Personal experiences & issue salience

Past studies have found that self-interest motivates concerns on policy issues. The appraisal theory posits that individuals pay attention and evaluate issues, situations and events in relation to their personal relevance (Scherer Reference Scherer2005). Studies have found that citizens tend to have higher levels of cognitive, behavioral and emotional engagement with issues that are personally relevant to them (Lazarus and Smith Reference Lazarus and Smith1988). For example, individuals who have had negative experiences with the criminal justice system are significantly more likely to be politically mobilized (Walker Reference Walker2014, Reference Walker2020). Similarly, Americans who have been personally impacted by environmental hazards or who have ancestral or historical connections to the land are more likely to report climate change as a salient policy issue (Brown and Perkins Reference Brown and Perkins1992; Kingsley et al. Reference Kingsley, Townsend, Henderson-Wilson and Bolam2013). In contrast, studies have found that when an issue is not personally relevant to individuals, it is unlikely to elicit cognitive or behavioral responses (Scherer Reference Scherer2005; Miller et al. Reference Miller, Krosnick and Fabrigar2017; Eberl et al. Reference Eberl, Tolochko, Jost, Heidenreich and Boomgaarden2020). Therefore, self-interest and personal experiences can have important implications on policy attitudes.

Extensive research has been conducted on the attitudes towards immigration from various groups. US Latinos tend to have more positive attitudes towards liberal immigration policies than non-Hispanic Whites, Africans Americans or Asian Americans (Cain and Kiewiet Reference Cain and Kiewiet1986; Gramlich Reference Gramlich2019). Studies have found that among Latinos’ connections to immigrants, whether personally being an immigrant or having relatives who were born abroad, are associated with more positive attitudes towards immigration (Bedolla Reference Bedolla2005; Abrajano and Singh Reference Abrajano and Singh2009; Kerevel Reference Kerevel2011; Len-Ríos Reference Len-Ríos2017). However, there have been significant fluctuations in the political attitudes of Latinos towards immigrants and immigration policies in the past decade. Scholars have argued that the evolving political attitudes of Latinos towards immigration have changed as a result of economic environments (Hainmueller and Hopkins Reference Hainmueller and Hopkins2014; Ybarra et al. Reference Ybarra, Sanchez and Sanchez2016), assimilation or acculturation (Michelson Reference Michelson2003; Abrajano and Singh Reference Abrajano and Singh2009), perceived group discrimination (Michelson and Pallares Reference Michelson and Pallares2001; Michelson Reference Michelson2003; Pedraza Reference Pedraza2014) and xenophobic rhetoric and environments (Pérez Reference Pérez2015; Sanchez and Gomez-Aguinaga Reference Sanchez and Gomez-Aguinaga2017; Wallace and Zepeda-Millán Reference Wallace and Zepeda-Millán2020 b). These sources reveal a contested puzzle for public opinion towards immigration among US Latinos, most of whom are immigrant themselves or have immediate connections to immigrants.

Studies of policy salience have continuously shown that personal experiences and self-interest are positively associated with perceptions of greater importance of the issue area (Lavine et al. Reference Lavine, Sullivan, Borgida and Thomsen1996; Cochran and Warren Reference Cochran and Warren2012; Demski et al. Reference Demski, Capstick, Pidgeon, Sposato and Spence2017; Howe Reference Howe2018; Bromley-Trujillo and Poe Reference Bromley-Trujillo and Poe2020). However, the immigrant experience among US Latinos has changed in the past decade. For example, Latin American and Caribbean countries are no longer the fastest growing source of migrants internationally or to the US (Noe-Bustamante and Lopez Reference Noe-Bustamante and Lopez2019). Moreover, recent patterns reveal a net loss of immigrants from Mexico, meaning that “more Mexican immigrants have returned to Mexico from the US than have migrated” to the US since the end of the Great Recession (Villarreal Reference Villarreal2014; Gonzalez-Barrera Reference Gonzalez-Barrera2015). While immigrants from Latin American countries still represent the majority of foreign-born residents in the US, the majority of recent immigrants in the US come from Asia (Smith Reference Smith2019; About Foreign Born 2020). These recent demographic changes call for a reexamination of the personal experiences of US Latinos and their public opinion towards immigration issues.

