Content-Type: text/html Public Trust or Mistrust?: Perceptions of Media Credibility in the Information Age By Spiro Kiousis Doctoral Student Department of Journalism University of Texas Austin, TX 78712 (512) 719-5572 [log in to unmask] Submitted to the AEJMC annual conference in New Orleans, LA August 4-7, 1999 * The author would like to thank Dr. Nick Lasorsa and Dr. Jim Tankard for their helpful comments on earlier versions of this paper. Public Trust or Mistrust?: Perceptions of Media Credibility in the Information Age Abstract This paper explores perceptions of news credibility for television, newspapers, and online news. A survey was administered to a randomly selected sample of residents in Austin, Texas, to assess people's attitudes toward these three media channels. Contingent factors that might influence news credibility perceptions, such as media use and interpersonal discussion of news, were also incorporated into the analysis. Findings suggest that people are generally skeptical of news emanating from all three media channels but do rate newspapers with the highest credibility, followed by online news, and television news respectively. Furthermore, opinions about news credibility seem to be correlated across media outlets. The data show a moderate linkage between interpersonal discussion of news and perceptions of media credibility for television news but not for newspapers and online news. Finally, a marginal association between media use and public perceptions of credibility for newspapers and television was found but not for online news. Submitted to the AEJMC annual conference in New Orleans, LA August 4-7, 1999 Public Trust or Mistrust?: Perceptions of Media Credibility in the Information Age Public Trust or Mistrust?: Perceptions of Media Credibility in the Information Age Background Introduction Media credibility research has been an integral part of mass communication scholarship since the field's earliest days. Whereas the seminal work on credibility concentrated on dimensions of source credibility (e.g., Hovland & Weiss, 1951), more contemporary literature has highlighted variations in the perceived credibility attributed to different channels (e.g., Rimmer & Weaver, 1987). Despite this expanded scope in research, the realm of online news has just recently been explored in media credibility analyses (e.g., Johnson & Kaye, 1998). Though helpful, such inquiries are restricted because they have normally been based on samples of Internet users. Indeed, Johnson & Kaye (1998), who performed a survey of news credibility on politically interested online users, recommend that "future studies could be conducted among the general population to determine the degree to which the Internet is viewed as credible and whether it is indeed judged as more trustworthy than traditionally delivered counterparts" (p. 335). Accordingly, the purpose of this project is to ascertain how credible people believe online news to be in comparison to television and print news through a probability sample of the general population in one city. To answer this broad research question, a survey was administered to a randomly selected sample of residents in Austin, Texas, to assess people's perceptions of news credibility. Specifically, media credibility was gauged by asking respondents how they rated news from television, newspapers, and the Internet in terms of factuality, profit making, privacy, community well-being, and trustworthiness. Contingent factors that might affect perceptions of news credibility were also examined, including media use and interpersonal communication patterns. Source Credibility Before delving into more pertinent literature on the credibility of various media channels,[1] it is critical that we review the general literature on source credibility because most studies probing channel credibility are anchored in this fundamental theoretical framework. The earliest research on credibility investigated how modifications in source characteristics influenced people's willingness to alter their attitudes on certain topics (e.g., Hovland et al., 1953; Hovland et al., 1949). Carl Hovland and his associates ran a series of experiments to determine what combinations of communicator qualities induced attitude change in subjects. Usually, the steps involved in their research were measuring subject attitudes on an issue, exposing them to manipulated messages based on different source qualities, and then measuring attitude change after the initial exposure. Source expertise and trustworthiness were envisaged as two central attributes of source credibility. Source, in these early investigations, was typically defined as an individual mass communicator, such as a newscaster or world leader. However, Hovland et al. (1953) did acknowledge that "the impact of a message probably depends also upon the particular publication or channel through which it is transmitted" (p.19). Therefore, even in these inaugural explorations, researchers differentiated between attitudes toward the communicator and opinions about the medium. This is a pivotal distinction upon which the current project hinges because aspects of source credibility are not to be confused with elements of medium credibility, although the two obviously do overlap. The mass communication theory that underscored much of the early source credibility work was the two-step flow model of communication. This theory posits that mass media indirectly affect audiences by exerting influence on a few "opinion leaders" in individual communities, who in turn, may shift the attitudes of others within those same communities (Lazarsfeld et al., 1944; Lazarsfeld & Merton, 1948; Berelson et al., 1954). The relevance of this model to credibility research is that it synthesized how interpersonal communication about news might shape people's perceptions of credibility. It suggests that reliance on interpersonal communication is associated with reliance on mass media, possibly transforming people's perceptions of credibility with the press (a question that will be scrutinized in the current project). It should be mentioned, however, that the two-step flow model has been strongly challenged by scholars (e.g., Rogers & Shoemaker, 1971; Lin, 1971; Chaffee & Hochheimer, 1985), yet is germane because it offers a conceptual base that merges the impact of media exposure and interpersonal discussion onto mass communication processes. Despite the many contributions flowing from the Hovland Model, scholars soon criticized its vision of source credibility as too simplistic and static (Markham, 1968). Berlo et al. (1970), for instance, presumed that source credibility was "multidimensional" and that the factors identified by Hovland had no theoretical foundation, nor had they withstood the rigors of empirical testing. Through a series of factor analyses of semantic differential scales, Berlo and his colleagues claimed that source credibility entailed safety, qualifications, and dynamism.[2] Meanwhile, Whitehead (1968) added competency and objectivity as vital components of credibility.