Selective Exposure to Cable News and Immigration in the U.S.: The Relationship Between FOX News, CNN, and Attitudes Toward Mexican Immigrants Homero Gil de Ziga, Teresa Correa, and Sebastian Valenzuela

In the past 2 decades, cable television and the Internet have greatly increased the availability of media content. The phenomenon has reinvigorated a longstanding debate about the effects of this media landscape, as people selectively get exposed to specific content. Based on U.S. national survey data, this article advances research in this area by analyzing the interplay between individuals' ideological predispositions, their selective exposure to cable news, and the relationship between selective exposure and their attitudes toward an issue with key policy-making implications: Mexican immigration. Results indicate conservative Republicans are more likely to watch FOX News, which is associated with negative perceptions of Mexican immigrants and higher support for restrictive immigration policies. Findings also suggest that liberals who get exposed to FOX News also show less support for Mexican immigration. In the past 2 decades, cable television and the Internet have exponentially increased the choice of media content available in U.S. households. For instance, as of 2006, there were more than 560 national cable programming networks (NCTA, 2010). By 2009, nearly 55% of Americans were using the Internet every day and spending, on average, 60 hours a month online, according to data from Nielsen and the Pew Internet and American Life Project (Smith, 2010).

This growing choice of content has led to a fragmentation of audiences. As a result, scholars are paying attention to media selective exposure and its potential effects on public opinion formation. Existing research has examined the impact of greater media choice on gaps in political knowledge, polarization of elections, reinforcement of extreme attitudes toward political figures, and the resurgence of a partisan, oppositional pressjust to name a few areas of inquiry (Bachmann, Kaufhold, Lewis, & Gil de Ziga, 2010; Garrett et al., 2012; Iyengar & Hahn, 2009a; Johnson, Bichard, & Zhang, 2009; Stroud, 2007, 2008; Valentino, Banks, Hutchings, & Davis, 2009; Valenzuela, Kim, & Gil de Ziga, 2012). The literature, however, falls short in exploring the consequences of media choice and audience selectivity for public opinion formation on issues of great importance. For instance, how does the current media environment influence perceptions about particular social groups and people's attitudes toward controversial issues? When citizens choose to watch a particular news channel based on ideological grounds, does it have an effect on policy preferences? If so, what kind of effects? Study of the attitudinal consequences of selective exposure is an area that lags behind other aspects of the theory. Furthermore, there is an ongoing debate on whether selective exposure can be understood as a media effect (cf., Holbert, Garrett, & Gleason, 2010; Bennett & Iyengar, 2008; Stroud, 2008). On the one hand, scholars suggest that the occurrence of selective exposure is leading to an era of ''minimal effects'' because the media would have limited power to change beliefs when people get exposed to messages that match their predispositions (Bennett & Iyengar, 2008). On the other hand, research suggests that selective exposure to certain media has polarizing and reinforcing effects (Holbert et al., 2010; Stroud, 2008). This article advances the literature by analyzing the interplay between individuals' ideological predispositions and their selective exposure to cable news channels. Furthermore, this article advances how these predispositions and selective media exposure relate to people's attitudes toward certain social groups (Mexican immigrants). This topic may be of great value in today's U.S. context as it has implications for immigration policy. Thus, the ultimate purpose of the study is to examine how selective exposure and ideological reinforcement are associated with the public's attitudes toward social groups and the public's policy choices toward these same groups. For that purpose, we conducted a survey of a national sample of U.S. residents. The survey was designed specifically to measure ideological orientations and exposure to particular media channels, as well as respondents' views on Mexican immigrants and policy preferences toward Mexican immigration. We focused on Mexican immigration to the US for several reasons. The issue has been at the forefront of public debate in the last decade. For instance, the 2006 rallies for comprehensive immigration reform attracted millions of protesters across 102 cities (Balz & Fears, 2006; Dunaway, Branton, & Abrajano, 2010). Also, Arizona's 2010 legislation against undocumented immigrants has captured the attention of the media, the public, and policymakers (Riccardi, 2010). In addition, immigrants of Mexican origin constitute an important share of the U.S. population

