Journal of Communication, December 2003 What Do Americans Really Want to Know? Tracking the Behavior of News Readers on the Internet By David Tewksbury Evaluations of the health of contemporary political systems typically include some discussion of the modes through which people acquire public affairs information. In response to survey questions, Americans often profess an interest in current events news, but assessments of citizens’ political knowledge often find them wanting. Unfortunately, the limitations of previously available research methods have left researchers with an incomplete understanding of news audiences and their exposure patterns. Widespread adoption of the Internet for news reading may change that situation. The World Wide Web provides audiences with substantially more control over the news selection process than they enjoyed with the traditional media. With that enhanced control, it appears online readers are particularly likely to pursue their own interests, and they are less likely to follow the cues of news editors and producers. The present study takes advantage of this attribute of online news presentation to examine the topics people select at Web-based news outlets. In general, online news audiences choose to read public affairs news less frequently than survey research suggests. This result has implications for the long-term health of democratic nations. An enduring concern of mass communication researchers is the identification of the news topics that interest audiences (e.g., Price & Czilli, 1996; Stone, 1987). Identifying these topics has obvious utility for both media organizations and for researchers examining the processes through which citizens acquire public affairs and other information (e.g., Campbell, Converse, Miller, & Stokes, 1960; Delli Carpini & Keeter, 1996; Price & Zaller, 1993). For a number of historical and practical reasons, the vast majority of the quantitative work in this area has examined newspaper readership as opposed to the use of television and other electronic news sources (Stone, 1987). The academic arm of this research has most often employed newspaper reader surveys (e.g., Burgoon, Burgoon, & Wilkinson, David Tewksbury (PhD, University of Michigan) is an assistant professor in the Department of Speech Communication and Department of Political Science at the University of Illinois at Urbana-Champaign. The author thanks Larry Oathout for his assistance in coding the data used in this project. Funding for this study was provided by the UIUC Campus Research Board. Copyright © 2003 International Communication Association 694 Internet News Reading 1983; Jeffres & Atkin, 1996) with a supplemental focus on group discussions (e.g., Clark, 1979) and experiments (e.g., Rarick, 1967). The difficulty with much of the research in this area is that people are rather poor at reporting their own habits. News readers may vaguely recall hearing about a topic—perhaps from a friend—and so they may erroneously tell survey researchers that they read a story about the topic. Likewise, people may not be able to recall reading an article they scanned or a headline they quickly read. Observing readers going through a newspaper may allow researchers to more accurately rate interest in specific news areas, but this sort of research is artificial by necessity, with readers engaging the newspaper out of their element and potentially subject to Hawthorne and other disruptive effects. The upshot of all this is that communication researchers have an incomplete picture of how people receive the news. However, new technologies are changing the nature of news reading and providing new opportunities for studying that behavior. The interactive nature of the Internet allows people to efficiently select the news that interests them. Online news users are not bound by the linear format of electronic and even (to a some extent) print news media. Coupled with the expanded volume of news content available at most news sites on the World Wide Web, the interactivity of the medium should lead to greater selectivity on the part of news audiences (Chaffee & Metzger, 2001; Tewksbury & Althaus, 2000). At the same time, the Internet is also a powerful research tool, one that allows researchers to observe news reading behavior more reliably and less obtrusively than had been possible before. This study will take advantage of these two attributes of the medium to examine what online news audiences seek from the news. The results of that study should inform both research on news consumption and ongoing evaluations of the political commitment and competence of the American public (e.g., Kuklinski & Quirk, 2001). Viewing News on the Internet Early proponents of new communication technologies (e.g., Abramson, Arterton, & Orren, 1988) argued that the new media would raise the quality of democracies by removing many of the barriers between audiences and information. The chief mechanisms for democratization would be the wide availability of information in the new media and audiences’ consequently decreased reliance on centralized content producers (Corrado, 1996). As the Internet and its popular audience have developed over the past decade, these predictions have been challenged. In particular, early findings suggest that Internet audiences still heavily rely on the traditional news organizations for current events information when they go online (Margolis & Resnick, 2000). One bit of evidence that points in this direction is the observation that World Wide Web use often clusters around a few sites. With millions of sites to choose from on the Web, the average at-home Internet user goes online 22 times in a month but visits only about 48 different sites in that month (Nielsen//NetRatings, 2002). Further, the 10 most popular properties on the Web—a property refers to a collection of sites under a brand such as Yahoo!, AOL Time Warner, or Disney— receive visits from between 14% and 63% of all users every month (Nielsen//NetRatings, 695 Journal of Communication, December 2003 2002). Thus, it appears that the average user visits few sites (relative to the millions available online), and those sites tend to cluster into large media brands. Other evidence suggests that when people go to the Internet for news, they often visit the sites of the traditional media, not Internet-only outlets. One national survey found that 51% of those who use the Internet to follow current events had visited an online newspaper in the preceding 6 months (Cyberatlas, 1999). A late 1998 survey found that 80% of those