Abstract

High-speed wireless Internet technologies such as Wi-Fi and WiMAX allow users to go online at broadband speed, anywhere, anytime. Experimental projects in publicly accessible Wi-Fi systems and citywide networks are mushrooming in the United States and elsewhere. Grounded in the diffusion of innovations research paradigm, this study examines factors that influence the adoption of Wi-Fi in the workplace. Similar to the adoption of the Internet at home, results show that perceived advantages and compatibility of wireless Internet, larger number of fellow employees and family members already using wireless Internet, and higher frequency of communication with technicians about the wireless system—but less time spent reading newspapers—all lead to a higher likelihood of using the Wi-Fi powered Wireless Local Area Network (WLAN). Managerial implications of the findings are discussed.

Purpose of Study

Emerging high-speed wireless Internet technologies such as Wi-Fi—wireless fidelity—(the 2.4 GHz 802.11b wireless Ethernet standard) and WiMAX allow Internet users to go online at broadband speed, anywhere, anytime. Users can access the Internet via laptops and hand-held devices like personal digital assistants (PDAs), pocket PCs equipped with a wireless card, and Internet-enabled cell phones. The wireless network is basically an extension of the wired network because it provides broadband access to the Internet via access points known as "hot spots" built to the edge of the wired network. With this new way of accessing the Internet, users can check email, download files or music, access databases, and browse the World Wide Web while on the move. Experimental projects in publicly accessible wireless systems and city-wide networks of hot spots are mushrooming across the world (Tedeschi, 2004). There were 84,000 hot spots in the world in 2005; industry estimates put the total over 100,000 in 2006 (Best, 2005). The number of Wi-Fi units shipped increased sharply from 2.1 million in 2000 to 40 million in 2005 (Dell'Oro Group Report, 2005).

Wireless Internet offers expanding applications for all organizations, which increasingly rely on LANs (local area networks) for internal and external communications (Steinfield, 2002). Applications in higher education, for instance, include emailing, accessing LAN and Web-based course management systems (e.g., Blackboard), and exchanging files, as well as researching databases outside the office and classroom. The purpose of this study is to explore factors predicting the adoption of Wi-Fi in a workplace organizational setting.

The focus on the diffusion of wireless Internet among employees in the workplace is based on two theoretical considerations: First, LaRose and Hoag (1996) argued that the Internet was primarily an "organizational innovation" (p. 200). For instance, institutions of higher education and research institutes built Internet networks earlier than individuals who wired their homes. It is not a surprise that faculty, researchers, and college students were the earliest users of the Internet prior to its popularity in the general population in the 1990s (Jones & Johnson-Yale, 2005).

Second, a new information technology may be widely acquired by institutions, but only sparsely used within "adopting" organizations. Fichman (1995) described this phenomenon as an assimilation gap. Hiltz and Johnson (1989) called it a "usage versus acceptance" issue (p. 193). Thus, a focus on adoption of the newer Internet technology of Wi-Fi among employees in organizations will avoid the gap. To explore how a Wi-Fi powered wireless network adopted by an organization is actually used by employees, following past research on computer adoption (Lin, 1998), the study expands the status of adoption beyond the dichotomy of users and non-users by including a middle category of likely users. Grounded in the diffusion of innovations research paradigm (Rice & Webster, 2002; Rogers, 1995; Steinfield, 2002), the present study addresses the following issues concerning the adoption of wireless Internet in the workplace: (1) Does the diffusion curve apply to Wi-Fi? (2) What are the differences in personal characteristics, perceptions of Wi-Fi attributes and benefits, social pressures, technology cluster, and media use among early users, likely users, and non-users? (3) What are the predictors of using wireless Internet in the near future among likely and non-users?

Theoretical Framework, Literature Review, and Hypotheses

This study is theoretically framed in the diffusion of innovations research paradigm, which analyzes and explains how people respond to new innovations, including new information technologies (Rogers, 1986, 1995). The diffusion process is conceived as communication driven via certain channels over time among members of a social system. Thus, the innovation, communication channels, time, and social system are considered the four major elements in the paradigm. Rogers (1995) identified five types of adopters in a diffusion curve known as the "multi-step flow theory": innovators (individuals who are adventuresome, cosmopolitan, risk-taking, information seeking with a higher financial status; 2.5%), early adopters (individuals with the greatest degree of opinion leadership, respected by other members of a social group; 13.5%), early majority (the deliberate individuals who adopt new ideas just before the average member of a social system; 34%), late majority (the skeptical individuals who adopt new ideas just after the average member of a social system; 34%), and laggards (the traditional individuals who are the last in a social system to adopt an innovation; 16%).

Defined as "the decision to make full use of an innovation as the best course of action available" (Rogers, 1986, p. 122), the adoption decision can be individual and organizational. Adoption decisions made by individuals in a social system fall into the category of what Rogers (1995) characterized as "optional," while adoption decisions made by organizations tend to be "collective" and "authority" (p. 29). The distinction between individual decisions, and collective and authority decisions, thus lies in the unit of decision making, with the former being individuals and the later involving organizations. Not surprisingly, the process of organizational adoption, as Rogers (1995) suggested, is more complex due to an organization's structures, size, and resources. His adoption model in organizations highlights five stages in two phases: initiation (e.g., problem recognition and search for solutions) and implementation (e.g., putting an innovation to work in terms of restructuring, clarifying, and routinizing). Rogers further argued that adoption rate by authority decisions, although the fastest, may be "circumvented" (1995, p. 30) by employees in the implementation stage. Steinfield also argued that organizations may build new communication infrastructure, but "individuals ultimately choose to use one medium or another to communicate with others" (2002, p. 249).

