1995, p. 18).

One of the key factors in this social process is the group of those who seize upon innovations while these are still in their infancy. Rogers refers to this group as early adopters. Early adopters are respected opinion leaders who take the lead in their social system and are looked up to for information. Members of this group are not on the cutting edge of technology; they are not so far ahead of average people that their knowledge is intimidating, and they make up a fair percent of the distribution curve for diffusion of innovations (13.5%). However, they do adopt innovations fairly rapidly and provide a subjective analysis to others in their social networks.

In the case of the diffusion of communication technologies, Rogers identified three important ways in which adoption differs from other innovations: 1) there must be a critical mass of new users to persuade the average users of the technology's efficacy; 2) use of the innovation must occur regularly and often for the diffusion effort to succeed; 3) the technology should be very malleable and should be easily repurposed by individual adopters.

Other researchers have stressed the importance of social networks in diffusion of communication technology innovations. The strength and importance of these social networks can be evaluated, as suggested by Simpson (2005) by examining a related field of inquiry, that of social capital.

Social Capital

Putnam defines social capital as “connections among individuals – social networks and the norms of reciprocity and trustworthiness that rise from them” (19). This implies that social capital is an inherent component in creating and maintaining relationships with others and in collective action. Social capital is a combination of formal and informal ties, the former precipitated in the form of local (community) organizations and the latter in friendship, kind and “strong tie” networks. Social capital is thus an abstract concept that is an amalgamation of a set of characteristics that deal with personal and impersonal or semi-personal relationships with others. Pigg and Crank (2004) in their comprehensive review of the literature on the relationship between information technologies and social capital identify five recurring dimensions: networks, resources for action, reciprocity transactions, bounded solidarity and enforceable trust.

One of the most important distinctions that should be made in any discussion about social capital, as suggested by Coleman (1988), Portes (1998), or Putnam (2000), is that between immediate, direct social relationships and weaker social ties. While the former can be considered to be similar to Putnam’s bonding ties, tha latter can be assimilated with his bridging social ties and are usually established and maintained in more instrumental and formalized settings, such as a local community organization. While they are often found and examined together in research on social capital, bonding and bridging, or informal and formal ties can work independently of one another in terms of influencing behaviors.

Although not always directly addressed in diffusion of innovation studies, it becomes immediately obvious that social capital is and should be directly addressed by this line of research. As Simpson suggested in her larger discussion of “community informatics” (CI)-- technological initiatives that can impact community life--social capital can be seen an essential element for the successful diffusion of any technology in a given community:

A community’s receptiveness to a CI initiative is influenced by the extent to which the initiative matches the community’s aspirations, values and needs, or is perceived as contributing to the future well being of the community. Community aspirations, values and needs, and a shared sense of the future well-being of the community are aspects of social capital. The extent to which social capital exists in a community is therefore a critical factor in the community’s receptiveness to CI initiatives and its acceptance of the technology, and consequently the likelihood of the CI initiative to succeed and be sustained. (Simpson, 2005, p. 113).

This argument echoes Rogers suggestion that social networks are of crucial importance in diffusion of innovation and it makes it all too natural to consider social networks not as an isolated phenomena, but as part of a larger class of social experiences. As Pigg and Crank (2004) emphasize, networks are part and parcel of the social capital world. More precisely, they can be considered types of social capital. The social networks Putnam (2000) or Coleman (1998) talk about, those established within a church, a parent-teacher association, a neighborhood, or those that grow out of friendship or kinship ties are the same social connections that we use when we need support for understanding and adopting many objects and technologies. Neighbors and friends not only volunteer together for the same church charity or parent teacher association, they also talk about and advise each other about home improvement projects, gardening, car repair or new gadgets and devices. Often purchasing new objects is a direct product of imitation or emulation of those that are closer in one’s social circle.

The relationship between social capital and technological diffusion is not just intuitive or theoretical, but also increasingly discussed in the empirical literature. Social capital has been shown to play an important role in diffusion of innovations and is integral to how new technologies emerge from fringe status symbols to mainstream consumer goods.

For example, Burt (1999) brings up the issue of trendsetters in relation to social capital and diffusion. In his view, local opinion leaders—who by definition have a higher level of social capital—are instrumental in the diffusion process. These community members are vital in the diffusion process because they can cross boundaries between social networks and are “brokers” of information. By being able to carry information from group to group they reap the knowledge benefits of multiple social networks and are able to use their high amounts of social capital to help the process of diffusion. These opinion leaders may belong to groups that are both interest-based and geographically-based, which can enable them to pass information learned from their interest-based groups to their neighbors or co-workers.