who said they went online to get news and information on current events, public issues, or politics at least 1 or 2 days a week reported visiting the websites of television networks, national newspapers, cable news networks, and magazines (Pew Research Center, 1999). A study of political campaign news consumption on the Internet reported essentially the same pattern (Pew Research Center, 2000). This tendency translates into a substantial flow of visitors to traditional news outlets online. The 10 most frequently visited news sites (e.g., MSNBC, CNN, USA Today) collectively had over 33 million visitors in May 2000 (PC Data, 2000).1 Thus, focusing on audience use of the Internet sites run by the traditional news brands may yield useful insights into Americans’ online news consumption. Whereas people seem to focus on the traditional news outlets online, it appears that the news they select on those sites may differ from what they may have received from the offline versions. The wealth of information available on the Internet and, even more importantly, the hyperlink and menu structure of Webbased news sites provide users with extraordinary levels of control over the news consumption process. In general, these attributes of the Internet offer audiences the ability to select the news topics that most interest them. In an analysis of print newspaper audiences, Graber (1988) reported that readers use a mixture of their own news interests and cues embedded in the presentation of news to guide their exposure decisions. In an important sense, it may be that online news selection is guided more purely by readers’ interests than is the case with the traditional media (Fico, Heeter, Soffin, & Stanley, 1987). In a study of print and Web-based readers of The New York Times, Tewksbury and Althaus (2000) found that online readers are less likely than their print counterparts to read international, national, and political stories. The general trend among online readers appears to be more focused attention to topics of personal interest and less to topics that interested news editors, a result that has implications for readers’ issue agendas (Althaus & Tewksbury, 2002). Sunstein (2001) and others have discussed this process in the context of normative concerns about the potential for the Internet to polarize citizens as people seek out the information they desire and the views they prefer. These speculations need to be put in some context, of course. Prior research has found evidence that people often use online news to supplement, not replace, their core news consumption (e.g., Althaus & Tewksbury, 2000). A common finding in surveys of Internet users is that people report going online to follow up on 1 In the present study, a news site refers to a World Wide Web page whose primary offering is news content. This, of course, excludes the major portal sites (e.g., Yahoo!, MSN, AOL), which often post a handful of headlines and appear to be relatively popular sources of news (Barringer, 2001). 696 Internet News Reading news they originally received offline (e.g., Pew Research Center, 1999). In addition, Chyi and Lasorsa (1999) found that readers of local newspaper sites usually subscribe to the print versions. Readers’ online viewing would naturally be supplementary to their print reading, with readers perhaps following local events or investigating particular topics. Readers of national online sites were much less likely to be subscribers to the corresponding newspapers (Chyi & Lasorsa, 1999), however, so their reading of those sites might be more comprehensive in character. Still, it may be that a substantial portion of the online news audience already reads newspapers or attends to other electronic news media. It seems reasonable to assume that online news selection by these readers would reflect the topics that most interest them because they would likely have had their basic information needs met by exposure to traditional news media. The online behavior of people for whom the Internet is the primary news source, on the other hand, should more fully reflect the joint influence of news consumption norms and personal interests. Consuming Public Affairs News Contemporary versions of democratic theory suggest that citizens have a duty to acquire information about events in the public domain and to use that information in their political decisions (Berelson, 1952; cf. Schudson, 1998). This theoretical perspective can beg the question of whether most people are, in fact, exposed to political information. Public opinion researchers typically bemoan the average American’s awareness of public affairs (e.g., Delli Carpini & Keeter, 1996). Some observers place the blame for this state of affairs on the media (e.g., Patterson, 1993), arguing that news outlets fail to provide the basic information audiences want and need. Some researchers, on the other hand, have looked at the mass audience itself as the locus of responsibility (e.g., Converse, 1964; Price & Zaller, 1993). After all, it does not matter how much information the news media present if audiences do not seek it. Survey research has shown that most Americans are aware of and appear to accept an obligation to attend to public affairs news. McCombs and Poindexter (1983) found that 93% of respondents to a set of regional surveys agreed that everyone has a duty to keep informed “about news and current events” (p. 90). Acceptance of a civic obligation was strongly correlated with the frequency of newspaper reading and national television news viewing (McCombs & Poindexter), results also obtained in a 2000 replication (Poindexter & McCombs, 2001). In that more recent study, acceptance of the civic obligation was also positively related to self-reports of reading national and presidential campaign news on the Internet. In survey responses, people often report an interest in public affairs news (Bogart, 1989; Schramm & White, 1949) and a preference for general news and current events over sports, business, features, and other content (e.g., Bogart, 1989; Munro & Weaver, 1977; Stone & Boudreau, 1995; Weaver, 1978). For example, Bogart found that 80% of newspaper-reading adults reported opening the opinion and editorial pages but only 59% reported opening the sports pages. Similarly, 75% of respondents in one 1987 survey reported they usually read or looked at international news in the newspaper, and 46% reported the same for 697 Journal of Communication, December 2003 business and financial news (Bogart). At the same time, it is a truism among some researchers