By Laurie Phillips, Daniel Riffe and Robert McKeever
WJMCR 47 (June 2014)
Introduction | The Digital Divide | Specialized Information | Environmental Risk | Hypotheses | Method | Discussion and Conclusion
Using statewide telephone survey data from North Carolinians (N=406), this digital divide study oversampled rural households to explore urban rural differences in Internet access, use, and information seeking pertaining to environmental risk. The data suggest that the urban rural access divide has closed, and that urban and rural residents are making similar use of the Internet, particularly for environmental risk information. Moreover, survey respondents deemed the Internet more useful for environmental risk information than either television news or newspaper sources, but Internet use was not a predictor of respondents “sense of ‘environmental confidence,” a new variable introduced to the digital divide and online information seeking literatures through this study.
For years scholars have lamented the access gap between urban and rural residents and technology “haves” and “have nots” when discussing the digital divide and Internet access.1 Of course, urban rural access gaps mirror similar historical differences between urban and rural residents, such as their views of technology (the “rural mindset”),2 racial and poverty disparities,3 access to health care, transportation, and educational opportunity,4differences in income and economic well being,5 and exposure to kinds of environmental hazards or risk,6 like concentrations of solid waste facilities,7 waste from industrial scale agriculture,8 and resulting contamination of water and air.9 Wilhelm warned that many of the digital “have nots” are disproportionately poor, rural, minority, and older, and they are also denied “job opportunities, social service information, and lifelong learning opportunities that build the capacity of all Americans.”10Consider the rapid growth of online information seeking and associated research. A 2001 national survey found that 40% of respondents with Internet access sought information about health online.11 By 2004, Luo and Najdawi noted that 40% of Americans used online health information to make health care decisions,12 while Pew projected that 113 million American adults sought online health information and that for 20% of adults the Internet was their main source for science news.13 Shortly after, scholars found that 50% of national survey respondents went online “at least once a week” for health information, 40% sought science news online,14 and 12% 13% went online for news about national or international environmental problems.15 Moreover, recent research has shown that 22% of Internet users search online for information about environmental health hazards,16 while other scholars note the overarching challenges associated with research about online news seeking.17 This shift in perceived Internet utility is rendered more evident when contrasted with a study examining Internet related beliefs of rural Ohioans, which found striking numbers of digital””have nots” unable to conceive of using the Internet to conduct a variety of tasks. Though the study was conducted just one decade ago, when respondents were asked what they would do with Internet access, only 22% envisioned looking for employment, 24% making social connections, and 25% seeking news online compared to 46% seeking news among respondents already enjoying home access.18 While Zickuhr found that 15% of American adults are still not online,19 Pew Research Center’s 2013 Home Broadband Report found that “70% of American adults now have a high speed broadband connection at home, and the urban rural gap has nearly closed: 70% of urban adults have home access compared to 62% of rural adults.20 Early concern over the digital divide forecast just these kinds of urban rural disparities in information access.21 Although today Internet is part of American life no matter one’s locale,”22 and the range of information, big data, and other opportunities afforded by Internet access increases exponentially, questions remain about how rural residents use the Internet and other media sources to serve their needs, particularly when rural residence is associated with what environmental justice advocates see as an inequitable burden of environmental health risks. To echo Riffe, Lacy and Varouhakis, whether basic Internet access and potential information access vary along social structural lines” remains an important question for study.23 Accordingly, the purpose of this media dependency study is three fold: 1) to explore the existence and nature of any remaining urban rural digital divide; 2) to examine urban and rural differences in how people use the Internet and other media sources for seeking information about environmental risks, risks that many see as disproportionately concentrated in rural areas;24 and 3) to determine how these factors relate to “environmental confidence” (knowledge and sense of empowerment). Data are from a telephone survey of North Carolinians (N=406) in which rural counties were deliberately oversampled because of the concentration of environmental problems in those counties.25
The “Digital Divide” and Internet Access
