Political Identity Convergence:On Being Latino, Becoming a Democrat, and Getting Active
The majority of Latinos in the United States identify with the Democratic Party, a tendency with broad political implications as Latinos become an increasingly large segment of the population. Little research, however, has delved into the origins of this preference. In this research, we contrast two explanations for Latinos’ Democratic proclivities: an instrumental explanation grounded in ideological policy preferences and an expressive identity account based on the defense of Latino identity and status. In analysis of data from two large national datasets, the 2012 Latino Immigrant National Election Study and American National Election Study focused on Latino immigrants and citizens respectively, we find strong support for the expressive identity explanation. Hispanic and partisan identities have converged among Latinos in the United States to create a large number of Latino Democrats regardless of citizenship status. Those who identify strongly as Latinos and see pervasive discrimination against Latinos are the strongest Democrats, a process that further intensified over the course of the 2012 election. A strong partisan preference increased political campaign activity, though this activity level was modest overall. Relatively few Latinos had worked on a campaign or given money to a candidate; somewhat larger numbers had tried to convince others about a candidate or worn a button or displayed a sticker. Finally, some support was evident for an instrumental account. Latino support for government-provided health insurance in 2012 consistently increased support for the Democratic Party.
Latino identity, partisanship, political participation, social identity
Latinos are a growing segment of the U.S. electorate, and their political proclivities matter. The 2012 presidential election underscored Latinos’ increasing political clout, an election year in which they were part of a nontraditional coalition of young people, women, Asians, and blacks that provided a decisive victory to Democrat Barack Obama over his Republican opponent Mitt Romney. The election outcome, coupled with the likely growth of the Latino voter population from 11 percent of the eligible electorate in 2012 to 16 percent in 2030, sparked considerable debate on the future of party politics in the United States (Taylor et al. 2012). Latinos have historically identified with the Democratic Party, roughly 57 percent over the [End Page 205] last fifteen years, more than double the number who identify as Republicans (25 percent in the 2000 Latino Voter Study conducted by Knight-Ridder, see Alvarez and García Bedolla 2003). Shaun Bowler and Gary Segura (2011) note that although Latino support for Democrats has fluctuated to some extent over time it remains generally stable. Strong Democratic identification also translates into support for Democratic candidates, some 71 percent of Latinos voting for Obama in 2012 and 67 percent in 2008, according to network exit polls (New York Times 2012, 2008).
Evidence of Latinos’ general Democratic proclivities is pervasive. Figure 1 demonstrates that in every year since 1984, Latinos have been far more likely to identify with the Democratic than the Republican Party. Far less research has been conducted, however, on why Latinos are more likely to be Democrats than Republicans. Country of origin and past political history each play a role. Data collected over the past fifteen years—the 1989–1990 Latino National Political Survey (LNPS) and the 2000 Latino Voter Survey (LVS)—indicate that roughly 67 percent of Mexicans and between 64 and 69 percent of Puerto Ricans are Democrats, and that between 66 and 69 percent of Cubans are Republican (Alvarez and García Bedolla 2003; Uhlaner and Garcia 1998). Some evidence indicates that more recent Cuban migrants are less Republican than those who came to the United States in the immediate aftermath of the Cuban revolution in 1959 (Garcia 2011). Overall, country of origin influences Latino partisanship. Nonetheless, attributing trends in partisanship to country of origin is a static way to understand current and future trends in Latino political proclivities. We look to factors inherent within contemporary American politics as an alternative and more dynamic account of Latino partisan preferences.
Political party policy stances and political ideology are an ongoing and dynamic aspect of American political life. From an instrumental perspective, partisanship is a running tally of party performance, ideological beliefs, and proximity to the party in terms of preferred policies (Fiorina 1981; Franklin and Jackson 1983). Michael Alvarez and Lisa García Bedolla (2003) view Latino partisanship as instrumental in an account that also explains why partisanship differs among Latinos of different national backgrounds. From this perspective, partisanship rests on a preference for a political party’s policy stances on social or economic issues. Alvarez and García Bedolla demonstrate, with data from the 2000 LVS, that issues and ideology provide a strong foundation for partisanship among Latinos. Holding a liberal stance on a mix of issues such as abortion, affirmative action, school vouchers, health care, gun control, and tax cuts distinguishes Democrats from Independents and Republicans. Carole Uhlaner and Chris Garcia (1998) argue along similar lines that Cubans who left Cuba after the Mariel boatlift incident (during a period of Cuban economic downturn) are more likely to ground partisanship in economic considerations.
Interestingly, immigration attitudes do not [End Page 206] explain Latino partisan preferences in Alvarez and García Bedolla’s (2003) data. This is at odds with numerous suggestions that Latinos’ position on immigration explains their support for Democratic candidates (Garcia 2011). For example, Shaun Bowler, Stephen Nicholson, and Gary Segura (2006) argue that anti-Latino and anti-immigrant propositions sponsored by the Californian Republican Party in the 1990s reversed Latino flight from the state Democratic Party. Sophia Wallace (2012) also argues that immigration reforms feature heavily in Latino support for Obama despite some disappointment in his immigration policy. From Wallace’s perspective, Latino support for Obama arises because his administration has taken small measures designed to improve the daily life of Latino immigrants.
From an instrumental perspective, then, Latinos’ preference for the Democratic Party stems from their generally liberal position on a range of social and economic issues. This support for the Democrats does not appear to be a direct defense of self-interest because less well-educated Latinos are only somewhat more likely to be Democrats than their better-educated counterparts are; no connection can be made between low income and increased Democratic identification (Alvarez and García Bedolla 2003). In other words, Latino support of liberal economic policies and expanded government welfare programs is not concentrated among those in lower-income households or those who are less well educated, suggesting that support for such policies, and the Democratic Party that promotes them, is not a matter of simple self-interest. This is a somewhat puzzling finding given that the instrumental model would predict a link between low income, support for social welfare policies, and the Democratic Party.