[3] Indeed, one overarching pattern that has resonated in credibility research is that academics have failed to fully agree on the core dimensions of the concept. More recent work on source credibility monitors public perceptions of political candidates and individual journalists (e.g., Andsager, 1990; Pfau & Kang, 1991). The primary method for such inquiries has been the controlled experiment in the Hovland tradition. Slater & Rouner (1996), for example, submit that the internal characteristics of messages, such as their aesthetic presentation or their actual content, can also sway perceptions of source credibility. Hence, credibility appears to be a variable that can be studied within the context of the communicator, channel, or message itself. In addition, discussions attempting to deal with the methodological problems of operationalizing the intricate concept continue to linger (Infante, 1980). While a great deal of knowledge has been accumulated from the experimental approach of designating credibility as an attribute of the source, to understand differences in public opinion concerning channel credibility, we must also visit the vast survey research dedicated to this topic. Medium Credibility Westley and Severin (1964) are credited with conducting the first comprehensive analysis of news credibility across media outlets. In their classic study, the scholars noted that certain demographic variables (e.g., age, education, & gender) mediate people's perceptions of news credibility. Moreover, they discriminated between perceptions of media credibility and media preference. In other words, people did not always feel their most preferred medium was the most credible. They also found that television news was typically deemed more accurate than print news. Some of the more important issues that surfaced from this examination were: (1) to what extent did media use influence news credibility ratings; (2) was the trend for high television credibility consistent; and (3) were people's general perceptions of credibility uniform across media channels (questions pursued in this study as well). Media Credibility Scales Paralleling the literature on source credibility, news media credibility research has also been plagued by methodological concerns. The primary debate revolves around how many factors news credibility scales actually measure. The Gaziano & McGrath (1986) news credibility index is the most popular, although some controversy has been generated from its application. In their initial analysis, Gaziano & McGrath performed a factor analysis that located one dominant factor of news credibility. Meanwhile, Meyer (1988) alleged that their scale actually gauges two properties of news credibility: believability and community affiliation. West (1994), attempting to compare the reliability of both models using covariance structure modeling, found that Meyer's believability dimension was reliable but the community affiliation dimension was not. In addition, West observed that the Gaziano & McGrath scale gauged more than one factor. On the other hand, Rimmer & Weaver (1987) report a Cronbach's alpha score of .90 for a credibility scale based on Gaziano & McGrath's work, implying that it does measure a single construct. Thus, media credibility seems to be a concept that can be measured through a composite index or can be subdivided into several smaller subvariables, depending upon the combination of questions utilized (a problem addressed in the methods section). Media Use The early work on news credibility established that increases in media use were usually accompanied by enhanced credibility ratings for whatever channel was being scrutinized (Westley & Severin, 1964; Greenberg, 1966; Shaw, 1973; Cobbey, 1980). In contrast to this previous scholarship, however, Rimmer and Weaver (1987) conjecture that media use is not strongly correlated with media credibility, asserting that the type of question employed to estimate media use may be distorting results. Specifically, they maintain that affective questions of use (e.g., "Which medium do you prefer?") are normally being reported and that actual frequency of media use measures are not necessarily connected with news credibility ratings. Wanta & Hu (1994), in an investigation of agenda setting, found no significant linkage between exposure (frequency of use) and credibility but did find a relationship between reliance (how dependent respondents said they were on a particular medium for information) and credibility.[4] Nevertheless, the impact of media use remains enigmatic, especially considering the rise of interactive media. Interpersonal Discussion Since mass media use has been linked to news credibility evaluations, it seems that interpersonal communication behavior would also be relevant to forming such opinions, yet it has only received minimal attention in empirical investigations. Though media use is sometimes positively related to perceptions of news credibility, one might surmise that a similar relationship would exist with interpersonal discussion of news because interpersonal communication behavior is often connected to media use (Atkin, 1972; Chaffee & McLeod, 1973; Lenart, 1994). For example, Kraus & Davis (1976) describe a complementary linkage between mass and interpersonal communication patterns, where people are believed to supplement the information they secure from mass media with similar information from interpersonal channels. However, the complementary patterns of use do not necessarily lead to parallel perceptions of credibility. In fact, some researchers have concluded that expanded interpersonal communication undermines the influence of mass media on attitudes (e.g., Erbring et al., 1980; Atwater et al., 1985). Specific to credibility, Chaffee (1982) states that "several surveys have reported null or even negative correlations between channel use and credibility; generally interpersonal communication has been more prevalent in these cases" (p.63). Moreover, McLeod et al. (1968-69) concur that interpersonal communication patterns may account for the limited impact of media use on news media credibility. Therefore, the literature suggests that there is an inverse correlation between interpersonal communication and news media credibility, although few studies have ever empirically tested this hypothesis. Channel Credibility Competition Several analysts indicate that television news is more credible than newspapers (Carter & Greenberg, 1965; Lemert, 1970; Gaziano & McGrath, 1986). Indeed, Westley & Severin's (1964) seminal study details that the "Roper's report showed more people believe the news they learn from television than the other media..." (p.325), a trend that has remained relatively stable in Roper polls over time. In a more streamlined study of media credibility and news coverage of natural disasters, Major & Atwood (1997) found that television credibility did not decay in comparison to newspaper credibility when predicted natural disasters failed to occur, hinting that opinions of television credibility may be more stable than newspaper credibility perceptions. Newhagen & Nass (1989) reason that the discrepancy in credibility ratings is partially caused by the alternative standards people utilize to