As Dillman noted,26 the digital divide has been studied by sociologists for nearly forty years, but the digital information ageís effects on rural America have roots that extend to the Industrialization era of the late 19th century.27 Nearly a quarter century ago before Internet access became a widespread household commodity, Dillman predicted that technology would ease tensions between urban and rural life;28 geography would no longer be a limiting factor and many would move into rural America if amenities were made available, a sentiment echoed by Johnson,29 despite rural residents’ tendency to adopt technologies at a slower pace than urban residents. Recently, Pew found lingering racial and ethnic differences in home Internet access: 74% of non Hispanic whites are broadband users, compared to 64% of non Hispanic African Americans and 53% of Hispanics. When accounting for smartphones, racial and ethnic differences tell a different story: 80% of non Hispanic whites have home broadband or smartphone Internet access, compared to 79% of non Hispanic African Americans and 75% of Hispanics.30 Previously, a study of North Carolinians found that race predicted home Internet access, with rural African Americans nearly 30 percentage points below whites. Gender differences also existed, with males more likely to report home Internet access,31 and other studies revealed that urban rural differences in age, income, education,32 and household size led to lower rural Internet access rates and “digital inequality.”33 Echoing previous findings, Hindman uncovered an increasing access gap between urban and rural residents over a decade ago, though income, age, and education were the strongest predictors of access disparities ñ not geographic location.34
Seeking Specialized Information Online
While previous research leaves open the question of whether urban rural digital divides still exist in various places, the question also remains of how such a divide whatever its current “width” relates to people’s use of technology for specialized information types. Recall that Riffe found variation within 13 rural counties he surveyed, including perceptual differences between rural residents with and without Internet access.35 Among the 30% of respondents without any Internet access were citizens lacking knowledge about the Internet or unable to see its functional value. Collectively, surveys have found widespread, increasing Internet use for specialized information: current affairs,36 health37 science,38 and business.39 While rural residents reported lower Internet use in general and less online health information seeking specifically,40 Collins and Wellman challenged broad generalizations about rural residentsí online health information seeking behavior, calling online health information a necessity for rural residents and illustrating the potential differences among rural communities.41 4 Sources of Environmental Risk Information These rapid, dramatic shifts in online information seeking behavior, coupled with the concentration of particular types of local environmental problems in rural areas, beckon the question: Does an urban rural disparity in Internet access also lead to differences in how people seek information about environmental risk? Studies have confirmed that traditional media (newspapers, radio and television) are still relied upon sources for news,42 though the bulk of recent research points to growing reliance on online sources.43 Of course, social structural and individual differences matter. Moreover, education and issue familiarity play a part in preference for television news coverage of environmental issues,44 with television preferred for stories involving victims of environmental problems.45 Using “perceived environmental risk” and “information sufficiency” indices, Riffe and Hrach found that race and income were significantly correlated with environmental risk perception, and income and education were related to belief that one has sufficient information (also referred to as “subjective knowledge”).46 As perceived risk increased, respondents tended to be more critical of local mediaís environmental coverage. Comparing rural residents’ newspaper, television and radio consumption for health related news pertaining to local environmental dangers, Riffe found reliance predominantly upon television.47 While residents living in areas with multiple local environmental hazards rated television coverage better than newspaper coverage, television viewers did not consider themselves as knowledgeable as did newspaper readers.48 When television newspaper differences were explored in greater detail along five dimensions (causes, solutions, costs, victims and responsibility), survey respondents revealed that newspapers provided more thorough coverage, particularly in topics considered to be more difficult to cover, such as solutions and costs.49 Interestingly, those who stated that both newspaper and television news were poor sources for environmental issues believed the severity of local environmental risks was significantly higher.50 Environmental news scholars now look beyond traditional media to include not only online versions of traditional media outlets but also online only