But another possible explanation for Latino partisanship has not been well tested as yet. We refer to this as an expressive approach in which partisanship is viewed as a social identity (Huddy, Mason, and Aaroe 2015). This account of partisanship is inherently dynamic, partisans showing increased or decreased enthusiasm for their party in direct proportion to the strength of their partisan identity and potential partisan status threats and gains. The strongest partisans respond most emotionally to partisan threats and reassurances, and therefore work hardest for the party. One way in which expressive partisan identities form is in the merger of partisan identities with broad demographic identities based on race, ethnicity, gender, and class. This can occur when one political party better represents and respects group members than another. Lilliana Mason (forthcoming) has demonstrated the merger of evangelical and Republican identities, black, secular, and Democratic identities, and the particular political potency of partisan identities formed through identity convergence.
Expressive Partisanship and Convergent Identities
Social identity theory provides a strong foundation for the study of expressive partisanship among Latinos. First, the approach can account for the conditions under which Latinos develop a cohesive political outlook, revolving around Latino or Hispanic identity. Second, it can account for the political engagement of Latino Democrats once Latino and Democratic identities merge. The general approach to the study of social identities that Henri Tajfel (1981) developed is agnostic to the nature of the group under study. From a social identity theory perspective, once group members identify as Latinos or Democrats they are motivated to protect and advance their group’s status as a way to maintain their positive distinctiveness (Huddy 2001). In developing the theory, Tajfel and John Turner (1979) placed key emphasis on the need among group members “to differentiate their own groups positively from others to achieve a positive social identity” (Turner et al. 1987, 42). In that sense, Latinos who identify strongly with fellow ethnics are motivated to protect the social status of Latinos from Anglo prejudice and disrespect. Likewise, Democrats are motivated to defend their party against partisan threats and electoral loss. The more strongly individuals identify with their group, the more strongly they defend the group’s status (Huddy 2013).
Political cohesion is especially likely when multiple identities, one of which contains strong political content, converge. Sonia Roccas and Marilynn Brewer (2002) [End Page 207] develop the concept of identity complexity to capture this. They measure the extent to which different social groups were perceived to share characteristics and members, and discover that individuals who are members of highly overlapping groups (those in which members have similar characteristics or include many of the same people) are more reactive to group-based threats than members of groups that are not seen as overlapping. Thus a Latino Democrat will react more to threats against Latinos than a Latino Republican would because Latino and Democratic identities overlap. Political identities have exhibited this type of fusion in the United States in recent years as political partisans become sorted more fully along the lines of political ideology (Levendusky 2009). Mason (2015) examines this process and finds that convergent partisan and ideological identities lead to greater political activism and increased emotional reactivity to group threats. Factors such as group threat can promote identity convergence. For example, in Northern Ireland threat reduced social identity complexity, leading to increased overlap between an identity as Catholic and Irish, and Protestant and British (Schmid et al. 2008).
It is possible to follow the roots of identity convergence all the way back to seminal voting studies that introduced the idea of cross-pressured voters (Lazarsfeld, Berelson, and Gaudet 1944; Campbell et al. 1960). Early electoral studies indicated that partisans who identified with groups associated with the opposing party were less likely to vote. Seymour Lipset went so far as to call these cross-pressured voters “politically impotent,” suggesting that “the more pressures brought to bear on individuals or groups which operate in opposing directions, the more likely are prospective voters to withdraw from the situation by ‘losing interest’ and not making a choice” (1960, 211). Further research found that these voters would be less strongly partisan (Powell 1976) and that such “cross-cutting cleavages” would mitigate social conflict (Lipset 1960; Nordlinger 1972). More recent work has begun to suggest that, in fact, cross-pressures do reduce the strength of partisan affiliation and levels of political activism (Brader, Tucker, and Therriault 2013; Mason 2015; Mutz 2002).
As Democratic and Latino identities move into alignment, Democrats are likely to become increasingly sensitive to ethnic threats, and Latinos to become increasingly sensitive to partisan threats. The more aligned the two identities, the more important it is to a group member that the in-group prevails and maintains status because declining status of one group means declining status of the other. If your party loses an election, your ethnic group loses some of its positive distinctiveness. In that case, according to Tajfel, a group member has two options: to leave the group or to work to make it better. As partisan identity grows stronger, and more identities line up behind it, leaving the party becomes less possible and action becomes necessary. As political identities come into alignment, the effects of identity on political action should thus increase.
African Americans provide a powerful example of identity convergence, involving party and race. They exhibit an impressive degree of racial identity and loyalty, are staunch Democrats, and are far more likely than whites to vote for black Democratic candidates (Reese and Brown 1995; Philpot and Walton 2007; Sigelman and Welch 1984). The electoral effect of group loyalties is most pronounced among African Americans who identify with both the Democratic Party and their racial group (Tate 1994; Dawson 1994). In exit polls conducted during the 2008 Democratic presidential primaries (pitting Barack Obama against Hillary Clinton), respondents in thirty-one states were asked whether race was the single most important factor, one of several important factors, or not important in their vote choice. This is admittedly a crude way to get at the influence of racial loyalties because not everyone is aware of or willing to admit that their vote was affected by such considerations. Nonetheless, roughly 30 percent of black men and women said that race was important to their vote, and they voted overwhelmingly for Obama. Moreover, in a 2008 Democratic primary poll conducted in Pennsylvania by Time magazine, blacks strongly supported Obama based on their concern about racial discrimination in [End Page 208] American society (Huddy and Carey 2009). The fusion between black and Democratic identity is palpable.
Political Action and the Defense of Group Status
One of the real strengths of an expressive approach to partisanship is its ability to explain the link between partisanship and political activity, and the conditions most likely to foster partisan political action (Huddy, Mason, and Aaroe 2015). Researchers have documented the past influence of partisanship on electoral engagement and voter turnout (Abramson and Aldrich 1982; Campbell et al. 1960; Rosenstone and Hansen 1993). But they have paid far less attention to the origins of partisan-driven political engagement. The link between partisanship and political engagement is critical to the study of Latino political behavior and civic incorporation. It may seem odd at a time of growing partisan incivility in the United States to promote partisanship as a path to civic engagement but that is indeed the reality (Iyengar, Sood, and Lelkes 2012). Electoral engagement increases as one becomes a more staunch partisan (Huddy, Mason, and Aaroe 2015). And, as noted, the link between identity strength and political action is even larger when several identities converge, as we believe is the case for Latino and Democratic identities. The link between partisanship and engagement is a compelling reason to examine closely the origins of Latino partisanship.