Based on the appraisal theory, we predict that connections to immigration is a self-interest that continues to have an impact on Latinos’ policy views. Hence, we present the Personal Experience Hypothesis 1, which posits that regardless of the recent demographic changes of the group, foreign-born Latinos are more likely to report immigration as a salient policy issue than their counterparts. This is a result of this subgroup of Latinos not only having direct experience with the immigration policy environment in the US but also potentially benefiting from more protective immigration policies.

Group identity & issue salience

Other scholars argue that while personal relevance can be associated with issue salience, there are stronger predictors of issue salience, including group identity. Although 3 in every 4 US Latinos are immigrants or have at least one immediate relative who is foreign-born, studies have shown significant fluctuations in the public opinion and political attitudes of US Latinos towards multiple issues. Specifically regarding immigration and immigration policy, US Latinos have a wide range of attitudes that have oscillated in the past decade. In 2013 and 2016, for example, about 40% of US Latinos said they worried about deportations, compared to over half of them in 2008, 2010 and 2018 (Lopez et al. Reference Lopez, Gonzalez-Barrera and Krogstad2018 a). Similarly, in the early 2000s, over 40% of Latinos believed that there were already too many immigrants coming to the US annually, compared to only 25% of them in 2018 (Noe-Bustamante and Gonzalez-Barrera Reference Noe-Bustamante and Gonzalez-Barrera2019). Additionally, while in 2016, 31% of US Latinos were immigrants, and only about 20% of Latino voters identified immigration as important in the same year (Pew Research Center 2017; Gallup n.d.). Hence, alternative explanations of immigration salience among Latinos are worth exploring, particularly in the recent elections.

Studies have found that group identity can help explain the changing political attitudes of ethno-racial minorities, including US Latinos (Dawson Michael Reference Dawson Michael1994; Masuoka Reference Masuoka2006; Sanchez Reference Sanchez2006; Michelson Reference Michelson2016; Schildkraut Reference Schildkraut2015, Reference Schildkraut2016). Group identity “refers to an individual’s awareness of belonging to a certain group and having a psychological attachment to that group based on a perception of shared beliefs, feelings, interests, and ideas with other group members” (McClain et al. Reference McClain, Carew, Walton and Watts2009, 474). This concept is based on the Social Identity Theory (SIT), which aims to explain inter-group dynamics and differences “as a function of group-based self-definitions” (Islam Reference Islam2014, 1781); in this way, group identities are based on the identification of people’s “in-group” against “out-group” individuals who do not belong to such group, with warmer feelings towards in-group members (Sidanius et al. Reference Sidanius, Feshbach, Levin and Pratto1997; Gibson and Gouws Reference Gibson and Gouws2005; McClain et al. Reference McClain, Carew, Walton and Watts2009; Islam Reference Islam2014; Schildkraut Reference Schildkraut2015).

Among Latinos, group identity has served as a strong predictor of public opinion towards some policy issues, including attitudes towards immigration. Sanchez and Medeiros (Reference Sanchez and Medeiros2016), for example, find that group identity predicts Latino public opinion on healthcare. Sanchez (Reference Sanchez2006) argues that in addition to nativity, measurements of group consciousness predict Latino’s political attitudes towards Latino-salient policy areas, such as immigration and bilingual education. Johnson (Reference Johnson1998) argues that instead of personal experiences or the legal conceptualization of nativity, group identity shapes US Latinos’ opinions on issues that are salient for the group, such as affirmative action programs and policies. These studies suggest that group identity can serve as a stronger predictor than personal experiences.

We contend that the group identity of US Latinos is increasingly important, particularly in the recent presidential elections as it has increasingly become politicized. Migration rates from Mexico and Latin America, for example, have continuously been on decline since the Great Recession (Burke Reference Burke2017). Furthermore, there has been a substantial rise in second and third-generation Latinos in the US (Flores Reference Flores2017). However, Latinos continue to be racialized as outsiders regardless of their nativity. For example, a provision of Arizona’s SB 1070Footnote ⁵ , which was not upheld by the Supreme Court, requires local law enforcement officials to investigate the immigration status of anyone arrested or stopped; this provision has led to racial profiling of Latinos and people of color in Arizona (Alegria Reference Alegria2014; Peard Reference Peard2018). The increasing number of detention and deportation of immigrants from Latin America also has had detrimental implications among US-born children, youth and families (Vargas and Ybarra Reference Vargas and Ybarra2017; Juárez et al. Reference Juárez, Gómez-Aguiñaga and Bettez2018). Moreover, US-born Latinos continue to be a targeted group during political campaigns. In 2015, when Donald Trump announced his candidacy for the Republican nomination, he condemned Mexican and Latin American immigrants as rapists and drug traffickers (Sanchez and Gomez-Aguinaga Reference Sanchez and Gomez-Aguinaga2017). Also, throughout his presidential campaign, Trump slammed prominent Latino political figures such as federal judge Gonzalo Curiel, the US-born son of Mexican immigrants who oversaw the lawsuit against Trump University, and Susana Martinez, the first Latina governor of a US state (Gomez-Aguinaga and Sanchez Reference Gomez-Aguinaga and Sanchez2020).