According to the Pew Hispanic Research Center (2009), the number of Mexican immigrants living in the US has increased by 17 times since 1970, reaching a record 12.7 million in 2008, by far the largest immigrant minority in the country. Lastly, existing research on the effects of media choice and selectivity on public opinion formation has focused almost exclusively on elections, candidates, and political parties. In this context, Mexican immigration represents fertile new ground for research. Because the evidence suggests that audience selectivity is particularly evident in the cable news networks, FOX News and CNNwith conservative Republicans preferring FOX News while liberal Democrats lean toward CNN (Iyengar & Hahn, 2009; Stroud, 2007)we investigate partisan selective exposure and its association with attitudes toward Mexican immigration by gauging exposure to these cable networks in particular.

Partisan Selective Exposure on Cable Television

Selective exposure is the process by which people deliberately select information channels that match their predispositions and beliefs (Stroud, 2007, 2008). Although this concept is not new and has been subject of scholarly scrutiny for decades, it has garnered renewed attention as the media environment fragments and people have more opportunities to choose their media. Developed from cognitive dissonance theory (Festinger, 1957), early research in the 1960s posited that selective exposure helped people to reduce cognitive dissonance by looking for information that agreed with their opinions and avoided messages that challenged them (e.g., Klapper, 1960). Subsequent studies, however, challenged those assumptions by arguing that human evolution could not have happened by seeking out redundant information only (McGuire, 1968). Although in the current media landscape the occurrence of selective exposure remains a contested area of research (Kinder, 2003; Zaller, 1992), scholars have argued that rather than avoiding dissonance, selective exposure is a strategy to process information in a more effective way (Smith, Fabrigar, & Norris, 2008; Stroud, 2008). The development of cable television and online platforms has led to a fragmentation of the media that compete for the creation of niche audiences and give people more media choices (Prior, 2007). Because people have a limited capacity to process mediated information (Lang, 2000), and processing attitude-consistent information requires less cognitive effort than counter-attitudinal messages (Edwards & Smith, 1996), it is more efficient to select information that matches one's beliefs and predispositions, as convergent pieces of information also facilitate a smoother cognitive assimilation and information processing (Cho, Gil de Ziga, Shah, & McLeod, 2006). Selective exposure can occur in many areas, including information on child development (Adams, 1961) and general news exposure (Knobloch, Dillman Carpentier, & Zillmann, 2003). Research has found, however, that in politics selective exposure is more likely to occur because individuals tend to have stable political predispositions. Thus, political ideology or partisanship is an accessible shortcut to choose an information channel (Chaffee, Saphir, Grap, Sandvig, & Hahn, 2001; Iyengar & Hahn, 2009; Johnson et al., 2009; Knobloch-Westerwick & Meng, 2009; Stroud, 2007, 2008). For this reason, this study integrates socio-political ideology with selective exposure. It has been found that selective exposure is particularly salient with cable news networks and certain online sites (Jamieson & Cappella, 2009) which provide increasingly polarized content to match their audiences' ideological preferences. Up to the 1980s, the news consistently offered a ''point-counterpoint'' approach to news related to notions of fairness, balance, and objectivity (Iyengar & Hahn, 2009). However, as cable news and Internet sites erupted, this balanced approach faded away (Prior, 2007). The increasing competition led news organizations to create niche audiences by catering to audiences' predispositions (Mullainathan & Schleifer, 2005). In the cable industry, FOX News, CNN, and MSNBC describe themselves as news outlets reporting with a sense of equilibrium and fairness in their views, representing both sides of the political spectrum when covering any given story. However, studies suggest that they are not so balanced, particularly FOX News. For instance, a comparison between FOX News, Associated Press, and UPI revealed that FOX News leaned significantly toward conservative and Republican beliefs compared to the other two news organizations (Groeling & Baum, 2007). Similarly, content analyses have found that FOX News showed a pro-conservative slant compared to the other cable outlets in coverage of the Iraq War (Aday, Livingston, & Herbert, 2005) and the 2004 presidential campaign (Project for Excellence in Journalism, 2004). Finally, Groseclose and Milyo's (2005) study revealed that CNN's program News Night leaned toward the left compared to FOX News' Special Report. Not surprisingly, the ideological sorting of the cable networks has transferred to their audiences. Using survey data, Iyengar and Hahn (2009) found that while ''conservatives and Republicans preferred to read news attributed to FOX News and to avoid news from CNN and NPR, democrats and liberals exhibited exactly the opposite syndrome'' (p. 19). Using cross-sectional and panel survey data to generalize to the population and demonstrate causal links, Stroud (2007) found that during the 2004 presidential election conservative Republicans were more likely to read newspapers endorsing Bush, listen to conservative talk radio, and watch FOX News. On the other hand, liberal Democrats were more likely to read newspapers endorsing Kerry, listen to liberal radio, and watch CNN and MSNBC.