that people are not really as interested in politics as in other news content. For example, Bennett (2001) noted that one survey found that people reported greater interest in certain “soft” news categories (e.g., crime and sports) than in local, national, or international political news. It is unclear how much confidence researchers should place in survey measures of popular interest in public affairs news. If Americans remain underinformed about public affairs events (e.g., Delli Carpini & Keeter, 1996), there is an inconsistency between what they often say they attend to and what they take away from news exposure. A number of factors may be affecting such a phenomenon (Delli Carpini & Keeter); one possibility is that many survey questions may fail to accurately measure respondents’ true news interests. Although people understand the civic obligation norm, they prefer other news to public affairs content. Therefore, they may skim current events news, but they give most of their attention to other topics. As a result, they can report having seen the former, but they retain relatively little of its content.2 One way to assess the plausibility of this suggestion is to examine the news topics Americans select to read. If some studies have been overestimating popular interest in public affairs news, unobtrusive observation of topic selection should reveal a somewhat tepid interest in that content, relative to other available news options. The present study will combine information from two sources to examine the news-viewing behavior of online audiences. Data useful for examining relationships between aggregate patterns of Internet and traditional news use come from a national telephone survey. These data will provide general information about news exposure habits, establishing the context for looking at how a subset of Americans—Internet news readers—selects specific news topics. Data for looking at online news-viewing patterns come from records of the addresses of World Wide Web pages viewed by a sample of American Internet users; these records were coded for the content of the pages viewed. Together, these data will be used to build a profile of the aggregate news preferences of the Internet audience. To be sure, these data will tell researchers only so much. It would have been preferable to assess directly the traditional news consumption of the Internet users and examine relationships between offline and online behaviors. Unfortunately, no mechanism for such an analysis currently exists. The present study examines the best data available to answer important questions about the nature of online news audiences. Method Survey Data The Pew Research Center for the People and the Press conducts a biennial national survey of Americans’ news consumption habits. The data for the present 2 Researchers have noted that measuring attention to news rather than exposure often increases the predictive power of media use (e.g., Chaffee & Schleuder, 1986). 698 Internet News Reading study come from a sample of 3,142 adults questioned between April 20 and May 13, 2000. Respondents were selected through a random digit sample of telephone numbers. Each respondent was assigned a demographic weight; this weight was applied in all analyses reported here. Among the measures contained in the survey were questions that asked respondents whether they read a newspaper regularly (coded 1 for the present analysis) or not (coded 0) and whether they listened to news on the radio regularly or not (coded 1 and 0, respectively). Subsequent questions asked respondents to report the frequency (response options were regularly, sometimes, hardly ever, and never) with which they watched network news programs on television, local television news programs, and CNN. Responses of “regularly” were recoded to 1; all other valid responses were recoded to 0. An index of traditional media use was created by summing responses to these five items. Later in the survey, respondents were asked to report how closely they followed certain topics “either in the newspaper, on television, or on radio.” Among the topics were “news about political figures and events in Washington,” “sports,” “business and finance,” “international affairs,” and “entertainment.” Responses of “very closely” or “somewhat closely” were recoded to 1; all other valid responses were recoded to 0. A pair of questions was used to determine whether respondents read news online. Respondents who said they “ever go online to access the Internet or the World Wide Web or to send and receive email” were asked to report the frequency with which they accessed news online. All respondents were categorized as online news readers (if they reported receiving news online at least 1 day a week), other Internet users, or other people. People who said they read news online were asked to report whether they “sometimes go online to get” news that fell into certain categories. The topics used in the present analyses were news about politics, international affairs, sports, business, and entertainment. Online News-Viewing Data Page view records were obtained from Nielsen//NetRatings (hereafter NetRatings), an Internet ratings company. NetRatings maintains at-home and at-work panels of Internet users; they record the address (URL) of each World Wide Web page these people view. For the present study, they provided records of occasions on which panelists visited 1 of 13 preselected news sites in the months of March and May 2000. Each of these records included a user identification number, the page address, the date it was accessed, some limited user demographic information, and a demographic weighting factor. NetRatings provided 179,864 such records from 9,209 panelists. The present data will help extend prior news reading studies in several ways. The first advantage of the present data is their level of precision. Web page addresses provide information about what readers chose when given options from some menu or set of stories. Thus, the patterns for story selection appear to accurately tap content preferences in a naturalistic environment. In addition, NetRatings attempts to recruit a representative panel of Internet users in the United States. Thus, the present study may avoid the tendency in some prior research to rely on convenience or regionally limited representative samples. Also, by aggre699 Journal of Communication, December 2003 gating story selection into general categories and over a 2-month period, the results should not be greatly affected by the particular selection of news available