These attacks, along with the politization of US Latinos and the ongoing xenophobic rhetoric against immigrants and racial minorities, have been proven to influence Latinos’ ethnic identity and their political views (Pérez Reference Pérez2015; Schildkraut Reference Schildkraut2015; Michelson Reference Michelson2016; Wallace and Zepeda-Millán Reference Wallace and Zepeda-Millán2020 b). Furthermore, they have had tangible spillover effects on the lives of US Latinos who, regardless of their nativity, have reported increasing feelings of exclusion and discrimination (Lopez et al. Reference Lopez, Gonzalez-Barrera and Krogstad2018 b) and have experienced a growing number of hate crimes as reported by the FBI (Brooks Reference Brooks2019). These issues show the continuous politicization and racialization of Latinos regardless of their nativity or citizenship status. The rise in discrimination directed towards Latinos is important to our theory as scholars have found a meaningful link between discrimination and political behavior for Latinos (Sanchez Reference Sanchez2006, Reference Sanchez2008; Schildkraut Reference Schildkraut2015). In fact, perceived discrimination is one of the components or dimensions of group consciousness, one of the dominant measures of group identity (Miller et al. Reference Miller, Gurin, Gurin and Malanchuk1981). Although we do not have the capacity to include all dimensions of group consciousness in our analysis, our study does have the ability to include perceived discrimination across all years.

Linked fate is a specific form of group identity that has been particularly useful in explaining why racial and ethnic minorities will often make political decisions based on the collective interests of their racial or ethnic group (Dawson Michael Reference Dawson Michael1994). While Dawson Michael’s (Reference Dawson Michael1994) linked fate theory was originally developed specifically for the African American experience in the US, scholars have found that although the mechanisms that drive linked fate may be different for Latinos (Sanchez and Masuoka Reference Sanchez and Masuoka2010; Sanchez et al. Reference Sanchez, Masuoka and Abrams2019; Gomez-Aguinaga Reference Gomez-Aguinaga2021 a), linked fate influences Latino political behaviour (Schildkraut Reference Schildkraut2013; Bejarano et al. Reference Bejarano, Brown, Gershon and Montoya2021). We rely on this particular from of group identity in our analysis given the advantage it has of being “a sophisticated and parsimonious alternative” to the operationalization of racial group consciousness (McClain et al. Reference McClain, Carew, Walton and Watts2009). Linked fate has also the practical benefit of being a common measure across each data set in our analysis, whereas other forms of group identity – such as group consciousness – are not.

Based on the Social Identity Theory, the racialized and politicized experiences of US Latinos, and existing literature on the implications of group identity on public opinion, we present Hypothesis 2, which predicts that group identity serves as a predictor of identifying immigration as a salient policy issue among Latinos. Our analysis has the advantage of being able to account for the two most dominant forms of Latino racial or ethnic group identity, linked fate and group consciousness. This is an important distinction given that measures for the two forms of group identity have been found to be empirically distinct from each other when applied to Latinos (Sanchez and Vargas Reference Sanchez and Vargas2016).