Research investigating political emotion helps shed light on why strongly convergent identities generate political action. Anger and enthusiasm are highly relevant political emotions known to increase political engagement (Groenendyk and Banks 2013; Marcus, Neuman, and Mackuen 2000; Smith, Cronin. and Kessler 2008; van Zomeren, Spears, and Leach 2008; Valentino et al. 2011). Both emotions are felt more intensely by strong group identifiers and are especially likely to arise during a political campaign. A threatened electoral loss and related loss of power and status generates the action-oriented emotion of anger, whereas reassurance of electoral success and status gains arouse the action-oriented emotion of enthusiasm. And both electoral success and threat are pervasive in competitive elections. Thus we would expect Latino Democrats with highly convergent identities to be the strongest and most politically active partisans. They will take greatest umbrage at a status threat directed at either Latinos or Democrats, and should be more politically active as a consequence. Symbolic grievances concerning perceived ethnic discrimination or other forms of potential ethnic status threat may be especially prone to identity-based intensification, in which the strongest identifiers are angrier and take the greatest offense in response to a specific campaign situation or event.
Republican politicians have pushed Latinos increasingly toward the Democratic Party in recent years. For example, over the last decade, Republicans have taken a series of actions and positions that have alienated and offended many Latinos. These moves are not so much economic as symbolic, involving antiimmigrant legislation, negative rhetoric and portrayals of Latinos, offensive political campaign ads, and an unwillingness to consider immigration reform legislation. The House passage of HR 4437, the Border Protection, Antiterrorism, and Illegal Immigration Control Act of 2005, which included harsher penalties for illegal immigrants, sparked the 2006 immigration reform protests. Most recently, Republicans have advocated repeal of President Obama’s Deferred Action for Childhood Arrivals (DACA) program and voted to deport unaccompanied minors who have been arriving at the border from Central America. According to social identity theory, these are exactly the kinds of threats likely to push Latinos toward the Democratic Party or further enrage Democratic Latinos to take political action.
In this study, we examine the connections of Latino identity, partisanship, and political engagement among two groups of Latinos: citizens included in the main and oversample component of the 2012 American National Election Study (ANES) Time Series and Web component, and immigrants included in the 2012 Latino Immigrant National Election Study (LINES) survey. The inclusion of immigrants is central to this project. The majority of the Latinos in the LINES survey are not citizens, but [End Page 209] they are politically important and vastly understudied. In one sense, we might expect non-citizen immigrants to be more likely than citizens to hold merged partisan and ethnic identities because a Latino identity provides a simple cue to the political system. All one needs to know is that Democrats are more supportive than Republicans of Latino concerns. In that sense, Latino identity is a convenient decisional heuristic. But identity politics may be equally effective among citizens who pay close attention to American politics.
In summary, we examine several hypotheses in this research. First, we assess both the degree to which Latino-Hispanic and Democratic partisan identities have converged among American Latinos and the degree to which Democratic identity is further strengthened among Latinos who believe anti-Latino discrimination is pervasive in American society. Second, we assess the degree to which a strong Democratic identity increased political action in the 2012 presidential campaign (because it involves the convergence of an ethnic and political identity) and contrast that with the effects of a strong (nonconvergent) Republican identity, which we expect to be weaker. Third, we examine the degree to which the 2012 campaign increased the link between Latino identity, Democratic partisanship, and political engagement. The election was characterized by presidential candidate Mitt Romney’s anti-immigrant rhetoric in the Republican primaries, something that could have aroused Latino anger and heightened engagement. Latinos may have also felt increased enthusiasm for Democrats after the election because their support was viewed as instrumental to Obama’s victory, leading to further identity convergence.
We assume throughout that Latino identity is a social identity built on cultural, familial, and geographic factors. We test the notion that Latino identity is convergent with a Democratic identity and in that sense also has political aspects. But we see it as being grounded in a broader array of factors. In that sense, it is distinct from ideology and related stances on economic and social issues, allowing us to contrast its effects on partisanship with such instrumental political factors.
The Latino Immigrant National Election Study is based on data from a sample of adult immigrants from Spanish-speaking countries in Latin America. A sample of 855 Latinos was interviewed by telephone in the preelection survey. Interviewing began on October 4, 2012, and was completed on November 5, 2012. Of those 855 adults, 435 were reinterviewed in the postelection survey for a reinterview rate of 51 percent. An additional new sample of 451 Latino respondents was also interviewed in the postelection survey. The postelection telephone interviews occurred between November 12 and December 20, 2012. This resulted in a total of 886 Latino respondents interviewed in the postelection survey.
Respondent contact information was obtained from the marketing research firm Geoscape; both landlines and cellular numbers were randomly selected for national coverage. Sampling was not conditional on naturalization status. The overall response rate (AAPOR RR4) was 0.320. Professional bilingual interviewers conducted the surveys, most of which were conducted in Spanish. To a large extent, questions were developed to mirror questions asked in the 2012 ANES to facilitate comparison between Latino citizens and noncitizen immigrants.
All LINES data reported in this manuscript were weighted. Weights were created by raking the data to conform to marginal distributions of education, age, and gender of Latino immigrants based on data from the American Community Survey (ACS). The raking calculates weights based on an iterative proportional fitting procedure.