on 1 day or even 1 week. Finally, the data were gathered through unobtrusive observation. Thus, they should be superior to survey and other readership study methods in which people report prior behavior or make generalizations about their own habits (Bogart, 1989). User sample. For the at-home panel, NetRatings follows recruitment procedures similar to those employed by Nielsen Media Research in its television audience research. American households are probabilistically sampled via a random digit dialing (RDD) procedure, screened for Internet access, and asked to join the panel.3 Households officially join the panel by installing NetRatings’ tracking software on computers and by completing registration procedures that include identification of household members and their demographic profiles. The at-work panel is recruited via two methods. The first follows individual sampling via an RDD procedure that generates home addresses. Selected households receive a mailed and telephoned invitation to participate at their place of employment. NetRatings supplements this recruitment method with a cluster approach, probabilistically selecting employers and employees for recruitment. All participants have a unique login name and password, which must be provided at the outset of a World Wide Web use session. This is the only overt reminder that Web use is monitored. The at-home panel had approximately 44,000 members in March and 57,000 in May 2000. The at-work panel had about 6,700 members in March and 7,200 in May.4 A comparison of the demographic profile of Internet users in the Current Population Survey (CPS)—a joint project of the Bureau of the Census and the Bureau of Labor Statistics—with the profile of the at-home panel in March 2000 reveals few large differences.5 However, people of Hispanic origin were slightly underrepresented in the NetRatings panel (4.4% of the sample versus 6.3% in the CPS data), as were African and Asian Americans (6.4% and 2.3%, respectively, versus 7.2% and 5%, respectively, in the CPS data). Predictably, the average atwork panelist is older, better educated, and more likely to be male and nonHispanic than the average Internet user in the CPS data. News site URL records. A number of factors were applied to select the 13 sites used in this analysis. One important consideration was audience reach. In order for this study to reasonably examine American online news-reading habits, the news sites selected must include some sites that reach large audiences. Accordingly, four of the sites were among the five most frequently visited news sites in May 2000 (PCData, 2000). Perhaps even more important for the goals of this study was geographical coverage. Six of the sites (CNN, ABC news, USA Today, the BBC, Nando Net, and, arguably, The New York Times) have national or interna3 NetRatings does not publish its response rate. 4 For the sake of simplicity, the at-home and at-work panels are pooled in all analyses. NetRatings notes that their sampling methods do not allow for the measurement of Web use from home and work for people who go online at both locations. 5 The profile in May 2000 is essentially the same. 700 Internet News Reading tional emphases and likely draw readers from across the country. This broad focus was supplemented with seven sites (Dallas News, Chicago Sun-Times, Detroit News, OregonLive, San Diego Union-Tribune, San Jose Mercury-News, and Louisville Courier-Journal)—-all online versions of print newspapers—that can be expected to reflect reading emphases at local news outlets. The final factor was the ease with which an analyst could determine story content from the URL structure employed by each site. Some sites (e.g., The New York Times Online) used relatively transparent URL-naming systems, but others (e.g., MSNBC) used systems that provided little to no information about content. The selected sites featured naming systems that (in varying degrees) facilitated news topic coding. It is possible that the selection process injected some bias into the results reported here, a possibility explored in the discussion. Each record was categorized by the primary content of the page viewed. The categories assembled for the present analysis were drawn from a relatively standardized typology of newspaper sections, fleshed out with more specific topics when possible. This list of news topics was supplemented with categories that captured the interactive features often available at online news sites (e.g., site search functions and message boards). The sites’ classification structure was the primary basis for the assignment of stories, menus, and topical indexes to specific categories. In all cases, the categories were mutually exclusive. Most of the content categories that appear in the analysis are self-explanatory, but a few represent an aggregation of smaller categories. Specifically, the “features” category includes (in the order of page view frequency) pages about fashion, home, travel, food, automobiles, and religion, as well as website recommendations, comics, and gambling (e.g., lottery) information. The “other news” category captured URLs that could be identified only to the point of having news content; no further information was available. “Interactive elements” included (in the order of page view frequency) games and online crossword puzzles, message boards and forums, site search, online surveys, shopping, and electronic messages to the site. An aggregate category of public affairs news included page views of political news, international news, opinion and editorial content, metro and local news, and U.S. national news. A single trained analyst performed all of the coding. The reliability of the coding scheme was separately determined for each site. The reliability coefficients (Cohen’s kappa) had a range of .81 and .94 and a mean of .88 (SD = .04). Naturally, some URLs could not be coded at all (n = 729). These page views and all users for whom these were the only views (n = 14) have been eliminated from all analyses. This reduced the number of usable views to 179,131 and the number of available panelists to 9,195. Results General News Consumption Patterns The initial step in the analysis is a comparison of how online news readers and other Americans use the traditional news media. To begin, the total number of traditional news media the survey respondents reported using was no different for 701 Journal of Communication, December 2003 Table 1. Traditional News Use Among Online News Readers and Other Americans Online newsreader (38.9%) Other Internet