From Coverage to Action

Abstract

The past decade has witnessed a proliferation of media stories about immigration as a result of increases in authorized and unauthorized immigration to the United States. Scholars know little about how this coverage influences political participation across different groups in society. This study employs an experimental design to test the effects of different media frames on immigration in spurring political participation among recent immigrant-rooted communities and non-immigrant-rooted communities. The authors find strong mobilizing effects among Latinos, particularly for frames that highlight social costs and national security concerns, and weak to no effects on Asians, African Americans, and whites.

Keywords

immigration Latinos experiment media political participation frames mobilization

Immigration issues have become contentious in American politics as a result of a dramatic increase in authorized and unauthorized immigrants to the United States. Perhaps the apex of concern over immigration occurred shortly after the House of Representatives passed the so-called Border Protection, Antiterrorism, and Illegal Immigration Control Act, or HR 4437, in December 2005. The bill made it a felony to be an undocumented immigrant and provided criminal sanctions against persons or groups assisting undocumented immigrants. Following its passage, in the spring of 2006, an estimated 3.5 to 5 million immigrants and immigrant rights advocates took to the streets across the nation in protest of HR 4437 (Pantoja, Menjivar, and Magana 2008). This was the single largest civil rights action in American history, and scholars of Latino and immigrant politics scrambled to explain the size and extent of the protests. The most popular, though untested, explanation is that the media played a critical role in spurring immigrants and their sympathizers to action (Barreto et al. 2009).

Research on media and politics suggests that increased media coverage of immigration issues will increase its saliency among the public (Iyengar and Kinder 1987) and in particular among those most directly affected by the issue, namely immigrants (Abrajano and Singh 2009). In addition, the content, tone, and source of immigration news stories will sway evaluations, either positively or negatively (Brader, Valentino, and Suhay 2008). Finally, and most relevant for our purposes, news about immigration may increase the likelihood of political participation (Brader, Valentino, and Suhay 2008). But will such coverage affect all groups equally? If media coverage plays a role in spurring participation as some scholars contend, why did we not observe much of a countermobilization or equal numbers of sympathetic non-Hispanic whites or African Americans joining Latino immigrants at the rallies? That is, why might reading news about an issue affect some populations and not others?

One key reason that we put forward in this article is that the effects of a media message will vary by the characteristics of the consumer. We draw from literature in social psychology, political advertising, and race and ethnic politics to explain why media stories on immigration are salient for Latinos, and to a lesser extent Asians, and spur them to political action, while the same stories have little to no effect on other groups. In so doing, we extend research exploring the nexus between media and political participation to the realm of immigration and different racial and ethnic subgroups.

We employ an experimental design with a student sample and a national Internet sample to isolate the effects of five different immigrant-focused media frames, (1) economic positive, (2) economic negative, (3) social positive, (4) social negative, and a (5) national security story, on the propensity to participate among Latinos and Asians, or immigrant-rooted groups who have large numbers of foreign-born persons, and non-Hispanic whites and African Americans, groups with smaller foreign-born populations. Our concern is why certain media messages spur some groups to political action, while the same messages have little to no effect on other groups. The five different immigrant-focused frames we examine vary in content and tone, and we analyze whether stories that focus on material interests, symbolic cultural issues and identity, or national security have stronger mobilizing effects, and whether this varies across different populations. We argue that stories that focus more on symbolic threats and national security will be more effective in mobilizing immigrant-rooted groups.

Our study explores the role of the media in spurring certain groups to political action over immigration issues. While we do not look directly at the role of the media in the political protests in 2006, our results shed light on how the media can increase the likelihood of participation among certain populations. Furthermore, our study has more general implications for how media coverage might influence participation on other issues that may arise on the political scene. In particular, our findings have implications for media mobilization of issue publics.

Media and Political Participation

There is an extensive literature examining the determinants of voter participation and other forms of political participation. Since our concern is with the role of media in shaping political participation, our review is limited to this research. However, we discuss at length other relevant literature informing our argument and propositions in the next section.

There are two potential ways in which media may increase the likelihood of participation. First, when exposed to media of various kinds, citizens might learn about an issue or a candidate, which will reduce information costs and therefore increase the likelihood of participation (e.g., Rosenstone and Hansen 1993; Geer 2006). Second, exposure may affect psychological engagement by increasing the salience of the issues at stake and the relevance of participation (e.g., DeFrancesco Soto and Merolla 2006).

Much of the literature has focused on the effect that political advertisements have on participation, which are crafted by political elites with the goals of getting supporters to turnout and influencing the voting decisions of those who are undecided. The evidence has been mounting for the mobilizing effects of political advertisements (e.g., Clinton and Lapinski 2004; Geer 2006; but see Gerber and Green 2000 for more moderate effects), though there is a debate as to whether negative advertisements mobilize or demobilize the electorate (for a review, see Lau et al. 1999). The mobilizing effect of ads has also been found for certain groups of Latinos (DeFrancesco Soto and Merolla 2006; Panagopoulos and Green 2008; Abrajano 2010).

Less work has looked at the link between news coverage and political participation, especially among groups with large foreign-born populations such as Latinos and Asians. Among whites and African Americans, media use, especially newspaper readership, has been linked to an increase in electoral participation (e.g., Latimer 1983; Tolbert and McNeal 2003; Prior 2005). If we move outside of the mainstream media outlets, scholars have found that media targeting African Americans are an important catalyst for increased participation (McKenzie 2004). The effects of mainstream or ethnic media on Asian and Latino political mobilization are less known. For example, it is speculated that Spanish-language media (television, radio, and newspapers), in 2006 in several cities throughout the United States, were at the forefront of informing and motivating Latinos to participate in protests that took place in reaction to HR 4437 (Barreto et al. 2009).

Furthermore, little is known about how media coverage of a particular issue affects propensities to participate and whether this varies across different segments of the population. With respect to coverage of immigration, scholars have primarily looked at the effects of media coverage on attitudes toward immigrants and immigration policy (e.g., Domke, McCoy, and Torres 1999; Brader, Valentino, and Suhay 2008; Abrajano and Singh 2009). Also, the focus of much of the existing work has been on turnout in elections. There are many other avenues of political participation that have yet to be explored for immigrant-rooted communities (Latinos and Asians), as well as whites and Africans Americans. One important work in this regard was an experimental study by Brader, Valentino, and Suhay (2008), who found that whites exposed to a story highlighting the costs of immigration that depicted a Latino immigrant (rather than a European immigrant) were more inclined to send a message to their member of Congress to restrict immigration levels.

