Communication Communities or “CyberGhettos?”: A Path Analysis Model Examining Factors that Explain Selective Exposure to Blogs1
This study used an online panel of Internet users to examine the degree to which blog users practice selective exposure when seeking political information. The research employed a path analysis model to explore the extent to which exposure to offline and online discussion of political issues, and offline and online media use, as well as political variables and demographic factors, predict an individual's likelihood to engage in selective exposure to blogs. The findings indicate that respondents did practice selective exposure to blogs, predominantly those who are heavy blog users, politically active both online and offline, partisan, and highly educated.
Supporters tout blogs as superior to traditional media in promoting democracy because blogs foster a sense of community among users who can share their viewpoints with likeminded individuals (Papacharissi, 2004; Trammell, Tarkowski, Hofmokl, & Sapp, 2006). Critics, however, contend that because of the strong ideological perspectives of blogs, they are creating communication ghettos where people go to support their own opinions and attack opposing ones, leading to increased polarization of political views (Stroud, 2006,2008).
While the emergence of the Internet in general, and blogs in particular, have made it easier for individuals either to explore a variety of views on a topic or to examine only those that support their viewpoint, only a handful of studies have examined selective exposure among Internet users (Bimber & Davis, 2003; Garrett, 2006; Mutz & Martin, 2001; Stroud, 2006,2007a,2007b,2008). Fewer still have centered on selective exposure to blogs (Authors, 2008b). We could find no study that has systematically examined factors explaining selective exposure among blog users.
This study relies on an online panel of Internet users to investigate the degree to which blog users practice selective exposure to political information and examine factors that explain selective exposure. More specifically, this study examines the extent to which blog users say that they seek out information that supports their beliefs and avoid information that challenges their beliefs. This study also employs a path analysis model to explore the degree to which exposure to offline and online discussion of political issues, offline and online media use, as well as political attitudes and demographic factors predict whether an individual will engage in selective exposure.
This study ranks reliance on political Web sites as the most important predictor of selective exposure followed by reliance on online discussion (chatrooms/instant messaging and electronic e-mail lists/bulletin boards). We predict offline political discussion will be directly related to selective exposure and indirectly linked through online political discussion. We also assess both the direct and indirect effects of reliance on traditional and online media, offline and online discussion, and political variables, as well as demographic measures on whether people practice selective exposure to political blogs.
Selective exposure to blogs needs to be explored in more depth because of the democratic implications of people searching out only congenial points of view. While some researchers contend selective exposure may motivate people to participate in politics (Schudson, 1995), others worry that selective exposure may endanger democracy. Some researchers claim that engaging in partisan selective exposure through blogs will cause people to develop more polarized and fragmented political views (Stroud, 2006). Selective exposure may lead to less tolerance of opposing viewpoints (Mutz, 2002) and produce a less informed electorate because democratic theory requires that citizens gather and critically evaluate a wide range of views before making decisions (Delli Carpini & Keeter, 1996).
Selective exposure arose out of cognitive dissonance theory, the belief that people enter a state of discomfort or dissonance when they are exposed to information that is inconsistent with their own beliefs (Festinger, 1957). One way to reduce the dissonance was through selective exposure: actively seeking out information that supports one's beliefs and avoiding sources that challenge those beliefs. However, because researchers found more support for cognitive dissonance than selective exposure as a way to reduce that dissonance, research interest in selective exposure dwindled (Kinder, 2003; Sears & Freeman, 1967; Zaller, 1992).
Selective exposure, however, has experienced renewed interest as researchers have focused on selective exposure less as a way to reduce dissonance and more as a means of processing information effectively (Smith, Fabrigar & Norris, 2008; Stroud, 2006,2007; Ziemke, 2000). As the sheer volume of available information increases, information processing researchers have suggested that both users' ability and motivation to process information decreases (Petty & Cacioppo, 1986). People need to simplify their information processing tasks (Taylor, 1981), and selective exposure has been identified as one way to simplify that task. Researchers have found that it requires considerable cognitive energy to process information that challenges one's predispositions (Edwards & Smith, 1996), while it requires less effort to process supportive information (Edwards & Smith, 1996; Ziemke, 1980). In addition, searching out information that supports one's already-held views may be considered rewarding because users judge such information as more convincing and more credible than information that challenges their views (Fischer, Jonas, Frey, and Schulz-Hardt, 2005; Melican & Dixon, 2008).
Researchers have consistently found that users practice selective exposure for partisan information (Stroud, 2006,2008) because they are more motivated to process information that they perceive is more relevant to their personal beliefs (Donsbach, 1991; Price & Tewsbury, 1997). Political partisanship has been shown by researchers to be a construct easily accessible to memory and therefore one that helps guide exposure decisions (Green, Palmquist & Schickler, 2002; Lau, 1989). Indeed, researchers have consistently found that partisan leanings influence what media and media messages people seek out (Bartlett, Drew, Fahle, & Watts, 1974; Chaffee, Spahir, Grap, Sandvig & Hahn, 2001; Meffert, Chung, Joner, Walks & Garst, 2006; Ziemke, 1980). Partisans are more likely to seek out information on candidates (Chaffee et al., 2001; Donsbach, 1991; Meffert et al., 2006) and issue positions (Graf & Aday, 2008; Iyengar & Hahn, 2009; Mende, 2008; Taber & Lodge, 2006) that support their already strongly held views.