News consumption & issue salience

Media scholars have argued that, beyond personal experiences and group identity, news media and news consumption can drive perceptions of issue salience among the public (Dearing et al. Reference Dearing, Rogers and Rogers1996; Lowry et al. Reference Lowry, Nio and Leitner2003; Diefenbach and West Reference Diefenbach and West2007). The theory of agenda setting emphasises the news media’s ability to promote the salience of certain issues or events among the public through content coverage (McCombs and Shaw Reference McCombs and Shaw1972; Zaller et al. Reference Zaller1992; McCombs and Valenzuela Reference McCombs and Valenzuela2007; Dumitrescu and Mughan Reference Dumitrescu and Mughan2010). According to this theory, the more attention a topic receives in the media, the more likely the public is to think the issue is important (Cohen Reference Cohen1963; McCombs and Shaw Reference McCombs and Shaw1972; Peake Reference Peake2001; Jones and Baumgartner Reference Jones and Baumgartner2004). Studies have repeatedly found that judgements about salient issues among the public change with news coverage, even in today’s environment of abundant information and social media (Mazur and Lee Reference Mazur and Lee1993; Miller and Wanta Reference Miller and Wanta1996; Iyengar and Reeves Reference Iyengar and Reeves1997; Althaus and Tewksbury Reference Althaus and Tewksbury2002; Feezell Reference Feezell2018). For example, multiple studies have found that exposure to crime news is a stronger predictor of crime salience than personal experience with crime (Einsiedel et al. Reference Einsiedel, Salomone and Schneider1984; Gross and Aday Reference Gross and Aday2003; Lowry et al. Reference Lowry, Nio and Leitner2003; Diefenbach and West Reference Diefenbach and West2007). Similar patterns exist with news coverage of terrorism (Ullrich and Cohrs Reference Ullrich and Cohrs2007; Spencer Reference Spencer2017) and mass shootings (Chyi and McCombs Reference Chyi and McCombs2004; Muschert and Carr Reference Muschert and Carr2006; Schildkraut and Muschert Reference Schildkraut and Muschert2014).

Existing studies on agenda setting, however, are mostly centered on mainstream media (McCombs and Shaw Reference McCombs and Shaw1972; Cohen Reference Cohen1963; Peake Reference Peake2001; Abbe et al. Reference Abbe, Goodliffe, Herrnson and Patterson2003; Baumgartner and Jones Reference Baumgartner and Jones2010). This issue is problematic given that the emergence and growth of new information technologies have given raise to alternatives to traditional media, including ethnic media (Jeffres Reference Jeffres2000; Viswanath and Arora Reference Viswanath and Arora2000). Ethnic media are commonly defined as “broadcast, print, and digital communication” alternatives to mainstream media designed to “serve a particular cultural or racial group” (Jeffres Reference Jeffres2000). This area of study is relevant for the study of the political attitudes of the American electorate not only because of the rapid socio-demographic changes in the country but also because ethnic media, including Spanish-language news media, are known for having agendas that substantially differ from mainstream media (Lozano Reference Lozano1989; Tirodkar and Jain Reference Tirodkar and Jain2003; Lin and Song Reference Lin and Song2006; Dunaway et al. Reference Dunaway, Branton and Abrajano2010; Kerevel Reference Kerevel2011; Gomez-Aguinaga Reference Gomez-Aguinaga2021 b). Specifically, studies have found that Spanish-language news media generates a larger volume of news related to certain policy issues, such as immigration and health care, compared to English-language media (Branton and Dunaway Reference Branton and Dunaway2008; Abrajano and Singh Reference Abrajano and Singh2009; Dunaway et al. Reference Dunaway, Branton and Abrajano2010; Kerevel Reference Kerevel2011; Villar and Olson Reference Villar and Olson2013; Gomez-Aguinaga Reference Gomez-Aguinaga2021 b). These differences are important not only because of language differences but also because the growing consumption and diverging agendas of nontraditional media (Bendixen and Associates 2005; Allen Reference Allen2009; Lopez Reference Lopez2013).

Concurring with the theory of agenda setting, a handful of studies of ethnic media across the globe have found that consumers of nontraditional media are more likely to identify different policy issues as salient when compared to their counterparts who predominantly or exclusively consume traditional media (Tirodkar and Jain Reference Tirodkar and Jain2003; Ojo Reference Ojo2006; Villar and Olson Reference Villar and Olson2013). While specific examination of the issue salience has been limited in the US context among racial and ethnic minorities, scholars have examined the extent to which the consumption of ethnic media – and SL news – influences the public opinion of the Latino electorate. In the early 2000s, Branton and Dunaway Reference Branton and Dunaway2008 found that SL news media in the US generated more and more positive coverage of immigration news compared to traditional media; this issue, at the same time, resulted in more positive attitudes towards immigration among Latino consumers of Spanish-language news media. Similarly, Abrajano and Singh (Reference Abrajano and Singh2009) found variations on content among Spanish and traditional news media, which were associated with more positive attitudes towards immigration policies among Latino consumers of SL news. Besides the lack of specific studies analyzing issue salience among the Latino electorate by news consumption, existing studies on political attitudes and behaviour have been limited to cross-sectional data (Dunaway, Branton and Abrajano Reference Dunaway, Branton and Abrajano2010; Kerevel Reference Kerevel2011; Gomez-Aguinaga Reference Gomez-Aguinaga2021 b). While we are unable study to conduct longitudinal analyses due to data limitations, the availability of multiple post-election surveys from the past two decades allows us to test the theory of agenda setting among US Latinos across different points in time. This contribution is important not only because the public opinion of immigration among the Latino electorate has constantly shifted but also because it can help us understand the complexity and homogeneity of Latino voters from a more comprehensive approach (Beltrán Reference Beltrán2010; Hainmueller and Hopkins Reference Hainmueller and Hopkins2014; Ybarra et al. Reference Ybarra, Sanchez and Sanchez2016).