user (14.9%) Traditional news media use Read newspaper Watch network TV news Watch local TV news Watch CNN Listen to radio news 65.3%a 27.2a 53.2a 24a 53.1a 58.3%b 22.4b 46.6b 18.4b 47.4b 61.5%b 35.2c 60.9c 19.8b 40.3c News followed in traditional media National political figures & events International affairs Sports Business & finance Entertainment 67.5a 66.5a 55.7a 63.7a 57a 54.5b 54.9b 47b 44.9b 56.5a 55.6b 54.9b 49.7b 38.8c 55.2a Media use Not an Internet user (46.2%) Note. Demographic weight applied to all analyses. Cells not sharing subscripts differ from other cells within the row, p < .05. Internet news readers (M = 2.23, SD = 1.29) than for people who do not use the Internet (M = 2.17, SD = 1.29), although both of these groups used a slightly larger number of news types than did other Internet users (M = 1.93, SD = 1.27).6 The data reported in the top portion of Table 1 reveal that these apparent similarities conceal some real differences in the media used by online news users and people without Internet access. Online news readers are more likely to report reading a newspaper, watching CNN, and listening to radio news but less likely to report watching network and local television news. This suggests that online news readers are somewhat more frequent consumers of long-format programming and news with relatively fewer constraints on the time of day that the exposure occurs. Relative to the other respondents, online news readers were more likely to report exposure to public affairs content (assessed here in terms of national political figures and events and international affairs) in the traditional media, and they were particularly likely to report an interest in business and finance content.7 In summary, then, the survey data suggest that online news readers are frequent users of traditional media and—particularly germane to the present study—they follow public affairs content in those media. 6 The overall F ratio for this ANOVA was 16.36 (2, 5522), p < .01. Other Internet users differed from the other respondents in Scheffé post hoc tests, p < .01. 7 These relationships between the Internet user category and the topical interests remained in multivariate models that controlled respondents’ sex, education, and income. 702 Internet News Reading The final area of analysis is a look at what online news readers said they read online. Online news readers were particularly likely to report sometimes going online to get business news (64.1%). They were somewhat less likely to report going online to retrieve news about international news (56.2%) and political affairs (50.9%), but 70% of them reported viewing at least one of these two topics. About half of the online news readers reported going online to get sports (50%) and entertainment (48.2%) news. Thus, these people reported online news reading habits that mirror what they reported about traditional news use. In both cases, they exhibited an interest in business and public affairs news. This would suggest that an examination of actual news selection habits online would show that people frequently seek public affairs news, a finding that would be consistent with previous newspaper reading research. Online News Consumption An examination of news content viewing online can determine whether news audiences are reading public affairs news on the major Web news sites. The relevant data are presented in Table 2. In this display, five general content categories—site menu or index, news, interactive elements, advertisements, and other elements—account for all of the page views. Within the news category, topic selection is separately measured for the news views. As one might expect, the vast majority of all views refer to news content. A good many of the remaining page requests were to the sites’ primary home pages or other general indexes. Audiences could receive a bit of information about current events from home pages, but headline scanning may not qualify as exposure to public affairs content. Of the news views, sports content was the most common category by a wide margin.8 It is followed by U.S. national news, but popular attention to public affairs was not as high as one might expect from the survey data. National, world, politics, opinion and editorial, and state and local news together accounted for only 24% of the news content views—a number comparable to sports.9 A common approach in prior research that has demonstrated an audience preference for public affairs has been to report the percentage of some sample that professes an interest in this topic (e.g., Bogart, 1989). The total category page viewing measure examined thus far may obscure a pattern in which a relatively small number of users access a topic multiple times and so distort the level of attention the topic receives. Table 2 also reports the percentage of viewers who selected each topic at some point in the 2-month period and the average number of views each received from its visitors. The notable concentration in views of 8 Some portion of that number may be due to the greater frequency with which menus showed up in the sports category. Most sites had separate sections for each major sport, and within those sections were submenus for particular teams, conferences, and tournaments. Still, the viewing of menus as destinations and as transitions was common across all content categories. Overall, 27% of the nonsports news pages were menus or indexes; for sports, the portion was somewhat higher (39%). So, the popularity of sports may be partially explained by this phenomenon. 9 It is important to note as well that the national news category includes a wide range of content (e.g., legal news and natural disasters). 