A longitudinal study on formal social capital and the Internet revealed that the practice of internet adoption occurred more quickly among individuals with higher social capital and that Internet use does not reduce time spent interacting in social networks, which indicates a positive relationship between social capital and the practice of Internet adoption .

In a comparison of the implementation of information networks in two Minnesota cities, Oxendine et al. found that the city which promoted a formal collaborative approach and equal access for its citizens saw a more developed community electronic network as compared to the city that took an entrepreneurial approach when implementing the technology. The researchers cited the trust and cohesion, resulting from the stronger social capital endowment of the more successful city, as the responsible factor.

Riedel, Dresel, Wagoner, Sullivan, & Borgida examined the implementation of an electronic network in a rural community and found that initial adoption of technological advances is done by those with greater material resources. Yet, those most responsible for the diffusion of the technology were the community members with larger resources of social capital.

It is also important to remember that as Baym, Zhang, & Lin found, face-to-face interaction is higher quality than online communication when thinking about communication technology; therefore, physical proximity to other group members is a factor when considering the role of social capital in diffusion of innovations in the communication technology arena.

Based on these studies, it is not unreasonable to extrapolate the processes detected for communication technology in general to adoption of wireless network related social practices. Our theoretical premise will be that social capital, both in its formal and informal forms, is part of diffusion of encryption. To explore this issue we have collected over time (2 phases, at a 10 month interval) data related to diffusion of wireless networks in the residential environment of a case-study urban community, for which we also know, at a satisfactory level of detail, its level of formal and informal social capital endowment and its socio-demographic composition.

To elucidate the relationship between social capital and diffusion of wireless networks encryption practices we advance two research questions and two hypotheses. The research questions are exploratory in nature and aim to determine the extent to which wireless networks have become a part of the residential arena in the target location, and how prevalent the practice of encryption has become, over time, in the same location:

• RQ 1: At what rate do wireless networks diffuse in the residential arena?

• RQ 2: At what rate does encryption, as a social practice, diffuse in the residential arena?

The role of the research questions is to determine the extent of the phenomena of diffusion and to ascertain that a concern for encryption is a reality for residential users of wireless networks.

Describing the process of diffusion cannot, however, explain its determining factors. Our main interest is not only to determine if wireless-related practices are diffusing, but also what social factors impact them. Among them, of particular importance is social capital in both its forms: formal and informal. Our hypotheses address exactly this matter, one proposing that formal and the other that informal social capital has a direct influence on diffusion of encryption practices. Since our research is conducted with macro social units of analysis (neighborhoods, see Methods section for details), the hypotheses take the following format:

• H1: The units of observation with the highest level of formal social capital are the most likely to evidence a shift to encryption practices.

• H2: The units of observation with the highest level of informal social capital are the most likely to evidence a shift to encryption practices.

Methods

Datasets

We answer the research questions and we address the hypotheses presented above using data about wireless technology diffusion collected from a small-medium American Midwestern town: X. The city was chosen as a study site for two reasons. First, the research was partially supported by University of Y, which is located in X. The university sponsored this research project to investigate the role of new technologies in fostering the development of the local economy and society. Second, X reflects the transformation of many smaller-medium size towns, at the border between the Midwest and the South, from sleepy regional agro-industrial centers to new economy magnets. X, for example, was very successful in the last 20 years in attracting and retaining high-tech and high-power manufacturing businesses. Z, the printer manufacturer, is located here and one of the main W (car brand) factories is found in the immediate vicinity.

X also presents other socio-demographic and urban advantages. It is a self-contained urban area, presenting a good combination of business, residential, and public areas, representative overall of the mid-American urban landscape (Table 1). Its population includes a mix of social groups, from college students to farmers, from high-tech professions to blue-collar workers. The racial distribution resembles that of the US, in general, although its Hispanic population is smaller than the national average. The town also matches the national average in terms of marital status, owner-occupied homes and some professional categories (service and office occupations). Although the city is slightly wealthier, better educated, and more Internet connected, than the national average, the differences are not very great (see Table 1), departures being confined within a 10% band.

Table 1. X reflects many general United States socio-demographic characteristics.