Therefore, this study examines uncertainty in the implementation of an adopted wireless Internet technology within an organization by focusing on six influences on individuals' adoption decision: individual, technology characteristics, perceived benefits, social, technology cluster, and media use. Influences that are unique to organizational adoption such as organizational structures are not included.

Individual Influences

Individual influences refer to personal characteristics of potential adopters and users of a new media technology such as age, gender, education, income, marital status, and household size. The generalization is that those who are younger, better educated, and up-scale will be more receptive to a new information technology and tend to be early adopters (Dutton, Rogers, & Jun, 1987; Fulk, 1993; Rogers, 1986). In addition, males are more likely than females to be adopters of new media (Rice & Webster, 2002; Rogers, 1995).

In examining the diffusion of personal computers, past studies found significant demographic differences among adopters, likely adopters, and non-adopters (Dickerson & Gentry, 1983; Lin, 1998). In addition, Crispell (1994) found that the homes most likely to have computers were those with married couples and children under 18.

Similar findings were reported in research on the diffusion of the Internet. Using what they called "social locators," Atkin, Jeffres, and Neuendorf (1998) found a young, educated profile for early adopters that mirrored the profile of personal computer adopters. An Internet adoption study (Busselle, Reagan, Pinkleton, & Jackson, 1999) among university faculty and staff reported gender and age as significant predictors of Internet use, with younger males being heavier users. Katz and Rice (2002) analyzed the effects of age, gender, income, and education in the adoption of the Internet using longitudinal data collected in 1992 and 2000. Specifically, users in the 1992 survey were young. Despite the increases in the proportion of users aged 40 and older in 2000, this proportion was still below the proportion of people aged 40 and older in the general population. In terms of gender differences, more male users were found in the 1992 survey. The majority of users had a college degree; the proportion of non-college graduates was only 28% in the first survey. Early users also tended to have a higher income.

Studies of the adoption of the Internet in other countries have reported comparable findings. Zhu and He (2002) found that age, gender, and education were significant predictors in China. In their study of Internet adoption in South Korea, Rhee and Kim (2004) found significant differences between Internet users and non-users in age, gender, education, marital status, and income. These findings provide the basis for the first hypothesis:

In addition to demographics, an individual's innovativeness (the degree to which an individual is receptive to new ideas, Midgley & Dowling, 1978; Rogers & Shoemaker, 1971) plays an important role in adoption of innovations. The generalization is that the more innovative a person is, the more likely it is that he or she will adopt a new information technology (Rogers, 1995). Leonard-Barton and Deschamps (1988) identified personal innovativeness as a strong intrinsic motivation not only to adopt an innovation earlier, but also to use it more in organizations.

Past studies of personal computer adoption (Lin, 1998) showed that rating the need for innovativeness as measured by willingness to learn new things, exploring new technologies, and keeping up with new technologies was highest among adopters. In a study of Internet adoption, Atkin, et al. (1998) used the following questions to measure the extent to which respondents were pro-technology: "I consider myself a modern person who is usually up-to-date on new technologies, and I enjoy trying out new technologies in order to introduce them to my friends" (p. 483). Further, innovativeness was found to have the power to discriminate between Internet users and non-users. Therefore, it was further hypothesized that:

Influences of Technology Attributes

Rice and Webster (2002) proposed that objective characteristics or attributes of new media technologies affect their adoption and use. These attributes include transmission quality, video-image quality, accessibility of a medium such as speed and ease, and reliability (Goodhue & Thompson, 1995; Webster, 1998). Empirical research on the adoption of high-definition television (HDTV) focused on picture sharpness, sound quality, and screen size as predictors (Dupagne, 1999). This research found that the perceived importance of these attributes, picture sharpness in particular, was positively related to awareness of and interest in HDTV. In addition, the importance of screen size was a positive correlate of HDTV purchase intention. In explaining the likelihood of people subscribing to interactive television in Hong Kong, Leung and Wei (1998) found that connectivity and expanded choice of interactive TV predicted subscription intention.

In the context of a wireless Internet network that supplements a wired network, high-speed Internet access via a wireless, ubiquitous telecommunications network is the touted feature. Informed by past research, this study focused on three attributes of wireless Internet: access without a wire, mobility, and the capability to use a variety of receiving devices (e.g., laptops, PDAs, and pocket PCs) to access the Internet. Specifically, it was hypothesized that:

Influences of Perceived Benefits of Innovation

Rice and Webster (2002) argued that the subjectively-perceived benefits of a new media technology are equally important, if not more important, than objective attributes in influencing adoption. Rogers (1995) listed five key perceived benefits of an innovation: 1) relative advantage (the degree to which an innovation is perceived as better than the idea it supersedes), 2) compatibility (the degree to which an innovation is perceived as being consistent with existing values, past experiences, and needs), 3) complexity (the degree to which an innovation is perceived as difficult to understand and use), 4) trialability (the degree to which an innovation may be experimented with on a limited basis), and 5) observability (the degree to which the results of an innovation are visible to others). These benefits apply in the context of wireless Internet at its first stage of deployment, especially compatibility and observable results.

Others have suggested that adoption and use of new information technologies may be considered an indication of social status, and thus that image (the degree to which use of the innovation is perceived to enhance one's image in one's social unit) should be a perceptual factor (Tornatzky & Klein, 1982; Zhu & He, 2002). Agarwal and Prasad (1997) proposed that these perceptual factors were the major predictors of initial use of the World Wide Web. In their study, higher benefit evaluations led to a stronger intention to continue using the Web.