703 Journal of Communication, December 2003 Table 2. Online News Content Selection Patterns Number of viewsc Site element % of all viewsa % of all viewersb Mean SD 15 73.2 26.0 13.4 10.9 10.7 10.2 7.0 6.1 5.4 3.6 1.5 1.4 1.2 .1 2.5 7.7 1.0 3.1 100% 46.7 84.4 29.2 21.9 21.2 28.3 29.8 22.0 17.4 16.3 8.9 10.4 3.3 4.8 .5 6.4 21.6 6.4 18.8 5.78 15.56 13.41 7.44 7.20 5.70 6.75 4.75 5.24 4.99 6.19 3.51 6.53 3.93 2.28 3.41 6.41 2.86 2.96 11.97 47.69 32.60 15.52 24.84 13.32 19.51 11.16 13.15 11.28 12.71 7.64 15.59 11.02 2.41 7.61 23.25 4.38 4.56 Site menu or index News Sports Business & money Arts & entertainment Features U.S. national Technology & science World Politics Weather Health Opinion & editorial State & local Obituary Other news Interactive elements Advertisements Other elements Total a Percentage of all views that fell within each category. Within the news category, cells represent the percentage of all news views that fell in a particular topic. b Percentage of viewers who selected each category. c Number of views by users who selected at least one page in the respective category. sports news in the first column is explained here, in part, as high levels of repeat content viewing.10 In these data, public affairs news and technology and science look much stronger than in the first column. For example, world and political news received at least one page view from substantial portions of the audience. Together, national, world, politics, opinion and editorial, and state and local news received page views from 45.6% of those who selected a news topic, with a mean number of views of 9.4 (SD = 28.19). This paints a normatively better portrait of news consumers than was evident in the first column of the table, but public affairs news-viewing levels were still below what the newspaper research literature suggested might be the case. 10 The sports category saw 12 users view content more than 200 times in the 2-month period. National news (4), arts and entertainment news (4), and interactive content (8) also saw a number of users view content over 200 times. 704 Internet News Reading Online news consumption habits may also be described in terms of their breadth and stability. The first of these describes the number of topics that news readers accessed. Of those who viewed a page with news content (i.e., not merely the home page or some nonnews page element), a slight majority (54%) viewed only one topic in the period studied here, and an additional 19.7% viewed only two topics (M = 2.28, SD = 2.13). Naturally, there was some constraint imposed by the coding scheme that grouped news into 14 potential topics, but there was really very little evidence of reading breadth in these data. Reading stability can be measured as the number of days on which people read news, both in general and in terms of specific topics. News reading was not particularly frequent, again, as a majority (54.1%) of panelists viewed news on only 1 day in the study period (M = 3.93 days, SD = 6.93). Thus, the average and modal reader stuck to one topic in the news sites studied here and viewed it on only 1 day. People who read more often typically read a broader range of topics. Indeed, measures of daily reading frequency and the number of topics people read were highly correlated (r = .7, p < .01). These patterns were visible in the over-time stability of viewing of the specific topics. As a general rule, the more commonly selected topics (see Table 2) were also those viewed more frequently over time. For example, the mean number of days business (M = 3.76 days, SD = 6.31), sports (M = 3.72, SD = 6.34), U.S. national (M = 3.27, SD = 5.46), and entertainment news (M = 2.92, SD = 4.85) were viewed represented both four of the five most frequently revisited topics and four of the five most frequently viewed topics in Table 2. Similarly, health (M = 2.19, SD = 3.37), state and local (M = 1.96, SD = 3.67), and obituary news (M = 1.37, SD = .86) were among the least frequently visited topics by both measures. An interesting exception to this pattern was observed with opinion and editorial content (e.g., editorials, letters to the editor). Selections of this content represented only 1.4% of all views in Table 2, but readers of this content were relatively devoted to it. Thirty-five percent of its readers viewed it on more than 1 day, and the mean number of days they did was the highest seen here (M = 3.82, SD = 7.63). Discussion The chief conclusion one may take away from this research is that online news readers do not select public affairs content as often as they select other news content. The results confirm that fewer than half of the readers accessing news on the sites sampled here viewed a public affairs topic at least once in March and May 2000. This finding is particularly notable, given that March fell in the middle of a contested presidential campaign primary election period.11 It is also notable in light of the fact that respondents to the Pew survey were very likely to have reported getting public affairs news online. A good example of the apparent inconsistency can be found in exposure to international news. Just over half of the 11 That observation begs the question of whether political news viewing was attuned to the primary election schedule. Addressing this question must wait for some future analysis. 705 Journal of Communication, December 2003 survey respondents reported sometimes getting international news online, but only about 17% of the viewing sample was observed doing this at least once in the 2-month period. There are a number of potential explanations for this general finding. The first is that people do things differently online than they do offline. This is certainly a plausible suggestion. If, as the survey data suggest, online news readers are also regular consumers of traditional news media, they may be getting ample public affairs news offline and supplementing it with other information online. In order for that explanation to completely explain the findings here, it must contain the assumption that the World Wide Web is perceived to be a better place for other news than for public affairs content. It is not entirely clear why this should be the case, however. The Internet is ideally suited to bring users up-to-date information on a range of topics. It is not clear why entertainment and features content should account for about as many page views (together, about 22%) as current affairs content. It seems likely that these topics should be about equally likely to be the subject of breaking news online, so it is not apparent why public affairs news does not enjoy the advantage one might expect on the basis of previous research. If people were separating their topical reading by medium, it would surface in the survey data as a negative relationship between self-reports of attending to a topic in the traditional media and attending to it online. A post hoc analysis of the relationships between these variables revealed significant positive relationships. Internet news readers who reported following a topic in the traditional media were likely to have reported getting news on that topic online as well. For example, only 26.1% of online news readers who do not regularly followed politics in the traditional media followed it online, whereas 56.3% of online readers who followed politics in the traditional media also followed it online, χ2(1, 2145) = 172.59, p < .01. However, it is worth noting that 43.7% of online news readers who followed politics in the traditional media did not claim to follow it online. Thus, there is only mixed evidence of medium-specific topic selection in these data. A second potential explanation is that the set of 13 sites studied here could misrepresent the average Internet news-reading experience, but it is not clear how that sort of bias might work. Most of the page views studied here were to sites that typically feature a broad representation of the available news. Previous studies of newspapers have observed a strong correspondence between online and offline editions (e.g., Chyi & Lasorsa, 2002; Martin, 1998), with organizations often accused of shoveling their print content onto the Internet. Thus, there is little reason to believe that something about the content of these major sites dramatically affected the results of this study. On the other hand, one might suggest that the disproportionate influence exerted by the most popular sites here (e.g., CNN) might have meant that the viewing patterns at smaller sites, at which people were perhaps paying more attention to public affairs news, were being obscured. One way to test this suggestion would be to calculate a separate page view distribution that gives equal weight to all 13 sites and then compare that to the distribution reported in the first column of Table 2. In an analysis not shown, such a comparison showed no improvement in the relative frequency with which public affairs news was selected. 706 Internet News Reading A third possibility, of course, is that people do not, in fact, have a deep and abiding interest in public affairs news, at least not as deep as one might gather from the results of some survey studies. Prior research (e.g., Fico et al., 1987; Tewksbury & Althaus, 2000) has suggested that online news formats encourage people to follow their particular interests once they are online. Thus, the results here may speak in some measure to the core interests of online news audiences. In addition, the breadth and stability data suggest that online news readers display a relatively narrow range of interests. The average reader observed here viewed one topic and read about it on 1 day on these sites and in these months. As a result, people do not seem to be broad readers of the news on the Internet. This pattern seems to reflect the unique attributes of the Web that allow people to select the content they want and to ignore the rest. This finding has implications for the frequently heard concern that media audiences may be increasingly fragmented in the emerging new media environment (Katz, 1996). The study results also may have implications for future patterns of public affairs news exposure. Studies have consistently found that people generally do not replace their use of the traditional media when they begin receiving news on line (e.g., Althaus & Tewksbury, 2000). However, this research has typically not been longitudinal in nature. It is possible that American news audiences are in the midst of a cohort replacement process in which successive groups of news readers are gradually shifting aggregate news exposure away from traditional outlets and toward new media. A parallel process has been underway in traditional media use in the United States for many years. As successive age cohorts have entered and left the media audience, television use has risen and newspaper reading has fallen (Peiser, 2000). If that process is at work with respect to Internet news and its offline counterparts, the present findings may portend a gradual lessening of popular exposure to public affairs news. That possibility carries normative significance when placed in the context of the expectations that democratic theory places on citizens (e.g., Berelson, 1952). Among the limitations of this study is its inability to speak to the relationship between a site’s design and its use. The location of news links on a page can be predictive of the frequency with which the news content is selected (Eveland & Dunwoody, 1998), and there are undoubtedly a number of other factors that can influence selections. Unfortunately, the page view records were analyzed 1 year after they were collected. The location of particular stories on a home page can vary from day to day, and more importantly, many of the sites had undergone at least partial redesigns in the intervening period. Thus, it was not possible to precisely reconstruct the appearance of the sites on the days the panelists used them. As a consequence, the present data speak most directly to aggregate patterns of news attention. This is assuming, of course, an adequate diversity of content and designs across the sample of sites. Another question for this study is the extent to which the news-reading habits observed here may be generalized to Internet users and the general population. The NetRatings sample’s demographic characteristics are generally reflective of the Internet population. However, Internet news users are likely different than the 707 Journal of Communication, December 2003 general news population on a number of dimensions. For example, one study of early Internet adopters found they were more interested in technology, business, national, and international news than were most Americans, but they were less interested in weather and local news (Radio and Television News Directors Foundation, 1996). Thus, it may be that some portion of observed differences between the present data and earlier studies of news preferences and habits may be due to the unique nature of the online audience. However, it may not be wise to place substantial weight on that explanation. The online audience is getting both larger and more diverse over time (Cyberatlas, 2001). Thus, there are fewer differences between those offline and online today than there were only 7 years ago. Future studies need to expand on the results reported here and explore relationships between individual-level characteristics (e.g., demographics) of online news readers and their news selection patterns. Future research should also examine a greater segment of Internet users’ Web behavior, and it should more carefully investigate patterns of online news exposure. One goal of such research might be the development of general profiles of online news readers and their interests. Such a typology might shed some light on where people go for their news online and how separate content areas fit within larger news consumption patterns. References Althaus, S. L., & Tewksbury, D. (2000). Patterns