Based on the theory of agenda setting, the increasing prominence of Spanish-language media and the fact that Spanish-language news media covers immigration stories to a much greater extent than mainstream media, we expect that the consumption of Spanish-language news media will have an impact on the perceptions of issue salience among US Latino voters. More specifically, we expect that Latino consumers of Spanish-language news media are more likely to identify immigration as a salient issue than their Latino counterparts who consume mainstream media:

Data and methods

To test our hypotheses, we employ the 2008, 2012 and 2016 Collaborative Multi-Racial Post-Election Surveys (CMPS), a widely used survey among social scientists.Footnote ⁶ The surveys were administered over telephone in 2008 and online in 2012 and 2016. Moreover, they were conducted in either English and Spanish and were weighted to reflect a host of census demographics. The survey instruments are particularly appropriate because they provide an opportunity to observe similar items over time with similar control variables available for analysis. To ensure similarity across data sets, we limit our analysis to Latino registered voters. Although this limits our ability to generalise about Latinos, it provides a more rigorous test of our hypotheses since citizens, to some degree, may be less concerned with the issue of immigration than noncitizens.

Our main dependent variable is immigration salience. We measure immigration salience using survey items that ask respondents to identify the most important problem facing the Latino community. All three surveys include items that are reasonably close to this particular question wording.Footnote ⁷ In the 2016 and 2012 CMPS surveys, the items are close-ended, with 14 randomized responses, whereas respondents can select up to two answers. In the 2008 CMPS survey, the questions are open-ended with an opportunity to also select up to two answers. Respondent’s answers are then matched against a precoded list of 14 issues or problems – a strategy that reduces code varianceFootnote ⁸ (Lavrakas Reference Lavrakas2008). Respondents are coded as 1 if they indicated that immigration was important as a first or second answer. Respondents who did not mention immigration were coded as 0.

Next, we include three clusters of independent variables. The first cluster accounts for respondents’ personal stake in immigration, measured with nativity; this variable, referred to as foreign-born in the tables and figures, is based on a survey item that asks respondents if they were born in the US, the island of Puerto Rico, or in another country. Respondents who were born in another country or the island of Puerto Rico are coded as 1. Respondents born in the mainland US are coded as 0 (Valle Reference Valle2019). According to Table A.1, the distribution of our nativity measure shows that less than half of Latino likely voters are foreign-born.Footnote ⁹

Following Social Identity Theory (SIT), the second cluster of variables measures politicized group identity, including perceptions of discrimination – a key dimension of group consciousness – and Latino linked fate. Perceptions of Discrimination is measured with the following survey item: “How much of a problem do you think discrimination against (Hispanics/Latinos) is in preventing (Hispanics/Latinos) in general from succeeding in America?” Response items include: 1=Not a Problem, 2=Minor Problem, and 3=Major Problem. Linked fate is perhaps a stronger measure of Latino identity (McClain et al. Reference McClain, Carew, Walton and Watts2009). Latino Linked Fate is measured by the following survey question: “Do you think what happens generally to Latino people in this country will have something to do with what happens in your life?” The survey question in 2008 is somewhat different, however. Specifically, it asks, “How much does your ’doing well’ depend on other [Insert Group /S3] also doing well?” Although we prefer all questions to have the same wording across surveys, we recognize that this is not always possible. Still, we take comfort in knowing that both survey items have been well utilized by scholars to measure Latino’s perceptions of linked fate (Sanchez et al. Reference Sanchez, Masuoka and Abrams2019). Linked fate is an ordinal variable whereby 1=No, 2=Not Very Much/Little, 3=Some, 4=A Lot. Across the three surveys, 62% in 2008, 30% in 2012 and 51% 2016 reported saying they either have “Some” or “A Lot”. While the percentage of Latinos who say they either have some of or a lot of linked fate has remained above 50% in 2008 and 2016, the results from the 2012 CMPS are somewhat of an outlier across the election cycles in our analysis.Footnote ¹⁰