Table sources: US Census for all data except for Internet connectedness.

*X survey conducted by author, August-September 2002.

**Pew Internet national survey, September 2002.

Three types of data about X are utilized in this study. First, there are data about presence, characteristics, and over-time diffusion of wireless networks. This dataset was collected by the first author of the paper. Second, there is basic geo-demographic information (education, income, home ownership, zoning information, etc.) about X, aggregated at neighborhood level, obtained from the US Census Bureau and from the local municipal authority. Third, there is data about density of social connections (social capital) and participation in community organizations for each of the 57 X neighborhoods included in the study. This type of information was obtained through a random digit dialing survey, conducted by the first author of this paper.

The first type of data, regarding wireless networks, will be utilized for constructing the main dependent variables employed in the analyses. The two other datasets are used for deriving the main dependent variables utilized in testing the hypotheses. In what follows we will briefly describe the methodologies utilized for obtaining each type of data.

Wireless networks. A wireless monitoring methodology was designed around a population-weighted spatial sample of the urban street grid. This builds on and improves on previous wireless monitoring methodologies, which used non-random, convenience, or exhaustive urban street samples . We start by drawing a sample of 238 spatially-random locations (1 for each 1000 inhabitants) throughout the entire urban area of X, with denser populated areas being assigned more locations. The unbiased sample of locations is drawn by randomly selecting values from the known ranges of the x and y geographic coordinates of the study area. This is accomplished using an ArcView scripting utility. ArcView is a mapping program and spatial analysis software platform. The script assigns sampling location to pre-defined geographic units (neighborhoods) according to their population density . The random locations are then connected by a 400 mile-long shortest-path route (Figure 1). The path is also constructed via ArcView, most specifically by using the Network Analysis extension. The shortest path algorithm ensures that given a network grid (in this case the street framework), each point will be connected to its closest neighbor and the resulting path will be the shortest possible.

A two person research team (driver and navigator) drove this route twice (August 2003 and May 2004), locating via Global Positioning Services (GPS) on an ArcView map the wireless networks detected along the way (Figures 1 and 2). The ten-month interval between measurements corresponded to the length of one academic year. Both the GIS application and the GPS device were connected to a computer laptop mounted on a passenger vehicle. The computer picked up the wireless signals using an Orinoco wireless card, connected to a 8db booster antenna, and logged relevant information (station identification name and unique numeric code [MAC address], encryption status, signal-to-noise ratio) using Netstumbler, a wireless network monitoring software (http://www.netstumbler.com). Encryption status, which is a central variable in our study, is represented as a binary variable: 1 if the access point broadcasts its signal using password-protected encryption, and 0 if the signal is broadcast with no cautionary measures added to it.

Because the monitoring process is a continuous one, more than one reading was obtained for each access point (wireless network). However, the final map and dataset locates each individual access point at a unique location, that where the signal was the strongest.

Figure 1. Monitoring route and sampling points

After the two rounds of data collection, a total of 3406 unique wireless networks (access points) were identified. Of these, 754 were identified in Phase 1 and 2652 in phase two. The two groups partially overlap, 436 access points identified in Phase 1 being re-identified in Phase 2.

Wireless data was post-processed for further analysis in two ways. First, each wireless network was assigned an assumed type of use, of three possible: residential, business, and public. Use status was derived from the official zoning status of the land parcel the wireless network was identified on. Land parcel information was obtained in a digital format from the X-F County Urban Government GIS office. Assignment of use values to the wireless points was done using ArcView, with each access point’s geographic coordinates being mapped onto the zoning map.

Figure 2. Locations for strongest signal point where wireless networks were identified (green, phase 1; red, phase 2).

In a second processing step we aggregate the data at neighborhood level, using US Census Bureau geographic units, census tracts, for defining the neighborhoods. For each of the 57 Census defined neighborhoods we calculated: absolute number of access points found during Phase 1 and Phase 2, number of access points that were found during both phases, number of residential, business or public access points found for each phase, and prevalence of encryption among access points for each phase and type of use.