American National Election Studies Time Series and Panel: Latino Sample
Hispanic respondents from the ANES are drawn from both the nationally representative sample and the Hispanic oversample. Data were collected both face-to-face (FTF) and over the Internet (Web). Preelection interviews were collected beginning two months before the [End Page 210] 2012 elections and postelection interviews until two months after the elections. Overall, 472 (141 foreign-born) Hispanic respondents were gathered through FTF methods and 533 (232 foreign-born) through the Web for a total of 1,005. Further, 438 of the FTF sample and 482 of the Web sample completed the postelection survey for respective reinterview rates of 93 percent and 90 percent and a postelection sample of 920.
Web interviews were conducted through GfK Knowledge Networks, which is based on a randomly selected sample drawn from both an address-based sampling frame and a random-digit dialing (RDD) frame of nationally representative telephone numbers. FTF sampling was conducted by Abt SRBI under the oversight of ANES. The forty-eight states were first stratified and then random residential addresses were selected based on information contained in the Delivery Sequence File maintained by the U.S. Postal Service. The Hispanic oversample was collected from tracts in which 20 percent of the population was Hispanic. The overall response rate for the FTF sampling was 38 percent and for the Web sampling 2 percent.
All ANES data are weighted using the overall population weights created by ANES. Weights for the FTF population were created using: age, a cross-classification of age and sex, race-ethnicity, a cross-classification of race-ethnicity and sex, educational attainment, a cross-classification of race-ethnicity and educational attainment, marital status, income, census region, home ownership, and nation of birth. Internet weights were developed using: cross-classification of race-ethnicity and educational attainment, a cross-classification of age and sex, metropolitan status, household Internet access, income, marital status, and home ownership.
Demographic differences are dramatic between the foreign and native-born U.S. Latino populations and are mirrored in the LINES and ANES samples as seen in table 1. Based on estimates of the entire Latino population from the ACS, native-born Latinos are slightly younger than foreign-born (an average age of thirty-seven to forty-four). Native-born Latinos are also far better educated and more likely to speak English than their foreign-born counterparts. More than half (52 percent) of all native-born Latinos have at least some college, versus just over a quarter (27 percent) of their foreign-born counterparts. Interestingly, both groups of Latinos live in relatively low-income households. More than half of all Latinos regardless of nativity live in households earning less than $20,000 per year (ACS).
The native-born ANES sample is very similar to the ACS statistics on native-born Latinos and in that sense quite representative. In contrast, the foreign-born ANES sample stands out in part because they are citizens, a feature of the ANES design. They are also somewhat wealthier, better educated, and a little older than the entire foreign-born Latino population. Latinos in the LINES data mirror the foreign-born population more accurately when it comes to citizenship, age, and education. They are less representative when it comes to income, however. They earn far less than the foreign-born population, and fully 50 percent live in households earning $20,000 or less, versus only 24 percent among foreign-born Latinos in the ACS. The ANES foreign-born sample is somewhat more likely to speak English (although a near majority speak mostly Spanish). And the ANES foreign-born sample, especially those obtained on the Internet, had been in the United States for longer than respondents in the LINES sample. These differences should be kept in mind when drawing comparisons between the two studies.
Except where noted explicitly, identical measures were used in the LINES and ANES studies. Typically, these measures appeared on the same survey (pre- or postelection) in the two studies. Unless noted otherwise, all measures were rescaled to vary from 0 to 1.
The expressive approach to partisanship and the study of political action requires a more finely differentiated measure than the traditional measure of partisanship, which captures [End Page 211] minimal variation in partisan strength between strong identifiers, not so strong identifiers, and leaning independents. In past research, we have developed a direct measure of partisan identity (Huddy, Mason, and Aaroe 2015). These identity questions were not available in the current study, and we thus created a new fine-grained measure of partisan preferences in the pre- and postelection LINES and ANES studies. We regard this as a de facto measure of identity but refer to it throughout as a measure of partisan preference because it contains component measures of party preferences and identification.
In the preelection study, we developed a reliable measure of partisan preference by additively combining three measures of partisanship: self-reported partisan identification (which was not asked in the postelection survey) and Democratic and Republican feeling thermometers (α = 0.71, LINES; α = 0.86, ANES). Those who reported no partisanship affiliation [End Page 212] in response to the traditional question were coded as Independents. Second, a postelection measure of partisanship was created by combining three measures: the party to which one feels closest, liking for the Democratic Party, and liking for the Republican Party (α = 0.71. LINES, α = 0.84, ANES). Finally, a joint measure of partisanship was created from all six items for those in both waves of the panel (α = 0.82, LINES; α = 0.91, ANES).
As seen in table 2, a majority of Latinos identified as Democrats in both the LINES and ANES in terms of self-placement in the preelection survey. All subsamples of Latinos gained [End Page 213] a mean score well above 0.5 on the party preference scale, indicating a preference for the Democratic Party both before and after the election. The native-born ANES Latino Web sample was the weakest subgroup of Democrats across the two studies, only 51 percent identifying as Democrat versus 60 percent of the foreign-born ANES Web component and 61 percent of the LINES sample.
Hispanic Identity and Perceived Discrimination
Respondents were asked in the postelection survey how important Hispanic identity is to them on a 5-point scale that ranged from extremely to not at all important. This was rescaled from 0 to 1, 1 indicating that Hispanic identity was extremely important. Respondents were also asked about linked fate with other Hispanics (“Does what happens to Hispanics in the United States have something to do with what happens in your life?”). These two questions were combined despite modest correlations (r = 0.11, LINES; r = 0.17, ANES) to create a scale of Hispanic identity.1
For the most part, Latinos were strongly identified as Hispanic with over 80 percent of the LINES sample, roughly 60 percent of the foreign-born ANES sample, and just under 60 percent of the native-born ANES FTF sample saying it was extremely or very important to them to be Hispanic. Native-born Latinos in the ANES Web sample were the only subgroup in which a majority did not feel this way. Only 38 percent of this group said Hispanic identity was extremely or very important to them, whereas 26 percent said it was not at all or only a little important.