evaluate television as opposed to newspapers. People tend to judge the journalists delivering the news on television, but they assess the institution of the newspaper itself when analyzing print media. Therefore, opinions of television should be more favorable because survey respondents ally news anchors to television credibility, in contrast to the nameless institution they link with newspaper credibility. In short, Newhagen & Nass (1989) assert that television news credibility is most influenced by local broadcasts, demonstrating that television use might best be calibrated by monitoring both local and national viewing patterns (an approach adopted for this study). A final factor that may explain the conflicting ratings is the superior technological quality of television news in comparison to print news. As alluded to earlier, Slater & Rouner's (1996) discussion of message quality proposes that peripheral characteristics about messages (e.g., their aesthetic quality) may enhance source credibility. As a result, in the case of television vs. print news, the ascent of television credibility ratings may have mirrored the improvements in the technology of news transmittal over this medium. Of course, the influence of web news (with its blending of newspaper and television news attributes) on media credibility endures as a relatively unchartered frontier. There is also contradictory evidence with regard to how the credibility of Internet information compares to traditional media, though evidence is scarce at this point. In one of the first investigations of news credibility on the Internet, Johnson & Kaye (1998) documented that online media are perceived as more credible than print media, but both are judged as "somewhat" credible. However, it is difficult to generalize their results because their sample consisted only of politically interested Internet users and because their study focused solely on public affairs information. On the other hand, some research has found traditional media to be judged more credible than the Internet (e.g., Pew Research Center, 1996). Elsewhere, analysts have found mixed evidence concerning the credibility of traditional and online media channels within the same survey (e.g., Mashek et al., 1997). In an experimental setting, Sundar (1998) discovered that the credibility of online news could be enhanced by including quotes in stories (a technique used by traditional media outlets), suggesting that channel credibility perceptions are similar across channels. Due to these inconclusive findings, more empirical work is needed to understand the dynamics of news credibility attitudes across media channels. Uniformity of Credibility Perceptions Whereas previous scholarship elucidates that levels of perceived news credibility vary across channels, there is also some evidence that the general direction of these opinions is related. For example, Gaziano & McGrath (1986) and Newhagen & Nass (1989) convey that television and newspaper credibility are positively correlated. Further, Chaffee (1982) and Wanta (1997) articulate that people often seek out mass media with homogeneous content in order to reinforce information they have just learned. As a consequence, this homogeneity in content may partially explain why people's opinions of news credibility are fairly consistent, since the news content they consume is also often homogenous itself. This by no means indicates that such perceptions would be exactly the same, just that they would be correlated. Mapping out Credibility According on the discussion above, we see that the boundaries between channel and source credibility are somewhat blurred but are linked given their common conceptual foundation. To sketch out their relationship, we argue that credibility is defined as a characteristic of the media channel, which in turn, is a component of the larger concept of source credibility. This depiction is simultaneously based on the evolution of news credibility research from the larger body of source credibility scholarship and on Hovland et al.'s (1953) contention that source credibility is at least, to some degree, shaped by the channel disseminating the message. Figure 1 illustrates this conceptualization. - - - Figure 1 here - - - This explication is more expansive than prior models (that either do not distinguish between the terms or do not recognize their common theoretical framework) because it allows for a separation of the concepts but admits that they do overlap. Accordingly, this project essentially measures opinions about media channels, but perceptions of individual sources (e.g., news anchors) will inevitably also impact such opinions. Still, we are not exhaustively gauging channel credibility per se because not all print, broadcast, and online media are accounted for. While comparing credibility differences within channels would be helpful (e.g., newspapers vs. magazines for print media), we felt the broader channel comparison were more suitable for an exploratory study such as this. Significance of the Study One key question that comes to the forefront at this juncture in news credibility research is how will the proliferation of new technologies shift public perceptions of news credibility? Will the credibility of newspapers continue to plummet or will the presence of competing news sources cut into people's trust of television news? The primary contribution of this study will be to focus on such issues. Secondly, it is hoped that this project can also serve as replication for previous trends that have been documented in media credibility research. Ultimately then, this project should bolster our comprehension of news credibility in the information age. Hypotheses & Research Questions According to the literature comparing television and newspaper credibility ratings (e.g., Carter & Greenberg, 1965): H1: Television news will be judged as more credible than newspapers. Based on the conclusions of Gaziano & McGrath (1986) and Chaffee (1982) regarding people's tendency to seek out homogenous content from mass media: H2: People's perceptions of news credibility will be positively correlated across all three media. (Please note that this hypothesis does not contradict Hypothesis 1 because it speculates only on the direction of media credibility ratings, not the differences among them). Building on the literature suggesting that interpersonal communication behavior reduces trust in media outlets (e.g., McLeod et al., 1968-69): H3: The amount of interpersonal communication respondents devote to discussing news will be negatively correlated to their perceptions of news credibility. To assess the relationship between media use and perceived credibility and to provide credibility ratings across media outlets, the following research questions were also advanced: RQ1: Will media use be related to perceived credibility? RQ2: Which type of news medium will respondents rate most credible? Method Data Collection The data were obtained from a cross-sectional survey conducted on a sample of randomly selected residents in Austin, Texas, between February 26 and March 8, 1998. Students from two undergraduate and one graduate