sources. Lacy, Riffe, and Varouhakis conducted a statewide survey of environmental news source preference across five geographic levels: global, national, state, city, and local community. As per previous studies, a majority of respondents cited television most frequently for environmental news, but only for three out of five levels (global: 64%, national: 61%, and state: 58%); newspaper coverage was consumed most frequently at the city (61%) and community (57%) levels. One fifth (19%) of respondents identified the Internet as their source for news about national or international environmental problems, leading the scholars to conclude that, though the Internet was not then the most relied upon source for environmental news at any level, in the near future it would become a key source.51 Riffe and Reimold extended the study to a national sample, finding the same pattern: Slightly fewer (12% 13%) respondents identified the Internet as their national or international environmental news source.52 The Internet proved to be popular for younger consumers, but the data again indicated that online only media had not yet displaced traditional media for environmental news. A 2008 survey, however, found online sources viewed as more useful than television or newspapers with online use predicting greater 5 environmental health risk perception though levels of exposure and attention to online content were lower than for television or newspapers.53 Hypotheses Building off of the bodies of literature discussed above, this study examines Internet access, use, and information seeking among urban and rural residents of North Carolina. That state represents an ideal case study of any remaining digital divide in the 21st century: while its major urban areas are associated with technological, medical, and financial innovation, its eastern third is very lightly populated, agricultural, and plagued by infrastructure problems and lack of health care access, and much of the western part of the state is rural and mountainous Appalachia. Moreover, proponents of “environmental justice” often cite North Carolinaís rural areas as environmental hotspots that have high concentrations of poor and minority residents who lack the collective political clout to deal effectively with an unfair share or burden of environmental risks.54 Based on previous research on the urban rural digital divide and factors related to Internet access, use, and people’s use and assessment of sources of environmental risk information, several hypotheses and research questions were posed. First, despite increasing growth of Internet access generally, recent evidence from the Pew Home Broadband 2013 Report55 and previous research suggest continuing urban and rural differences in access.56 Thus, H1: A significantly greater proportion of urban residents will report having home Internet access than will rural residents. In addition, the study examines urban and rural differences in reported Internet and other media use: H2: There are urban rural differences in media and Internet use. Based on previous research, we anticipate that Internet use may relate to respondent race, gender, age, income and education: younger white males with higher levels of income and education report greater access.57 Because of possible urban rural differences in these demographics, we pose our second research question: H3: Differences in home Internet use relate to individual (demographic) differences among urban and rural respondents? In addition, the study examines Internet use for specialized “environmental risk News” An earlier study contrasting rural Appalachian residents with and without Internet access found that those without access had limited knowledge of the functional capacity of the Internet.58 Further, Hale et al. found urban rural differences in online information seeking, with rural residents (with access) significantly less likely to seek health related information.59 Though Collins and Wellman called going online for health information a necessity for a particularly isolated rural community,60 we nonetheless predict: H4: Urban residents will report significantly greater Internet use for environmental risk information than will rural residents. The study also seeks to explore possible differences in usefulness that urban and rural residents ascribe to different sources of environmental news. Previous research on health information seeking has shown individual differences (e.g., younger, female, higher education and income),61 and Riffe and Reimold found age significantly related to preference for online sources when comparing environmental risk information sources.62 Thus: 6 H5: Differences in frequency of environmental risk information seeking relate to individual differences among urban and rural respondents? Within that information seeking context and based on previous research,63 we predicted: H6: Online sources of information about environmental health risks will be judged significantly more useful than other sources. Because we have no reason to propose a directional hypothesis, H7: Urban or rural residence relate to perceived usefulness of sources of information on environmental health risks? Finally, respondent rating is only one measure of