Researchers have found more support for the claim that individuals search out information that supports their views than avoid information that runs counter to them (Chaffee et al., 2001; Donsbach, 2001; Graf & Aday, 2008). However, Knoblach-Westerwick and Meng (2008) found that those who viewed issues as important to them personally, who had greater interest in politics and a stronger party affiliation also sought out counterarguments, perhaps because they wanted to know what the other side thinks and they believed they had strong defensive mechanisms to counter attacks on their positions.
Most selective exposure studies have looked at traditional media. However, increasingly researchers are focusing on the Internet because its wealth of partisan information and user ability to control what sites to visit should lead to increased self-exposure.
The Internet and Selective Exposure
The renewed interest in selective exposure has paralleled the growth of more partisan sources of information such as cable television, radio talk shows, and the Internet, which make it easier for users both to find information that supports their predispositions (Galston, 2003; Mutz & Martin, 2001; Sunstein, 2001) or to avoid opposing political information altogether (Atre & Katz, 2005; Prior, 2002,2005). The Internet, in particular, may promote selective exposure because it provides such an abundance of information that selective exposure is not only possible, but also necessary. While traditional media seek to balance political viewpoints in stories, the Internet provides a wealth of information that caters to a partisan point of view (Albrecht, 2006). In addition, the interactive nature of the Internet provides the user more control over what sites to visit (Garrett, 2006). Not surprisingly, then, researchers find that people practice selective exposure when they seek out partisan information online (Authors, 2008a, 2008b; Bimber & Davis, 2003; Garrett, 2006; Mutz & Martin, 2001; Stroud, 2006,2007a,2007b,2008).
Most Internet researchers have focused on the Internet as whole. However, few scholars have looked at components of the Web such as blogs, which contain more partisan content and therefore should lead users to increased selective exposure.
Blogs and Selective Exposure
Blogs have received limited attention from selective exposure scholars, but evidence suggests that selective exposure would be stronger for blogs than for the Internet in general. While nonpartisan blogs certainly exist, studies suggest that most political blogs advocate a political viewpoint and therefore attract likeminded individuals who share the blogger's opinions (Choi, Watt & Lynch, 2006; Johnson, Kaye, Bichard & Wong, 2007; Tremayne, Zheng, Lee & Jeong, 2006). Indeed, blog users claim that a major advantage of blogs over mainstream media is that they can discuss issues from specific ideological perspectives and therefore provide more opinionated and thoughtful analysis of topics (Bruns, 2006; Johnson, et al., 2007; Johnson & Kaye, 2004; Wall, 2006). Similarly, blogs may foster a sense of community because users can locate blogs that represent their point of view and can share their thoughts with those on the blog that hold similar viewpoints (Kaye, 2007; Kaye & Johnson, 2004; Perlmutter, 2008; Trammell, Tarkowski, & Sapp, 2006).
The authors (2008a) specifically explored selective exposure to blogs and found that more than half of blog users visited sites that shared their point of view as compared to only 22.2% who sought out blogs that challenged their points of view. This supports earlier studies that suggest that selective exposure is stronger for media with more partisan content (Bennett, 2002; Chmielewski, & Kruger, 2005; Hollander, 1996; Stroud, 2006,2007,2008).
H1: Those who visit blogs for political information will seek out sites that support their political views rather than sites that challenge their views.
This study also examines predictors of selective exposure, with offline and online political discussion advanced as the major predictors of selective exposure to blogs.
Selective Exposure and Offline Political Discussion
The Pew Internet and American Life Project found that the offline world mirrors the online world. That is, activities that interest and engage people in the offline world spill over to the Internet (Rainie & Horrigan, 2005). Therefore, those who enjoy discussing politics offline are likely to want to engage in political debate in online forums and in blogs (Dutta-Bergman, 2006).
Little attention has been paid to whether offline discussion will lead to selective exposure to blogs. Researchers studying political homogeneity note that in situations where people can self-select conversational partners, such as in volunteer organizations and at church, people will engage in discussions with likeminded people rather than those they disagree with (Brundidge, 2006,2008; Scheufele, Nisbet, Brossard, & Nisbet, 2004; Scheufele et al., 2006). That is, people only attend to information that supports their beliefs because they largely affiliate with individuals who support their views rather than consciously seek out people with similar views (Sears & Freedman, 1967). Therefore, if people talk to people who share their views offline, they are likely to seek out likeminded individuals online. Similarly, Mutz (2006) found that when people increase their social networks they seek out people with similar views. Indeed, the authors (2008) found that offline discussion proves a weak, but positive, predictor of selective exposure to blogs, suggesting that blog users carry over their offline discussions of politics with likeminded individuals to the online world.
H2: Reliance on offline discussions will be positively related to selective exposure to political information on blogs.