course administered the survey from the University of Texas at Austin. All students were trained in interviewing protocol and on the ethics of survey research. A technique called "plus-one random digit dialing" was implemented to acquire a representative sample (Poindexter & McCombs, 1997). Specifically, the sample was based on telephone numbers in the "Greater Austin" Southwestern Bell phone book. Respondents were selected by utilizing a computer program that randomly generated page, column, and line numbers in the phone book for interviewers to locate. Once found, 1 was added to the last digit of the base telephone number and the interview was attempted (this step ensured that unlisted phone numbers would also be included in the sample). If unsuccessful, the interviewer would try back up to five times before repeating the process for a new number. Comparisons to 1990 Census data show that the collected sample adequately represented the area's population (see Appendix A for breakdowns of sample and 1990 Census data). The efficiency of reaching potential households in the survey was 66%.[5] Of those reached, 818 people finished the survey for a completion rate of 61%.[6] All figures fall within acceptable parameters for survey data. Media Credibility Index To test the hypotheses and probe the research questions, survey respondents were asked their opinions on news credibility for television, newspapers, and online news. While media credibility has been gauged in several ways, five indicators that consistently emerge in research are: how factual a medium is, the extent to which it is motivated by money, whether it invades people's privacy, what is its concern for the community, and whether it can be trusted (Meyer, 1988; Gaziano & McGrath, 1986; Wanta, 1997; Johnson & Kaye, 1998). Thus, these items comprised the media credibility index employed for this study. This abbreviated scale was used because some researchers have suggested that the original Gaziano & McGrath scale might be redundant and lengthy (e.g., Meyer, 1988; Rimmer & Weaver, 1987). At the same time, these questions were comprehensive enough to touch on the believability and community affiliation aspects of credibility employed by other scholars (e.g., Meyer, 1988; Wanta & Hu, 1994). For the items mentioned above, a composite index of credibility was created because adequate reliability scores were attained only when aggregating all questions together rather than subdividing them into groups (to be presented below). In particular, respondents were asked whether they strongly agreed, agreed, disagreed, or strongly disagreed with the following statements for these media: television news, daily newspapers, and online news. Interviewers asked the questions in the following manner. We'd like you to think about the DAILY NEWSPAPER you are most familiar with. Please tell me whether you strongly disagree, disagree, agree, or strongly agree with each statement. THE DAILY NEWSPAPER: Is factual. Is concerned about making profits. Invades people's privacy. Is concerned about the community's well being. Cannot be trusted. Now, please tell me whether you strongly disagree, disagree, agree, or strongly agree with these statements about TV news. TV NEWS: Is factual. Is concerned about making profits. Invades people's privacy. Is concerned about the community's well being. Cannot be trusted. Now, please tell me whether you strongly disagree, disagree, agree, or strongly agree with these statements about ONLINE news. ONLINE NEWS: Is factual. Is concerned about making profits. Invades people's privacy. Is concerned about the community's well being. Cannot be trusted. Based on respondent agreement or disagreement with these statements, a four-point Likert scale (0 to 4) was developed for each question to reflect the various aspects of credibility. For example, someone who said they "strongly agreed" with the statement that newspapers were factual would yield a score of 4 for that particular item. These scales were then collapsed into a credibility index for each medium (television, newspaper, and web). Because certain questions were reversed to avoid response sets (Poindexter & McCombs, 1997), items were recoded as necessary so that higher scores would signify higher levels of perceived credibility. Cronbach's alpha for the newspaper, television, and web credibility indices were .60, .63, and .49 respectively. The .49 score for web credibility is probably lower because these scales were originally intended for newspaper and television news, not Internet news.[7] To measure media use behavior, the following questions were posed to respondents. About how often do you watch local evening TV news? 1) never or seldom, 2) 1-2 days a week, 3) 3-4 days a week, 4) nearly every day, 5) every day. About how often do you watch network evening news? 1) never or seldom, 2) 1-2 days a week, 3) 3-4 days a week, 4) nearly every day, 5) every day.[8] How often would you say you read the daily newspaper? 1) never or seldom, 2) 1-2 days a week, 3) 3-4 days a week, 4) nearly every day, 5) every day. How often do you use the world wide web? 1) never, 2) 1-5 hours a week, 3) 6-10 hours a week, 4) 11-15 hours a week, 5) 16-20 hours a week, 6) more than 20 hours a week. (This media use item was worded differently because another researcher involved in the survey was analyzing Internet use in terms hours per week. Hopefully, this should not disturb the current examination because the item still measures frequency of use).[9] To measure interpersonal discussion of news, the following question was given to participants: How often do you discuss news with your friends and family? 1) never or seldom, 2) 1-2 days a week, 3) 3-4 days a week, 4) nearly every day, 5) every day. Thus, each of the items measuring media use and interpersonal communication formed Likert scales for the data analysis phase of the research project. To ascertain respondent perceptions about the definition of online news, the following question was provided to participants: When you think of online news, which of these do you most think of? (1) news websites, (2) news and discussion groups, (3) chatrooms, (4) email listservs, (5) other. (respondents were asked to choose one) This question was asked to outline what facets of the Internet people considered to be online news. Data Analysis For the first hypothesis, a correlated groups t-test was chosen to compare the mean scores of the television and newspaper credibility indices. This test was selected because the assumption of independent samples for an ordinary t-test would have been violated. This allowed us to enjoy the advantages of applying a t-test without resorting to a less powerful statistical procedure. For Hypotheses 2 and 3 and Research Question 1, Pearson r correlations were utilized to analyze the data. Research Question 2 was analyzed using a repeated measures ANOVA test. This test essentially extends the model of the correlated groups t-test and was therefore selected to evaluate all three credibility scales in a simultaneous analysis. Before continuing, it should be mentioned that significance levels were adjusted for this analysis because of the large sample size. Specifically, because a large sample increases the possibility of Type 1 error (the probability of rejecting the null hypothesis when it should not be), we adopted a stricter threshold (p<.001) for determining statistical significance. As a result, findings at the p<.05 or p<.01 level were viewed as "approaching" significance or being "marginally" correlated. Results Before attending to the specific hypotheses and research questions, it would be beneficial to examine some of the exploratory statistics for the key variables included in the data analysis. Table 1 is offered below to summarize the individual variables. - - - Table 1 about here - - - The large number of people who "didn't know" about their opinions toward online news might be explained by the sparse time respondents spent on the Internet despite their 80 percent access rate to it. For example, over half of the sample (62.5 percent) spent less than five hours per week on the web or did not use it at all, limiting their exposure to online news.