The third cluster captures Latinos who get their news from Spanish-language media outlets. Spanish-language media is an ordinal variable, whereas 4=Mainly Spanish 3=Both Spanish and English, 2=Mainly English, 1=Never. We include “Never” in our coding sequence since respondents who get their news mainly in English will be more likely to receive media cues than those who say they do not get their cues from the media at all. Unfortunately, the 2012 CMPS does not include a specific question that asks about Latino’s preference for Spanish language media. Since language preference is correlated with the language in which Latino’s consume their news, we created a new variable that equals the product of two survey items: a survey item that captures Latino’s primary language (4=English Dominant, 3=Bilingual, 2=Spanish Dominant) and a survey item that captures whether Latino’s consume their political news by any amount in at least one of the following mediums: television, radio, newspapers magazines, Internet news, social media and blogs. The distribution of the Spanish-language news consumption variable indicates growing popularity (and accessibility) for Spanish-language media. In 2016, 20% of Latino likely voters said they consumed their news mainly in Spanish compared to 17% in 2012 and 10% in 2008.

Additionally, we account for a host of other factors that may potentially explain Latino’s attitudes towards immigration as an important policy issue to the Latino community. First, we include a control variable that accounts for Latino’s partisan affiliation (Hajnal and Lee Reference Hajnal and Lee2011; Ocampo et al. Reference Ocampo, Garcia-Rios and Gutierrez2021). We expect Latinos who identify with a political party to be less likely to identify immigration as a salient issue compared to Independents since they are easier to persuade with elite messaging (Iyengar, Kinder et al. Reference Iyengar and DR1987; Albertson and Gadarian Reference Albertson and Gadarian2015). Our partisan measure is based on a survey item that generally asks if respondents are Republican, Democrat, Independent or something else. Partisan is a dichotomous variable, whereas a coding of 1 indicates Latinos that identify as Democrat or Republican and a coding of 0 indicates political Independents or something else.Footnote ¹¹ Second, we control for Ideological Strength and Political Interest since Latinos who are ideologically more conservative or liberal or keep up with politics may have more a interest in policy issues, such as immigration. Ideological Strength is based on survey items that ask respondents to identify how they think of themselves ideologically. We then constructed a four-point scale whereas 4 indicates respondents who are ideologically “Strong”, 3 indicates respondents who ideologically liberal or conservative, 2 indicates “Slightly” or “Somewhat” ideological, and 1 indicates respondents who say “Moderate”, “None of These” or “Do Not Think in These Terms”. Political Interest is based on survey items that ask respondents about their level interest in politics either more generally or about a specific presidential election. In the 2016 survey, respondents were asked the following: “Some people are very interested in politics while other people can’t stand politics, how about you? In the 2012 and 2008 surveys, respondents were asked the following: ‘Thinking back to October and November of this year, how closely did you follow the news about the 2012/2008 presidential race?’ A coding of 4 indicates ‘Very Interested’ or ‘Very Closely’, 3 indicates ‘Somewhat Interested’ or ‘Somewhat Closely’, 2 indicates ‘Not That Interested In Politics’ or ‘Not Too Closely’, and 1 indicates ‘Not Interested At All In Politics’ or ‘Not Closely At All’”.

Finally, the analysis includes a host of demographic characteristics, including gender (female), age, college education, national origin (Mexican, Puerto Rican, Cuban), Catholic identification (Branton Reference Branton2007; Knoll Reference Knoll2009a, Reference Knoll2009 b), US-Mexican Border State (Branton and Dunaway Reference Branton and Dunaway2008; Dunaway et al. Reference Dunaway, Branton and Abrajano2010) and the percentage change in the Latino population from 2000 to 2010 by zip code (Wilkinson Reference Wilkinson2014; Morin et al. Reference Morin, Sanchez and Barreto2011; Frasure-Yokley and Wilcox-Archuleta Reference Frasure-Yokley and Wilcox-Archuleta2019; Gomez-Aguinaga et al. Reference Gomez-Aguinaga, Sanchez and Barreto2021). Table A.1 in Appendix A shows summary statistics for all independent and control variables. Appendix B in Online Appendix shows the list of all survey items in the analysis.