We are also interested in exploring whether media coverage of immigration influences political participation. However, rather than just consider non-Hispanic whites, we want to look at how immigration coverage differentially affects participation across groups, especially among communities most affected by the issue. As we noted in the introduction, we expect to find that reading news stories about immigration will make immigrant-rooted communities more inclined to participate than non-immigrant-rooted communities. In the next section, we develop our argument for why immigration coverage will make some groups more likely to participate. We also discuss the types of content that may be more or less likely to lead to political engagement across groups.

Who Is Mobilized by Media Coverage of Immigration?

The general argument articulated in the previous section is that any exposure to media coverage reduces information costs and should thus increase participation. However, this argument does not consider how well individuals are paying attention to and processing the information. Research on political persuasion in social psychology finds that information is more likely to be processed systematically and have lasting effects when individuals are motivated to process the information (e.g., Petty and Cacioppo 1996; Chen and Chaiken 1999). One factor that influences motivation to process information is how relevant the information is to an individual (Petty, Cacioppo, and Goldman 1981). Information will be more relevant when it is high in personal importance and/or is linked to one’s identity (e.g., Johnson and Eagly 1989). Individuals for whom the issue is high in relevance will be more likely to pay attention to the coverage and process the information, as will individuals who view the information as being linked to their identity. Exposure to the information should then decrease costs and increase psychological engagement with the issue, making participation more likely.¹

Empirical support for this argument has been observed in many studies of the effects of political advertisements on political participation. More specifically, several studies have shown that ads targeted to a particular group are more effective in increasing turnout than ads not targeted to the group since these messages are higher in personal relevance (e.g., Clinton and Lapinski 2004; DeFrancesco Soto and Merolla 2006; Green and Vavreck 2008; Panagopoulos and Green 2008; Abrajano 2010). For example, in a field experiment, Green and Vavreck (2008) found evidence that Rock the Vote ads had a mobilizing effect among the target audience, young voters, and null effects among older voters. With respect to Latinos, a field experiment by Panagopoulos and Green (2008) found that radio ads on Spanish-language stations boosted turnout among those with Spanish surnames in uncompetitive districts in the 2006 congressional election but had no effect among those without Spanish surnames. Abrajano (2010) and DeFrancesco Soto and Merolla (2006) observed similar effects for targeted political advertisements, with Spanish-language advertisements having stronger effects on the likelihood of turnout among Latinos compared to general ad appeals, particularly for Spanish-dominant individuals. Thus, political ads that target particular segments of the population are most effective at inducing mobilization among the targeted group.

We expect a similar process to be at work with news coverage on immigration, with the impact of such stories being most relevant to those who have a stake in the issue, most notably, immigrant-rooted communities. Since the issue should be of higher relevance to Latinos (37 percent are foreign-born according to 2010 census figures) and Asians (61 percent are foreign-born according to 2010 census figures), the effects of immigration media coverage on political participation should be greater among these groups compared to non-Hispanic whites and African Americans. There may be subsets of non-Hispanic whites and African Americans for whom the issue is high in relevance, and we explore this possibility in our analyses. Thus, we arrive at our first hypothesis:

Hypothesis 1 (H1): Media coverage of immigration will increase the likelihood of participation among Latinos and Asians relative to whites and African Americans.

Furthermore, we expect the effects to be stronger among Latinos than Asians, at least in the context of when our study was being conducted. It is likely that Latinos felt a greater sense of being the targets of the immigration debate surrounding HR 4437. One key provision in the bill was building a 700-mile-long fence along the southwestern border with Mexico. As the fastest growing immigrant community, Latinos are often associated with the term illegal immigrant, while this is not the case for Asians (Brader, Valentino, and Suhay 2008; Branton and Dunaway 2008; Knoll, Redlawsk, and Sanborn 2010). Knoll, Redlawsk, and Sanborn (2010) indicate that this association is the result of elite discourse in which candidates use the term illegal alien followed by some reference to the Latino community. The public’s connection of “illegal immigration” with Latinos makes sense since 80 percent of unauthorized immigrants come from Latin America, while the second largest region, Asia, constitutes only 9 percent of all unauthorized immigrants in the United States (Hoefer, Rytina, and Baker 2010). Despite having a sizable foreign-born population, the number of Asians who are undocumented nationally or as a proportion of their population is significantly smaller than that of Latinos; thus, we anticipate that policies targeting undocumented immigrants will be less salient to Asians.

Media discourse also closely links immigration issues with the Latino population. For example, in a content analysis by Abrajano and Singh (2009) of media coverage surrounding Bush’s immigration reform proposal from January to February 2004, 6.7 percent of the articles covered Bush’s attempt to court the Latino vote while none discussed other immigrant groups. We conducted our own content analysis of fifty randomly selected stories on immigration in the New York Times and the Los Angeles Times from June 2007 to June 2008. For each article, we did a count of the number of times the article mentioned Latino or Hispanic or referenced a Latin American country on one hand and Asian or referenced an Asian country on the other. For the Los Angeles Times, fifty articles contained 209 mentions of the former and only 24 for the latter. The comparable counts for the New York Times are 212 for Latino/Hispanic/Latin American country and only 12 for Asian/Asian country. In this type of environment, we expect that immigration media coverage will be of higher relevance to Latinos than Asians, and thus should have stronger mobilization effects among Latinos. Hence, our second hypothesis,

Hypothesis 2 (H2): Exposure to immigrant-focused media will have a stronger effect on Latino levels of political participation than on Asian levels of political participation.