Based on the above literature, the first hypothesis that will be tested in the study states:

H1: Individuals will select cable news channels that support their political ideology. Specifically, conservative Republicans will use FOX News more often than CNN, while liberal Democrats will use CNN more often than FOX News.

Limited or Strong Media Effects?

Whether growing levels of partisan selective exposure reflect a media effect has been the subject of debate among communication scholars. Recently, Bennett and Iyengar (2008) warned that ''the increasing level of selective exposure based on partisan preference ::: presages a new era of minimal consequences, at least insofar as persuasive effects is concerned'' (p. 725). In other words, people who are exposed to messages that match their own beliefs are less likely to change those beliefs and, as a consequence, the media will have limited power to change attitudes and persuade users. Holbert and colleagues (2010), however, have suggested that selective consumption of media leads to attitude strengthening and reinforcement, which are strong media effects. This line of reasoning has been supported by both cross-sectional and time-series analyses conducted by Stroud (2007, 2008, 2010), who has found that people's political attitudes become more polarized over time after repeated exposure to politically and ideologically consistent media messages. From a normative perspective, the polarizing and reinforcing impact of selective exposure is not innocuous for the democratic process. Exposure to messages that only reinforce preexisting beliefs leads to an echo-chamber effect, in which media use triggers attitude extremity and polarization (Mutz & Martin, 2001; Stroud, 2008). Thus, this study not only seeks to examine to what extent citizens selectively get exposed to a particular cable news media outlet but also the potential effects the exposure may have on positions toward relevant policy issues, such as people's perception on Mexican immigration. More specifically, this study attempts to build on this line of research by examining the attitudinal effects toward Mexican immigration among both CNN and FOX News viewers, as well as the polarizing or reinforcing effects of watching these news outlets.

Immigration in the U.S. Press

Immigration, particularly from Latin American countries, has been one of the most salient issues in the U.S. national agenda over the past 2 decades and has become the focus of heated debate among policymakers. For example, California's Proposition 187, approved by voters in 1994 but rejected by the federal courts, planned to cut off social services for undocumented immigrants and triggered an intense public discussion over the issue of immigration. According to Pew Research Center polls, in 2007 nearly 55% of Americans said that illegal immigration should be a top priority for government (Keeter, 2009). Furthermore, during the 2008 primary elections, some Republican caucuses argued immigration was more important than any other issue (Brader, Valentino, & Suhay, 2008). The issue of immigration in the US is not ethnically neutral. The largest immigrant population comes from Latin America, and the majority of Latin American immigrants are Mexican (Pew Research Center, 2009). Of those who identify as Mexican or Mexican-American, four out of ten were born abroad and arrived in the US in 1990 or later (Pew Research Center, 2009). Public attitudes toward immigrants are not ethnically neutral either. Experimental studies have shown that individuals' attitudes toward immigration are more negative when the news features Latino immigrants, rather than European immigrants (Brader et al., 2008). Several factors predict people's views on immigration. Individuals with lower education, those who are older, and females tend to have more negative attitudes toward tolerant immigration policies (Espenshade & Calhoun, 1993; Simon, 1987). A stronger and more consistent predictor is political ideology. In the US, conservative Republicans tend to have a more negative view toward immigrants and immigration (both of the documented and undocumented types) and are more opposed to open immigration policies compared to liberal Democrats (Bierbrauer & Klinger, 2002; Doherty, 2006; Huddy & Sears, 1995). From a social justice perspective, liberals are more sensitive toward the context of immigrants in need whereas conservatives have a higher motivation to punish norm deviance and violations and are more likely to attribute personal responsibility for their plight (Skitka & Tetlock, 1993). In order to confirm these expectations, the following hypothesis will be tested:

H2: Individuals' political ideology will be related to their views on Mexican immigration. Specifically, conservative Republicans will have more negative attitudes toward Mexican immigration than liberal Democrats. People's views toward immigrants and immigration are fed and shaped by the media. The media consistently under represent and stereotype certain immigrant groups, particularly Muslims and Latinos (Correa, 2010). In the case of Latinos, an exhaustive content analysis conducted by the Pew Research Center found that only 2.9% of the news in the US contained references to Latinos. Studies have also found that they are depicted as a burden for society in issues related to illegality, crime, and affirmative action (Subervi, Torres, & Nontalvo, 2005). Furthermore, scholars have analyzed that the news media rhetorically associate the ''flow'' of immigrants with negative metaphors such as invaders, destructive floodwaters, and pollutants that contaminate American ''purity'' (Cisneros, 2008). Regarding Mexican immigration, a study of the U.S. network news coverage between 1971 and 2000 revealed that in the 1970s and part of 1980s immigration and border problems were depicted as a few states' problems. Over time, Mexican immigration coverage was portrayed as part of the national agenda and became increasingly associated with violence and economic costs for the US. (Johnson, 2003). This negative image of immigration in general and Mexican immigration in particular, has pervaded the news media in general, including FOX News and CNN (Cisneros, 2008). For instance, in line with its ideological inclination, a study of FOX's Bill O'Reilly, who is considered a journalist by 40% of the American public (Annenberg Public Policy Center, 2005), concluded that he portrayed immigrants as evil ''illegal aliens'' and ''foreigners'' (Conway, Grabe, & Grieves, 2007). Therefore, it is possible that persistent exposure to polarized channels leads viewers to think about Mexican immigration in a way that is consistent with those outlets. However, because it is not entirely clear whether FOX News portrays Mexican immigration in a more negative light than CNN, we pose the following research question: RQ1: Do individuals who watch FOX News exhibit more negative attitudes toward Mexican immigration than individuals who watch CNN, even after controlling for individuals' political ideology? Furthermore, if partisan selective exposure reinforces individuals' attitudes, particularly when they hold more extreme political views, it is necessary to examine for possible polarization effects toward Mexican immigration. Hence, RQ2: Is partisan selective exposure associated with more polarized attitudes toward Mexican immigration?

Methods

Data The data used in this study are based on a U.S. national survey collected between December 15, 2008, and January 5, 2009, by a research unit hosted at the University of Texas at Austin.1 To overcome the limitations of Web surveys and assure an accurate representation of the national adult population, the research unit based this particular sample on two U.S. census variables: gender and age. The procedure of matching online samples with census data to provide a more accurate representation of the population has been validated by previous research (Gil de Ziga & Valenzuela, 2010, 2011; Iyengar & Hahn, 2009). The survey instrument was administered using Qualtrics, a Web survey software, and was pilot-tested before actual fieldwork. After matching a 10,000 random draw to these demographic characteristics, a total of 1,432 email addresses were invalid. Of the remaining 8,568 participants, 1,159 responded on all items and 323 had missing values for some of the variables of interest in the analysis. Accordingly, based on the American Association of Public Opinion Research's (AAPOR) RR3 calculation, the response rate was 22.8% (AAPOR, 2008, pp. 34-35).2 This relatively low response rate falls within the acceptable range for panel Web-based surveys (Sax, Gilmartin, & Bryant, 2003). Compared to U.S. Census data, our sample had more females and was slightly better educated. Nevertheless, the demographic breakdown of our sample was similar to that of surveys conducted by the Pew Research Center and other organizations that employ random digit dialing (Pew Internet & American Life Project, 2009), which seems to lend support to how well our sample statistics estimate overall U.S. population parameters.