Results

Table 2 presents “partial” Logistic regression models, which include the three variable clusters measuring Latino identity, Spanish-language news consumption, and personal experience based on nativity. The “fully specified” models include the three variable clusters and additional control variables. All models are based on the 2008, 2012 and 2016 CMPS surveys.Footnote ¹² To interpret the substantive effects of our main independent variables on the dependent variable (immigration salience), we present in Figures 1, 2, and 3 the marginal effects or the change in the predicted probability from their minimum to maximum values while holding all other variables to their respective averages. In the plots, all independent and control variables are ordered according to their effect size from positive to negative.

Table 2. Logit results. Perceptions of immigration salience among Latino likely voters

Note: *p < 0.1; **p < 0.05; ***p < 0.01.

Figure 1. Determinants of immigration salience in 2008 (marginal effects).

Figure 2. Determinants of immigration salience in 2012 (marginal effects).

Figure 3. Determinants of immigration salience in 2016 (marginal effects).

In Table 2, we limit our discussion to the fully specified models (models 2, 4 and 6) since the results are mostly consistent with the models without the added control variables (models 1, 3 and 5). First, the models show support for Hypothesis 1, as our Foreign Born variable is significant across all three models. The positive coefficient suggests that foreign-born Latinos are more likely to say immigration is an important issue to the Latino community. Figures 1, 2, and 3 show foreign-born status influenced immigration attitudes by 16.8 percentage points in 2008, 9.2 percentage points in 2012 and 7.2 percentage points in 2016. Still, it is worth noting that nativity remains a substantive predictor; in fact, a top two predictor of immigration salience in all three models. Thus, immigrant experiences, at least for foreign-born Latinos, play an on-going role in influencing Latino’s perceptions of issue salience across election years.

In line with our second hypothesis, the findings further indicate that group identity has played a prominent role in shaping Latinos’ attitudes towards policy issues. In 2008 and 2012, for example, Latinos who said discrimination prevents Latinos from succeeding in the US were more likely to say that immigration was an important issue to the Latino community. According to Figures 1 and 2, perceptions of discrimination increased the probability of saying immigration is an important issue by 5 and 12.4 percentage points, respectively. In 2012, perceptions of discrimination had the greatest effect on immigration attitudes, which was at least in part attributed to immigration policies, such as Arizona SB 1070 and subsequent copy-cat legislation, that led to concerns over the racial profiling of Latinos.Footnote ¹³ By 2016, perceptions of discrimination failed to achieve statistical significance. However, Figure 3 indicates that Latino linked fate – an arguably stronger measure of politicized group identity – played a greater role in shaping attitudes compared to our discrimination measure. In that year, Latinos who said their individual circumstance was dependent on Latinos as a whole were more likely to say immigration is important by 2.1 percentage points.

The three models also show consistent support for our third hypothesis. Table 2 indicates that Spanish-language news consumption is positive and significant in all models. Figures 1 and 2 show that consuming news mainly in Spanish versus nothing at all increased the probability of saying immigration is a salient policy issue by 7.9 percentage points in 2008 and 6.7 percentage points in 2012. In the 2016 model, Spanish-language media had the greatest substantive effect on immigration attitudes, increasing the probability of reporting immigration as important to the Latino community by 10.7 percentage points. During the 2016 presidential election, for instance, Spanish-language media outlets, such as Telemundo and Univision, routinely reported on Trump’s negative rhetoric towards Mexicans and anti-immigrant platform, which included among other things, a border wall. Moreover, Trump’s attack on the media, including journalist Jorge Ramos, also may have heightened the media’s agenda setting effect.Footnote ¹⁴

In sum, the findings show support for our expectations regarding personal stake, group identity, and Spanish-language news consumption. However, there is a possibility that immigration attitudes are driving both group identity and Spanish-language media consumption. For example, concern for immigration policies affecting the Latino community may influence respondents to consider their relative group status and ties to the Latino community more generally. Additionally, it may also be the case that Spanish-language media outlets are responsive to Latino’s concerns for immigration-related issues. Due to these possible endogeneity concerns, we estimate a full match (Ho et al. Reference Ho, Imai, King and Stuart2007) using the Matchit package in R. The main findings with respect to our identity measures (linked fate and perceptions of discrimination) and Spanish-language media consumption across all three election years continue to hold. In other words, it is likely that both factors are indeed influencing perceptions of immigration salience among Latino voters.Footnote ¹⁵