of Internet and traditional news media use in a networked community. Political Communication, 17, 21–45. Althaus, S. L., & Tewksbury, D. (2002). Agenda setting and the “new” news: Patterns of issue importance among readers of the paper and online versions of the New York Times. Communication Research, 29, 180–207. Abramson, J. B., Arterton, F. C., & Orren, G. R. (1988). The electronic commonwealth: The impact of new media technologies on democratic politics. New York: Basic Books. Barringer, F. (2001, August 27). Growing audience is turning to established news media online. New York Times, p. C1. Bennett, L. W. (2001). News: The politics of illusion (4th ed.). New York: Addison Wesley Longman. Berelson, B. (1952). Democratic theory and public opinion. Public Opinion Quarterly, 16, 313–330. Bogart, L. (1989). Press and public: Who reads what, when, where, and why in American newspapers (2nd ed.). Hillsdale, NJ: Erlbaum. Burgoon, J. K., Burgoon, M., & Wilkinson, M. (1983). Dimensions of content readership in 10 newspaper markets. Journalism Quarterly, 60, 74–80. Campbell, A., Converse, P. E., Miller, W. E., & Stokes, D. (1960). The American voter. New York: Wiley. Chaffee, S. H., & Metzger, M. J. (2001). The end of mass communication? Mass Communication & Society, 4, 365–379. Chaffee, S. H., & Schleuder, J. (1986). Measurement and effects of attention to media news. Human Communication Research, 13, 76–107. Chyi, H. I., & Lasorsa, D. (1999). Access, use and preferences for online newspapers. Newspaper Research Journal, 20(4), 2–13. 708 Internet News Reading Chyi, H. I., & Lasorsa, D. (2002). An explorative study on the market relation between online and print newspapers. Journal of Media Economics, 15, 91–106. Clark, R. (1979). Changing needs of changing readers: A qualitative study of the new social contract between newspaper editors and readers. Charlotte, NC: American Society of Newspaper Editors. Converse, P. E. (1964). The nature of belief systems in mass publics. In D. A. Apter (Ed.), Ideology and discontent (pp. 206–261). New York: Free Press. Corrado, A. (1996). Elections in cyberspace: Prospects and problems. In A. Corrado & C. M. Firestone (Eds.), Elections in cyberspace: Toward a new era in American politics (pp. 1–31). Washington, DC: Aspen Institute. Cyberatlas. (1999). Newspaper sites draw news traffic. Retrieved December 1, 2000, from http:// cyberatlas.internet.com/big_picture/traffic_patterns/article/0,,5931_152021,00.html Cyberatlas. (2001). Online consumers now the average consumer. Retrieved October 20, 2001, from http://cyberatlas.internet.com/big_picture/demographics/article/0,,5901_800201,00.html Delli Carpini, M. X., & Keeter, S. (1996). What Americans know about politics and why it matters. New Haven, CT: Yale. Eveland, W. P., Jr., & Dunwoody, S. (1998). Users and navigation patterns of a science World Wide Web site for the public. Public Understanding of Science, 7, 285–311. Fico, F., Heeter, C., Soffin, S., & Stanley, C. (1987). New wave gatekeeping: Electronic indexing effects on newspaper reading. Communication Research, 14, 335–351. Graber, D. (1988). Processing the news: How people tame the information tide (2nd ed.). New York: Longman. Jeffres, L. W., & Atkin, D. J. (1996). Dimensions of student interest in reading newspapers. Journalism & Mass Communication Educator, 51(3), 15–23. Katz, E. (1996). And deliver us from segmentation. Annals of the American Academy of Political and Social Science, 546, 22–33. Kuklinski, J. H., & Quirk, P. J. (2001). Conceptual foundations of citizen competence. Political Behavior, 23, 285–311. Margolis, M., & Resnick, D. (2000). Politics as usual: The cyberspace “revolution.” Thousand Oaks, CA: Sage. Martin, S. E. (1998). How news gets from paper to its online counterpart. Newspaper Research Journal, 19(2), 64–73. McCombs, M. E., & Poindexter, P. (1983). The duty to keep informed: News exposure and civic obligation. Journal of Communication, 33(2), 88–96. Munro, J. B., & Weaver, D. H. (1977). Patterns of newspaper readership. ANPA News Research Report, 4, 1–4. Nielsen//NetRatings. (2002). Hot off the net. Retrieved May 20, 2002, from http://www.nielsennetratings.com/hot_off_the_net.jsp Patterson, T. E. (1993). Out of order. New York: Knopf. PC Data. (2000). Top monthly Web sites by category (single month). Retrieved June 27, 2000, from http://www.pcdataonline.com Peiser, W. (2000). Cohort trends in media use in the United States. Mass Communication & Society, 3, 185–205. 709 Journal of Communication, December 2003 Pew Research Center for the People and the Press. (1999). Online newcomers more middle-brow, less work-oriented; The Internet news audience goes ordinary. Washington, DC: Author. Pew Research Center for the People and the Press. (2000). Internet election news audience seeks convenience, familiar names. Washington, DC: Author. Poindexter, P. M., & McCombs, M. E. (2001). Revisiting the civic duty to keep informed in the new media environment. Journalism & Mass Communication Quarterly, 78, 113–126. Price, V., & Czilli, E. J. (1996). Modeling patterns of news recognition and recall. Journal of Communication, 46(2), 55–78. Price, V., & Zaller, J. (1993). Who gets the news? Alternative measures of news reception and their implications for research. Public Opinion Quarterly, 57, 133–164. Radio and Television News Directors Foundation. (1996). News in the next century: Profile of the American news consumer. Washington, DC: Author. Rarick, G. R. (1967). Field experiments in newspaper item readership. Eugene: Division of Communication Research, School of Journalism, University of Oregon. Schramm, W., & White, D. E. (1949). Age, education, economic status: Factors in newspaper reading. Journalism Quarterly, 26, 149–159. Schudson, M. (1998). The good citizen: A history of American civic life. Cambridge, MA: Harvard University Press. Stone, G. C., & Boudreau, T. (1995). 1985, 1994: Comparison of reader content preferences. Newspaper Research Journal, 16(4), 13–28. Stone, G. C. (1987). Examining newspapers: What research reveals about America’s newspapers. Newbury Park, CA: Sage. Sunstein, C. R. (2001). Republic.com. Princeton, NJ: Princeton University Press. Tewksbury, D., & Althaus, S. L. (2000). Differences in knowledge acquisition among readers of the paper and online versions of a national newspaper. Journalism & Mass Communication Quarterly, 77, 457–479. Weaver, D. H. (1978). Recent trends in newspaper readership research. Bloomington: School of Journalism, Center for New Communications, Indiana University. 710