We suspect that reaction to media frames is driven by a combination of material self-interest (e.g., members of my family or friends are immigrants or may be undocumented) and/or symbolic identities (e.g., anti-immigrant stories are an attack on my political or social identity as a Democrat, liberal, or Hispanic) since both dimensions can affect the relevance of the information to the individual. Some media messages may be more effective in cueing material or symbolic interests, and we believe that the latter may have stronger mobilizing effects.² As we indicated earlier, we examine two frames (positive and negative) revolving around the economic consequences of immigration, two frames (positive and negative) over the social and symbolic consequences of immigration, and a national security frame. We chose the first two (economic and social/symbolic stories) because they are the most commonly found in the media and are important factors in shaping attitudes about immigrants and immigration policy (Citrin et al. 1997). We chose to include a national security story since it has become a relevant frame in the post-9/11 context.

Which of these frames might lead to higher levels of political participation? Our discussion centers on Latinos, as they are the ones most likely to be affected by these stories (H2). In the case of negatively toned frames, emotions such as anxiety, fear, or anger will likely be activated. In the case of positively toned frames, emotions such as enthusiasm will be activated. The theory of affective intelligence notes that participation is induced by both anxiety and enthusiasm (Marcus, Neuman, and MacKuen 2000); thus, both types of coverage may boost Latino participation. However, more recent research finds that certain negative emotions, particularly anger, have stronger mobilizing effects than positive emotions such as enthusiasm (Valentino et al. 2011; White 2011). Furthermore, research on Latino reactions to threatening ballot initiatives (Pantoja, Ramírez, and Segura 2001) suggests that contexts of perceived threat may be particularly mobilizing. We can therefore rule out the positively framed articles as being most effective. We are not suggesting that they will have no impact, only that the impact will be less in magnitude than that of the negatively framed articles.

Turning to the effects of different types of negative content (economic, social, and national security), the economic negative frame is most likely to cue material interests. We suspect that Latinos will be less affected by this frame since negative economic arguments about immigration are unlikely to have an immediate direct impact on the economic standing of the broader Latino community as a result of employers favoring non-Latino workers or the possible denial of social and public services. However, Latinos may still perceive a threat to future material well-being from such messages. The social negative and national security frames, on the other hand, have a more chilling effect on Latinos broadly as these frames are more likely to cue symbolic considerations. Both frames present immigrants and Latinos (both are seen as synonymous among the public) as a direct threat to the personal and social security of Americans, while the social negative frame also suggests a threat to American culture. Latinos may perceive such messages as being likely to activate white fears about immigrants, which may create an atmosphere where discrimination toward Hispanic immigrants and nonimmigrants seems justified. Researchers and media pundits believe such a context was present in California in the mid-1990s and Arizona a decade later. As a consequence, Latino rates of naturalization and political engagement surged in this context (Pantoja, Ramírez, and Segura 2001; Avalos, Magaña, and Pantoja 2010). Thus, our final hypothesis states,

Hypothesis 3 (H3): Exposure to the social negative and the national security frame will lead Latinos to higher levels of political participation relative to the other frames.

Experimental Design

To test our arguments, we conducted two experiments. The first was administered in the fall of 2007 with student respondents, and the second was done through the Internet with a national adult sample in the spring of 2008. Participants were randomly assigned to the control group or one of five treatment groups. After exposure to the treatments, we asked respondents a battery of questions about immigration and political participation. This design enables us to see how immigration media coverage influences the likelihood of political participation among Latinos and Asians, relative to whites and African Americans.

Our first study took place in November 2007 in two large political science classes at a public university in California. Respondents were given extra credit for participating in the study. The average age of our respondents was 19.7 years, 60 percent were female, and 84.8 percent were born in the United States. Our student respondent pool was very diverse, with 20.1 percent of the sample identifying as white, 32 percent as Latino, 30.4 percent as Asian, 9.2 percent as African-American, and 8.3 percent as other. On political disposition measures, the respondents were slightly left of center in terms of ideology (M = 3.4 on a 7-point scale) and more Democratic; 51.2 percent identified with the Democratic Party. Only 18.3 percent identified with the Republican Party, 21.2 percent identified as an Independent, and 9.3 percent stated other partisanship. About 44.5 percent of the sample had a friend or family member in the United States in an undocumented status. Our student sample is therefore much younger, diverse, and left-leaning than the non-Hispanic white population. Furthermore, Latinos and Asians in our sample have higher levels of education and are all dominant English speakers, compared to their counterparts in the greater population.

The second study was conducted in May 2008 using an Internet panel from Survey Sampling International (SSI) with a national adult sample. The average age of the respondents was 51 years, 73 percent were female, and 85 percent were born in the United States. The respondent pool was also quite diverse in that we asked for an oversample of Latinos, with 50.5 percent white, 38.1 percent Latino, 4.3 percent African-American, 3.5 percent Asian, and 3.6 percent other.³ On political disposition measures, 44.1 percent of the subjects identified as Democrats, 22.8 percent as Republicans, 22.3 percent as Independents, and 10.8 percent stated other partisanship. The average income was $40,000 to $59,999 and, on average, respondents had some college education. Last, 15 percent of respondents had a friend or family member who is undocumented. If we compare our sample to the 2008 NES study, it looks similar with the exceptions that the SSI sample has more females and is on average one category higher on education (a 7-point scale) and income (an 8-point scale). Our sample is also more diverse, by design. As with the student study, all of the Latinos and Asians speak English. Our results therefore can speak to only English-speaking Latinos and Asians.

Respondents in both studies were randomly assigned to the control (students n = 76; Internet n = 181), economic negative (students n = 78; Internet n = 192), economic positive (students n = 72; Internet n = 158), social negative (students n = 79; Internet n = 158), social positive (students n = 79; Internet = n = 169), and national security (students n = 74; Internet n = 140) experimental conditions. We tested whether random assignment worked as intended across a host of measures, including gender, race, age, education, income, nativity, ideology, partisanship, likely voters, and whether the participant knew someone in an undocumented status. In the student sample, participants were not evenly distributed across the experimental conditions with respect to whether they knew someone in the United States in an undocumented status. In the Internet study, the participants were not evenly distributed across conditions with regard to income, education, identification with the Republican Party, and nativity. To ensure that our conclusions are not affected by these variables, we control for them in later analyses.⁴

In the study with the student sample, we were introduced by the professor and asked respondents to take part in a study that was about attitudes toward immigration. After introductions, the professor left the room. The students were told that if they participated in the study or did an alternative exercise, they would receive extra credit in class. For those who chose to take part in the study, we then handed out the survey packet. Panelists with SSI received an email invitation to participate in the study. If interested, they were instructed to click on a link to participate in the online survey (which was created and hosted with identifying information removed). For both studies, the first part of the survey asked some basic sociodemographic and political predispositions questions. These questions were followed with a news article (or not for the control group) that respondents were instructed to read. After reading the article, respondents completed a short survey, which included questions on immigration and political participation.