perceived usefulness of sources. “Usefulness” of sources should be correlated with respondentsí “environmental confidence,” or sense of efficacy (i.e., empowerment) and subjective knowledge (i.e., feeling sufficiently knowledgeable about the environment), as suggested by Riffe and Hrach (2009). Thus: H8: Differences in environmental confidence relate to media use and individual differences among urban and rural respondents? Method To test the hypotheses , a statewide telephone survey using CATI equipment was conducted of adults (18 and older) by trained interviewers at a university calling center. A starting sample of 2,000 random landline numbers was purchased, and 1,438 numbers were deemed viable (excluding fax machines, pagers, non working and business numbers, non English speaking households, and households with no adult present). Respondents were selected within households by interviewing the adult with the next birthday, a common procedure for random sampling surveys.64 Calling was completed over a four week period with up to six callbacks attempted. Rural counties (as classified by the state) were deliberately oversampled because: 1) proportional sample phone surveys over represent urban residents; and 2) the studyís focus is on a potential urban rural divide.65 Interviews lasting approximately 15 minutes were initiated with 428 respondents and completed with 406, for a response rate of 29% by the American Association of Public Opinion Researchís (AAPOR) formula 3.66 Just over half (n = 209) of the studyís total respondents lived in rural areas. The sampling error for the two sub samples (urban versus rural) was 6.7% at the 95% confidence level. Measures Internet access: To operationalize this construct, respondents were asked to indicate if Internet access was available to them at home, work, or anywhere else. Information Seeking: To characterize their general information seeking behavior, respondents were asked how often they looked for information about the environment in the past 30 days (daily, several times per week, weekly, a few times per month, once a month, less often than that, or never). Adapting the media dependency approach used by Riffe, Lacy, and Varouhakis, “regular” information seeking was operationalized as seeking information “at least weekly” (combining “every day”” “several times a week” and “about once a week”), in contrast to “less often than weekly.”67 “First Source” for Environmental Risk News: At the outset respondents were asked: “If you needed to find out about environmental problems where you live, where would you 7 get the information? Callers recorded, verbatim, the “first source” named and responses were coded as “Internet”” “Government”” and “Other” Use: Respondents reported daily use (i.e., in minutes) of national television news, local television news, local newspapers, and Internet news. Usefulness: In addition, respondents rated each source (television, newspaper, and Internet) in terms of how well it does at informing readers or viewers about environmental health risks (1 = very poor, 5 = very good). Environmental Confidence: Subjects’ subjective knowledge (how much they think they know) and sense of personal efficacy regarding the environment were measured with a series of Likert type items (1 = strongly disagree, 5 = strongly agree) adapted from previous research:68 1. I consider myself knowledgeable about health risks related to the environment. 2. If a friend asked me about health risks related to the environment, I could give my friend advice. 3. If I had to take action to deal with health risks related to the environment, I would know what to do. 4. I am confident I can protect myself from environmental health risks. 5. I consider myself well qualified to participate in solving environmental issues. 6. I pay close attention to news and information about health risks related to the environment. Principal components analysis (KMO = .835) yielded a single dimension explaining 51.4% of the variance in the data ( = .80). For subsequent analysis, the six items were summed and then averaged to form the measure of “environmental confidence” (EC). Demographic Information: Each respondent’s age, gender, and race were recorded. Household income was measured with seven ordinal categories in $15,000 increments ranging from “less than $15,000 per year” to “more than $90,000 per year.” Employment status (full time, part time or unemployed) was reported. The last grade of school completed (eighth grade or less, some high school, high school graduate, some college, college graduate, postgraduate study) was dummy coded into those with a “high school education or less” and those with at least “some college experience..” To further characterize the urban and rural subsamples, respondents were also asked about their political party and ideology (very conservative, somewhat conservative, moderate, somewhat liberal, very liberal or other). Results Demographics and Descriptive Statistics. Rural counties were oversampled in order to ensure representation of rural residents, who ultimately constituted 51.9% of the final sample, slightly more than the stateís actual rural population (50.3%, according to the stateís rural economic development center). Mean