Respondents were also asked in the postelection survey about how much discrimination exists against Hispanics in the United States today on a 5-point scale ranging from none to a great deal. Responses were recoded so that 1 represented a great deal of perceived discrimination. Perceived discrimination against Hispanics was pervasive. More than half of the LINES sample and roughly 40 percent of the ANES sample saw a lot or a great deal of discrimination. A small minority of Latinos in both studies said there was only a little or no discrimination.
Respondents were asked in the postelection survey whether they had participated in each of five political activities during 2012: trying to convince another about politics, attending a rally, displaying a sign or button, working for a candidate or party, or donating to a candidate or party. These five items were combined to create an index of the number of political activities engaged in during the 2012 election. The scale ranged from 0 to 1 (α = 0.37, LINES; α = 0.61, ANES). Levels of activity are low overall, with the exception of trying to influence another’s vote, something to keep in mind when considering later analyses of the determinants of activity.
Ideology and Policy Issue Stance
Ideological self-placement was measured in both the pre- and postelection surveys. Respondents were asked to place themselves on a 7-point scale from extremely liberal to extremely conservative. Respondents who did not respond were asked to choose one of the labels. Those who responded liberal to the second question were given a score of 3 on the 7-point scale, and those who chose conservative were given a score of 5. Those who did not respond were coded as moderate and placed at the scale midpoint (4).
In the preelection survey, respondents were asked three questions concerning their support for the Dream Act, status checks by state and local officers to determine the immigration status of anyone they suspect of being illegal, and whether illegal immigrants should be criminalized, deported, legalized with penalties, or legalized without penalty. These three items were combined but do not form especially [End Page 214] strong scales (α = 0.18, LINES; α = 0.50, ANES). In the postelection survey, respondents were asked whether they felt that the level of immigration should be increased, decreased, or maintained.
Support for government health care was assessed by two questions in the preelection survey. Respondents were asked how strongly they supported or opposed the 2010 Affordable Care Act and then asked their position on a scale that ranged from 1, support for a government health insurance plan that covers everyone, to 7, support for private health insurance. This was also rescaled from 0 to 1, 1 being the most supportive of the government plan. The two questions were combined to form a measure of support for government-provided health insurance (α = 0.14, LINES, and α = 0.43, ANES).
Respondents were asked whether they supported legal abortion and under what circumstances.
Gay Rights Policy
Respondents were asked if gay couples should be allowed to adopt children, their support for gay marriage, and whether they favored or opposed laws to protect homosexuals from job discrimination. Responses were rescaled and combined to form a scale that ranged from 0 to 1, with 1 indicating greater support for gay rights (α=.64, LINES; α=.74, ANES).
Political Mobilization and Patriotism
Several other factors that could influence levels of political activity were also assessed. To ensure that mobilization efforts were controlled in models analyzing political action, respondents were asked in the postelection survey whether they had been contacted by a political party or some other political entity during the 2012 election. Respondents were also asked in the postelection survey about their feelings for the United States and the American flag. These two questions were combined to form a measure of patriotism (α = 0.16, LINES; α = 0.60, ANES).
In addition, a series of additional questions were asked to assess immigrant status and background. Country of origin was asked of Latinos in the LINES but not the ANES study.
The LINES study, an entirely foreign-born sample, is a productive place to begin an examination of the elements that inform Latino partisan identity. In the first column of table 3, a Democratic Party preference is regressed onto Hispanic identity and perceived discrimination against Hispanics, key facets of an expressive account of partisanship. These analyses are confined to the postelection survey because it contained the ethnic identity and discrimination measures. The dependent variable in column 1 is a partisan preference scale that ranges from 0 (strongly pro-Republican) to 1 (strongly pro-Democratic). Among foreign-born Latinos in the LINES study (column 1), even when controlling for ideological identification, Hispanic identity and perceived discrimination against Hispanics are both strongly tied to an affiliation with the Democratic Party in the postelection survey. The coefficients for Hispanic identity and perceived discrimination are both similar in magnitude to the effect of ideology, usually one of the strongest determinants of partisan identity. In an immigrant sample, therefore, simply identifying as Hispanic or viewing Hispanics as subject to discrimination are linked to stronger support for the Democratic Party. This analysis also confirms the widespread finding that Mexican, Central American, and Dominican Latinos are more likely than Cubans (the omitted category) to gravitate toward the Democrats. These analyses provide preliminary support for the convergence of Latino and Democratic affiliations.
In additional analysis of immigrant Democrats in the LINES study (those who show at least some preference for Democrats over Republicans on the preference scale), Hispanic identity and perceived discrimination further intensify a preference for the party. In this analysis (column 2, table 3), substantial missing data on ideology lead us to omit it from the [End Page 215]
[End Page 216]
analysis in order to increase sample size.2 In this model, Latinos who strongly identify as Hispanic and perceive ethnic discrimination strongly prefer the Democratic Party to the Republican. The interaction between Hispanic identity and perceived discrimination against Hispanics is large and significant, showing that a Latino with a strong Hispanic identity who perceives ethnic discrimination is nearly 48 percent more likely to prefer the Democratic Party to the Republican than a weakly identified Latino who does not perceive discrimination.
Figure 2 depicts the marginal effects of perceived discrimination against Hispanics on Democratic Party identity strength at different levels of Hispanic identity. At low levels of identity, perceived discrimination dampens Democratic identity, but as Hispanic identity strength increases, perceived discrimination increases a preference for the Democratic Party. This is the expected pattern of findings if status politics and identity threat are driving strongly identified Latinos toward the Democratic Party. Thus, among Latino immigrants, identity politics plays a powerful role in driving support for the party. Both Hispanic identity and perceived Hispanic discrimination increase a preference for Democrats over Republicans (despite no interaction between identity and discrimination in the sample as a whole). Also, a preference for the Democratic Party is especially strong among Democrats who identify strongly with and perceive discrimination against Hispanics. Evidence in these analyses indicates that Hispanic and Democratic Party identities have converged as hypothesized. The interaction between Hispanic identity and discrimination also helps rule out the notion that Democratic identification increases Hispanic identity, the reverse causal order to that hypothesized. It is difficult to see why a Democratic identification would only intensify Hispanic identity among those who perceive ethnic discrimination.