[10] To view the range of responses for credibility, the percentage distribution of credibility ratings is presented in Table 2. - - - Table 2 about here - - - As far as definitions of online news, the most popular answers were news websites (327 respondents=53%), email listservs (180 respondents=29%), news & discussion groups (52 respondents=8%), chatrooms (36 respondents=6%), and other (28 respondents=5%).[11] Thus, a majority of people agreed on its meaning, but definitions were certainly not unanimous. In essence, this finding indicated that when we asked about people's perceptions of online news, we were mainly monitoring attitudes towards news websites. H1: Television news will be judged as more credible than newspapers. According to the descriptive statistics reported above, it appears that this hypothesis was not supported because newspapers were perceived to be more credible than television news (see Table 1), but a correlated groups t-test was also performed to resolve if the difference was statistically significant. The correlated groups t-test show that the difference is indeed significant (t=11.39; df=780; p<.001). Therefore, Hypothesis 1 was not supported but was significant in the opposite direction. H2: People's perceptions of news credibility will be positively correlated across all three media. Pearson r correlations show that a positive relationship exists between people's opinions of credibility across all three modalities. In particular, the correlation between television and newspaper credibility is .58 (p<.001; n=781). The correlation between television and web news credibility is .45 (p<.001; n=467). Finally, the correlation between newspaper and web news credibility is .43 (p<.001; n=468). As a result, Hypothesis 2 seems to be strongly supported by the data. H3: The amount of interpersonal communication respondents devote to discussing news will be negatively correlated to their perceptions of news credibility. Pearson r correlations were again calculated to determine if this inverse relationship exists. Based on the significance level of .001 advanced earlier, interpersonal discussion about news and television credibility appear is only modestly associated (r=-.11; p<.01). No linkage existed between interpersonal discussion about news and perceptions of credibility for newspapers and web news. Thus, the data only moderately confirm Hypothesis 3 for television news credibility but do not for newspaper and web news credibility. RQ1: Will media use be related to perceived credibility? The data reveal that newspaper readership is marginally correlated to newspaper credibility (r=.09; p<.05). Likewise, the correlation between viewing local broadcast news and television news credibility is also marginally associated (r=.07; p<.05). Conversely, no linkage occurred between national network news viewing and television credibility. Moreover, no relationship existed between web use and web news credibility. As a consequence, the data display some very small linkages between media use and credibility for local media outlets (i.e., local newspaper and nightly broadcast news), although other variables certainly seem to intervene in the relationship. Table 3 encapsulates these findings. - - - Table 3 about here - - - Collectively, media use seems to be only minimally connected to perceptions of credibility for newspapers and television news but not online news. RQ2: Which type of news medium will respondents rate most credible? This research question attempts to identify the statistical significance of all three credibility indices simultaneously. A repeated measures ANOVA was executed to estimate the differences between means. [12] The results are reported below in Table 4. - - - Table 4 about here - - - The repeated measures ANOVA supplies evidence that the means are significantly different. To pinpoint exactly which means are different, paired group t-tests were utilized for post-hoc comparisons, a technique often recommended by other researchers (Hochberg & Tamahane, 1987). These comparisons confirm that all three means are significantly different from one another. Specifically, the t-values are: 11.39 (df=780; p<.001) for newspaper vs. television credibility; -3.70 (df=466; p<.001) for television vs. online news credibility; and 4.23 (df=467; p<.001) for newspaper vs. online news credibility. Thus, it seems that respondents thought newspapers were most credible (avg.=2.16), followed by Internet news (avg.=2.07), and television news (avg.=1.92) respectively. (Please note that means are slightly different from Table 1 because the ANOVA procedure only examines cases where data were available for all three indices). It should be remembered, however, that all three medium scores fall within the "moderately credible" category so the means are not radically different. Additional interpretation of the data will be advanced in the subsequent section. Discussion Methodological Concerns Before discussing the results, it is imperative that we review some of the investigation's methodological caveats. Perhaps the most troubling problem is that the sample size was much smaller for online news opinions than for the other two media channels. Still, this should be expected because the Internet is a relatively new medium that has not yet penetrated the entire population. The fact that Austin has one of the highest diffusion rates for Internet access made a traditional survey possible and is therefore, invaluable since most polls conducted about the Internet are self-selected with participants normally responding online. Another limitation involves the generalizibility of the sample because Austin residents are typically more educated than the rest of the nation yet are similar to other urban areas (Census, 1995). Of course, this should not discount the results, but we should be careful in drawing any sweeping conclusions about media credibility perceptions. Replication in rural settings would be helpful to substantiate findings. A final shortcoming that might be raised is that the online news questions inevitably were partially influenced by attitudes toward