604 Journal of Broadcasting & Electronic Media/December 2012

Control Variables

Demographics. According to extant literature, demographic variables may have an influence on many of our variables of interest. Research shows a stable statistical relationship between people's demographic characteristics and whether they consume more or less news (Althaus & Tewksbury, 2000; Kaufhold, Valenzuela, & Gil de Ziga, 2010; Reagan, 1987). Demographics also have an effect on attitudinal variables such as support for immigration (Davidov, Meuleman, Billiet, & Schmidt, 2008; Hood & Morris, 1997). In this study a set of controls was introduced to eliminate potential confounding relationships in our analyses. In addition to respondents' gender (67% females), age (M D 45.79, SD D 11.31) and race (84% whites); education and income were also included in the models. Education was measured with a 7-point scale ranging from less than high school to doctoral degree (M D 4.11, Mdn D 2-year college degree, SD D 1.50). For income, each respondent chose 1 of 15 categories of total annual household income (M D 6.05, Mdn D $50,000 to $59,999, SD D 4.03).

Socio-political Ideology.

Building on previous work in the context of political communication (Allsop & Weisberg, 1988; Keum et al., 2006; Shah, McLeod, & Yoon, 2001), we have combined the notion of political affiliation or party identification with a measure of citizens' ideological preferences on economic and social issues. The goal was to achieve a comprehensive measure that broadly captured whether respondents have a socio-political ideology. This combined ideological measure was operationalized in three different items. The first item measured their party identification using an 11-point scale ranging from strong Republican (8.7% of respondents) to strong Democrat (13.2% of respondents). The other two items registered respondents' preferences on social and economic issues using an 11-point scale, ranging from very conservative to very liberal. Therefore this index includes all three items: 1) people's political affiliation, 2) citizens' ideological preference on economic issues, and 3) citizens' ideological preference on social issues. In this way, the validity of the scale is established as it has been previously tested in the literature. It is also exhaustive as it registers different dimensions of what it means to be liberal or conservative, which is central to this study. The scale is reliable as reflected by the Cronbach's achieved (Cronbach's D .88, M D 18.01, SD D 8.05).

News Exposure.

One of the main goals in this study was to test the relationship between media exposure, specifically to cable news (FOX News/CNN), and its effects on people's support for Mexican immigration. Therefore, the media use controls employed here were as exhaustive as possible, and included a broad variety of news uses that also included digital media and social media sources, social network sites, citizen journalism sources, and blogs. All of which have been associated with political attitudes and behaviors (Gil de Ziga, 2009; Gil de Ziga et al., 2011). Respondents were asked to rate on a 7-point scale how often they used the following media to get information about current events and public issues: network TV news, cable TV news, local TV news, radio news, print newspapers, online newspapers, print news magazines, online news magazines, news reports generated by regular people, blogs, and social network sites. The items were reverse-coded, so that a higher number indicated more news consumption, and combined into an additive index (Cronbach's D .73, M D 39.92, SD D 9.87).

Criterion Variables

Cable News Use. Once the overall use of media has been residualized, this study introduces two measurements that registered the level of exposure to FOX News and CNN. Both items were operationalized as dummy variables by asking subjects to report which cable news network they watch most often to get information about current news, issues, and events: FOX News (Yes D 28.2%, M D .36, SD D .43); CNN (Yes 31.8%, M D .41, SD D .42). Support for Mexican Immigration. This variable attempts to capture respondents' attitudes toward immigration. Previous research has measured this concept with items that register both the degree to which immigrants may contribute to different values in one's country (see for instance, O'Rourke & Sinnott, 2006), the implications of immigration as a process, and how the government should react to them (Espenshade & Calhoun, 1993; Fennelly & Federico, 2008). Building on this research, the index included 10 items asking on a 10-point Likert-type scale to what extent the government should promote Mexican immigration, legalize it, and increase raids and deportation sweeps (recoded). It also tapped respondents' agreement or disagreement levels to statements about how much Mexican immigration contributes to U.S. values, economy, culture, education, security, and workforce (Cronbach's D .92, M D 43.8, SD.