The influence of Latino identity and perceived Latino discrimination on Democratic partisan preference in the LINES study may arise because noncitizen immigrants (the bulk of the LINES sample) do not know enough about American politics or the stance adopted by the parties on specific issues to base their partisanship on instrumental concerns. In column 3 of table 3, we turn to the 2012 ANES [End Page 217] sample of Latino citizens (immigrant and nonimmigrant), to determine whether identity politics plays an equally powerful role among citizens. Model 3 generally replicates findings observed among Latino immigrants, with one exception, that a liberal ideology has a far stronger influence on a preference for the Democratic Party among Latino citizens in the ANES than among Latino immigrants in LINES. A change from conservative to liberal increases a preference for the Democratic Party by fully half of the party preference scale, whereas the effect of Hispanic identity and perceived discrimination against Hispanics is essentially the same as that observed in the LINES sample. Even controlling for the large effect of ideology in the ANES study, a Latino with a strong sense of Hispanic identity is significantly more likely than someone with a weak identity to prefer the Democratic Party. Similarly, perceived discrimination against Hispanics significantly pushes Latinos closer to the Democratic Party. Interestingly, the magnitude of these effects is not that different among immigrants in the LINES and foreign-and native-born citizens in the ANES. Even after accounting for a potentially strong instrumental component of partisan affiliation among Latino citizens via ideology, status and identity influence partisan affiliations. Once again, no interaction is evident between Hispanic identity and perceived discrimination when looking at the ANES sample as a whole.
Identity politics—based on an interaction between Hispanic identity and perceived discrimination—does, however, intensify a preference for the Democratic Party among Latino Democrats in the ANES. Model 4 in table 3 is confined to Latinos in the ANES who express a preference for the Democratic Party over the Republican Party. In this model, even after controlling for ideology, someone who strongly identifies as Latino and perceives widespread ethnic discrimination is 45 percent more positive about the Democratic Party than a comparable Latino who perceives no discrimination and has weak Hispanic identity. This model replicates the same finding among Democratic supporters in the LINES study. Figure 2 depicts this relationship graphically. Again, in line with the expectations of social identity theory, strongly identified Latinos who perceive discrimination against members of their ethnic group are the strongest Democrats. In essence, a strong ethnic identity seems to motivate Latinos to defend their group against discrimination, and in contemporary American politics this translates into intensified support for the Democratic Party. Latino and Democratic identities grow increasingly aligned as group-based threats are perceived more clearly.
Finally, column 5 of table 3 demonstrates that among the minority of Latinos who prefer the Republican to Democratic Party, a strong Hispanic identity or the perception that there is widespread discrimination against Hispanics does nothing to intensify the strength of Republican support. In the (admittedly small) subsample of Latinos who call themselves Republicans, the strength of their attachment to that party lies in their ideological leanings and to a smaller extent in demographic characteristics. Latinos who are wealthier, older, and male are more strongly Republican than Democratic. Conservatism has an even greater effect on a preference for the Republican Party among Latino citizens in the ANES. For conservative Latino citizens, Hispanic identity has nothing to do with their support of the Republican Party.
Identity politics appears to have played a substantial role in the development of Democratic partisanship among Latinos, at least after the 2012 election. The election may have intensified the role of identity-based partisanship because it involved Republican anti-Latino rhetoric and potentially boosted Latino pride in effecting a Democratic victory. To assess this possibility, we draw more fully on both the pre- and postelection survey waves to examine the potential intensification of Latino and Democratic identity convergence. Examination of the LINES panel also allows us to contrast the role of issues and identity politics in driving partisan preferences because most issue questions were included in the preelection survey whereas identity questions were in the postsurvey.
Thankfully, both the LINES and the ANES data include an assessment of partisanship before and after the election, which allows us to simplify our model and analyze the role of ethnic [End Page 218] identity and policy issues in affecting changes in party preference intensity over the course of the election. To analyze potential change, we confine analyses to the panel respondents interviewed at both time points. This results in a substantial reduction in the LINES but not ANES sample sizes. In table 4, the direction and strength of party preference is assessed after the election, whereas controlling for party preference is measured before the election. In all five models of table 4, preelection party preference is by far the strongest predictor of postelection party preference, as expected. However, both Latino identity and issue stances also influenced partisanship.
In model 1 of table 4, the full Republican-to-Democratic scale of partisanship is predicted after the election, using the full scale measured before the election. In the LINES data, Hispanic identity is aligned increasingly with Latino preference for the Democratic Party over the course of the election after controlling for their party preference before the election. The analysis in this model also makes clear that party preferences are not based solely on identity politics. Support for government-provided health insurance and gay rights both intensified support for the Democratic Party among Latino immigrants in the LINES study.
Latino identity also intensified Democratic preferences among those who initially preferred Democrats over Republicans in the preelection survey. Support for government health insurance, legalized abortion, and gay rights also intensified a preference for the Democrats. As seen in model 2 of table 4, however, the strongest factor that accounted for the intensity of Democratic support after the election was intensity prior to the election.
In fact, the results from model 1 are also replicated in model 3, in the ANES sample of Latino citizens. Party support after the election is linked most strongly to the choice of party preelection, but Hispanic identity also plays a role. As Hispanic identification increased in the ANES sample of Latinos, Latino postelection support for the Democratic Party also increased. In addition, support for health care and gay rights intensified support for Democrats as it did in the LINES sample. The main difference between the LINES and ANES samples is that ideology matters more in the ANES sample. Among Latinos in the ANES, ideology is a significant predictor of partisan direction postelection, even when controlling for preelection partisanship, indicating that liberals became more supportive of Democrats and conservatives more supportive of Republicans over the course of the election.
Even when only looking at Democrats from the ANES sample in model 4 of table 4, and controlling for preelection Democratic identification, a stronger identification with Hispanics leads to a stronger affiliation with Democrats over the course of the election in the ANES sample. Health-care and abortion attitudes also increased Democratic support over the course of the election among ANES Latino Democrats.