To track over-time trends in encryption practices, we focused on the subset of access points that met the following criteria: were found on residential land parcels and were identified both in phase 1 and phase 2 of the study. These 312 access points were then divided into three categories, according to their encryption status in Phase 2, compared to Phase 1. A first category included the 43 (14%) access points that became encrypted between in phase 2, after not being encrypted in phase 1, a second the 265 (85%) access points that remained unchanged between phases (encrypted or unencrypted), and third containing the 4 (1%) access points that were encrypted in phase 1 but ceased to be encrypted in phase 2. This variable was used for computing the main dependent variable for testing the hypothesis. For each of the 57 neighborhoods we generated a score, reflecting how many access points became encrypted in the neighborhood. The score summates both positive and negative values, such that if in a neighborhood there were both access points that became and that ceased to be encrypted, the final figure reflects the net outcome, subtracting from the total of access points that became encrypted those that ceased to be encrypted.

Geo-demographic information. To explain these changes in encryption practices over time we use neighborhoods (census tracts) as units of analysis and data from the US Census Bureau and from the X-F County urban government as control variables. The analysis is done at neighborhood, rather than individual level of analysis, for two reasons. First, it is impossible to associate individual access points to specific socio-demographic variables. We do not know who the specific owners of the access points are and for ethical reasons did not attempt to identify them. Second, the most reliable source of information, the US Census Bureau, is only available at aggregate (neighborhood level).

The aggregate-level variables we chose to use in the study reflect the characteristics that can define the rate of adopting and utilizing communication technologies. These include measures of population stability and composition, such as: neighborhood age, measured as average age of homes, population density, proportion of single vs. family or owned vs. rented homes or proportion of non-Hispanic white population. A second class of variables include social characteristics, such as median family income in 1999 and proportion of college educated population. Finally, variables in a third category reflect the physical structure of the neighborhoods and their location in the city: proportion of neighborhood area that is dedicated to residential use, and neighborhood distance from the civic center (center to center measure). All these variables are used mainly as controls, to eliminate the possibility of a spurious relationship between the main predictive variables related to social capital, and the dependent variable, expansion of encryption.

Social capital measures. To reflect the dual, individual and collective, nature of the social capital concept, two measures are used. First, we employed an informal social capital measure. This was constructed at neighborhood level from answers to a battery of questions obtained from on a random sample of 801 X respondents (response rate 50%). The survey was conducted by the authors in September 2002, one year prior to Phase 1. The informal social capital index, which was previously used in other research conducted by the authors (citations deleted for review), has a satisfactory reliability (α = .80) and reflects amount of informal social capital found in a neighborhood. The index is a measure of everyday acts of neighborliness that denote potential of initiating and density of informal social ties. The concrete questions are:

Do you strongly agree, agree, neither agree, nor disagree, or strongly disagree with the statement: 1. You are interested in knowing what your neighbors are like (M=3.85, SD=1.166). 2. You enjoy meeting and talking with your neighbors (M=4.34, SD=.888). 3. It’s easy to become friends with your neighbors (M=4.05, SD=1.100). 4. Your neighbors always borrow things from you and your family (M=1.97, SD=1.302). How many of your neighbors do you know well enough to ask them to (respondent specifies a number): 1. Keep watch on your house or apartment (M=3.91, SD=4.526)? 2. Ask for a ride (M=4.23, SD=6.392)? 3. Talk with them about a personal problem (M=1.74, SD=3.821)? 4. Ask for their assistance in making a repair (M=2.49, SD=2.832)?

The higher the score on any of these items, the more likely that the respondent will be involved with other members of his or her community and the higher the level of informal social capital in that community. The index is calculated by capping the “number of neighbors” variable at 10 (10 and higher values were recoded as 10), to reduce the skewness of the data, typical in such variables. We further divided this variable by two, to bring it to a 1-5 range, similar to that utilized in the other four variables where the answers in the format “Strongly agree/Strongly disagree” are ranked on a 1 through 5 scale. The informal social capital scores were obtained by summating the scores for the 8 variables. Their theoretical range spans the interval 4 (respondent knows no one in the neighborhood and strongly disagrees with all the “agree/disagree” statements) to 40 (respondent knows more than 8 people in each category and strongly agrees with all the “agree/disagree” statements). Further, the data was combined at neighborhood level. This was accomplished by combining (averaging) the individual scores of all the respondents living in the same neighborhood into one, synthetic, neighborhoods level of informal social capital. Average neighborhood scores range between 16-28.