Online Political Discussion and Selective Exposure
Researchers suggest that online discussion groups differ in the degree to which they attract heterogeneous users. Therefore, reliance on different types of online groups will have different effects on selective exposure. Heavy blog use has proved to be the strongest predictor of blog credibility as researchers suggest heavy users are more likely than infrequent users to know which blogs support their views on candidates or issues, and these blogs are considered the most credible (Banning & Trammell, 2006; Johnson & Kaye, 2004,2008; Johnson et al., 2007).
Newsgroups typically are set up with their bias stated up front (Hills & Hughes, 1998; Wojciewszak & Mutz, 2009) and are therefore populated with groups of likeminded individuals, making it easy for individuals to find sites featuring people who share their views. The authors (2008b) found those who visited news groups that mirrored their political views are also likely to visit political Web sites that support their perspectives, and therefore newsgroups are indirectly related to selective exposure through reliance on political Web sites.
On the other hand, those who visit political chatrooms are more likely to encounter a diverse group of people than in newsgroups. Chatroom is a synchronous form of communication, so people may be less able to predict whether they will be talking with likeminded people than with blogs or newsgroups and therefore will be exposed to a greater variety of viewpoints (Brundidge, 2006,2008). In addition, the authors (2008a) discovered that chatroom reliance reduced selective exposure to political Web sites.
H3: Reliance on a) blogs and b) newsgroups will lead to selective exposure to information on blogs.
H4: Those who rely on chatrooms for political information will be less likely to practice selective exposure.
This study also examines the extent to which reliance on newspapers, broadcast television, cable television, talk radio, and political Web sites predict selective exposure to blogs. This study predicts that reliance on more partisan sources such as cable television news, talk radio, and political Web sites will better predict select exposure.
Media Reliance and Selective Exposure
Results for media use and selective exposure differ depending on which media are examined. Traditional news media and their online counterparts embrace balance and objectivity. Therefore, reliance on traditional media may be unrelated to selective exposure (Best, Chmielewski, & Kruger, 2005; Mutz & Martin, 2001).
Selective exposure may be more strongly linked to more partisan sources such as talk radio, cable television news (Stroud, 2007,2008) or political Web sites (Authors, 2008a, 2008b; Stroud, 2007) as users can listen to hosts or visit sites that hold the same views they do.
H5a: Reliance on broadcast television news will be unrelated to selective exposure to blogs for political information.
H5b: Reliance on online broadcast television news sites will be unrelated to selective exposure to blogs for political information.
H5c: Reliance on newspapers will be unrelated to selective exposure to blogs for political information.
H5d: Reliance on online newspaper sites will be unrelated to selective exposure to blogs for political information.
H5e: Reliance on talk radio will be positively related to selective exposure to blogs for political information.
H5f: Reliance on cable television news will be positively related to selective exposure to blogs for political information.
H5g: Reliance on cable television news sites will be positively related to selective exposure to blogs for political information.
Political Variables and Selective Exposure
Political variables have proven strong predictors of selective exposure, particularly exposure to partisan content. However, while some political measures may increase selective exposure, others may squelch it. Studies assert that political participation, political interest, and political partisanship increase the likelihood individuals will practice selective exposure. On the other hand, political tolerance and willingness to speak out on controversial issues may suppress selective exposure.
Researchers have found that selective exposure effects are more powerful for those with more strongly held attitudes and beliefs (Brannon, Tagler, & Eagly, 2007; Smith et al., 2008). Not surprisingly, then, political partisanship has proven to be one of the strongest predictors of selective exposure to political information as strong party supporters or those who hold strong issue positions will logically search for information that supports their point of view (Bimber & Davis, 2003; Garrett, 2006; Mutz & Martin, 2001; Stroud, 2006,2008).
Strong political trust, political knowledge, and political participation may be the major pillars of a strong democratic system. However, ironically, they have all been linked to increased, not reduced, selective exposure. Those who are politically interested tend also to be highly partisan and therefore will seek out more partisan sources of information (Mutz, 2006; Stroud, 2006,2007b). The politically knowledgeable, because they are already well versed on issues and candidates, often know who they will support. Therefore, the politically knowledgeable do not need to compare information from both sides before making a decision and are more likely to visit sources that reinforce their existing positions (Stroud, 2006).
Researchers who have examined selective exposure to both offline and online political participation have found because participation is linked to both increased media use and political interest that political participation also leads to selective exposure (Huckfeldt et al., 2004; Mutz, 2002; Stroud, 2006).
However, when the authors (2008a) examined political predictors of selective exposure to blogs, those who were liberals and strong partisans practiced selective exposure, while political interest was not linked to selective exposure.
Other political measures such as political tolerance and willingness to speak out in public may cause people to seek out multiple perspectives on topics. Those who are politically tolerant are more willing to listen to, and consider, information that runs counter to their own perspectives (Hardy, Scheufele, & Wang, 2005; Mutz, 2002; Pettigrew, 1997). Similarly, willingness to speak out in public means a willingness to talk to people with different points of views, which in turn should be linked to relying on sources with a variety of perspectives (Brundidge, 2006).
When the authors (2008b) examined predictors of selective exposure to political Web sites, willingness to speak out indeed led to less selective exposure. However, the politically tolerant were more likely to seek out only information supporting their predispositions. The authors speculated that while the politically tolerant may be more willing to listen to views that counter their own perspectives, they seek out information online that supports their views.