ages of urban and rural subgroups (Urban M = 54, SD = 15.75; Rural M = 53, SD = 16.27) were comparable. Overall, more rural respondents were white (78.2%) than were urban respondents (70.4%), and the total sample matched 2010 Census information (74.4% white in the sample, compared to 72.4% of the population). Despite our using the next or most recent birthday method, to select respondents, women were over represented,69 consisting 60.2% respondents as opposed to 50.8% of the population.70 8 Employment status differed significantly (x2 (1, N = 406) = 14.57, p < .001) between the urban (68.2% employed) and rural subgroups (49.5% employed). Similarly, the urban and rural groups differed on education ( 2 (1, N = 407) = 17.932, p < .001): 78.8% of urban residents attended beyond high school, compared to 59.3% for rural residents. Rural residents average household income was between $45,000 and $60,000 per year, while urban residents averaged between $60,000 and $75,000 per year. No significant urban rural difference existed for political party, but rural residents were more likely to be conservative than urban residents ( 2 (2, N = 393) = 11.352, p < .005). H1 predicted an urban rural digital divide in home Internet access, with significantly greater urban access. However, most respondents in the sample had home access (84.3%), and an additional 3.7% had access at work or elsewhere. The slight difference between urban (85.9%) and rural (82.8%) home Internet access was not significant (x2 (2, N = 407) = 5.153, p = .076). Consequently, H1 was not supported: There was no urban rural Internet access gap for this sample of North Carolina, a conclusion offered with more confidence because of our over sampling of rural residents. Table 1 reports descriptive data for the entire sample on a number of key variables, including time spent using the Imternet. H2 asked about urban rural differences in use of the Internet and other traditional media, measured as self reported minutes per day. An independent samples t test found no significant differences between urban and rural residentsí Internet use (Urban: M=32.2, SD=51.995; Rural: M=31.90, SD=38.07; t (404) = 0.136, p=.892 (two tailed), local television news use (Urban: M=54.2, SD=109.06; Rural: M=51.78, SD=55.552; t = 0.275, p=.62), national television news viewership (Urban: M=42.28, SD=48.666; Rural: M=49.66, SD=69.772; t = 1.229, p=.063), or time spent reading newspapers (Urban: M=21.90, SD=27.362; Rural: M=21.14, SD=22.648; t = 0.305, p=.153). Thus, in addition to having comparable home Internet access, rural residents make comparable use of information sources, the first hint that the gap in perceived Internet utility Riffe reported a decade ago (2003) has also narrowed. (TABLE 1) H3 asked how differences in home Internet use relate to individual differences and urban and rural residence. To answer this research question, regression analysis was conducted using time spent viewing news on the Internet as the dependent measure. Predictors included respondentsí residence (0=urban and 1=rural) and five standard demographic variables: age and income, and dummy coded education (1=some education beyond high school), gender (1=male), and race (1=white). The model was checked for outliers based on Tabachnick and Fidellís guidelines,71 but the four outliers found did not exceed the number expected from a normally distributed sample, nor did they exert undue influence. Thus, they were retained for the regression analysis. Results from the regression analysis (see Table 2) indicated that 5.8% of the variance in the time viewing Internet news could be explained by predictors in the model, F(6, 334) = 3.446, p < .005. Most telling given our studyís focus ñ is that urban rural residence did not contribute significantly, when respondent age, gender, income, race, and education were present. Of the variables included in the model, the only significant predictor was income. Recall that urban respondents averaged $60,000 $75,000 household income per year, compared to $45,000 $60,000 for rural respondents. (TABLE 2) 9 H4 predicted that urban residents would use the Internet more for environmental risk information than would rural residents. Recall that respondents were asked the “first source” they would consult about environmental problems. We compared the proportions of respondents in each sample who identified the Internet as the source they would consult for environmental information, using the z test for proportions. H4 was supported: the proportion of urban respondents (37.4%) naming the Internet was significantly larger (z = 2.054, p <.01) than the rural proportion (28.1%). Rural respondents were significantly more likely (44%) than urban residents (35%) to name government or government agencies (z =1.89, p < .05). Greater reported reliance among urban residents on an Internet source may reflect greater concentration in urban areas of multiple agencies and offices, each with Web sites that might be consulted. To provide context for comparing the usefulness of different information sources for urban and rural respondents. H5 asked how the frequency of information seeking about environmental risks was related to individual differences. As noted