In the adoption of personal computers at home, Lin (1998) found that relative advantages were a significant predictor of adopting home computers when other predictors such as cost were controlled. A study of Internet use among academics (Busselle, et al., 1999) found significant differences in perceptions of the benefits of Internet between light and heavy users. Those who were more positive in viewing the Internet as less complex and more advantageous tended to use it more. Complexity and observability were also found to be factors influencing the adoption of wireless telephony in Hong Kong (Leung & Wei, 1999), and Internet adoption studies in China reported similar results (Zhu & He, 2002). Internet users perceived the Internet favorably in terms of such benefits as advantage, compatibility, ease of use, and demonstrability of results. The above research led to the fourth hypothesis:

Social Influences

The adoption process is considered primarily a communication process in which various forms of social influence are at work (Rogers, 1995). Rice (1993) proposed that social influences may be particularly relevant in the early stages of adoption when uncertainty is high. Social influences originate from peer groups, co-workers, schools, families, or other salient social groups, and may take the form of social support or social pressures (Markus, 1994; Steinfield, 2002). Social influences have also been described as producing a bandwagon effect (Zhu & He, 2002).

Peer use, consulting, and advice were found to be significant factors affecting information system adoption in organizations in an empirical study by Leonard-Barton and Deschamps (1988). Schmitz and Fulk (1991) found that attitudes toward, and use of, email by surveyed respondents' close colleagues had a positive impact on their attitudes and use. Perceived popularity of the Internet was found to be a major predictor of using the Internet in China (Zhu & He, 2002). The more Internet users in an individual's family and in the same profession, the more likely he or she would be to adopt it. Rhee and Kim (2004) examined social influences on Internet adoption in South Korea and also found that social support of family members was positively related to Internet use. In a case study of IT adoption at a U.S. university, McMillan and Hyde (2000) highlighted the pressure from administrators and colleagues on faculty for adoption. The present study focused on family and peers as the primary groups influencing the adoption and use of Wi-Fi in the workplace. It was hypothesized that:

Further, change agents who introduce an innovation into a social system are an important source of social influence in the diffusion process. Rogers (1995, p. 313) characterized change agents as "linkers" between change agency and client system (adopters). They are responsible for forwarding feedback from potential adopters to decision makers. At the same time, they are particularly influential in forming and changing attitudes toward innovations and converting non-adopters into adopters. Thus, contacts with change agents are considered to be a predictor of adoption of an innovation. The generalization is that the more contacts with change agents, the higher the adoption likelihood (Rogers, 1995). In addition, early adopters tend to have a greater number of change agent contacts than late adopters (Rogers, 1995). Talks with salespeople about cell phones (Leung & Wei, 1999; Vishwanath & Goldhaber, 2003; Wei, 2001) have also been found to correlate positively with using cell phones. It was thus hypothesized that:

Influences of Functionally-Similar Technologies Adopted

Adoption research examining the interdependent relationships among a set of innovations that diffuse concurrently indicates that adoption of one information technology is related to the adoption of another that is functionally similar (Atkin, 1993; Reagan, 1987). Rice and Webster (2002) also proposed that use of one or more media is an influence on adoption and use of other media in organizational settings. Rogers (1995) proposed the construct of technology cluster to explain adoption behavior. He defined the construct as a cluster that "consists of one or more distinguishable elements of technology that are perceived as being closely interrelated" (p. 15).

In an empirical study, Reagan (1987) reported that adoption of innovations such as personal computers predicted adoption of other media innovations. Similar findings were reported for the adoption of computers at home (Dutton, et al., 1987). Using a technology adoption index based on 14 media technologies (including cable TV, word-processing systems, and video game systems), Lin (1998) found that users of personal computers owned more communication technologies than non-users, and that the index was the strongest predictor of adoption likelihood. In studies of Internet adoption, people owning more technologies were found to be heavier Internet users (Busselle, et al., 1999; Lin, 2001). In organizational settings, Markus (1994) found that emailing was related to increased use of the telephone.

Considering that the wireless Internet upgrades and expands existing wired networks, adoption of computers (particularly laptops and PDAs) and prior access to the Internet would be logical factors that must be in place. Thus, it was anticipated that the more information technologies employees own, the higher their intention to adopt and use the Wi-Fi powered WLAN (wireless local are network) in the workplace.

Impacts of Media Use

The influence of mass media use has been articulated in the diffusion research paradigm. The effects of media exposure may be particularly strong at the earliest stages of the adoption process in creating awareness of an innovation. In general, early users tend to be heavier users of mass media than are later users (Rogers, 1995). Past adoption research focused on level of media use and reported mixed findings. A study of adoption of home computers (Lin, 1998) found that TV viewing was a significant predictor, but use of radio, magazines, and newspapers was not. Other studies (Busselle, et al., 1999) found no relationship between use of mass media and computer use. The same pattern was found in research on Internet adoption; media use variables failed to show predictive power (Rhee & Kim, 2004; Zhu & He, 2002). A recent study on adoption of online applications (Lin, 2001) found reading newspapers to be a significant but negative predictor of adopting communication-related applications, while reading magazines was a significant predictor of adopting marketing-oriented services.