Statistical Analysis

In order to test the proposed hypotheses and research questions, examining the role of watching FOX News and CNN, we employed zero order Pearson's correlations as well as hierarchical regression analyses. In the regressions, the variables were entered causally in separate blocks; with the demographic variables included first (age, gender, etc) and socio-political orientations and media use added as a Downloaded by [University of Texas at Austin] at 10:49 13 December 2012 606 Journal of Broadcasting & Electronic Media/December 2012 second block. The third block consisted of cable news use (FOX News and CNN)

Results

The first hypothesis in this study deals with the proposition that individuals' sociopolitical ideology will be related to which cable news outlet they watch most often. Results from Pearson's partial correlations support the hypothesis, yielding consonant outcomes with extant literature (i.e., Iyengar & Hahn, 2009). The more conservative a person is, the more inclined they will be to watch FOX News (r D .38, p < .001), and the less likely they will be to watch CNN (r D .18, p < .001). Also, subjects who are more liberal prefer to watch CNN, and prefer to watch less FOX News on cable television (r D .23, p < .001; r D .37, p < .001, respectively). The second hypothesis, that individuals' political ideology will be related to their views on Mexican immigration, also was supported. Specifically, conservative individuals will have more negative attitudes toward Mexican immigration (r D .25, p < .001; D .158, p < .001) than those with a more liberal socio-political ideology (r D .30, p < .001; D .205, p < .001; see Table 1 and Table 2). It should be noted that Fox News and CNN variables should not be correlated to one another employing Pearson's correlation test as they are both dichotomous variables. A Spearman's coefficient of correlation should be calculated or alternatively, the Phi value with a Bonferroni test to correct for the appropriate p-value may also be used as an option (see Hayes, 2005, pp. 263-264). In this instance, we tested these alternative correlations for comparison purposes and changes were almost imperceptible as the Table 1 Zero-Order and Partial Correlations between Socio-Political Orientations, Cable News Use and Support for Mexican Immigration Variables 1 2 3 4 5 1. Conservative .77*** .40*** .20*** .30*** 2. Liberal .76*** .41*** .25*** .35*** 3. FOX News .38*** .37*** .63*** .29*** 4. CNN .18*** .23*** .62*** .21***5. Immigration .25*** .30*** .26*** .18*** Note. Zero order correlations are on top diagonal and partial correlations at the bottom diagonal. Partial correlation controls are age, gender, education, income, race and media use. N D 781 (zero order); N D 740 (partial).

Cable News Use Predicting Support for Mexican Immigration Immigration Block 1: Demographics Age .141*** Gender (female) .014 Education .242*** Income .084* Race (white) .013 R2 (%) 13.2*** Block 2: Media & SP Orientations News Use .089* Conservatives .158*** Liberals .205*** R2 (%) 9.8*** Block 4: Cable News Use CNN .031 FOX News .134*** R2 (%) 2.1*** Total R2 (%) 25.1*** N D 742. Cell entries are final-entry OLS standardized Beta coefficients. *p < .05; **p < .01; ***p < .001. coefficients remained very similar. For instance, FOX News and CNN are correlated at a .63 with Pearson's correlation test and a Spearman's correlation test yielded a .65. According to Hayes (2005) the interpretation of the sign between these two tests is the same, however the interpretation of the size of the effect is a bit more ambiguous as ''the square of the Spearman's coefficient cannot be interpreted as a measure of the percent of variance in one variable explained by variation in the other.'' The Phi test provided similar results although the comparison is even less interpretative. Given these results we have left the original analyses in the table for the sake of comparability and to provide readers the possibility to calculate and interpret the magnitude of the relationship (for more on comparative analyses see Gil de Ziga, Jung, & Valenzuela, 2012). In addition to these hypotheses, two research questions were posed. The first question investigated whether individuals who watch FOX News exhibit more negative attitudes toward Mexican immigration than individuals who watch CNN, even after controlling for the effect of individuals' political ideology. The regression model to predict support for Mexican immigration explained a total of 25.1% of the variance in the dependent variable. Demographic controls explained the largest portion of variance in the model, 13.2% (R2 D .132, p < .001), followed by the media use and socio-political orientations block, which added 9.8% of incremental variance explained (R2 D .982, p < .001), and cable news use with 2.1% (R2 D .214, p < .001). It is important to note that the incremental variance explained by exposure to cable news block is above and beyond the effect (explained variance) that all demographic variables and other socio-political orientations have on people's support for Mexican immigration. Among demographic variables, age ( D .141, p < .001), education ( D .242, p < .001) and income ( D .084, p < .05) were all important predictors of the dependent variable. Younger people and citizens with higher levels of income and education tend to support Mexican immigration. General media use and sociopolitical orientations also have a sound predicting power over the levels of support for Mexican immigration. While being liberal tends to be positively associated with supporting Mexican immigration ( D .242, p < .001), being conservative and, in a milder way, getting exposed to media in general negatively predict pro-immigration attitudes ( D .158, p < .001; D .089, p < .05). After all these controls, results also indicate that selective exposure to FOX News is negatively associated with supporting immigration ( D .134, p < .001). The more individuals watch FOX News, the less likely they will be to support Mexican immigration. On the other hand, getting the news from CNN has no relationship whatsoever with shaping any pro- or anti-immigration attitudes among U.S. citizens ( D .031, p < .11), once all controls are included in the model (see Table 2). The second research question explored the ways in which partisan selective exposure may lead individuals to more polarized attitudes toward Mexican immigration. Thus, the sample was divided into two subgroups: those who reported a conservative sociopolitical inclination and those who reported to be liberal. Results indicate that after controlling for demographic variables and general media use, being exposed to FOX News was associated with a polarizing and reinforcing effect for both conservatives ( D .200, p < .001) and liberals ( D .114, p < .01). Conservatives selectively exposed to FOX News showed a reinforcing effect in terms of their negative support for Mexican immigration. On the other hand, liberals who got selectively exposed to FOX News were also less likely to endorse pro-Mexican immigration attitudes. The model explained 14.3% of the variance (R2 D .142, p < .001) for the group of conservative individuals and 16% (R2 D .160, p < .001), for liberal subjects. Overall, for both groups, the same set of demographic and variables