H6: Those who are a) liberal b) report a strong party affiliation c) are politically knowledgeable and d) politically interested will be more likely to selectively expose themselves to blogs for political information.
H7: Those who are more politically active both a) offline and b) online will be more likely to selectively expose themselves to blogs for political information.
H8: Those who a) express less tolerance toward opposing viewpoints and are b) less willing to speak out in public will be more likely to expose themselves selectively to blogs that support their views.
The researchers posted an Internet survey examining the practice of selective exposure from September 13 to September 23, 2007. Politically interested Internet users were solicited from an online panel of 20,000 opt-in consumers operated by a media research lab at a major university in the southwestern United States. The researchers sent e-mail invitations to 5,000 panel members requesting their participation if they used the Internet for accessing political information and were eligible voters in the United States. Respondents were also offered an opportunity to participate in a cash drawing as a further incentive. This sample yielded 772 completed surveys, a 15.4% response rate.
Selective exposure to blogs
Respondents used a 5-point scale to indicate how frequently they visited political weblogs that provide information about specific issues or policies that they agreed with and disagreed with, ranging from “1” (never visited) to “5” (very often visited). Responses for “disagree with” were reverse coded. The two items were summed to form an index of selective exposure to political weblogs (high number means high selective exposure).
Reliance on online sources for political information
Respondents were asked a series of questions regarding the extent to which they relied on four online sources for political information on a 5-point scale where “1” meant “don't rely on at all” and “5” meant “heavily rely on”: blogs (M = 1.71, SD = .99), broadcast television news sites (M = 2.72, SD = 1.2), cable television news sites (M = 2.85, SD = 1.2), newspaper sites (M = 2.84, SD = 1.3), and politically oriented Web sites (M = 2.2, SD = 1.2).
Reliance on offline sources for political information
Similarly, respondents were asked the extent to which they relied on offline sources for political information on the same 5-point scale: broadcast television news (M = 3.09, SD = 1.2), cable television news (M = 3.11, SD = 1.2), newspaper (M = 3.24, SD = 1.2), and talk radio programs (M = 2.44, SD = 1.3).
Reliance on offline and online discussion
Offline discussion This was a single item measure. Respondents were asked the extent to which they relied on face-to-face discussion with others for their political information on a 5-point scale where “1” meant “don't rely on at all” and “5” meant “heavily rely on.”
Online discussion Three items were employed for this measure. On the same 5-point scale, respondents were asked the extent to which they relied on blogs, electronic mailing lists/bulletin boards, and chatrooms/instant messaging for political information. Reliance on political weblogs was the focal variable.
Political variablesStrength of party affiliation was measured on a scale of 0 to 10 where “0” meant “very weak party affiliation” and “10” meant “very strong party affiliation.” Respondents showed fairly strong party affiliation with an average of 6.8 (SD = 2.7).
Overall, 41.2% considered themselves moderate, 21.8% conservative, and 22.5% liberal. Moreover, 31.6% identified themselves as Republicans, 36.4% Democrats, and 27.6% Independents, with the remainder responding “Libertarian” or “Green Party.”
General political interest was measured on the same scale of 0 to 10 ranging from “0”“absolutely not interested in politics” to “10”“absolutely interested in politics.” Respondents are fairly interested in politics in general (M = 7.6, SD = 2.5).
Offline political participation was measured by an index of five items adapted from the National Election Studies, where “1” meant “never” participated and “5” meant “very often.” Respondents were asked how often they had engaged in a variety of political activities such as wearing a campaign button, helping or giving money to a candidate, attending political meetings, and persuading others to support a candidate during the past 2 years (Cronbach's alpha = .88).
On the same 5-point scale as offline political participation, online political participation was an additive measure of seven items taken from the Pew Research Center surveys. Respondents were asked how often they participated in any of the following activities when they went online: (1) registering their own opinions by participating in an online poll, (2) getting information about a candidate's voting record, (3) sending or receiving e-mails about the candidates or campaigns, (4) contributing money online to a candidate running for public office, (5) looking for more information online about candidates' positions on the issues, (6) finding out about endorsements or ratings of candidates by organizations or individuals online, and (7) checking the accuracy of claims made by or about the candidates online (Cronbach's alpha = .89).
Political knowledge Respondents were asked four questions taken from the Pew Research Center surveys: If Americans bought more foreign goods than they sold to people overseas, or sold more goods than they bought, or about the same; if they know which political party has a majority in the U.S. Congress; what was the new minimum wage; whether the U.S. Supreme Court Chief Justice is more liberal or conservative; and if in 2006, more Iraqi civilians or more U.S. soldiers were killed as a result of the fighting in Iraq. Correct answers were recoded as “1” and incorrect ones as “0” so that the higher the number, the more knowledgeable the respondents (Cronbach's alpha = .64).
Political tolerance This was an additive measure of six items taken from Sniderman and associates (1989). Respondents were asked to nominate their least-liked group and were asked if their least-liked group should be allowed to enjoy certain civil rights (be allowed to run for public office, teach in universities, make a speech, have books in public libraries, and be able to hold public rallies in one's city) based on “yes” and “no” answers. Answers were recoded so that “1” meant “high tolerance” and “0”“low tolerance.” Then the six items were summed to form a political tolerance index (Cronbach's alpha = .76).