Considering that the diffusion of wireless Internet is at an early stage, and given the wide media coverage of Wi-Fi projects in cities like Philadelphia and San Francisco, it was hypothesized that:

Whether use of the Internet displaces or supplements use of traditional media remains inconclusive, but past research on effects of broadband use on narrowband Internet in Japan (Ishii, 2004) found that broadband users spent more time using the Internet per week than did narrowband users. Also, broadband users used more advanced online services. Considering that the wireless Internet network supplements the existing wired network, it provides more time-enhancing opportunities for users to go online. Therefore, it was anticipated that:

Finally, while all of the influences reviewed above contribute to explaining technology adoption through their respective net effects, it is theoretically interesting to explore which influence makes a greater contribution in multivariate analyses. Therefore, the relative contributions of all predictors (e.g., demographics, innovativeness, importance of wireless Internet attributes and perceived benefits, social influences, and media use) to account for the variance in adoption of Wi-Fi was explored in the following research question:

Method

A probability sample of employees at a public university in the southeastern United States was used to collect data for hypothesis testing. The university was chosen as the site for data collection because it deployed a Wi-Fi powered Wireless Local Area Network in 2004. The wireless network was campus wide, available to all faculty and students in buildings for common use such as the library, the student union, and the stadium, as well as other buildings designated for teaching and faculty office. Using the 2004-2005 employee directory as the sampling frame, every other academic employee in the university's 14 colleges and schools was systematically selected, for a total of 559 employees. They represented all the departments and programs including liberal arts, business, medicine, nursing, science, engineering, and law. Face-to-face interviews were supplemented with a mail survey. Trained undergraduate students interviewed the selected faculty from November 2004 to January 2005. A total of 268 respondents completed the survey, yielding a response rate of 47.9%.

The mean age of employees in the sample was 47.8 with a standard deviation of 10.9. In terms of gender, 67% were male and 33% were female. Most respondents (88.3%) had a Ph.D. or J.D. degree; 9.7% held a Master's degree. The remaining 1.9% had completed a B.A. degree as their highest level of education attainment. A substantial majority (86.3%) of the respondents were white. Asian Americans were second (7.9%), followed by African-Americans (2.5%) and Hispanics (1.2%). By checking with the university faculty database, we determined that the sample is representative of the population.

Measures

Adoption of Wireless Internet

Following Lin's (1998) three categories of users (e.g., non-users, likely users, and users), two questions were used to measure the status of wireless Internet adoption. First, respondents were asked if they had used the wireless Internet at work using a dichotomous scale (1=yes, 2=no). Those who answered yes were requested to report the length of use in months. Those who answered no were further asked about their likelihood of using it in six months on a 1-5 point scale where 1 meant "not likely" and 5 meant "very likely." Respondents who answered "somewhat or very likely" were classified as "likely users;" the rest were grouped as "non-users."

Individual Influences

Measures of demographic characteristics of respondents included age, gender, education, and race. Personal traits in terms of innovativeness were also measured. Conceptually, personality traits refer to people's inherited dispositions toward new ideas and risk taking (Lin, 1998). Operationally, the following four items were employed as measures: (1) willingness to learn new ideas, (2) willingness to explore new technology, (3) ability to keep up with computer technology, and (4) willingness to take risk. The 5-point scale ranged from 0 to 4, where 0 meant "least characteristic of me" and 4 meant "most characteristic of me." Results of an exploratory factor analysis showed a one-factor solution (Eigenvalue=3.02, accounting for 75.41% of variance). These results were combined into a composite scale after achieving sufficient reliability (M=3.87, SD=.83; a=.89).

Importance of Wireless Internet Attributes

Respondents were asked to rate the importance of being free of wires, mobility (Internet access anywhere, anytime, on campus), and versatility (the ability to connect using various devices). The scale ranged from 1 "not important at all" to 5 "extremely important."

Perceived Benefits of Wireless Internet

On a 5-point Likert scale where 1 meant "completely disagree" and 5 meant "completely agree," multiple items were used to measure the following five dimensions of perceived benefits of wireless Internet: 1) relative advantage (i.e., Wi-Fi can make me more productive. Wi-Fi will enhance the quality of overall Internet use. Using Wi-Fi will save me time. Wi-Fi frees me up from a fixed location to go online. Wi-Fi will make going online very convenient.), 2) compatibility (i.e., Use of Wi-Fi suits the way I work, and my lifestyle. Wi-Fi offers me more options to go online and seamless connectivity with wired networks.), 3) ease of use (i.e., Wi-Fi is easy to learn and use. It will be difficult to learn how to use Wi-Fi. Instructions for using Wi-Fi are hard to follow. It will be too much trouble to use WI-FI.), 4) observability (i.e., It's easy to explain the benefits of Wi-Fi. The benefits of Wi-Fi are obvious.), and 5) social status (i.e., People who use Wi-Fi appear to be superior to others. Use of Wi-Fi has become a symbol of social status.). Each of the five subscales showed a high internal reliability, providing the basis for creating five composite subscales. The surveyed respondents valued the benefit of relative advantage the most (M=3.91, SD=.84, a=.88), followed by observable results (M=3.85, SD=.86, a=.82), compatibility (M=3.72, SD=.90, a=.82), ease of use (M=3.76, SD=.72 a=.72), and social status (M=2.21, SD=.89, r=.34, p<.001).

An examination of the inter-correlations of the five subscales indicated that relative advantage and compatibility were highly correlated (r=.90), indicating that the respondents made no distinction between relative advantage and compatibility. To avoid multicollinearity in subsequent analyses, the two subscales were combined into a single composite scale to represent advantage and compatibility (M=3.81, SD=.84, a=.94). The rest were retained as separate subscales.

Technology Cluster

Respondents were given a list of 11 information technologies including desktop, laptop, pocket PC, PDAs, cell phone, broadband, DVD player, video game player, digital camera, and fax. The scale was dichotomous (1=yes, 2=no). The total was then summed to indicate the extent of a respondent's ownership of information technologies. The index ranged from 0 to 11, with 0 indicating owning none of the above 11 information technologies, and 11 indicating owning all of the 11 information technologies (M=7.84, SD=2.72, a=.72).

Social Influences

Social influences on diffusion of new information technology were measured in three questions. First, respondents were requested to report the number of family members using wireless Internet. They were then asked to estimate the number of colleagues in their department and at the university going online via the new wireless Internet network. The intercorrelations among the three measures were low; therefore they were used as three separate variables in subsequent analyses. In addition, respondents were asked to report if they had talked with a technician about wireless Internet at the university as a measure of change agent contact on a dichotomous scale (1=yes, 2=no).