Treatments

All of those assigned to a treatment group read an article about immigration policy reform (see the article text at http://prq.sagepub.com/supplemental/). The article was made to look like a real newspaper article on an online news website, such as the LA Times. The first two paragraphs of all of the articles were identical and are reprinted below:

As recently as 10 years ago, only five states—New York, Texas, Florida, Illinois, and California—were dealing with serious immigration problems. Today, immigration affects all 50 states. According to recent estimates, there are an estimated 12 million undocumented workers living in the U.S.

There is widespread consensus that the system is broke and needs fixing. Several attempts at bi-partisan reform have failed on the U.S. Senate floor. The Senate was divided between members seeking to give some legal status to 12 million undocumented workers and lawmakers who oppose such a move. The reform bill is tabled for now, since consensus could not be reached.

The articles then varied with respect to tone and what types of issues were highlighted. In the economic negative condition, respondents read that immigrants take jobs away from U.S. workers, push wages down for those who have jobs, and place a financial burden on state and local governments. The economic positive condition also highlighted economic considerations but placed immigrants in a positive light. Respondents in this condition read that business owners rely on foreign workers, that immigrants take jobs Americans are not willing to do, that immigrants push up the wages of U.S. workers, and that they do not place a burden on state and local governments.

The next two types of articles highlighted social and symbolic factors related to acculturation, history, and society. In the social negative condition, respondents read that today’s immigrants are not assimilating into U.S. culture, that their children are becoming isolated from mainstream America, and that given financial barriers, they are contributing to social problems. The social positive condition touched on similar themes but presented immigrants in a positive light, focusing on the social, economic, and military contributions immigrants have made in the United States.

The final article highlighted the problems that illegal immigration poses to national security. Respondents in this condition read that the border is very easy to cross, that failed border security was one of the reasons cited for 9/11, and that more advanced technology and a stronger fence are needed to protect national security. There was no direct positive message that seemed relevant in this condition, so we focused only on a negatively toned article.

Dependent Variables

We employ two different types of political participation measures: inclination to participate in various civic activities and actual political participation. We asked respondents to indicate on a 10-point scale how likely it is that they will engage in the following four activities: (1) sign a petition for immigration reform, (2) write a letter to your member of congress expressing your views about immigration reform, (3) donate money to an organization dealing with immigration reform, and (4) participate in an immigration reform protest. We combined the four measures into an additive scale of participation, which ranges from 0 to 40, with higher values indicating higher inclinations to participate. These measures hang together well. The Cronbach’s alpha for the student study is .81 and for the Internet study is .78. The mean on this measure is similar for both samples, 20.29 for the student sample and 20.44 for the Internet sample.

Since individuals are likely to give higher reports of willingness to participate, we also developed a measure of actual participation. We gave respondents the opportunity to fill out a postcard on immigration reform at the end of both surveys, which enabled them to send either a pro- or anti-immigration message. The instructions on the student survey were as follows: “We would now like to give you an opportunity to express your views on immigration to the two U.S. Senators from California. If you are interested in sending a message to them, please fill out the post card on the next page. . . . If you choose not to fill out the post card, turn in your questionnaire to the front of the room.” The instructions were similar for the Internet study, though the senators were not identified since it was a national sample and respondents clicked on a link if they wanted to fill out the postcard and a different link if they did not want to fill one out. We created a dichotomous variable in which a one indicates that the person filled out the postcard. About 75 percent of the student sample filled out the postcard, while the percentage was much lower on the Internet study, 36.5 percent. More students likely filled out the postcard since the study took place during class and the investigators were present. There is much less social pressure on an Internet survey, so it is probably a slightly better behavioral measure (see a discussion of these issues in Morton and Williams 2010).

Does Any Coverage of Immigration Increase Participation?

We first explore whether receiving any media story on immigration increases the propensity to participate among immigrant-rooted communities (H1), particularly for Latinos (H2). Figure 1 presents mean values on the participation additive scale broken down by whether the respondent read media coverage on immigration (labeled article in the figure) and by race and ethnicity.⁵ Looking first at the student sample, the only group for which we observe any visible differences is Latinos. Mean participation is 21.88 among Latinos in the control group and 26.02 among Latinos who read media coverage on immigration, and these differences are statistically significant according to a t-test (p = .10). Among Asians and non-Hispanic whites and African Americans, there are no significant differences between those in the control and media-exposed groups (p = .78 and p = .67, respectively). We did expect Asians to participate at higher rates given exposure to immigration coverage, but this does not appear to be the case. We observe a similar pattern in the Internet sample, though the substantive differences are more modest. Mean participation is 18.52 among Latinos in the control group compared to 21.87 among Latinos in the article group, and these differences are statistically significant according to a t-test (p = .01). Meanwhile, there are no significant differences in a t-test between the control and article group among non-Hispanic whites and African Americans (p = .35). We did not have a big enough sample of Asians to do a comparison between treated and control groups in the Internet study.

Figure 1.

Mean participation by exposure to immigration article and racial and ethnic group, student sample and Internet sample.

Figure 2 shows the percentage of individuals who filled out the postcard by whether they read an immigration newspaper article and by race and ethnicity for each study. A similar pattern emerges with this dependent variable. In the student sample, we observe significant differences between the control and article-exposed groups only for Latinos. While only 64.7 percent of Latinos in the control group filled out the postcard, 86.5 percent of Latinos who read an immigration article filled out the postcard, and these differences are significant according to a difference in proportions test (p = .02). There are no significant differences between the two groups for Asians (p = .866) or non-Hispanic whites and African Americans (p = .737). The differences between groups are more modest in the Internet sample, in which 27.5 percent of Latinos in the control group filled out the postcard compared to 38 percent of Latinos in the article group, but they are significant according to a difference in proportions test (p = .10). There is not a significant difference between the control and article group among non-Hispanic whites and African Americans (p = .745).