A second variable, also obtained through the survey, reflects the formal facet of the social capital concept and is expressed as average number of community organizations (including church affiliation) a typical neighborhood respondent has membership in. In the survey, respondents were asked to indicate if they are members of a number of 14 types of community organizations, spanning the spectrum from parent teacher associations and bible study groups to sports/arts, service, professional and church affiliation (M=1.4, SD=1.6, Range=0-8). The individual averages were then averaged at neighborhood level, such that for each of the 57 neighborhoods we obtained an indicator of formal social capital (i.e., how many associations an average neighborhood resident is a member of (M=1.3, SD=.6, Range=0-2.6).

Inter-item correlation between the items is presented in Table 2.

Intercorrelations Between Factors Influencing Wireless Encryption Practices

_____________________________________________________________

_______________________________________________________________

Note. For factors 1 through 10, N = 57. For factor 11, N = 55.

*p<.05. **p<.01.

Analysis

Our first research question asks: RQ 1: At what rate do wireless networks diffuse in the residential arena?

The results of the monitoring process indicate that wireless access points diffuse at a fast pace and that encryption has become a more frequently encountered practice in Phase 2, compared to Phase 1. Over the study period (August 2003 – May 2004) the prevalence of wireless networks has increased by 250%. While in August 2003 we identified 754 access points, in May 2004 the number increased to 2652 (Figures 2 and 3). Assuming that the effective coverage of an access point is about 200 ft from the wall of a building, we estimate that 15% of the households in Lexington had a wireless network during phase two of our study (May 2004). This figure was computed by dividing the number of access points identified in Phase 2 by the number of households found within 200 feet of the monitoring route. The figure is probably higher, but not by much, than the national average. The Forrester Research Technographics Survey , announced in June 2004 that in December 2003 (five months before the completion of the present study), 15% of American homes had a computer network of some kind (wired and wireless).

A significant finding of our study is also that a large majority of the residential networks found in X were residential, their proportion increasing over time. During the first phase of the study, conducted in August 2003, two-thirds of the networks were located in private residences; in May 2004 their proportion increased to three-quarters (Figures 2 and 3). During the same period, the relative proportion of business access points decreased, from 21% to 16%. This indicates that wireless networking is about to become a home appliance, incorporated into the everyday life of many middle-class homes, as illustrated in figures 3-4.

Figure 3. Phase 1 access points: type of use.

Figure 4. Phase 2 access points: types of use.

The answer to the second question – at what rate does encryption, as a social practice, diffuse in the residential arena? – is that encryption became more prevalent in 2004 compared to 2003 across all types of use. All wireless computer networks, and especially those located in residential areas, broadcast their signal using encryption at a higher rate in 2004, compared to 2003 as indicated by figures 5 and 6.

Figure 5. Phase 1 access points by use type and encryption status.

Overall, the encryption rate has increased from one-fifth to one-third. Looking at residential and business networks separately, 28% of the residential access points were encrypted in August 2003, compared to 18% in May 2004, a 55% increase in only 10 months. In the business arena, the increase was also significant, but not as high: from a 24% encryption rate, in August 2003, to a 32% rate in May, 2004, which translates into a 33% increase.

Figure 6. Phase 2 access points by use type and encryption status.

In addition, as already mentioned, a greater proportion of the residential wireless networks identified both in Phase 1 and 2 have become encrypted over time (14%) than those who ceased to be encrypted (1%).

This trend indicates that as wireless networks become more prevalent, individual and institutional users become more aware of the fact that this technology entails a number of weaknesses and that encryption can be a solution for these weaknesses. However, the fact that a majority of the networks are still not encrypted indicates that more is to be done until network security becomes an everyday concern in American households or businesses.

To determine the factors that impact diffusion of encryption, and to test the two hypotheses of this study, stating that the units of observation with the highest level of social capital are the most likely to evidence a shift to encryption practices – we ran and fit a multiple regression algorithm, using backward elimination of non-significant variables. The model includes number of wireless access points that became encrypted between Phase 1 and 2, regressed on: informal and formal social capital, percentage homes that are owned, percentage homes that are occupied by singles or by families, percentage population that is college educated, percentage neighborhood area dedicated to residential use, average home age, population density, and neighborhood distance from the civic center. The results indicate that while level of formal social capital has an effect on diffusion of encryption practices, informal social capital, while going in the predicted direction, does not. Neighborhoods with higher levels of formal social capital (β=.33, p<.05) and with higher income (β=.3, p<.05) are more likely to have seen an increase in number of residential access points identified in both phases that become encrypted. Level of informal social capital has no significant effect on encryption practices, the variable being eliminated by the stepwise procedure with a β value of .163 at a p value of .257. In this model, the two significant variables explain about 20% of the variance found in the dependent variables (Adjusted R²=.21). Since formal and informal social capital are correlated (r=.46), which increases the risk of colinearity, we also ran a multiple regression with backward elimination using only the informal social capital variable, as main predictor, alongside the other control variable. The results indicate that in absence of formal capital, informal social capital has an effect only at a p< .1 level of significance (β=.23, R²=.18) and that the only predictor significant at p<.05 is education (β=.39). In conclusion, the results indicate that while we cannot accept H₂, they support H₁.