Willingness to speak up This was a composite measure taken from Hardy, Scheufele and Wang (2005) of three items on a scale of 0 to 10 where “0” meant “not at all likely” and “10” meant “very likely.” Respondents were asked three questions about their willingness to speak up during a forum on a controversial issue such as gay marriage, express an opinion that is different from those of others at the meeting, and listen to an opinion that is different from theirs at the meeting (Cronbach's alpha = .63).
Demographics (age, gender, education, race, and income) were treated as exogenous variables in the path analysis model.
Almost 60 percent (59.8%) of the sample respondents were female. The mean age was 44 years old (SD = 12.7). On average, respondents reported obtaining a 4-year college degree (SD = 1.1). These variables closely resembled the panel demographics as provided by the research lab used in this study.1 Respondents also reported a mean household income between $65,001 and $80,000. In addition, 80.8% of the respondents were whites, 7.2% Hispanics, 4.6% blacks, 3.8% Asians, and .5% Native Americans. The race variable was recoded as a dummy variable where white was “1” and nonwhites as “0.”
To test the first hypothesis, frequencies will be examined to determine if people do practice selective exposure by searching out sites that support their point of view rather than seek out blogs that present a balanced viewpoint, no perspective on the news or challenges the users' viewpoints.
Path analysis techniques were used to examine the direct and indirect paths to selective exposure to political weblogs, to shed light on the tenability of our causal models, and to test hypotheses two through eight.
To obtain the path coefficients, three regression analyses were conducted. The first regression analysis employed selective exposure to political weblogs as the dependent variable, with the following four blocks of independent variables entered in this order: block 1: reliance on online discussion (weblogs, electronic mailing lists, and chatrooms) and face-to-face discussion; block 2, reliance on offline and online communication sources (reliance on broadcast television news, cable television news, newspapers, talk radio programs, online broadcast news sites, online cable television news sites, and online newspaper sites); block 3: political variables (political interest, offline and online political participation, political knowledge, tolerance, willingness to speak in public on controversial topics, strength of party affiliation, and ideology); block 4: demographics (gender, age, education, income, and race).
The above regression analysis included all variables. However, for ease of analysis, only reliance on political weblogs was included in the second regression analysis. Reliance on political weblogs was the dependent variable for the second regression analysis. It was regressed on the following four blocks of independent variables that were entered in this order: block 1: reliance on chatrooms/instant messaging, reliance on electronic mail lists/bulletin boards, reliance on face-to-face discussion; block 2: reliance on online and offline communication sources for political information; block 3: political variables; and block 4: demographic variables.
The third regression analysis had the reliance on online and offline discussion measures as the dependent variables. The following three blocks of independent variables were entered in this order: block 1: reliance on online and offline communication sources; block 2: political variables; block 3: demographic variables.
The path coefficients are shown in Tables 1–3.
Table 1. Effects of Reliance on Online and Offline Discussion on Selective Exposure to Political Weblogs
|Electronic mailing lists/bulletin boards||−.03||.08***||.08***|
Table 2. Effects of Reliance on Online and Offline Communication Sources on Selective Exposure to Political Weblogs
|Broadcast news sites||− .01||.01||0|
|Cable television news sites||− .09||− .01||− .10|
|Newspaper sites||− .02||.01**||.01**|
|Talk radio programs||− .003||.05||.05|
|Broadcast television news||− .08||− .07||− .15|
|Cable TV News||.03||− .05||− .02|
|Newspapers||− .02||.01||− .01|
Table 3. Effects of Political Variables on Selective Exposure to Political Weblogs
|Strength of Party Affiliation||.09*||.02||.09*|
|Offline political participation||.09*||.01||.09*|
|Online political participation||− .01||.06***||.06***|
|Political tolerance||.01||− .002||.01|
|Willingness to speak||− .04||− .07||− .11|
Effects of Online and Offline Discussion on Selective Exposure
The first hypothesis predicts that those who visit blogs for political information will seek out sites that support their political views more than ones that oppose them. Frequency analysis shows that more than half (53%) of those who visit blogs for political information visit blogs that provide political information that they agree with, as compared to only 22.2% who sought out blogs that challenged their points of view. Therefore, H1 is supported.
Hypothesis 2 states that reliance on offline political discussion will be positively related to selective exposure to political information on blogs. As indicated in Table 1, reliance on face-to-face discussion has no effect on selective exposure to political blogs (β = −.02, p = n.s. ). Therefore, H2 is not supported.
Hypothesis 3 asserts that reliance on a) blogs and b) electronic mailing lists and bulletin boards will lead to selective exposure to information on blogs. As seen in Table 1, reliance on political blogs has a direct effect on selective exposure to political weblogs (β = .5, p < .001), so H3a is supported.
Though there is no direct link between reliance on electronic mailing lists/bulletins and selective exposure to political weblogs, reliance on e-mail lists/bulletins influences reliance on political blogs (β = .163, p < .001), which, in turn, influences selective exposure to political blogs (β = .5, p < .001) with an indirect of .08 (p < .001). Thus, H3b is partially supported.