Figure 2.

The percentage of individuals filling out the postcard by exposure to immigration article and racial and ethnic group, student sample and Internet sample.

We find mixed support for our first hypothesis in that only one of the two immigrant-rooted communities, Latinos, becomes mobilized after reading an immigration newspaper article. It is somewhat surprising that the effects are null for Asians, given the large proportion of foreign-born persons among them. Nonetheless, high rates of citizenship acquisition among Asian Americans (Wong and Pantoja 2009) and the fact that the number of undocumented immigrants is small relative to that of Hispanics appear to weaken the salience of policies targeting undocumented immigrants, at least in our sample. The weak effects among Asians suggest mixed support for H1 since we find effects for Latinos but not Asians. However, the findings lend strong support to H2, that the effects would be more pronounced among Latinos.

One limitation to the basic difference in means and difference in proportions tests is that they do not test whether there are significant differences between those who are and are not from immigrant-rooted communities, nor do they take into account the variables that were unevenly distributed across experimental conditions. To take these factors into account, we turn to multiple variable analyses. To test whether immigrant-rooted communities, particularly Latinos, react differently to media coverage, we interact the article variable with a dummy variable indicating that the respondent is Latino as well as a dummy variable indicating that the respondent is Asian for the student sample. We do not have a big enough sample of Asians in the Internet study, so we include only an interaction between Latinos and the article dummy variable. The baseline in both studies is therefore non-immigrant-rooted communities, or African Americans and non-Hispanic whites. In the student sample, we also control for whether or not the respondent has a family member or friend who is undocumented since this variable was not evenly distributed across conditions. In the Internet sample, we control for income, education level, partisanship, and whether the participant is U.S.-born for the same reason. We use ordinary least squares for the participation scale since it is an additive scale running from 0 to 40 and probit for the postcard measure since it is dichotomous. The results for the student analysis are presented in the top half of Table 1, while the results for the Internet study are in the bottom half of the table. Since interaction terms are not directly interpretable, we calculate the slope and standard error of the treatment for each group to the right of the regression results.

Table 1.

Ordinary Least Squares Regression on Participation Scale and Probit on Filling Out Postcard, Student and Internet Samples.

	Student sample
	Participation		Postcard
	Coeff. (SE)	Slope of treatment (SE)	Coeff. (SE)	Slope of treatment (SE)
Article	0.617 (1.695)	0.617 (1.695)	0.072 (0.256)	0.072 (0.256)
Latino	2.330 (2.714)		−0.298 (0.398)
Asian	0.614 (2.310)		0.023 (0.346)
Latino × article	3.842 (2.887)	4.459* (2.336)	0.695* (0.432)	0.767** (0.348)
Asian × article	−1.396 (2.561)	−0.779 (1.920)	−0.043 (0.384)	0.029 (0.287)
Family friend undocumented	3.667** (0.909)		0.326** (0.143)
Constant	16.566** (1.561)		0.416* (0.234)
N	455		458
R²/pseudo-R²	.171		.036
	Internet sample
	Participation		Postcard
	Coeff. (SE)	Slope of treatment (SE)	Coeff. (SE)	Slope of treatment (SE)
Article	−1.314 (1.121)	−1.314 (1.121)	0.001 (0.140)
Latino	−2.665* (1.634)		−0.209 (0.209)
Latino × article	4.608** (1.802)	3.293** (1.411)	0.250 (0.229)	0.251 (0.181)
Income	0.058 (0.204)		0.034 (0.025)
U.S.-born	−2.462** (1.006)		−0.025 (0.126)
Democrat	1.653** (0.784)		0.084 (0.099)
Republican	4.098** (0.934)		0.293** (0.115)
Education	0.037 (0.263)		−0.033 (0.033)
Constant	21.165** (1.751)		−0.394* (0.218)
N	912		928
R²/pseudo-R²	.040		.010

p ≤ .10. **p ≤ .05.

The findings for these more complicated models confirm what we already observed in Figures 1 and 2. The article variable is statistically significant among Latinos across both sample for the participation additive scale (see slope of treatment column and row with Latino x article), in the study sample for filling out a postcard, and is just outside of conventional significance levels in the Internet sample for filling out a postcard (p = .16, two-tailed). Meanwhile, the article variable is not significant for Asians for either dependent variable in the student sample (see slope of treatment column and row with Asian × article) or for nonimmigrant communities across both measures and both samples (see slope of treatment column and row with article).

With respect to substantive effects, Latino students who read an article are 4.46 units higher on the participation scale relative to their counterparts in the control group; the comparable effect for Latinos in the Internet sample is 3.29 units. These effects are modest for a 40-point scale, but respondents read only a short newspaper article. The slope is not directly interpretable for the postcard analyses since these are probit coefficients. We calculate the probability of filling out the postcard given exposure to an article using CLARIFY (Tomz, Wittenberg, and King 2001). We set all variables to their mean and dummy variables to their mode, such that the baseline respondent has mean income and education, is U.S.-born, and is an independent. Latino students who read an article are 22.2 percentage points more likely to fill out the postcard compared to their counterparts in the control group, while Latinos in the Internet study are 8.6 percentage points more likely to do so.

One could argue that the weak effects among non-immigrant-rooted communities are the result of the fact that we are not taking into account individuals for whom the issue may be high in relevance. Individuals who perceive more of a threat from immigration may be more affected by the media coverage and therefore more inclined to participate. We can identify such subgroups in the Internet sample since the sample is large enough in that study. One obvious candidate for the issue being high in relevance is partisanship. Republican congressional representatives in the 2006 election cycle sought to capitalize on their tough stance on immigration (Hulse 2006). Another potential candidate is low-income individuals, who may face job competition from immigrants (Citrin et al. 1997). We created a measure of low income by splitting our income indicator at the median. Finally, for a more general indicator, we look at individuals who worried about immigration. As we discussed earlier, anxiety about an issue can lead to higher levels of participation (Marcus, Neuman, and MacKuen 2000), and Brader, Valentino, and Suhay (2008) found evidence of this in their study of white reactions to immigration media coverage. We did a median split for a question that asked respondents to indicate to what extent they felt worried about immigration.