earlier, respondents were dichotomized as either seeking information “at least weekly” or “less often than weekly” an approach used in previous media dependency research.72 Fewer than half (43.8% overall) of respondents reported seeking environmental information “at least weekly,” with no significant difference found between urban (40.7%) and rural (46.7%) respondents. Table 3 reports results of regressing””at least weekly” environmental information seeking on urban rural residence and demographic predictors. Results from the logistic regression indicate that prediction of environmental risk information seeking based on these independent measures was not statistically significant, 2 (6) = 3.464, p = .749. Frequency of going online was not significantly related to rural residence or any of the selected demographic measures (H3). (TABLE 3) H6 predicted that online sources of environmental health risk information would be judged as significantly more useful than other sources. Pairwise t tests were conducted for the entire sample comparing usefulness of the Internet (M = 4.01), television news (M = 3.52), and the local newspaper (M = 3.32). H3 was confirmed: the Internet was viewed as significantly more useful than television news (t = 7.47, p< .001) and newspapers (t = 9.49, p<.001), with television news rated more useful than newspapers (t =3.91, p<.001). H7 goes a step further by asking how urban and rural residence relates to usefulness of sources of environmental health risk information (see Table 4). First, pairwise t tests were used between the urban and rural sub samplesí means, finding no significant urban rural differences in the means for each source. Additional pairwise t tests within the urban and rural sub samples mirrored the H3 results, showing that within each sample, respondents deemed the Internet significantly more useful than local newspapers and television for risk information. (TABLE 4) Finally, H8 asked how urban or rural residence relates to environmental confidence (EC) a measure tapping subjective knowledge about the environment as well as sense of efficacy in dealing with environmental problems. Regression analysis was conducted (see Table 5) using residence, media use (daily time spent online in minutes, reading newspapers, and watching national and local news on television), age, income, education, gender, and race in a model predicting environmental confidence. 10 Total variance in environmental confidence explained by the model was 12.5%, F at recencyviewing national television news (b = .156, p < .05) and reading the daily newspaper (b = .163, p < .05). This role for newspapers is intriguing, given the fact that survey respondents had rated newspapers the least useful for environmental information. Notably, urban rural residence, minutes spent watching local television news, and reading online news were not significant predictors. Put more emphatically: Despite rating online news as most useful, time spent with online news did not predict subjective knowledge and efficacy environmental confidence whether one lived in urban or rural settings. (TABLE 5) Discussion and Conclusion Data from this statewide survey suggest there is no urban rural Internet access gap or “digital divide” in North Carolina, in terms of home access, as well as the additional access available via other sites, including at work, thus contradicting previous statewide73 and nationwide survey findings.74 Moreover, in terms of self reported minutes per day of use, there was also no gap in Internet use between urban and rural residents: Rural Internet access and time spent online, much like time spent viewing local and national television news and reading newspapers, matches urban levels. However, because of demographic differences between our urban and rural samples, we nonetheless explored how differences in time spent on home Internet use relate to age, income, education, gender, and race. The strategy involved comparing significant predictors of time spent online in regression analyses. The predictive power of the overall model was significant but modest, with income as the only significant predictor of time spent online. However, none of the remaining predictor variables achieved significance. Given what has been shown previously about younger and male respondents spending more time online,75 this finding was unanticipated. The time online data add context t and hinted at the fact that rural respondents having achieved parity of access and time spent online may no longer differ qualitatively in the ways they use the Internet, contradicting past research.76 While the access gap appears to have closed for individuals in this study, evidence from other digital divide 00research suggests that this equalizing is a recent phenomenon.77 Given that recency urban and rural residents surveyed in this study were expected to perceive in home Internet access and the value afforded by Internet connectivity in different ways. Of the study’s demographic variables, only income significantly predicted respondentsí Internet use. The lack of the expected differences between younger rural respondents (digital natives) and their older counterparts certainly defies stereotypical depictions of