The fourth hypothesis predicts that those who rely on chatrooms/instant messaging for political information will be less likely to practice selective exposure. Table 1 shows that reliance on chatrooms/instant messaging has a direct negative effect on selective exposure to political blogs (β = −.16, p < .001), reliance on chat rooms indirectly influences selective exposure to political blogs through reliance on political blogs (β = .12, p < .001) with a total effect of −.04(p < .001), therefore H4 is supported.
The block of reliance on online and offline discussion accounts for 25.6% of the variance in selective exposure to political blogs, by far the largest of any block.
Effects of Reliance on Offline and Online Communication Sources
Hypothesis 5a asserts that reliance on broadcast television news will be unrelated to selective exposure to blogs for political information. Table 2 indicates that reliance on broadcast television news has no effect on selective exposure to political blogs (β = −.15, p = n.s. ), therefore supporting H5a.
Hypothesis 5b states that reliance on online broadcast television news sites will be unrelated to selective exposure to blogs for political information. Table 2 shows that reliance on broadcast television news sites exerts no influence on selective exposure to political blogs (β = −.01, p = n.s. ); therefore, H5b is supported.
Hypothesis 5c states that reliance on newspapers will be unrelated to selective exposure to political blogs (β = −.01, p = n.s. ). As seen from Table 2, it is indeed the case, so H5c is supported.
Hypothesis 5d predicts that reliance on online newspaper sites will be unrelated to selective exposure to blogs for political information. As seen in Table 2, reliance on newspaper sites influences reliance on e-mail lists/bulletin boards (β = .11, p < .01), which, in turn, influences reliance on blogs (β = .16, p < .001), which ultimately is linked to increased selective exposure to political blogs (β = .5, p < .001) with an indirect effect of .01, p < .01). However, because online newspapers were indirectly related to increased selective exposure, H5d is not supported.
Hypothesis 5e proposes that reliance on talk radio will be positively related to increased selective exposure when visiting blogs. Table 2 shows that reliance on talk radio does not have any effect on selective exposure to blogs (β = .05, p = n.s. ), so H5e is not supported.
Hypothesis 5f and 5g predicts that reliance on cable television news and cable new sites will be positively related to selective exposure while visiting blogs. According to Table 2, reliance on cable TV news (β = −.02, p = n.s. ) and their online counterparts (β = −.10, p = n.s. ) have no effect on selective exposure to political blogs; thus, H5f and H5g are not supported.
The block of reliance on online and offline media sources accounts for 1.7% of the variance in selective exposure to political blogs.
Effects of Political Variables on Selective Exposure
The sixth hypothesis under scrutiny predicts that those who are a) liberal b) report strong party affiliation c) politically knowledgeable and d) politically interested will be more likely to expose themselves selectively to blogs for political information. According to Table 3, ideology (liberals coded higher) has a direct positive effect (β = .08, p < .05) on selective exposure to political blogs, so H6a is supported. Strength of party affiliation has a direct positive effect on selective exposure (β = .09, p < .05), thus H6b is supported. Table 3 also shows that political knowledge indirectly influences selective exposure to political blogs (β = .02, p < .05) via reliance on electronic mailing lists/bulletin boards and e-mail lists/bulletin boards. Therefore, H6c is partially supported. Political interest does not exert any impact on selective exposure to political blogs (β = .08, p = n.s. ), so H6d is not supported.
Hypothesis 7 predicts that those who are more politically active both a) offline and b) online will be more likely to selectively expose themselves to blogs for political information. As seen in Table 3, offline political participation has a direct effect on selective exposure to political blogs (β = .09, p < .05), lending support to H7a.
Online political participation does not have a direct effect but has an indirect effect on selective exposure to political blogs (β = .06, p < .001) through reliance on political weblogs, electronic mailing lists/bulletin boards, and chatrooms/instant messaging. Therefore, H7b is partially supported.
The eighth and final hypothesis states those who a) express less tolerance toward opposing viewpoints and are b) less willing to speak out in public will be more likely to selectively expose themselves to blogs that support their views. According to Table 3, neither political tolerance (β = .01, p = n.s. ) nor willingness to speak in public (β = −.11, p = n.s. ) has an effect on selective exposure to political blogs, so H8a and H8b are not supported.
Political variables explained 3.2% of the variance in selective exposure to blogs for political information.
Demographics and Selective Exposure
Education proved to be the only demographic measure that had an impact on selective exposure to blogs for political information (β = .10, p < .05). However, age influenced reliance on political blogs (β = −.11, p < .01), which in turn, influenced selective exposure to blogs for political information (β = .5, p < .001) with an indirect effect of −.06, p < .01.
The block of demographic variables account for 1.2% of the variance in selective exposure to political blogs, the smallest influence of any block.
This research used a path analysis model to examine an online panel of Internet users to explore the degree to which those who visit blogs for political information practice selective exposure. The study assessed both the direct and indirect effects of traditional and online media reliance, offline and online discussion, political and demographic variables on the practice of selective exposure to political blogs.