The results for the participation measures among these subgroups of non-Latinos are presented in Table 2.⁶ The left half of the table shows the mean on the participation scale for those in the control and article groups along with the corresponding t-test, while the right half of the table shows the percentage filling out the postcard for those in the control and article groups along with the corresponding difference in proportions test. In short, we do not find that exposure to media coverage increases propensities to participate for either dependent variable for Republicans, those with a low income, or high-worry non-Latinos. We should also note that low-income non-Latinos are actually less likely to fill out the postcard given exposure to a media story on immigration (p = .10). Even when we take into account non-Latinos for whom the issue may be high in relevance, we find null mobilization effects. Up to this point, then, the only mobilizing effects we observe for immigration media coverage is among Latinos, the population we expected to be most affected (H2). It could be that only certain stories are relevant for Asians and non-immigrant-rooted communities, so in the next section we disentangle the effects of different types of media coverage on the issue.

Table 2.

Differences on the Participation Scale and Filling Out the Postcard by Exposure to Immigration Article among Republican non-Latinos, Low-Income non-Latinos, and Worried non-Latinos, Internet Sample.

	Mean participation			Percentage filling out postcard
	Control	Article	p value	Control (%)	Article (%)	p value
Republicans	23	22.66	.88	40.0	48.5	.43
Low income	22.03	19.33	.15	51.2	37.3	.10
Worried	22.21	23.05	.56	41.5	47.3	.40

The p value under mean participation is from a difference of means test, while the p value under filling out the postcard is from a difference in proportions test.

Does the Content of the Frames Matter?

We now turn to whether the content of immigration media coverage matters for increasing participation. Recall that we argued that the articles that focus on social costs and national security should be particularly effective among Latinos (H3). Figure 3 presents mean participation across the six experimental conditions by study and broken down by race and ethnicity, while Figure 4 presents the proportion of respondents who filled out the postcard.

Figure 3.

Mean participation by exposure to type of immigration article and racial and ethnic group, student sample and Internet sample.

Figure 4.

The percentage of individuals filling out the postcard by type of immigration article and racial and ethnic group, student sample and Internet sample.

Turning first to Figure 3, mean participation is higher among Latinos in all of the immigration article conditions compared to the control group for both the student and Internet samples, which is consistent with the results from the previous section. There are some minor differences across the treated conditions. In the student sample, participation is lowest for the economic positive condition (M = 22.77), similar for the economic negative, social positive, and national security conditions (around 26), and highest for the social negative condition (M = 29.83), and these differences are significant according to an ANOVA (p = .09), though in regression analyses only the latter three conditions are significantly different from the control group (see Table B1 at http://prq.sagepub.com/supplemental/). For the Internet sample, the strongest effects appear for the economic positive, social negative, and national security conditions, and these treatments are significantly different from the control group in a regression analysis (see Table B2 at http://prq.sagepub.com/supplemental/). The evidence is supportive of H3 since the national security condition and social negative condition have meaningful effects among Latinos across both samples.

If we look at the pattern of results for Asians in the student sample, mean participation is highest in the social negative condition, 20.18, and lowest in the economic negative condition, 13.57, with the mean around 17 or 18 for the other conditions. These differences, however, are not significant according to an ANOVA.

Finally, among non-immigrant-rooted communities in the student sample, mean participation is slightly higher than the control group (M = 17.79) for the economic cost condition (M = 19.5), social benefit condition (M = 18.82), and national security condition (M = 19.34), but these differences are not significant according to an ANOVA (p = .889). In the Internet sample, mean participation is higher than the control group only for those in the national security condition, 22.23, and the ANOVA reveals significant differences across conditions (p = .058). However, these differences do not hold up when a regression is run controlling for the variables that were unevenly distributed across experimental conditions (see Tables B1 and B2 at http://prq.sagepub.com/supplemental/).

We do not replicate the analysis in the previous section in which we interacted the experimental condition with identifying as Latino or Asian in the student sample since that would mean estimating ten interaction terms, which would severely inflate the standard errors. However, we are able to estimate a similar model with the Internet study since we have to interact each experimental condition only with identifying as Latino. In this analysis (see Table C at http://prq.sagepub.com/supplemental/), the findings are consistent with those discussed above in which the treatments do not have a significant effect among non-immigrant-rooted communities, while the economic positive (p = .07), social negative (p = .03), and national security (p = .02) conditions are statistically significant among Latinos.

We now turn to the results for our behavioral measure of participation in Figure 4. When looking at Latinos in the student group, all of the articles show a higher percentage of people filling out the postcard relative to the control group, 64.7 percent, and the differences are significant according to a chi-square test (p = .098).⁷ The differences appear to be particularly pronounced for the economic negative condition, in which 89.3 percent filled out the postcard, as well as the social positive condition and the national security condition, in which 85.7 percent and 96.4 percent, respectively, filled out the postcard. For Latinos in the Internet sample, a higher percentage of respondents filled out the postcards in the treatment conditions relative to the control, especially for the social negative and national security conditions, but the differences across all of the conditions are not significant according to a chi-square test (p = .267). If we run a probit analysis (see Table D2 at http://prq.sagepub.com/supplemental/), we find that the national security condition is significantly different from the control group (p = .10), while the social negative condition is just outside of conventional significance levels (p = .13). The only consistent finding across the two samples is that the national security condition mobilizes Latinos, which provides partial support for H3.

Among Asians, the percentage filling out the postcard is higher for those in the social positive and negative conditions as well as the national security conditions, but the chi-square test (p = .232) and probit analysis reveal no significant differences across conditions. Finally, among non-immigrant-rooted communities, a higher percentage of students filled out the postcard when they read the economic negative, 80.8 percent, social negative, 70.6 percent, and national security articles, 77.4 percent, compared to the control group, 69.7 percent, but these differences are not significant according to a chi-square test (p = .823). In the Internet sample, a higher percentage of respondents filled out the postcard for the social positive and negative conditions as well as the national security condition, but these differences are not significant using a chi-square test (p = .805) or a probit analysis.