Finally, among the few Republican identifiers in the ANES sample of Latinos, preelection Republican support and conservative ideology had the most significant influence on the strength of Republican identification after the election. More conservative Latinos preferred the Republican Party more strongly. As seen in table 3, Hispanic identification did not intensify Republican identification during the 2012 election. However, instrumental concerns about health care and gay rights had a significant effect on Latino Republicans, conservative positions on those issues increasing the intensity of support for the Republican Party.
In sum, evidence is ample that Latinos, regardless of citizenship status, are moving toward the Democratic Party in part because of ethnic identity politics, merging Hispanic and Democratic identities. A strong Hispanic identity differentiates Democrats from Republicans and when combined with an awareness of Latino discrimination creates especially strong Latino support for the Democratic Party. Identity-based status politics is thus alive and well. But identity politics is not the only basis for party preferences among Latinos. Instrumental support for government-provided health insurance and the Affordable Care Act increased support for the Democratic Party over the course of the 2012 election in both the LINES and ANES studies. And, on average, more Latinos supported than opposed these policies, providing a boost to the Democrats. [End Page 219]
[End Page 220]
Attitudes toward government-provided healthcare assistance also intensified support for the Republican Party among those opposed to it. Gay rights influenced party support over the course of the 2012 election, but Latinos are quite divided in this area, and thus it provides no net benefit to either party. Finally, despite popular conceptions to the contrary, Latino views on immigration had no effect on party preferences. Presumably, if immigration does influence Latino partisanship, it does so through Latino identity and perceived Republican negativity toward Latinos.
CAMPAIGN ACTIVITY IN 2012
We had expected the concordance of partisan and ethnic identity to be especially powerful politically, leading to greater political activity among Latino Democrats than Latino Republicans. To test this theory, we first looked at Latinos in the LINES study, regressing activity in the 2012 campaign (the most common activities of which were talking to others and wearing a sticker or button or displaying a sign) on party preferences folded to indicate strength and then unfolded to indicate strength and partisan direction. Consistent with an expressive view of partisanship (Huddy, Mason, and Aaroe 2015), folded partisanship was a powerful predictor of campaign activity. Our hypothesis of additional activity among Democrats was not supported, however. In model 1 of table 5, no added effect of the Republican to Democrat (direction) scale after controlling for partisan strength was evident.
The hypothesis that merged Hispanic and Democratic identities would more powerfully influence political campaign action than a strong Republican identity had merit, as shown in additional analyses. A preference for the Democratic Party interacted with Hispanic identity to increase political campaign activity in 2012, as seen in model 2 in table 5. When presented visually in figure 3, it becomes clear that Hispanic identity dampened political activity among Latinos who preferred the Republican Party rather than further boosted political activity among Democrats. A similar finding emerges when analyses are confined to the LINES Panel (see model 3, table 5). In sum, Hispanic identity plays a double role in driving Latino political activity. It strengthens Democratic identity and thus elevates action, and it reduces political activity among Republicans, presumably because it generates paralyzing conflict between one’s party and ethnic identity. This conclusion is confirmed by simple bivariate correlation coefficients. On balance, the correlation between Latino identity and 2012 campaign activity is modestly positive for Latinos in the LINES study who preferred the Democratic Party (r = 0.09) and sizeable and negative among those who preferred Republicans (r = -0.22).
Finally, we turn to Latinos in the ANES to further assess the influence of partisan preferences and identity politics on political activity in the 2012 campaign. Once again, as expected, folded partisanship increased political activity. Thus both strong Democrats and Republicans were more likely to take action during the campaign, as shown in model 1 of table 6. No evidence, however, indicated that strong Democrats took greater action than strong Republicans in the ANES data. Interestingly, in this same model, Latinos who identified strongly as Hispanic and perceived widespread ethnic discrimination were also more likely to have been active within the campaign.
The activating effect of ethnic identity and perceived discrimination were somewhat larger among Democrats than Republicans, as seen in models 2 and 3 of table 6. Interestingly, when analyses were confined to Latinos who regarded themselves as Democrats in the preelection survey (based on the standard partisanship question) the interaction between identity and discrimination was significant. This is generally consistent with findings observed in the LINES study of immigrants. At odds with the LINES study, however, the interaction is positive and sizeable among Republicans, though it did not reach significance because of its smaller sample size (model 3, table 6). When the interaction is plotted for Democrats in figure 4, it is clear that perceived discrimination against Latinos has little effect on activity among Latinos who lack a strong Hispanic identity but an increasingly positive effect among Democrats with a stronger Hispanic identity. Oddly, the same trend is apparent among Republicans, suggesting that they [End Page 221] were motivated to work for the Republican Party based on the same identity factors. Admittedly, the ANES sample includes relatively few Republican Latinos, but they largely supported Romney. We are forced to conclude that they were motivated to work on his behalf driven by issues of concern to Latinos, though further research is needed to fully understand this finding.
Finally, among Latinos who were classified as pure Independents in the preelection wave of the ANES (based again on the standard partisanship question), Hispanic identity and a sense of perceived ethnic discrimination both [End Page 222] weakly intensified political activity. Even among individuals with no clear partisanship, politically independent Latino citizens were propelled to take action. Roughly a third of Independents voted in the 2012 election and did so overwhelmingly for Obama, suggesting that heightened political activity among political Independents also favored the Democratic Party.
We also examined the data to see whether living in a swing state with increased levels of political ads and enhanced mobilization efforts increased political activity among Latinos, or among strongly identified Latinos, but it did not. The one exception was that strongly identified Latinos who preferred the Republican Party were less politically active in a swing state, suggesting demobilization. This is consistent with our initial hypothesis that Republicans who identified as Latino would be less active politically. This is apparently true in swing states with heightened levels of statewide political mobilization.
The current analyses raise questions about whether a similar dynamic can be observed among African Americans. The 2012 candidacy of Barack Obama, an African American, and the strong ties between African Americans and the Democratic Party should also arouse identity politics among blacks. From that perspective, Democratic partisanship should be linked to a black identity and perceived racial discrimination among African Americans.