The results indicate that respondents indeed practiced selective exposure when accessing information on political blogs. This supports past research indicating that individuals frequently seek out information that aligns with their values (Bimber & Davis, 2003; Garrett, 2006; Mutz & Martin, 2001), particularly concerning political issues (Stroud, 2006,2007b). Studies have indicated that selective exposure effects are stronger for more partisan content (Best et al., 2005). Therefore, those looking for partisan content can find it without difficulty on political blogs.
Past analyses have suggested that face-to-face discussion can either promote or reduce selective exposure depending on the setting (Brundidge, 2006; Scheufele et al., 2004,2006). This study found no connection between face-to-face discussions and selective exposure to political blogs. This may suggest that either people talk to a variety of different types of individuals in their day-to-day discussions, or at least those discussions do not focus on where people go online for political information.
The offline world holds opportunities for encounters with people of widely diverse viewpoints, whereas blogs (Perlmutter, 2008) and news groups (Hills & Hughes, 1998; Wojciewszak & Mutz, 2009) often attract an audience of likeminded individuals. These venues make it relatively simple for people to find others with similar viewpoints. Indeed, in the current analysis, those respondents who relied on blogs and news groups that support their political views were also likely to visit political blogs that mirror their perspectives. Specifically, reliance on blogs directly influences selective exposure, while newsgroups are indirectly related to the practice of selective exposure. This study also offers support for prior research suggesting that instant messaging and online chat are positively related to seeking more ideologically diverse sources (Brundidge, 2006,2008). Brundidge asserts that individuals who engage in online chat may be less able to anticipate the types of people they will be talking with and their views and, therefore, may have an increased likelihood to encounter a greater mix of viewpoints.
This research posited that an individual's reliance on online and traditional media would be unrelated to selective exposure to blogs because journalists attempt to present both sides of an issue. Reliance on more partisan sources such as talk radio and cable television news sites were hypothesized as positive predictors of selective exposure (Stroud, 2007; Best, et al., 2005), but such effects were not revealed in the current analysis. Reliance on traditional broadcast news, cable television news, and newspapers, as well as their online counterparts was predicted to be unrelated to selective exposure. Indeed, the findings yielded no direct relationships between media reliance and selective exposure to blogs. This is not entirely surprising due to the fact that several past analyses of selective exposure focused on how media reliance was linked to specific content, such as Bush and Kerry Web sites (Stroud, 2006,2007) or online foreign news (Best et al., 2005). This study took a broader approach to investigate how media reliance is linked to selective exposure in the blogosphere. It supports research that indicates little connection between media reliance and political selective exposure to blogs (Authors, 2008b).
Past studies have indicated that political variables are consistent predictors of selective exposure, with some acting to increase selective exposure while others diminish it. Research suggests that those who are strong partisans are often very interested in political topics, highly involved, and knowledgeable about political matters. They also tend not to look for information from a wide range of perspectives, but rather seek out information that is congenial with their existing viewpoints (Stroud, 2006). In fact, the current research findings indicate that strength of party affiliation, ideology, and offline political participation proved the strongest political predictors of selective exposure to political blogs. Political knowledge and online political participation were also indirectly associated with selective exposure to blogs. In this case, political interest was not linked to selective exposure to blogs. Essentially, those who are politically active, knowledgeable, and ideologically devout tend to seek out views that accord with their previous positions, thus aligning with past research (Brundidge, 2006).
Past analyses suggest that other political measures such as political tolerance (Hardy, Scheufele & Wang, 2005; Mutz, 2002; Pettigrew, 1997) and willingness to speak out in public (Brundidge, 2007; Hardy, Scheufele, & Wang, 2005; Scheufele et al., 2006) may be negatively related to selective exposure. This study found no relationship between these variables and selective exposure to blogs for political information. Past research (e.g., authors 2008b) suggest a possible disconnect between views that people are willing to listen to and what they will actively seek out. This indicates a need for further analysis and the refined measurement of this variable.
Significance of the Study
Researchers are rediscovering selective exposure theory as more information sources from cable news networks to political Web sites target specific ideological groups (Galston, 2003; Mutz & Martin, 2001; Sunstein, 2001). Several studies have examined whether or not people practice selective exposure when they go online, but this is one of the first studies to examine selective exposure relating to blogs. This study confirmed perceptions that, because most political blogs are partisan, users practice selective exposure and seek out blogs that support their political perspectives. This study appears to be the first to examine factors that explain selective exposure to blogs, discovering that people who rely heavily on blogs, are liberal and highly partisan are more likely to seek out blogs that support their pre-established views.
But this study has practical as well as theoretical implications. Studies of the 2004 election noted that campaign managers relied on blogs as an important campaign tool to attract voters, volunteers, and contributors (Lawson-Borders & Kirk, 2005; Sweetser Trammell, 2007; Williams, Trammell, Postelnicu, Landreville, & Martin, 2005). Researchers noted that one strength of blogs as a campaign tool is that they can be targeted to certain groups, particularly the young, and that the interactive nature makes users feel more connected to the candidate and thus more willing to volunteer or to donate money. Candidates, particularly Barack Obama, turned to social networking sites to involve young voters in the 2008 election. But this study suggests that campaign managers can use blogs to involve partisan activists effectively in the campaign. Blog users seek out sites that support their political viewpoints, with partisan supporters being the most likely to practice selective exposure. Therefore, while candidates can use social networking sites to engage younger, less politically active users, blogs can be employed to help bolster support of the party activists who are usually the backbone of any campaign.