If we run one model for the Internet sample in which we interact the experimental conditions with identifying as Latino, the results are in line with those presented above. Only the social negative and national security conditions are significant in increasing the probability of filling out the postcard and only among Latinos (p = .06 in both cases; see Table C at http://prq.sagepub.com/supplemental/). Overall, the only consistent patterns are that the articles that highlight negative social aspects of immigration or national security aspects are mobilizing and only among Latinos, which is consistent with H3. As a follow-up, we ran regression analyses to see if low-income, high-worry, and Republican non-Latinos are more responsive to the particular frames (see Table E at http://prq.sagepub.com/supplemental/). Among Republicans, only the national security condition comes close to increasing mean participation (p = .12). None of the articles appear to mobilize low-income or high-worry nonimmigrants. In summary, when we look at subgroups of non-Latinos for whom the issue may be higher in relevance, we find mobilizing effects only for the consideration that links immigration to another highly salient threat, national security, among Republicans, and it occurs only for self-reports of future participation.

Discussion and Conclusion

Our results indicate that media coverage on an issue can play a role in activating participation. However, not all individuals are more likely to participate given media exposure. Rather, the effects depend on how relevant an issue is to a particular community or individual. In looking at media coverage of immigration, we find mobilization effects of media coverage of immigration primarily among Latinos. We observe much weaker effects among another immigrant-rooted community, Asians. As we argued in our theoretical section, Latinos have been made a more salient target of the immigration debate, and we thus witness stronger mobilization in the face of media coverage for this group. Latinos were also more inclined to participate than Asians in the waves of protest that emerged in 2006 (Pantoja, Menjivar, and Magana 2008). For non-immigrant-rooted communities, we observe mobilizing effects only when immigration was tied to another salient threat, national security, and only among a particular subset, those who are Republican.

With respect to the types of content that may be most effective in increasing participation, our findings suggest that negative coverage that highlights social and symbolic considerations as well as national security concerns are most effective. Among the Latino student sample, this is the case for the likelihood of participating in a range of activities but not for the behavior measure. However, among the Internet sample, only the social/symbolic frames and the national security frame make it more likely that Latinos fill out the postcard. The national security frame was even somewhat effective among non-Latinos who identified as Republican.

What do the findings mean more generally? Overall, there is a great deal of variation across news outlets in terms of the extent and content of immigration coverage. For example, scholars find that coverage of immigration is far more substantial in Spanish-language media compared to English-language media and is more positive in tone in the former (Abrajano and Singh 2009; Branton and Dunaway 2008). Areas closer to the U.S.–Mexico border also receive more coverage about immigration and Latino immigration, with the coverage being more likely to focus on the negative aspects of immigration (Branton and Dunaway 2008). Therefore, we can expect Latino mobilization to occur in areas where Spanish-language media markets are strong or when immigration becomes the focus of English-language media. Overall, media coverage of immigration will more likely spur mobilization on behalf rather than against proimmigration policies given that it is Latinos who are most affected by these messages. In this context, immigrant rights organizations and other linkage institutions can make positive use of the environment to increase political incorporation among Latinos. Increased Latino empowerment in California occurred after the passage of Proposition 187 (Pantoja, Ramírez, and Segura 2001), and we suspect similar effects nationally as a result of the 2005 Border Protection, Antiterrorism, and Illegal Immigration Control Act, or HR 4437.

While our findings are supportive of many of our hypotheses, we recognize there are limitations to the approaches we have taken. First, we have only English-speaking Asians and Latinos in our sample. We do not know if our effects would obtain among non–English speakers and for non-English media. This is an important area for future research to address. Second, we only had a student sample of Asians. It would be useful to replicate the study with a national sample of Asians to make sure that the findings were not specific to Asian students. Third, the study would benefit from additional measures of actual participation, though these are difficult to obtain in an experimental setting. Finally, we looked at only one type of content at a time. In the news, individuals are often exposed to multiple cues in one message (Druckman 2004).

Despite the limitations, this study is an important step in understanding how media coverage influences participation and suggests avenues for future work. First, the findings suggest that not all groups are responsive to media coverage with respect to political participation. Rather, media coverage plays an important role in the political mobilization of communities for whom the issue is high in relevance. Future work should tease out the specific mechanisms underlying the different responses to media coverage across groups. Second, many types of content affected participation, but certain frames seemed to have stronger effects than others. Again, more work is needed to tease out when economic, social or symbolic, and national security frames may be more effective, for whom, and across different types of participatory acts. Future work should also examine whether the mobilization effects are a one-time phenomenon, lasting a single election, or whether their effects are lasting. We suspect that the effects will vary across individuals, and efforts to understand this variation could help explain broader rates of political engagement across the electorate and elections.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Authors receive funding from an internal university grant, called the Blais Challenge Award from Claremont Graduate University.

Notes

References

Abrajano

Marisa A.

2010. Campaigning to the New American Electorate: Advertising to Latino Voters. Palo Alto, CA: Stanford University Press.

Abrajano

Marisa

Singh

Simran

. 2009. “Examining the Link between Issue Attitudes and News Source: The Case of Latinos and Immigration Reform.” Political Behavior 31:1-30.

Avalos

Manuel

Magaña

Lisa

Pantoja

Adrian

. 2010. “Battleground Voters in a Battleground State? Latinos in Arizona and the 2004 Presidential Election.” In Beyond the Barrio: Latinos in the 2004 Election, edited by de la Garza

Rodolfo O.

DeSipio

Louis

Leal

David L.

, 159-70. Notre Dame, IN: University of Notre Dame Press.

Barreto

Matt

Manzano

Sylvia

Ramirez

Ricardo

Rim

Kathy

. 2009. “Mobilization, Participation and Solidaridad.” Urban Affairs Review 44:736-64.

Brader

Ted

Valentino

Nicholas A.

Suhay

Elizabeth

. 2008. “What Triggers Public Opposition to Immigration? Anxiety, Group Cues, and Immigration Threat.” American Journal of Political Science 52:959-78.

Branton

Regina P.

Dunaway

Johanna

. 2008. “Spatial Proximity to the U.S. Mexico Border and Newspaper Coverage of Immigration Issues.” Political Research Quarterly 62:289-302.