traditional news outlets because respondents looking at news in cyberspace may be visiting their favorite print newspaper or broadcast news website. One problem with this view, however, is that it assumes news websites are the only source of online news. Respondents' answers to the online news definition question showed otherwise (e.g., listservs represented 29 percent of definitions). In addition, this criticism seems to suggest that looking at online news is no different than experiencing it through traditional channels, when prior empirical work has shown this is not the case (e.g., Fredin, 1997). As a result, we did not perceive this to be a dramatic problem but do believe it should be acknowledged when interpreting the findings. Interpretation The data analyses seem to uncover many interesting relationships that need to be further discussed. First, one unforeseen outcome was that newspaper credibility ratings exceeded television news credibility ratings, in the face of the abundant empirical work that has recorded the opposite trend over past years (e.g., Lemert, 1970; Gaziano & McGrath, 1986). Another noteworthy finding was that Internet news was assigned more credibility than television news, especially when pondering this channel's infancy. However, some research has suggested that online material is judged more credible than information from television (Brady, 1996). One explanation for these results may be that the introduction of online sources has changed public opinion of traditional media credibility. Regardless of whether people actually have access to or use the web, its mere presence could have reduced trust in television while concomitantly increased trust in newspapers. The large body of "diffusion" literature submits that the dissemination of new technology often shifts opinions of older media (Rogers, 1983; Rogers & Shoemaker, 1971). An alternative interpretation may be that our sample deviated somewhat from the 1990 Census in terms of education, thereby accounting for these differences. The University of Texas at Austin, for instance, is now the largest university in the United States, and the booming high-tech industry in the surrounding area constantly attracts employees with higher educational levels. Thus, one could argue that this more educated sample would read more newspapers and use the Internet with greater frequency, which would justify the incongruity in credibility scores. A final interpretation could be that the differences were not that staggering simply because all three channels were judged "moderately credible." This corroborates the findings of Johnson & Kaye (1998), who discovered differences in news credibility across channels but a general skepticism of media (print & online for their study) across the board. Hence, though it is vital that we acknowledge there are multiple viewpoints pertaining to news credibility, we must also be aware that people seem to be somewhat apprehensive about all three media outlets. In summary, one could infer that survey respondents appear to moderately believe news that originates from all three media but do tend to give higher credence to newspaper stories before Internet news and Internet news before television news. The definition of online news seems to be a pervasive topic for audiences. Although the data show some agreement, consensus is far from being reached. Nonetheless, the definition breakdowns do demonstrate that the majority of people agreed on the meaning of the topic-they think it refers to news websites. Another substantial observation involved the detected correlation among perceptions of all three channels. This finding not only replicated the results of previous scholars (e.g., Gaziano & McGrath, 1986; Newhagen & Nass, 1989), who noticed a relationship between newspaper and television credibility, but extends this observation to the domain of Internet news credibility as well. Survey participants seem to perceive all three media in the same direction, indicating that people probably have an overall perception of news credibility that only slightly fluctuates across media. This verifies the t-test and ANOVA analyses that reported differences but exposed generally moderate perceptions of credibility for television, newspaper, and online news. Therefore, perceptions of news credibility are fairly uniform, but some distinctions are made among media outlets. The lack of support for the relationship between interpersonal communication and news credibility was not anticipated but might be explained by the social atmosphere or lack thereof that is present when people interact with these different media channels. For example, watching television is often thought of as a group experience where audiences integrate the information they secure from the medium into their daily conversations with others (e.g., Lull, 1990; McQuail et al., 1972). As a result, the increased amount of interpersonal communication undermines the credibility of the media channel, as proposed earlier by Chaffee (1982). In contrast, reading newspapers and going online are activities people ordinarily engage in alone, which diminishes the degree to which they promote interpersonal discussion. Accordingly, the individualistic media channels (newspapers and the Internet) found no linkage between news credibility and interpersonal communication. Of course, future work should explore whether such speculation is warranted. Generally, it appears that media use is marginally connected with perceptions of credibility. Measures of local television viewership were modestly associated with television credibility. Similarly, perceptions of newspaper credibility were modestly linked to newspaper readership behavior. On the other hand, correlations between credibility perceptions for network news and web use were not. In short, these data confirm the findings of Rimmer & Weaver (1987) because our "frequency" media use questions were not strongly allied to evaluations of news credibility. Nevertheless, respondents appeared to hold stronger credibility opinions about local news outlets than other types of media channels, an area that commands more attention in future scholarship. Implications Ultimately, it appears that further investigation of media credibility is demanded given the rapid growth of online news. The convergence of media is not only shaping media content but also seems to be linked to people's perceptions of media as well. The heightened position of newspapers in credibility ratings is encouraging for the newspaper industry, which has been waning in public opinion polls for many years. However, the overall mediocre ratings of media credibility should be disconcerting for professional journalists and mass communication researchers alike. More scrutiny of this pattern is necessary to determine if this observation was merely idiosyncratic to this survey, or has the rise of online media truly modified public perceptions of press credibility generally. Notwithstanding the limited linkage with interpersonal communication, future news credibility research should explore the consequences of interpersonal discussion to shed light on the