Exposure to FOX News Predicting Support for Mexican Immigration by Socio-political Orientation Group Immigration Conservatives Liberals Block

1: Demographics Age .150** .127** Gender (female) .035 .029 Education .247*** .271*** Income .092 .087 Race (white) .002 .013 R2 (%) 9.6*** 13.7*** Block 2: Media Use News Use .093* .108** R2 (%) 1.2* 1.3* Block 4: Cable News Use FOX News .200*** .114** R2 (%) 3.4*** 1.1* Total R2 (%) 14.3*** 16*** N D 199 (Conservative); N D 87 (Liberal). Cell entries are final-entry OLS standardized Beta coefficients. *p < .05; **p < .01; ***p < .001.

Discussion

The main purpose of this study was to examine the possible relationship between selective exposure to cable news and the support for Mexican immigrants and Mexican immigrationa key public issue in the USwhile considering people's sociopolitical orientations. Traditionally, partisan selective exposure has been identified as evidence of limited media effects (Bennett & Iyengar, 2008).08). More recently, partisan selective exposure has been found to have strong reinforcement and polarizing effects on individuals (Holbert et al., 2010; Stroud, 2007, 2008). Using survey data on exposure to FOX News and CNNprime examples of conservative and mainstream news outlets in the US, respectively (Iyengar & Hahn, 2009; Stroud, 2007)and perceptions about Mexican immigrants and support for particular immigration policies, this study provides a new basis for the line of research that associates selective exposure with a polarizing effect of media.