Discussion

The results presented in this paper indicate that wireless networking has become an important home technology, diffusing at a very rapid pace. While still in the “early adoption” phase, its 250% increase in the case-study location explored here indicates that wireless technology has a lot of potential and it has caught the attention of many everyday users. In the process of its diffusion, as we have noticed, wireless technology has become more prevalent in the residential arena that in the business market. The advantages entailed by this technology are many, including mobility, convenience, and enhanced status. However, the drawbacks are also significant, especially those related to the danger of abuse and privacy loss. The current study indicates that these concerns have become more prevalent over time, as more networks have started to use encryption to ward off abuse and intrusive technology access. However, overall, the residential market is still wide-open to these dangers, since despite the fact that more wireless networks were encrypted in the second phase of our study, less than one-third of them were protected using this methodology.

Our results indicate that an important predictor of adoption of encryption in the residential market is wealth and strong formal civic capital. Areas where people had higher incomes and where participation in community organizations was higher saw the greatest increases in adoption of encryption. While the relationship between income and encryption is somewhat intuitive, higher income households displaying defensive practices in more than one way (home surveillance coming to mind first), that between formal social capital and demands further interpretation and qualifications. First, individuals that are members of community organizations are also more likely to be attuned with the burning ideas and issues of the day. Privacy infringement and privacy protection are obviously very important current topics and they have percolated down to the level of our communities and neighborhoods. Second, community organizations are environments where ideas and practices of a wide diversity are transacted and discussed. Third, the social ties established through participation in community organizations are “weak” in nature, as defined by Granovetter (1973), and thus much more likely to “bridge” and connect people to new ideas and technologies. Our findings seem thus to suggest that social knowledge relevant to technological practices is one of the issues people learn about in these environments.

On the other hand, our paper did not indicate any significant relationship between individual (informal) social ties and diffusion of encryption. This means that technological ideas and practices are more likely to be reinforced by formal rather than informal social capital. This finding apparently represents a departure from what we currently know from diffusion of innovation research, where the main mechanism of technological propagation was always seen in the interpersonal arena. However, our finding is not strong enough to completely disconfirm this widely accepted tenet. In fact, as we noted, when running a regression model only for informal social capital, there was a positive effect, although at a lower level of significance than the conventional .05 threshold. The lack of significance could be caused by a number of factors, including the reduction in number of cases due to data aggregation at neighborhood level. Thus, more research is required for clarifying this point.

Further research on the topic is also demanded by the fact that our study does not offer a direct measure and in-depth picture of how wireless technologies diffuse at the individual level. Our units of analysis were neighborhoods, not individuals. To avoid the dangers of ecological fallacy, we should limit our claims to the simple idea that community social capital seems to foster a environment for more protective technology use practices. The concrete way in which this takes place should be studied at the individual/attitudinal level, through appropriate tools: individual-level qualitative and quantitative research.

Such research might also want to address the implications of social capital specific behaviors and beliefs not only for the diffusion of security practices, but also with respect to the viability of some new WiFi applications. One of the more promising ideas is that of the mesh-networks. These are networks that connect individual users via radio waves in a direct manner, without the mediation of access points and Internet gateways. In such scenarios, each computer is both a client and a hub of the wireless network. Such networks depend in great measure on stable ties and a great deal of trust between the users. Users need to communicate between them and to help each other, if needed, since the network needs every single node to contribute its share in order to avoid bottlenecks. At the same time traffic can be quite intense and can serve multiple purposes, from education and one-to-one communication via instant messaging or voice over internet software, to entertainment and file sharing. In the latter situation, if one member gets involved in illegal activities, such as piracy, the network as a whole can be found liable for the illicit activity.