discrepant findings found in this study across media channels. Such work might expand the concept of "media use" to a more generic term such as "communication use," thereby providing hybrid measures of media and interpersonal communication habits. Another implication is that further analyses are required to pin down definitions of online news. Will a universal definition be engendered or will conceptions continue to remain elusive? Such explorations can help us determine what criteria people apply to form opinions about Internet news credibility, patterned after the work of Newhagen & Nass (1989) on television news. The similarity of credibility perceptions for television, newspaper, and online news suggests that people may orient themselves to media content that is analogous across media channels, triggering parallel opinions of credibility. Perhaps more strategies should be developed to elevate press credibility as a whole rather than creating specialized recommendations for television, newspaper, and online news separately. A panel study comparing participant news exposure patterns, in terms of content, to their opinions of news credibility might help cultivate such knowledge. Further research might also explore whether the uniformity in credibility perceptions applies to sub-genres of news such as magazines and prime-time television news programs (e.g., Dateline). For instance, would differences be discerned if respondents were asked to give their opinions on local vs. network television news? In summary, the proliferation of new technologies has provided a fruitful forum for future academic work on media credibility. The similarities and deviations that arose from this study, in comparison to prior investigations, reflect the need for a resurgence of news credibility analyses. It is hoped that the present study can act as springboard for future intellectual discourse in this area of mass communication. Notes Figures & Tables Figure 1 [--- Pict Graphic Goes Here ---] TABLE 1: Credibility Scores Variable Means Std. Deviation Possible Range N TV News Credibility Index 1.94 .64 0-4 796 Web News Credibility Index 2.08 .62 0-4 473* Newspaper Credibility Index 2.17 .62 0-4 792 * Smaller N due to large number of respondents who answered "don't know" to questions about online news credibility. Higher scores indicate higher levels of perceived credibility. TABLE 2: Credibility Scores by Media Channels Low Cred. Moderate Cred. High Cred. Chi-Square N Newspaper 65 (8%) 588 (75%) 136 (17%) 612*** 789 (100%) TV 116 (15%) 588 (74%) 88 (11%) 598*** 792 (100%) Web 54 (12%) 353 (75%) 64 (14%) 367*** 471 (101%)* * Value exceeds 100 percent due to rounding *** p<.001 Note: The credibility categories were broken down as follows: 0-1.33=low credibility, 1.34-2.66=moderate credibility, and 2.67-4=high credibility. These categories were determined by simply dividing the four point range in credibility scores into three equal intervals (4/3=1.33 range per category). -------------------------------------------------------------------------------- ---------------------------- Table 3: Correlations between Media Use and Perceptions of News Credibility Newspaper Cred. TV Cred. Web Cred. Watch Local News r=.04; n=789 r=.07*; n=793 r=-.01; n=471 Watch Network News r=.01; n=788 r=-.01; n=792 r=.03; n=472 Read Newspapers r=.09*; n=791 r=0; n=795 r=.04; n=472 Use Web r=.02; n=562 r=-.02; n=565 r=.09; n=401 * p<.05 Table 4: Repeated Measures ANOVA for Mean News Credibility Scores Source of Variation SS DF MS F Sig. of F Within + Residual 179.83 922 .20 Credibility 13.71 2 6.85 35.14 p<.001 Note: Tests for Within-Subject Effects. Appendix A (Sample & Census Demographic Breakdowns) Age Austin Sample 1990 Census (percents) (percents) 18-24 23* 21 25-29 15* 15 30-34 14* 14 35-39 12* 12 40-44 10* 9 45-49 9* 6 50-54 5* 5 55-59 3* 4 60-64 3* 4 65+ 6 10 Race/Ethnic Group Caucasian/white 74 68 African-American/black 7* 10 Hispanic/Latino 12 19 Asian or Pacific Islander 4 .03 Native American 1* .003 Other 2 N/A Gender Male 48* 50 Female 52* 50 Income Under $10k 9 15 $10-$19k 11 18 $20-$29K 17* 16 $30-$39k 14* 13 $40-$49k 13 9 $50-$59k 9* 6 $60-$74k 10* 12 $75-$99k 9 5 $100k+ 8 4 Education Some high school 3 16 High school graduate 15 20 Some college/technical school 37* 34 College graduate 28 20 Some graduate school/graduate 7 10 * Sample statistic falls within confidence interval of plus or minus 3.5 percentage points at the 95% confidence interval. ** Special acknowledgment is given to Robyn Goodman for making the calculations for this appendix. Works Cited & Consulted Andsager, Julie L. (1990). Perceptions of credibility of male and female syndicated political columnists. Journalism Quarterly, 67, 485-491. Atkin, Charles K. (1972). Anticipated communication and mass media information seeking. Public Opinion Quarterly, 36, 188-99. Atwater, Tony, Salwen, Michael B., & Anderson, Ronald B. (1985). Interpersonal discussion as a potential barrier to agenda-setting. Newspaper Research Journal, 6(4), 37-43. Berelson, Bernard, Lazarsfeld, Paul, & McPhee, William. (1954). Voting. Chicago, Illinois: University of Chicago Press. Berlo, David K., Lemert, James B., & Mertz, Robert J. (1970). Dimensions for evaluating the acceptability of message sources. Public Opinion Quarterly, 33, 563-576. Brady, Dwight J. (1996). Cyberdemocracy and Perceptions of Politics: An Experimental Analysis of Political Communication on the World Wide Web. Paper presented at the annual meeting of the Midwest Association for Public Opinion Research, Chicago, Illinois. Carter, Richard, & Greenberg, Bradley. (1966). Newspapers or television: which do you believe? Journalism Quarterly 42, 29-34. Census (1995, March). Educational Attainment in the United States: March 1995. [online]. Available: http://www.census.gov/prod/2/pop/p20/p20489ta.pdf. Chaffee, Steven H. (1982). Mass media and interpersonal channels: competitive, convergent, or complementary? In Gumpert, Gary, & Cathcart, Robert (eds.), Inter/Media: Interpersonal Communication in a Media World. New York: Oxford University Press. Chaffee, S.H., & Hochheimer, J.L. (1985). The beginnings of political communication research in the United States: Origins of the "limited effects" model. In Rogers, Everett, & Balle, Francis (eds.), The Media Revolution in America and in Western Europe. Norwood, New Jersey: Ablex Publishing Co. Chaffee, Steven H. & McLeod, Jack. (1973). Individual vs. social predictors of information-seeking. Journalism Quarterly, 50, 237-45. Cobbey, Robin. (1980, December 15). Audience attitudes and readership. ANPA News Research Reports 29, 8-9. Erbring, Lutz, Goldenberg, Edie N., & Miller, Arthur H. (1980). Front-page news and real world cues: A new look at agenda-setting by the media. American Journal of Political Science, 24, 16-49. Fredin, E.S. (1997). Rethinking the news story for the Internet: Hyperstory prototypes and a model of the user. Journalism & Mass Communication Monographs 163, 1-47. Gaziano, Cecile, & McGrath, Kristin. (1986). Measuring the concept of credibility. Journalism Quarterly, 63, 451-462. Greenberg, Bradley. (1966). Media use and believability, some multiple correlates. Journalism