First, the results showed that conservative Republicans are more likely to watch FOX News and less likely to watch CNN than liberal Democrats who, in turn, are more likely to watch CNN and less likely to watch FOX News. Second, even after controlling for respondents' partisanship and ideology, watching FOX News was associated with negative perceptions of Mexican immigrants and higher support for restrictive immigration policies. And, most importantly, the FOX News effect was not constrained to conservative Republican respondents because liberal Democrats who reported watching FOX News had more anti-immigrant attitudes than liberal Democrats who did not. The results of the study are noteworthy at several levels of analysis. First, they speak in a loud voice about the negative implications that a more partisan press may have on public opinion. Both CNN and FOX News claim to follow notions of objectivity, presenting impartial information to the audience and reporting in a balanced manner when covering the news. This study indicates that it is not entirely the case for FOX News. In our exhaustive statistical models, we controlled for the effects of demographic variables, and more importantly, the effect of other general media use components. Being exposed to CNN had no relationship whatsoever on the attitudes people hold against Mexican immigrants and immigration, while being purposively exposed to FOX News was associated to negative, anti-immigration attitudes. Second, people's ideological predispositions do not act as a barrier against the negative effects of FOX News on people's perceptions of Mexican immigration. For conservative Republicans, FOX News acts as an echo chamber, reinforcing their already negative inclinations against Mexican immigrants. But even for liberals and Democrats, watching FOX News was significantly related to their support for Mexican immigrants. Another important finding was the negative association between general news consumption and Mexican immigration. Although it is a mild relationship when compared to FOX News, it was nonetheless statistically significant. That is, the more people consume news across different outlets, the less support they report toward Mexican immigrants and immigration. Why so? One possible explanation is that selective exposure may be taking place not only when individuals watch television news, but also when they read newspapers, listen to radio shows, or browse the Internet. Lacking a more detailed knowledge of the exact outlets used by our respondents, it is impossible to discern the exact nature of this negative association. Future research, then, should go beyond comparisons of cable news networks and examine possible reinforcement effects of particular newspapers, radio shows and Web sites. Further research is also needed to understand the effect of selective exposure to CNN over immigration attitudes. In the partial correlations, the relationship between watching CNN and supporting Mexican immigrants and immigration was positive. However, in the regression analysis the association was reduced to non-significance. This may be a statistical problem: when more controls are included in the regression model, we are also reducing valuable degrees of freedom. On a more substantive side, it may also reveal that the important effect of CNN occurs via an interaction with individuals' sociopolitical ideology. This is likely a better explanation because the partial correlations did not control for partisanship or ideology as we wanted to examine these in more depth in the subsequent analyses. Additionally, this study did not include analysis on the effects of individuals who selectively got exposed to MSNBC as a potential partisan (liberal) media outlet. It would be of great interest to first test whether or not selective exposure to this channel leads to any media effect on the audience, and second, if so, whether that effect is similar, stronger or milder than that exerted by FOX News and CNN. This is of course, another recommendation for future research. Despite the results of this study, the analysis has some limitations. As occurs with any research that employs survey data, we were constrained to self-reports of media use, which may yield inaccurate measures due to imperfect recall and social desirability bias. However, the national survey data provide more generalizability of the findings. Future research could complement this analysis by relying on actual exposure data, as provided by TV ratings and through manipulation of exposure in a controlled lab experiment. Another limitation is the cross-sectional nature of the data employed, which is not the best method for testing causal-effects relationships and cannot properly address issues of endogeneity between explanatory and outcome variables. Therefore, we have carefully talked about associations rather than effects throughout the article. To be more confident about the order of influences in our hypotheses, we have addressed the possibility that immigration attitudes and socio-political ideology are influenced by the same factors that affect exposure to FOX News and CNN by taking into account a host of control variables. Certainly, analyses of panel survey data and controlled experiments should be conducted in the future. Nevertheless, even acknowledging all these limitations, this study adds empirical evidence to the current literature on the consequences of selective exposure on issues that go beyond partisan politics such as attitudes toward Mexican immigrants and immigration. Being selectively exposed to information that aligns with one's own views (i.e., conservatives exposed to FOX News) exerts a reinforcement attitudinal effect against Mexican immigration. On the other hand, there's also a media effect for those who are selectively exposed to dissonant information to their views (i.e., liberals exposed to FOX News) since they also reported negative views about Mexican immigration. Thus, the findings of our study are consistent with research that sees selective exposure as a source of political polarization, adding empirical evidence to the debate on the subject. After all, media effects in today's fragmented media environment may not be so minimal.

Notes

1The selected panel members received the survey's URL through an email invitation. This invitation provided respondents with a time estimate to complete the survey and information about a draw monetary incentive for their participation. The first invitation was sent December 15, 2008 and three reminders were submitted in the following 3 weeks. 2The formula for RR3 is (complete interviews)/(complete interviews C eligible nonresponse C e (unknown eligibility)), where e was estimated using the proportional allocation method, i.e., (eligible cases)/(eligible cases C ineligible cases).