Chen

Serena

Chaiken

Shelly

. 1999. “The Heuristic-Systematic Model in Its Broader Context.” In Dual-Process Theories in Social Psychology, edited by Chaiken

Shelly

Trope

Yaacov

, 73-96. New York: Guilford.

Citrin

Jack

Green

Donald P.

Muste

Christopher

Wong

Cara

. 1997. “Public Opinion toward Immigration Reform: The Role of Economic Motivations.” Journal of Politics 59:858-81.

Clinton

Joshua D.

Lapinski

John S.

2004. “Targeted Advertising and Voter Turnout: An Experimental Study of the 2000 Presidential Election.” Journal of Politics 66:69-96.

10.

DeFrancesco Soto

Victoria M.

Merolla

Jennifer L.

2006. “Vota por tu Futuro: Partisan Mobilization of Latino Voters in the 2000 Presidential Election.” Political Behavior 28:285-304.

11.

Domke

David

McCoy

Kelley

Torress

Marcos

. 1999. “News Media, Racial Perceptions, and Political Cognition.” Communication Research 26:570-607.

12.

Druckman

James N.

2004. “Political Preference Formation: Competition, Deliberation, and the (Ir)relevance of Framing Effects.” American Political Science Review 98: 671-86.

13.

Geer

John G.

2006. In Defense of Negativity: Attack Ads in Presidential Campaigns. Chicago: University of Chicago Press.

14.

Gerber

Alan S.

Green

Donald P.

2000. “The Effects of Canvassing, Direct Mail, and Telephone Contact on Voter Turnout: A Field Experiment.” American Political Science Review 94:653-63.

15.

Green

Donald P.

Vavreck

Lynn

. 2008. “Analysis of Cluster-Randomized Field Experiments: A Comparison of Alternative Estimation Approaches.” Political Analysis 16:138-62.

16.

Hoefer

Michael

Rytina

Nancy

Baker

Bryan C.

2010. “Estimates of the Unauthorized Immigrant Population Residing in the United States: January 2009.” Office of Immigration Statistics Report. January. U.S. Department of Homeland Security, Washington, DC.

17.

Hulse

Carl

. 2006. “In Bellwether District, G.O.P. Runs on Immigration.” New York Times, September 6, A1.

18.

Iyengar

Shanto

Kinder

Donald R.

1987. News That Matters. Chicago: University of Chicago Press.

19.

Johnson

B. T.

Eagly

A. H.

1989. “The Effects of Involvement on Persuasion: A Meta-analysis.” Psychological Bulletin 106:290-314.

20.

Knoll

Benjamin R.

Redlawsk

David P.

Sanborn

Howard

. 2011. “Framing Labels and Immigration Policy Attitudes in the Iowa Caucuses: ‘Trying to Out-Tancredo Tancredo.’” Political Behavior 33: 433-454.

21.

Latimer

M. K.

1983. “The Newspaper: How Significant for Black Voters in Presidential Elections?” Journalism Quarterly 60:16-23.

22.

Lau

Richard R.

Sigelman

Lee

Heldman

Caroline

Babbitt

Paul

. 1999. “The Effects of Negative Political Advertisements: A Meta-Analytical Assessment.” American Political Science Review 93:851-75.

23.

Marcus

George E.

Neuman

W. Russell

MacKuen

Michael

. 2000. Affective Intelligence and Political Judgment. Chicago: University of Chicago Press.

24.

McKenzie

Brian D.

2004. “Religious Social Networks, Indirect Mobilization, and African-American Political Participation.” Political Research Quarterly 57:621-32.

25.

Morton

Rebecca B.

Williams

Kenneth C.

2010. Experimental Political Science and the Study of Causality: From Nature to the Lab. New York: Cambridge University Press.

26.

Panagopoulos

Green

2008. “Spanish-language Radio Advertisements and Latino Voter Turnout in the 2006 Congressional Elections: Field Experimental Evidence.” Paper presented at the 2008 annual meeting of the Midwest Political Science Association, Chicago.

27.

Pantoja

Adrian

Menjivar

Cecilia

Magana

Lisa

. 2008. “The Spring Marches of 2006: Latinos, Immigration, and Political Mobilization in the 21st Century.” American Behavioral Scientist 52:499-506.

28.

Pantoja

Adrian

Ramírez

Ricardo

Segura

Gary M

. 2001. “Citizens by Choice, Voters by Necessity: Patterns of Political Mobilization by Naturalized Latinos.” Political Research Quarterly 54:729-50.

29.

Petty

Richard E.

Cacioppo

John T.

1996. Attitudes and Persuasion: Classic and Contemporary Approaches. Boulder, CO: Westview.

30.

Petty

Richard E.

Cacioppo

John T.

Goldman

1981. “Personal Involvement as a Determinant of Argument-Based Persuasion.” Journal of Personality and Social Psychology 41:847-55.

31.

Prior

Marcus

. 2005. “News vs. Entertainment: How Increasing Media Choice Widens Gaps in Political Knowledge and Turnout.” American Journal of Political Science 49:577-92.

32.

Rosenstone

Steven J.

Hansen

John M.

1993. Mobilization, Participation and Democracy in America. New York: Longman.

33.

Tolbert

Caroline J.

McNeal

Ramona S.

2003. “Unraveling the Effects of the Internet on Political Participation.” Political Research Quarterly 56:175-85.

34.

Tomz

Michael

Wittenberg

Jason

King

Gary

. 2001. “CLARIFY: Software for Interpreting and Presenting Statistical Results.” Cambridge, MA: Harvard University.

35.

Valentino

Nicholas A.

Ted

Brader

Groenendyk

Eric W.

Gregorowicz

Krysha

Hutchings

Vincent L.

2011. “Election Night’s Alright for Fighting: The Role of Emotions in Political Participation.” Journal of Politics 73:156-70.

36.

White

Christopher.

2011. “The Mobilizing Effects of Anger During the Political Campaign.” Unpublished manuscript.

37.

Wong

Janelle

Pantoja

Adrian D.

2009. “In Pursuit of Inclusion: Citizenship Acquisition among Asian Immigrants.” In Bringing Outsiders In: Transatlantic Perspectives on Immigrant Political Incorporation, edited by Hochschild

Jennifer L.

Mollenkopf

John H.

, 260-76. Ithaca, NY: Cornell University Press.