Quarterly, 43, 665-732. Hochberg, Y., & Tamahane, A.C. (1987). Multiple Comparison Procedures. New York: Wiley. Hovland, Carl I., Lumsdaine, Arthur A., & Sheffield, Fred D. (1949). Experiments on Mass Communication: Studies in Social Psychology in World War II. New Haven: Yale University Press. Hovland, Carl I., & Weiss, Walter . (1951). The influence of source credibility on communication effectiveness. Public Opinion Quarterly, 15, 633-650. Hovland, Carl I., Janis, Irving L., & Kelley, Harold H. (1953). Communication and Persuasion. New Haven, Conn.: Yale University Press. Infante, Dominic A. (1980). The construct validity of semantic differential scales for the measurement of source credibility. Communication Quarterly, 28 (2), 19-26. Intelliquest Information Group, Inc. (1998). Latest Intelliquest Survey Reports 62 Million American Adults Access the Internet/Online Services. [online]. Available: http://www.intelliquest.com Johnson, T.J., & Kaye, B.K. (1998). Cruising is believing?: Comparing the internet and traditional sources on media credibility measures. Journalism & Mass Communication Quarterly 75, 2, 325-340. Kraus, Sidney, & Davis, David. (1976). The Effects of Mass Communication on Political Behavior. London, England: Pennsylvania State University Press. Lazarsfeld, Paul F., Berelson, Bernard, & Gaudet, Hazel. (1944). The People's Choice. New York: Duell, Sloan, and Pearce. Lazarsfeld, Paul F., & Merton, Robert K. (1948). Mass communication, popular taste and organized social action. In Bryson, Lyman (ed.), Communication of Ideas. New York: Harper & Brothers. Lemert, James B. (1970). News media competition under conditions favorable to newspapers. Journalism Quarterly. Lenart, Silvo. (1994). Shaping Political Attitudes, the impact of interpersonal communication and mass media. London, England: Sage. Lin, N. (1971). The Study of Human Communication. Indianapolis, Ind: Bobbs- Merrill. Lull, James. (1990). Inside Family Viewing: Ethnographic Research on Television's Audiences. New York: Routledge. Major, Ann M., & Atwood, L. Erwin. (1997). Changes in media credibility when a predicted disaster doesn't happen. Journalism Quarterly, 74, 797-813. Markham, David. (1968). The dimensions of source credibility of television newscasters. Journal of Communication, 18, 57-64. Mashek, John.W., McGill, Lawrence, T., & Powell, Adam C. (1997). Lethargy '96: How the Media Covered a Listless Campaign. Arlington, VA: The Freedom Forum. McLeod, Jack., Rush, Ramona R. & Friederich, Karl H.. (1968-69). The mass media and political information in Quito, Ecuador. Public Opinion Quarterly, 32, 575- 87. McQuail, D., Blumler, J.G., & Brown, J.R. (1972). The television audience: A revised perspective. In McQuail, D. (ed.), Sociology of Mass Communications, pp. 135-165. Harmondsworth, England: Penguin. Meyer, Philip. (1988). Defining and measuring credibility of newspapers: developing an index. Journalism Quarterly, 65, 567-572. Newhagen, John, & Nass, Clifford. (1989). Differential criteria for evaluating credibility of newspapers and TV news. Journalism Quarterly, 65, 567-574, 588. Pew Research Center. (1996). One-in-Ten Voters Online for Campaign '96. Available at http://www.people-press.org/tec96-1.htm, November 1996. Pfau, Michael, & Kang, Jong Geun. (1991). The impact of relational messages on candidate influence in televised political debates. Communication Studies, 42, 114-128. Poindexter, Paula M., & McCombs, Maxwell E. (1997). Making Research Relevant. Final draft of unpublished book. University of Texas, Austin. Rimmer, Tony, & Weaver, David. (1987). Different questions, different answers? Media use and media credibility. Journalism Quarterly, 64, 28-36, 44. Rogers, Everett M. (1983). Diffusion of Innovations, 3rd edition. New York: Free Press. Rogers, Everett M., & Shoemaker, F. Floyd. (1971). Communications of Innovations. New York: Free Press. Shaw, Eugene. (1973). Media credibility: taking the measure of a measure. Journalism Quarterly, 50, 306-311. Slater, Michael D., & Rouner, Donna. (1996). How message evaluation and source attributes may influence credibility assessment and belief change. Journalism Quarterly, 73, 974-991. Stone, Gerald C. (1987). Examining Newspapers: what research reveals about America's newspapers. Beverly Hills, CA: Sage. Sundar, S.Shyam. (1998). Effect of source attribution on perception of online news stories. Journalism Quarterly, 75, 1, 55-68. Wanta, Wayne. (1997). The Public and the National Agenda, how people learn about important issues. Mahwah, New Jersey: Lawrence Erlbaum Associates, Publishers. Wanta, Wayne, & Hu, Yu-Wei. (1994). The effects of credibility, reliance, and exposure on media agenda-setting: A path analysis model. Journalism Quarterly, 71, 1, 90-98. Westley, Bruce H., & Severin, Werner J. (1964). Some correlates of media credibility. Journalism Quarterly, 41, 325-335. Whitehead, Jack L. (1968). Factors of source credibility. Quarterly Journal of Speech, 54, 59-63. [1] For the purposes of this paper, the terms media channel and medium will be used interchangeably. [2] Some examples of the Berlo et al. semantic differential scales included asking whether a source was "just--unjust" (Safety Factor), "experienced--inexperienced" (Qualifications Factor), and "confident--not confident' (Dynamism Factor). [3] Whitehead's Competence Factor is similar to Berlo's Qualifications Factor and the Objectivity Factor rates how "objective--subjective" a source is. [4] In this examination, frequency questions were used allowing us to replicate Rimmer & Weaver's (1987) observations. Ideally, we would have also liked to incorporate preference and reliance questions as Wanta & Hu (1994) did, but space was limited and frequency questions had applications for other researchers involved in the survey. As a result, they were selected. [5] The formula for efficiency of reaching households is possible households / all numbers. [6] The formula for the completion rate of the sample is completions / eligible numbers. [7] As alluded to earlier, statistical tests revealed that eliminating certain questions would not improve the reliability of the scale. [8] As suggested above, local and national network news viewing were separated because other researchers have argued that opinions about television news credibility are most influenced by local news broadcasts (e.g., Newhagen & Nass, 1989). [9] Regardless, this question was somewhat problematic because it deals with general use of the internet and does not focus on using the internet specifically for news. However, a similar criticism could be made of typical newspaper use questions since people surely read newspapers for purposes other than acquiring news, including advertising, coupons, entertainment, etc. yet are deemed appropriate for media credibility analyses. Likewise, we feel a general measure of Internet use can serve as a proxy for estimating online news use and was therefore justified in this study. [10] We will consider this point further in the Discussion section. [11] The total exceeds 100 percent due to rounding. [12] A Credibility Factor with 3 levels was created with each index acting as a level.