The current study employed an online sample of politically interested Internet users and relied on a self-report survey analysis. Future researchers may wish to acquire a larger sample with a more diverse demographic and psychographic profile in order to examine the degree to which people seek out supportive information online and avoid information that challenges their views. Survey analysis does not provide a causal direction, but offers only one snapshot of respondent views. Therefore, future study is needed that examines data for subjects over a period of time to capture a more complete picture of selective exposure to Internet components. Analyses could also explore such behavior in other contexts, such as a political election. Future research may want to address selective exposure to even more specific political content such as Web sites for competing candidates or issue-oriented sites and blogs to provide a more detailed understanding of the relationship between political content and selective exposure.
Future research should also improve measures of political knowledge and willingness to speak up. A balanced mix of national political knowledge, local political knowledge, current events knowledge, and civic knowledge is needed to capture the full extent of the construct of political knowledge. Similarly, future research should examine a wide range of controversial issues and ask respondents whether they are willing to speak up under real social settings instead of hypothetical ones.
For gender, the research lab panel demographics were 39.9% male and 60.1% female. Similarly, our sample yielded 40.2% male and 59.8% female. The Media Research Lab panel age breakdown given was 18–24 (9.6%), 25–34 (34%), 35–44 (26.2%), 45–54 (17.8%), 55+ (12.3%) while our sample yielded 18–24 (.1%), 25–34 (25%), 35–44 (25.3%), 45–54 (24.9%), 55+ (24.7%). The education profile for the majority of respondents on the Media Research Lab panel indicated some college up to a 4-year degree (58.5%). The majority of our sample also reported some college up to a 4-b year degree (57.5%). Therefore, our sample closely resembled the population on gender and amount of education, but is older than the panel population as a whole.
About the Authors
Thomas J. Johnson (Ph.D., University of Washington, 1989) is the Marshall and Sharleen Formby Regents Professor in Convergent Media at the Texas Tech College of Mass Communications, His research interest are public opinion and political communication research, particularly the role of the media in presidential elections. More recently, he has concentrated on how people use the Web, blogs, and social media and what effect online media have on them. He is the author or coauthor of two books (a third in press), 11 book chapters (2 more in press), 37 refereed publications (3 in press) and more than 100 refereed conference papers. He has published The Rehabilitation of Richard Nixon: The Media's Effect on Collective Memory, Garland Publishing, and coedited Engaging the Public: How Government and the Media Can Reinvigorate American Democracy Rowman & Littlefield. He has co-edited a third book, International Communication in a Global Age, which is scheduled to be published by Routledge/Lawrence Erlbaum Associates in Fall 2009. His work has appeared in numerous journals, including Journalism & Mass Communication Quarterly, Mass Communication & Society, Political Communication, Journal of Broadcasting & Electronic Media, Newspaper Research Journal, Harvard International Journal of Press/Politics, International Communication Gazette and Journal of Computer-Mediated Communication.
Address: College of Mass Communication, Texas Tech University, Lubbock, TX 76409. 806-742-6500 ext. 253. Email: email@example.com
Dr. Weiwu Zhang is an assistant Professor of Public Relations in the College of Mass Communications at Texas Tech University since fall 2007, where he teaches public relations, public opinion, research methods, and mass communication theory courses. Before that he was on the communication faculty at Austin Peay State University for several years where he was Director of CATI-equipped Communication Research Center.
His current research interests include new media and politics, the role of media and public relations in civic engagement, news framing and its public relations influences, and theorizing public relations practitioners-reporter relationships.
Zhang's work and co-authored work have appeared/will appear in such journals as Journal of Computer-Mediated Communication, Public Relations Review, Social Science Computer Review, The Review of Communication, Journal of Communication, Communication Studies, Communication Research, and Journalism & Mass Communication Quarterly.
Address: College of Mass Communication, Texas Tech University, Box 43082, Lubbock, Texas, 79409-3082. Email: Weiwu.firstname.lastname@example.org
Shannon Bichard, Ph.D., is currently an associate professor of Advertising in the College of Mass Communications at Texas Tech University where she has served since 2001. Her research interests focus on public opinion and consumer behavior with an emphasis on online communication and engagement. She is author or coauthor of 30 conference papers, 13 refereed publications, and does consulting work for research needs in the Lubbock community. Her research has appeared in journals such as Journalism & Mass Communication Quarterly, Journal of Computer-Mediated Communication, Health Communication, and many others. Her teaching area is Advertising and she currently advises the AAF National Student Advertising Competition team. Bichard received her B.A. and master's degree from the University of Central Florida and her doctoral degree from the University of Florida.
Address: College of Mass Communication, Texas Tech University, Lubbock, Texas, Box 43082, Lubbock, TX 79409. Email: Shannon.Bichard@ttu.edu