technology use.78 Another intriguing possibility is that levels of uncertainty about the real environmental risks faced by rural residents contributed to the diminished effect of demographic differences in Internet use, differences traditionally associated with being urban, young, and male. Unease about perceived risks may have motivational effects that transcend individual differences, and the desire to increase personal confidence about environmental risk may take precedence over traditional, demographically based media consumption tendencies. The uniform use of Internet news sources by rural residents may help explain, albeit in unexpected ways, the role media use played in predicting changes in environmental confidence. The study also examined sources that urban and rural respondents identified when facing an environmental information need. As hypothesized, the proportion of urban respondents naming the Internet was significantly larger than rural respondents, while rural respondents identified government or government agencies, which suggests disparities between Internet access and “first source” reliance and the importance of distinguishing between the two in future research studies. Fewer than half our respondents both urban and rural reported regular environmental information seeking, and age was not a significant predictor of preference for online environmental risk information, as per earlier findings.79 Within that context, we predicted that online sources of environmental health risk information would be judged more useful than other sources. Indeed, the Internet was viewed as significantly more useful than television news and newspapers, confirming earlier findings and our predictions.80 For both urban and rural respondents, the pattern is identical: the Internet was viewed as more valuable than other sources for health risk information. The study sought additional perspective on how valuable environmental information sources are, moving beyond usefulness ratings and building upon earlier and “information sufficiency indices through the exploration of”environmental confidence combining subjective knowledge and sense of efficacy.81 In this environmental confidence measure, reported source use (not rated usefulness) was examined, and we found no significant difference in environmental confidence between urban and rural residents. The regression results were, again, modest in terms of variance explained, but they showed an unexpected pattern in the variables that served as significant predictors. In the model, national television news use and daily newspaper reading were significant predictors correlated with environmental risk perception, whereas income and education were correlated with subjective environmental knowledge — although that study did not focus on potential differences in urban versus rural residents.82 Based on the regression model, time spent using the Internet was insignificant and less predictive of environmental confidence than time spent watching national television news and time spent reading newspapers. Urban rural differences were negligible, too, a final indication perhaps that the digital divide has been minimized in quantitative (access and time online) terms. This study has a number of limitations, of course, such as its singular focus on one state, albeit one whose citizens are densely concentrated in a handful of urban areas and spread across sparsely populated rural areas. While the data from this survey suggest a change in the urban rural divide of North Carolina, this singular dataset is not indicative of a trend. The use of a telephone survey to target landline served homes always introduces unknown biases, as do all such samples. Future studies should incorporate cellphone only households, non English speaking households, and measures of Internet access via computers as well as mobile devices, including but not limited to smartphones and wearable technologies. Additionally, the survey completion rate, while not uncommonly low nowadays, is less than ideal. Nonetheless, the study contributes valuable empirical data about urban and rural patterns of information access and use, particularly as they relate to environmental risk. Indeed, the study’s oversampling of rural counties provides a useful look at non urban residents, who are typically underrepresented in statewide studies.83 Furthermore, the introduction of the environmental confidence variable may be of use in future research about environmental news and information seeking, implications for policy and policymakers, support for political actors, and quality of media performance. That oversampling gives us greater confidence in our conclusions that the digital divide is becoming increasingly narrow, in terms of access, time spent online, and perceived utility among urban and rural residents.
Laurie Phillips is an assistant professor in the School of Journalism and Communication at the University of Oregon. Daniel Riffe is Richard Cole Eminent Professor in the School of Journalism and Mass Communication at the University of North Carolina at Chapel Hill. Robert McKeever is an assistant professor in School of Journalism and Mass Communication at the University of South Carolina.