Media Use

Respondents were requested to report the number of hours per day that they spent watching television, and number of days per week that they spent reading a newspaper and a magazine. In addition, the average time in hours per day spent surfing the World Wide Web via a wired network was measured.

Findings

Descriptive Results

Most of the 268 respondents had a personal desktop computer (89.1%) and a laptop computer (76.8%). Fewer had a PDA (23.2%) or a pocket PC (20.2%). On average, the respondents spent 4.48 hours each day online via the organization's wired network, ranging from 0 to almost 24 hours. While online, they received an average of 40.51 (SD=44.72) emails and sent an average of 18.9 (SD=20.61) emails per day. The most time the respondents spent on online activity in the workplace besides emailing (M=4.64 hours with a SD of 5.58) was searching for information for teaching and research (M=3.58 hours with a SD of 4.82), while the least time was spent playing video games online (M=.05 hours with a SD of .28).

More than half (64.8%) of the respondents had heard of the new wireless Internet system at the organization, but only 21.4% had used it. This distribution fits the profiles of innovators and early adopters in the adoption curve. Using the wireless Internet, they checked email, surfed for teaching and research information, and accessed the university databases the most often. The rest of the respondents were classified as likely users (55.2%) and non-users (23.4%). The number of likely users is comparable to the early and late majority as classified by Rogers (1983), while the number of non-users is proportional to laggards in Rogers’ classification.

Mobility was rated the most important attribute of wireless Internet (M=4.29, SD=.99). Internet access without the use of wires was second (M=4.24, SD=.99), and the ability to use various devices to connect was considered the least important (M=3.95, SD=1.15).

Results of Hypothesis Testing

Taking the three groups of early users, likely users, and non-users as the grouping variable, a series of one-way ANOVA tests was run to test the hypotheses. H1 predicted that early and likely users of wireless Internet would be younger, better educated, and more often male than would non-users. As results in Table 1 show, the three types of users differed significantly in age, with early users being the youngest (46.27 years old) and non-users being the oldest (51.22 years old; F=3.68, p<.05). Results of post-hoc tests indicate that age differences between early users, likely users, and non-users were significant. Thus, age turned out to be a key influence in adoption of Wi-Fi in this workplace. This is consistent with earlier studies of personal computers (Dickerson & Gentry 1983; Lin 1998) and Internet adoption (Atkin, et al., 1998; Lin 2001) in the general population and the use of the Internet among academics (Busselle, et al., 1999; Jones & Johnson-Yale, 2005).

A Chi-square test between gender and the three types of users showed that gender and adoption status were not related. Nor were any differences found among the respondents in education. This is not surprising, since the respondents were from a population of highly-educated academics. H1 thus received partial support.

It was anticipated in H2 that early and likely users of wireless Internet would consider themselves to be more innovative than would non-users. When the three groups were compared on level of innovativeness, significant differences were found. Early users scored the highest on being innovative, followed by likely users and non-users (F=17.09, p<.001). Post-hoc tests indicate that the group-based differences in innovativeness were significant. The mean rating of innovativeness of early and likely users differed significantly from that of non-users. H2 was thus supported.

H3 predicted that early and likely users of wireless Internet would rate the attributes of wireless Internet such as freedom from wires, mobility, and versatility higher than would non-users. Based on ANOVA test results, the three groups differed significantly. Early users rated freedom from wires as more important than did likely users, whose mean ratings were higher than those for non-users (F=21.26, p<.001). Post-hoc mean tests show that the difference between early and likely users was not significant, but the differences between early users, likely users, and non-users were significant. Similar patterns were found in rating the importance of mobility (F=5.81, p<.01) and versatility (F=13.35, p<.001). Early users had the highest mean ratings; non-users had the lowest ratings. Non-users perceived mobility and the ability to use various devices to go online as less important than did early and likely users. The differences were significant based on post-hoc tests (see Table 1). These results supported H3.

H4 predicted that early and likely users of wireless Internet would perceive the benefits of wireless Internet to be higher than would non-users. Results of ANOVA tests on the relationships between perceived benefits and the three types of users show that early users, likely users, and non-users differed significantly in their perceptions of advantage-compatibility (F=58.06, p <.001), ease of use (F=14.57, p <.001), observable results (F=30.53, p<.001), and social status (F=6.24, p<.05). Post-hoc tests were performed to further examine the differences among the three groups. As shown in Table 1, early users' perceptions of the benefits of advantage-compatibility, ease of use, and observable results were the highest, followed by the perceptions of likely users and non-users. Post-hoc tests show that the differences between users, likely users, and non-users were significant. On the dimension of social status, however, likely users were the most positive, while non-users were the least positive. The mean perceptions of early users fell in between. The difference between the likely users and non-users was significant based on post-hoc tests, but early users and non-users did not differ in their perceptions of using wireless Internet as a symbol of social status. H4 was basically supported.

H5a predicted that the effects of social influence in going online wirelessly would be greater on early and likely users of wireless Internet than on non-users. This hypothesis was supported. As results of ANOVA tests in Table 1 show, early users reported having the largest number of family members, department colleagues, and university employees already using a wireless network, followed by likely users and non-users (F=17.78, p<.001 for family members; F=32.75, p<.001 for colleagues, F=6.62, p<.001 for university employees). Additional post-hoc tests revealed that the group differences along the three measures of social influences were significant. The number of people using wireless Internet in the family, at the department and in the university for early users and likely users differed significantly from that for non-users. These results suggest that the social influences of family and peers were strongest on early users.