An important research question that needs to be asked before starting such projects is if they will have the capacity to “take off” if they are powered only by interpersonal connections and informal diffusion. The findings of this paper suggest that such expectations might be too optimistic. Instead, community activists should orient their efforts toward co-opting existing community organizations, which could serve as diffusion channels for the new wireless skills or technologies and could reinforce the trust needed for supporting their initiatives.

In conclusion, this is a line of research that is extremely promising and the intellectual and practical payoffs numerous and substantial. We hope that the methodology presented here offers a way to further this research agenda and will offer practitioners useful knowledge for understanding an emergent technology that promises to significantly influence our daily lives.

References

Barber, G. M. (1988). Elementary statistics for geographers. New York: The Guilford Press.

Baym, N. K., Zhang, Y. B., & Lin, M. C. (2004). Social interactions across media. New Media & Society, 6(3), 299-318.

Burt, R. S. (1999). The social capital of opinion leaders. Annals of the American Academy of Political and Social Science, 566, 37-54.

Byers, S., & Korman, D. (2003). 802.11b access point mapping. Communications of the ACM, 46(5), 41-46.

Castells, M. (1996). The rise of the network society (Vol. 1). Padstow, Cornwall, Great Britain: Blackwell Publishers.

Fleishman, G. (2002, December, 12). The Wi-fi boom: So many nodes, so little security. The New York Times.

Franzen, A. (2003). Social capital and the Internet: Evidence from Swiss panel data. Kyklos, 56(3), 341-360.

Granovetter, M. (1973). The strength of weak ties. American Journal of Sociology, 78, 1360-1380.

IEEE. (1999). Information technology. Telecommunications and information exchange between systems. Local and metropolitan area networks. Specific requirements. Part 11: Wireless LAN medium access control (MAC) and physical layer (PHY) specifications, ANSI/IEEE 802.11, 1999 Edition: IEEE.

Jordan, R., & Abdallah, C. T. (2002). Wireless communications and networking: An overview. IEEE Antennas & Propagation Magazine, 44(1), 185-193.

Kempe, D., Kleinberg, J., & Tardos, E. (2003). Maximizing the spread of influence through a social network. Paper presented at the 9th ACM SIGKDD Intl. Conf. on Knowledge Discovery and Data Mining.

Oxendine, A., Borgida, E., Sullivan, J. L., & Jackson, M. S. (2003). The importance of trust and community in developing and maintaining a community electronic network. International Journal of Human-Computer Studies, 58(6), 671-696.

Parthasarathy, D., & Chopde, V. K. (2000). Building social capital: Collective action, adoption of agricultural innovations, and poverty reduction in the Indian semi-arid tropics. Retrieved September 10, 2004, from http://www.eldis.org/static/DOC8246.htm

Pigg, K., & Crank, L. (2004). Building community social capital: The potential and promise of information and communications technologies. The Journal of Community Informatics, Vol. 1(1), 58-73. Retrieved July 28, 2006, from http://www.ci-journal.net/viewarticle.php?id=15.

Putnam, R. (2000). Bowling alone. New York: Simon and Schuster.

Portes, A. (1998). Social capital: its origins and applications in modern sociology. Annual Review of Sociology, 24, 1-24.

Rao, B., & Parikh, M. A. (2003). Wireless broadband networks: The US experience. International Journal of Electronic Commerce, 8(1), 37-53.

Riedel, E., Dresel, L., Wagoner, M. J., Sullivan, J. L., & Borgida, E. (1998). Electronic communities: Assessing equality of access in a rural Minnesota community. Social Science Computer Review, 16(4), 370-390.

Rogers, E. M. (1986). Communication technology. The new media in society. New York: The Free Press.

Rogers, E. M. (1995). Diffusion of Innovations (4th ed.). New York: Free Press.

Schadler, T. (2004). Benchmark data overview. Consumer Technographics North America. (Annual Report). Cambridge, MA: Forrester Research.

Simpson, L. (2005). Community informatics and sustainability: Why social capital matters. The Journal of Community Informatics, Vol. 1(2), 102-119. Retrieved July 28, 2006 from http://www.ci-journal.net/viewarticle.php?id=47.

WWRF. (2001). The book of visions (Draft Report): Wireless World Research Forum. Retrieved December, 2001, from http://www.wireless-world-research.org/general_info/BoV2001-final.pdf