The results in Table 2 also indicate a number of substantive findings. For example, the surrounding ethnic environment – measured by the percentage change in the Latino population – is significant and positive. However, the finding is limited to the 2012 model, and the effect of living in growing Latino communities is substantively small.Footnote ¹⁶ The table also shows Ideological Strength to have a varied effect on attitudes across the three election cycles. In 2008, for example, liberals and conservatives were less likely to say immigration is a salient issue. In 2012, however, the difference in attitudes disappeared, as they were as likely to say immigration is an important issue as ideological moderates. By 2016, ideological liberals and conservatives were more likely to say immigration is an important issue. The shift in attitudes among Latino voters on both ends of the political spectrum is likely due to the nomination of Republican presidential candidates who increasingly relied on anti-immigrant rhetoric and hardline positions to appeal to their base. For instance, John McCain, the Republican Senator from Arizona, supported an immigration package that included a pathway to citizenship for undocumented immigrants during his presidential run in 2008 while Mitt Romney in 2012 and Donald Trump in 2016 attempted to appeal to a more conservative base by proposing more restrictive immigration policies, such as self-deportation and a border wall. Donald Trump also took his campaign a step further than his Republican predecessors by relying on inflammatory rhetoric towards immigrants. While the trend in these candidate positions likely alienated a large percentage of Latinos who tend to support progressive immigration policies, more recent evidence indicates that hardline positions appealed to a large segment of conservative Latinos (Alamillo Reference Alamillo2019; Gutierrez et al. Reference Gutierrez, Ocampo, Barreto and Segura2019; Ocampo et al. Reference Ocampo, Garcia-Rios and Gutierrez2021), especially in regions such as South Texas and Florida.

In addition to ideological strength, partisan is significant and negative in the 2016 model, suggesting that political Independents were more likely than Democrats and Republicans to be swayed by elite messaging (Iyengar, Kinder et al. Reference Iyengar and DR1987; Hajnal and Lee Reference Hajnal and Lee2011; Albertson and Gadarian Reference Albertson and Gadarian2015). In Table D4 of the Online Appendix, we replaced the partisan measures with controls for Latinos identifying as Democrat and Republican with Independents as the comparison group. Notably, the addition of the partisan-specific measures does not substantively change the main results. Similar to the results in the main analysis, Democrat and Republican are statistically insignificant in the 2008 and 2012 models. However, the negative and statistically significant coefficient for Democrat in 2016 model suggests the partisan finding in the main analysis is primarily driven by a difference in immigration attitudes between Independents and Democrats. More specifically, political Independents are more likely to say immigration is an important issue compared to Democrats, while the difference in attitudes between Independents and Republicans is negligible.

Lastly, the demographic controls reveal important variation in immigration attitudes. For example, age is significant in the 2008 and 2012 models. Age is negatively associated with immigration salience in 2008 and positively associated with immigration attitudes in 2012 and 2016. In 2008, for example, older Latinos were likely responding to other issues – particularly the issue of healthcare (e.g. Affordable Care Act) that was the centerpiece of Obama’s presidential campaign. By 2012 and 2016, though the issue of immigration rose to political prominence with the Obama passing DACA and making immigration reform a centerpiece of his campaign, placing greater emphasis on the policy issue. The findings also demonstrate national origin to be a significant predictor of immigration salience, as Puerto Ricans were less likely to say immigration to be an important issue in 2012 and 2016. Overall, the finding makes intuitive sense given that Puerto Ricans are citizens by birth and, therefore, more likely to consider other policy issues as salient. Conversely, Latinos of Mexican ancestry were more likely to say immigration is the most important issue facing the Latino community in 2016. Finally, Catholics, women, college-educated Latinos and Latinos living in states along the US-Mexico border were less likely to say immigration is a salient issue. While Catholics, women and college educated Latinos were likely concerned with other policy issues, such as morality issues and the economy, the unanticipated negative and significant coefficient for the US-Mexico Border variable is due to multicollinearity, as preliminary analysis indicates no association between US-Mexico Border and Immigration salience at the bivariate level.