H5b predicted that early and likely users of wireless Internet would contact change agents more often than would non-users. Based on ANOVA results, early users talked with technicians about wireless Internet 2.5 times more frequently than did likely users, and 25 times more often than did non-users (F=23.18, p<.001). Post-hoc tests show the differences among the three groups were significant. This result, consistent with previous studies, underscores the critical role of change agents in the adoption of wireless Internet. H5b was supported.

To test the effects of technology cluster on use of wireless Internet, H6 proposed that early and likely users of wireless Internet would own more information technologies than would non-users. Results showed that technology cluster was positively related to status of wireless Internet adoption (F=6.01, p<.01). Both early users and likely users owned more information technologies than did non-users. To better understand the differences among the three groups, post-hoc tests were run. Results indicate that the differences in owning information technologies among early users, likely users, and non-users were significant. H6 was supported.

H7a predicted that early and likely users of wireless Internet would use mass media more heavily than would non-users. As results of ANOVA tests show, the only significant difference in use of mass media among early and likely users and non-users was reading newspapers (F=3.92, p<.05). According to post-hoc tests, non-users spent more time reading newspapers than did likely adopters. In contrast, the difference between early users and non-users was non-significant. H7a was thus partially supported.

H7b predicted that early and likely users of wireless Internet would spend more time using the Internet via wired networks than would non-users. Early users of wireless Internet spent an average of 5.17 hours online at the workplace per day, compared with 4.38 hours spent by likely users and 3.67 hours by non-users. Nevertheless, the differences were not significant. H7b was rejected.

Finally, to address the research question concerning the relative contributions of individual influences, attributes and benefits of wireless Internet, social influences, technology cluster, and media use in predicting the likelihood of using wireless Internet, a hierarchical multiple regression was performed treating the likelihood of using wireless Internet as the dependent variable. The predictors were entered hierarchically. The results of this analysis are presented in Table 2.

The multivariate model yielded five significant predictors for predicting the likelihood of likely users and non-users converting in six months. Among these, perceived benefit of advantage-compatibility (B=.43) was the strongest, followed by talks with technicians about wireless Internet (B=.15), number of estimated colleagues using wireless Internet (B=.13), number of family members using wireless Internet (B=.12), and reading newspapers (B=-.11). The results indicate that the more respondents considered going online wirelessly to be an advantage and to be no different from using the wired network, the more often they talked with technicians about wireless Internet, and the more colleagues and family member already used wireless Internet, the more likely they would become users in the next six months. However, the more time they spent reading newspapers, the less likely they were to convert.

The contribution of the wireless Internet attributes block was the greatest (15.4%), while the contribution of the media use block was the smallest (1.2%). The perceived benefits block and the individual influences block contributed 14.5% and 10.1%, respectively. As the results in Table 2 show, the unique contribution of social influence as a block is almost five times that of the mass media block (5.8% vs. 1.2%). This result supports the generalization that personal communication channels are more important (and more effective) in converting non-adopters than are the mass media. All together, the model accounted for a total of 44.7% (adjusted R square) of variance. In general, these results show that the attributes and perceived benefits of Wi-Fi were more important than the influences of individual, social, and media use in predicting the likelihood of using Wi-Fi powered networks.

Discussion

A major objective of this study is to seek an understanding of the diffusion pattern in the case of adopting wireless Internet in the workplace. As reported earlier, 21.4% of the surveyed academic employees were found to be early users, 55.2% likely users, and 23.4% non-users. They fit the adopter categories of innovators, early adopters, early and majority, and laggards proposed by Rogers (1995). As time progresses, the adoption of wireless Internet will flow from early users to likely users, and then from likely users to non-users. Thus, a diffusion curve exists in the diffusion of a Wi-Fi powered WLAN.

Theoretically, the results of this study demonstrate that the adoption of high-speed wireless Internet in the workplace is subject to the joint influences of individual, social, and technological factors. The role of the objective attributes of wireless Internet and subjective evaluations of its benefits are particularly influential in the adoption and use of this new Internet technology. The more important the respondents rate freedom from wires, versatility, and mobility, the more likely they are to be early and likely users. Further, the higher the perceived benefits of advantage-capability, observable results, ease of use, and social status, the more likely they are to be early users or likely users.

Similar to studies of the adoption of computers and the Internet at home (Dutton, et al., 1987; Lin, 1998, 2001), the results of this study provide solid evidence in support of the impact of individual (i.e., age and innovativeness in particular) and social influences on the likelihood of accessing the Internet wirelessly in the workplace. Early and likely users are younger employees with a higher level of innovativeness; they talk with technicians about wireless Internet more often than non-adopters, and report a larger number of people around them in family and at work as users of wireless systems. Finally, the more functionally-similar information technologies respondents own, the higher their likelihood of adoption and use of wireless Internet in the workplace.

These results underscore the key role of advantage compatibility, change agent contacts, and social influences from family and colleagues in the diffusion of wireless Internet in organizations. The influence of media use on Wi-Fi adoption was limited because only reading newspapers turned out to be a significant predictor. In fact, reading newspapers is negatively related to the likelihood of using Wi-Fi. Although this result is consistent with previous findings for the adoption of online services (Lin, 2001), it failed to support the generalization that mass media play a positive role in the adoption process (Rogers, 1995). Three explanations possibly account for this particular finding. First, this study focused on the adoption likelihood of Wi-Fi as the dependent variable instead of a series of dependent variables such as awareness, interest, and trial, which are the critical steps in the adoption process articulated by Rogers. As he put it (Rogers, 1986), adoption involves "a process that occurs over time and consists of a series of actions" (pp. 163-164). Studies have shown that the impact of media use on awareness, interest, and adoption decision differed (e.g., Dupagne, 1999). Thus, reading newspapers may have an indirect effect on Wi-Fi adoption, as compared to a direct effect on awareness of Wi-Fi.