When partisan preferences among blacks in the 2012 ANES were subject to the same analyses as those shown in columns 3 through 5 of table 3, we observed parallel findings. Black identity (assessed as the importance of being black) and perceived discrimination against blacks are associated with a postelection preference for the Democratic Party and a stronger partisan identity among black Democrats. Likewise, black identity and perceived racial discrimination are associated with a greater preference for the Democratic Party in the ANES panel after controlling for preelection party preference and several issues (immigration, government health insurance, abortion, and gay rights). Moreover, the link between ethnic-racial identity and support for the Democratic Party is comparable among blacks and Latinos. This lends added support to an identity convergence model of American partisanship among racial and ethnic minorities. Racial identity and perceived racial discrimination have no additional effect on political activity among blacks, however. As for Latinos, strong partisanship elevates levels of political activity among all blacks, and identity thus has an indirect influence on political activity via partisan preferences. Black identity and political activity are not directly linked in 2012, however. This difference between blacks and Latinos may arise because racial identity has been a part of Democratic politics for quite some time, whereas it has emerged more recently among Latinos.
Overall, we find ample evidence that Hispanic and partisan identities have converged among Latinos in the United States to create a large number of Latino Democrats. This conclusion holds for immigrants in the LINES study and foreign- and U.S.-born American citizens in the ANES. In that sense, our conclusions are similar to those drawn by Sears, Danbold, and Zavala [End Page 223]
[End Page 224]
in this volume. Moreover, the same identity factors strengthen a preference for the Democratic Party once someone develops a pro-Democratic orientation. This strengthening of Democratic identity is most pronounced among those who see pervasive anti-Latino discrimination in the United States and identify strongly with their ethnic group. We were able to document this process of Democratic preference intensification in action in the few months between the 2012 pre- and postelection ANES surveys. Even taking prior party preferences into consideration, a strong Hispanic identity pushed Latinos even closer to the Democratic Party during the course of the election. All in all, Latinos exhibit clear evidence of identity politics at work as their Hispanic and Democratic identities become increasingly fused.
An increasing alignment between Hispanic and Democratic identities holds powerful implications for Latinos’ political engagement now and into the future. We found that a strong partisan preference increased political campaign activity, consistent with an expressive view of partisanship in which a strong partisan identity increases emotional reactivity to campaign events, in turn driving political engagement (Huddy, Mason, and Aaroe 2015). As identity politics moves Latinos even more firmly into the Democratic camp, they are also more likely to get involved in political campaigns. Sergio Garcia-Rios and Matt Barreto document a similar trend elsewhere in this issue, finding that those who felt linked fate with other Latinos (an element in our Latino identity scale) and consumed Spanish language television were more active in the 2012 campaign. Admittedly, Latino action in 2012 was fairly modest. Few Latinos had worked on a campaign or given money to a candidate. Somewhat larger numbers had tried to convince others about a candidate or worn a button or displayed a sticker. But this leaves much room for future Latino political activity.
We found less evidence, however, that Democrats were more politically active than Republicans. We had hypothesized that this would occur because Latino and Democratic identities had converged. There was some evidence along these lines: Latinos in the LINES study who identified strongly as Hispanic and preferred the Democratic Party were even more politically active than others. Democratic Latinos in the ANES who identified as Hispanic and perceived discrimination were more active within the campaign. But so were Republicans in the ANES (albeit to a lesser degree), suggesting something of a contrary movement to reclaim Latino concerns for their party. In the end, identity politics does drive greater Latino activity on behalf of Democrats simply because it has pushed many more Latinos toward the Democratic Party and helped intensify their attachment.
Finally, we began with a discussion of two very different approaches to partisanship: instrumental and expressive perspectives. In the end, both account to some degree for stronger Democratic proclivities among Latinos. We have demonstrated stronger Democratic support among Latinos with a strong Hispanic identity. This preference is even stronger among those who also perceive widespread anti-Latino bias. But we should also return to evidence that majority Latino support for [End Page 225] government-provided health insurance in 2012 also played a consistent role in increasing support for Democrats and weakening it for Republicans among Republican identifiers. Support for pro-immigrant policies did not intensify a preference for the Democratic Party above initial preelection partisan preferences. But the issues of both immigration and government health insurance were tightly connected to identity politics, raising questions about how easy it is to disentangle the two approaches among Latinos. The ANES is the best study in which to look at this because of the large sample retained within the panel (which included preelection questions on issues and postelection questions on Hispanic identity). In these data, support for pro-immigration and government health insurance was significantly correlated with Hispanic identity (r = 0.25 and r = 0.21 respectively) and perceived ethnic discrimination (r = 0.21 in both instances). This link persists even after removing the common effects of partisan preferences, suggesting that both issues have become entwined with identity politics.
In conclusion, perceived Republican animosity toward Latinos is pushing them ever closer to the Democratic Party. As some observers have noted, this is odd because many Latinos are religious and socially conservative, which might incline them in the other direction. But instead, recent discussions of immigration and government health insurance policies have become entwined with Latino identity politics, placing Latinos closer to Democrats than to Republicans. There is no sign that Republicans will relent any time soon and repair their image among Latinos. Unless things change, Latinos will move increasingly into the Democratic camp and work on behalf of Democratic politicians. A policy that legalizes current illegal immigrants will have an especially profound effect on American politics (as Democrats hope and Republicans fear) creating an even larger Democratic Latino community. Without a change in Republican course, Democrats will benefit from continued and intensifying Latino loyalty over the coming years. Identity politics is alive and well in contemporary American society.
Leonie Huddy is professor of political science at Stony Brook University.
Lilliana Mason is assistant professor of government and politics at the University of Maryland, College Park.
S. Nechama Horwitz is a doctoral student in political science at Stony Brook University.
1. In subsequent analyses, this combined variable has greater predictive validity than either variable alone.
2. The omission of ideology does not change the substantive results of the model in additional analyses not shown here.