Second, from a media displacement perspective, use of computers leads to a decrease in use of mass media, particularly TV viewing (Atkin, et al., 1998; Lin, 2001; Vitalari, Venkatesh, & Gronhaug, 1985). Thus, the negative role of reading newspapers on Wi-Fi adoption makes sense in that newspaper use displaces time spent online. Respondents who have a strong habit of reading newspapers are less likely to consider using wireless Internet. Third, measuring media use as an aggregated activity may confound the results. This study did not link media use to content (e.g., news vs. entertainment). It is plausible that the more respondents read about IT news, the more aware and knowledgeable they will be, and hence more likely to consider using a new technology. Thus, a content specific measure would be desirable to test the effect of media content on adoption.

Overall, when all influences are considered together in the hierarchal regression model, the model offers some insights into the relative contributions of the studied factors to the adoption of wireless Internet. Higher perceptions of advantage and compatibility of wireless Internet, more colleagues and family members already using wireless Internet, higher frequency of talks with technicians about the wireless system, but less time spent reading newspapers, all lead to a higher likelihood of becoming a wireless Internet user in the near future. The contributions of technological attributes and perceived benefits are greater than individual and social influences. The study thus expands the utility of the diffusion of innovations model to the adoption of new Internet technologies in the workplace.

The findings of this study have managerial implications for how organizations can make full use of expanded communication infrastructures. Specifically, results of this study produced evidence in support of the "assimilation" or "adoption vs. use" gap in organizations (Fichman, 1995). Organizational leaders cannot assume that "if you build it, they will use it." Only one-fifth of respondents used Wi-Fi at the time of this study. To narrow the gap, organizations can tap into the role of younger and better-educated employees as influencers to convert coworkers. In addition, tech-support personnel can influence adoption by hosting hands-on demonstrations in the workplace. These demonstrations could be incorporated into training workshops for employees about the tangible benefits of using the wireless Internet system.

This study has several limitations. The results should be interpreted with some caution due to the use of a relatively small sample from a single organization in higher education. A sample of highly-educated employees limits the generalizability of the results to non-academic organizations. Future research should consider larger samples of different types of organizations such as businesses and government agencies. In addition, the significant relationships between the studied predictors and Wi-Fi use should be interpreted with caution. For example, consider social influence on adoption likelihood. It is not clear whether respondents used WI-Fi first at work and then their family members started to use it at home, or vice versa. The one-shot survey design makes it difficult to ascertain bi-directional relationships or draw any causal conclusions. Further research should use longitudinal studies to establish causal relationships in predicting wireless Internet adoption.

Moreover, measures of social influence focused narrowly on numbers of Wi-Fi adaptors around respondents, stopping short of exploring the mechanisms of peer influence at the workplace. Broad measures are needed in further studies to include behavioral dimensions of social influence, such as experience in working together with a colleague using Wi-Fi and experience in sharing wireless Internet access with an adopter at work. Also, the measures of mass media use did not differentiate time spent attending to specific media content. Future research should measure both general media use and attention to media content devoted to news and entertainment. Additionally, due to a limit on questionnaire length, this study did not include other corporate control factors that might explain Wi-Fi adoption likelihood. For instance, no question was asked of respondents regarding whether their department paid for wireless cards or provided computers with wireless capability. Such institutional support would facilitate adoption.

Finally, Rice and Weber (2004, p. 207) proposed that "contextual influences" such as circumstances of use, time pressure, and geographic distances also affect the adoption of an information technology in organizations. To pursue a fuller model of Wi-Fi adoption, follow-up studies should incorporate these variables to measure how the actual tasks and situations in which Wi-Fi is thought to be useful influence the adoption of Wi-Fi powered wireless networks. In addition, given strong social influences on adoption in the workplace, future research could expand the model tested here by considering additional factors such as opinion leadership in order to increase the total variance explained by the model.

Acknowledgments

I am grateful to the Office of Information Technology (OIT) at the University of South Carolina directed by Dr. William Hogue, Vice President for Information Technology and Chief Information Officer, for a grant to conduct this study.

References

Agarwal, R., & Prasad, J. (1997). The role of innovation characteristics and perceived voluntariness in the acceptance of information technologies. Decision Science, 28 (3), 557-582.

Atkin, D. (1993). Adoption of cable TV amidst a multimedia environment. Telematics & Informatics, 10 (1), 51-60.

Atkin, D., Jeffres, L., & Neuendorf, K. (1998). Understanding Internet adoption as telecommunications behavior. Journal of Broadcasting & Electronic Media, 42 (4), 475-490.

Best, J. (2005, September). 100,000 for worldwide Wi-Fi this year. Silicon News. Retrieved on January 17, 2006 from http://networks.silicon.com/mobile/0,39024665,39152752,00.htm

Busselle, R., Reagan, J., Pinkleton, B., & Jackson, K. (1999). Factors affecting Internet use in a saturated-access population. Telematics & Informatics, 16 (1), 45-58.

Crispell, D. (1994). Computers at home. American Demographics, 16 (2), 59.

Dell'Oro Group Report. (2005, August). Wireless LAN Market to Double. Retrieved on January 17, 2006 from http://www.delloro.com/news/2005/WL080305.htm

Dickerson, M. D., & Gentry, J. W. (1983). Characteristics of adopters and non-adopters of home computers. Journal of Consumer Research, 10 (2), 225-235.

Dupagne, M. (1999). Exploring the characteristics of potential high-definition television users. Journal of Media Economics, 12 (1), 35-50.

Dutton, W. H., Rogers, E. M., & Jun, S. H. (1987). Diffusion and social impacts of personal computers. Communication Research, 14 (2), 219-250.

Fichman, R. G. (1995). The Illusory Diffusion of Innovation: An Examination of Assimilation Gaps. Unpublished paper, MIT Sloan School of Management, Cambridge, MA.

Fulk, J. (1993). Social construction of communication technology. Academy of Management Journal, 36 (5), 921-950.

Goodhue, D. L., & Thompson, R.L. (1995). Task-technology fit and individual performance. MIS Quarterly, 19 (2), 213-233.

Hiltz, S. R., & Johnson, K. (1989). Measuring acceptance of computer-mediated communication systems. Journal of the American Society for Information Science, 40 (6), 386-397.

Ishii, K. (2004). Internet use via mobile phone in Japan. Telecommunications Policy, 8 (1), 43-58.

Jones, S., & Johnson-Yale, C. (2005). Professors online: The Internet's impact on college faculty. First Monday, 10 (9). Retrieved on May 9, 2006 from http://www.firstmonday.org/issues/issue10_9/jones/index.html

Katz, J., & Rice, R. (2002). Social Consequences of Internet Use: Access, Involvement, and Interaction. Cambridge, MA: The MIT Press.

LaRose, R., & Hoag, A. (1996). Organizational adoptions of the Internet and the clustering of innovations. Telematics and Informatics, 13 (1), 49-61.

Leonard-Barton, D., & Deschamps, I. (1988). Managerial influence in the implementation of new technology. Management Science, 34 (10), 1252-1265.

Leung, L., & Wei, R. (1998). Exploring factors of Interactive TV adoption in Hong Kong: Implications for advertising. Asian Journal of Communication, 8 (2), 124-147.

Leung, L., & Wei, R. (1999). Who are the mobile phone have-nots? Influences and consequences. New Media & Society, 1 (2), 209-226.

Lin, C. A. (1998). Exploring the personal computer adoption dynamics. Journal of Broadcasting & Electronic Media, 42, 95-112.

Lin, C. A. (2001). Audience attributes, media supplementation, and likely online service adoption. Mass Communication & Society, 4 (1), 19-38.

Markus, M. L. (1994). Electronic mail as the medium of managerial choices. Organization Science, 5 (4), 502-527.

McMillan, J., & Hyde, M. (2000). Technological innovation and change: A case study in the formation of organizational conscience. Quarterly Journal of Speech, 86 (1), 19-47.

Midgley, D., & Dowling, G. R. (1978). Innovativeness: The concept and its measurement. Journal of Consumer Research, 4 (4), 229-242.

Reagan, J. (1987). Classifying adopters and non-adopters for technologies using political activity, media use and demographic variables. Telematics & Informatics, 4 (1), 3-16.

Rhee, K., & Kim, W. (2004). The adoption and use of the Internet in South Korea. Journal of Computer-Mediated Communication, 9 (4). Retrieved on October 27, 2005 from http://jcmc.indiana.edu/vol9/issue4/rhee.html

Rice, R. (1993). Using network concepts to clarify sources and mechanisms of social influence. In W. Richards Jr. & G. Barnett (Eds.), Advances in Communication Network Analysis (pp. 43-52). Norwood, NJ: Ablex.

Rice, R., & Webster, J. (2002). Adoption, diffusion and use of new media. In C. A. Lin & D. Atkin (Eds.), Communication Technology and Society: Audience Adoption and Uses (pp. 191-227). Cresskill, NJ: Hampton Press.

Rogers, E. M. (1986). Diffusion of Innovation (3rd ed.). NY: Free Press.

Rogers, E. M. (1995). Diffusion of Innovation (4th ed.). NY: Free press.

Rogers, E. M., & Shoemaker, F. F. (1971). Communication of Innovations: A Cross-Cultural Approach (2nd ed.). NY: Free Press.

Schmitz, J., & Fulk, J. (1991). Organizational colleagues, information richness and electronic mail: A test of the social influence model of technology use. Communication Research, 18 (4), 487-523.

Steinfield, C. (2002). Organizations and external telecommunications networks: The adoption of wide area networks and their impact on organizational structure and strategy. In C. A. Lin & D. Atkin (Eds.), Communication Technology and Society: Audience Adoption and Uses (pp. 229-254). Cresskill, NJ: Hampton Press.

Tedeschi, B. (2004, September 27). What would Benjamin Franklin say? Philadelphia plans citywide free Wi-Fi Internet for computer users. The New York Times, C8.

Tornatzky, L. G., & Klein, K. J. (1982). Innovation characteristics and innovation adoption-implementation: A meta-analysis of findings. IEEE Transaction on Engineering Management, 29 (1), 28-45.

Vishwanath, A., & Goldhaber, G. (2003). An examination of the factors contributing to adoption decisions among late-diffused technology products. New Media & Society, 5 (4), 547-572.

Vitalari, N. P., Venkatesh, A., & Gronhaug, K. (1985). Computing in the home: Shifts in the time allocation patterns of households. Communications of the ACM, 28 (5), 512-522.

Webster, J. (1998). Desktop videoconferencing: Experiences of complete users, wary users, and non-users. MIS Quarterly, 22 (3), 257-286.

Wei, R. (2001). From luxury to utility: A longitudinal analysis of cell phone laggards. Journalism & Mass Communication Quarterly, 78 (4), 702-719.

Zhu, J., & He, Z. (2002). Perceived characteristics, perceived needs, and perceived popularity: Adoption and use of the Internet in China. Communication Research, 29 (4), 466-495.

About the Author

Ran Wei is an associate professor in the School of Journalism & Mass Communications at the University of South Carolina. His research interests focus on the diffusion, use, and impact of new media technologies.
Address: 4008 Carolina Coliseum, Columbia, SC 29208 USA