One of the current realities of politics is the inability of Americans with contemporary immigrant ties to impact electoral and partisan politics in a significant way. In every election since 2000, it has become customary for a chorus of journalists, pundits, and political scientists to predict that this will be the year the Latino “sleeping giant” awakens and makes its presence felt at the polls. In the buildup to the 2012 presidential election, Time Magazine ran a cover story titled “Yo Decido (I Decide), Why Latinos Will Pick the Next President” (Scherer Reference Scherer2012). The “sleeping giant” narrative about Latinos has routinely appeared in the press for two decades with titles echoing the 1998 Mother Jones article “The Sleeping Giant Awakes” (Maharidge Reference Maharidge1998). More recently, similar clichés about the “sleeping giant” are being forcefully applied to the Asian community. Publications like The Wall Street Journal (Reeves Reference Reeves2014) and organizations such as UCLA's Asian American Studies Center (Marquez Reference Marquez2006) have declared that Asians are destined to make their big push into partisan politics.
Yet, time and again these predictions have failed to come to pass. One observable example that highlights the lack of political participation among Latinos and Asians is turnout. According to the Pew Research Center, Latino and Asian turnout remains among the lowest of any ethnic or racial group in the United States (Krogstad et al. Reference Krogstad, Lopez, López, Passel and Patten2016; Taylor et al. Reference Taylor, Cohn, Livingston, Funk and Morin2013). The U.S. Census Bureau reports that in the 2016 presidential election, Asian turnout was 49%, and the only group that trailed Asians was Latinos, at 48% virtually matching the turnout rates for both groups in 2012 (File Reference File2017). When compared with turnout rates of 60% or above for African Americans and whites in 2016, the numbers look particularly unfavorable (File Reference File2017). Regarding political participation, the incongruity between reality and prediction has persisted for over two decades and gives rise to the question: why is this occurring? Why are Latinos and Asians not making their presence felt in partisan and electoral politics?
This paper asserts that a large part of the answer lies with what can be called the “prolonged partisan socialization process” of second-generation Americans. Understanding the protracted partisan development of second-generation Americans is critical because partisanship helps to minimize the informational costs associated with entry into politics. However, because for many second-generation Americans in Latino and Asian communities, partisan attainment occurs at a much slower rate than has been previously understood, it is often assumed that those who have yet to acquire a partisan heuristic are opting out of the political process entirely.
ALTERNATIVE OBJECTS OF SOCIALIZATION FOR SECOND-GENERATION AMERICANS
Much of the political socialization literature maintains that political orientations are acquired during the formative stages of childhood. Political behavior is primarily understood to be influenced by parents, peers, and schools (Berelson, Lazarsfeld and McPhee Reference Berelson, Lazarsfeld and McPhee1954; Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Easton and Dennis Reference Easton and Dennis1967; Reference Easton and Dennis1969; Greenstein Reference Greenstein1960, Reference Greenstein1965; Hess and Torney Reference Hess and Torney1967; Hyman Reference Hyman1959; Jennings and Markus Reference Jennings and Markus1977; Jennings and Niemi Reference Jennings and Niemi1968; Reference Jennings and Niemi1974; Niemi and Sobieszek Reference Niemi and Sobieszek1977). Of particular interest to this project is the consensus surrounding the primacy of parental influence in the acquisition of partisanship, as posited by one of the leading theories of partisan attainment in the field of political science, the Michigan Model. The Michigan Model declares that parents are the most influential conduit in the transmission of partisan beliefs to their children (Beck and Jennings Reference Beck and Jennings1991; Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Jennings and Niemi Reference Jennings and Niemi1968; Reference Jennings and Niemi1974; Lewis-Beck et al. Reference Lewis-Beck, Jacoby, Norpoth and Weisbert2008; Niemi and Jennings Reference Niemi and Jennings1991; Niemi and Sobieszek Reference Niemi and Sobieszek1977). Hence, the partisanship of most Americans is primarily understood to be acquired during childhood, often during pre-adolescence.
However, this seminal description of partisan attainment has often overlooked large segments of the population in terms of both race and citizenship (Abrajano and Alvarez Reference Abrajano and Alvarez2010; Alvarez and García Bedolla Reference Alvarez and Bedolla2003; Andersen Reference Andersen1979; Cain, Kiewiet and Uhlaner Reference Cain, Kiewiet and Uhlaner1991; Dawson Reference Dawson2001; Hajnal and Lee Reference Hajnal and Lee2011; Levin Reference Levin2013; Uhlaner and Garcia Reference Uhlaner and Garcia2005; Wong Reference Wong2000). Specifically, much has been written about how contemporary immigrants do not directly comport with the leading theories of partisan socialization. However, with some notable exceptions, little has been said about the partisan socialization of their American-born offspring and how they comport with the traditional partisan attainment story.Footnote 1
This paper asserts that the conclusions concerning partisan attainment reached by a majority of the political socialization literature are in need of elaboration regarding a significant segment of the population. The lack of deliberation involves the under-examined and under-theorized children of immigrants. This paper contends that the children of immigrants are unlikely to acquire their partisan identity at home, and will instead find it in institutional settings they come into contact with as they age. Those classified as second-generation Americans in this paper are native-born individuals who are understood to have parents that were both born outside of the United States. If a person has at least one parent born in the United States, they are classified with at least third-generation status. The expectation is that those in the third-generation subgroup will comport with the traditional partisan transmission story.Footnote 2
This argument has several major implications. First, it means that this subset of parents is not transmitting their partisanship to their American-born children as has long been contended. Second, it indicates that contrary to our current understanding, children are likely finding their partisan identity outside of the home. Third, these children are likely discovering their partisan identity for the first time much later in life and not during their adolescence as has been theorized. Ultimately this suggests that a significant number of second-generation Americans are arriving late to partisan politics because the partisan socialization process is unfolding at a much slower pace than has previously been accounted for in the literature. Accounting for what I label the “prolonged partisan socialization process” will provide context for the low levels of political participation among those with contemporary immigrant ties.
A few scholars have noted that the traditional partisan acquisition story as understood by the Michigan Model is largely unavailable to the offspring of immigrants (Abrajano and Alvarez Reference Abrajano and Alvarez2010; Alvarez and García Bedolla Reference Alvarez and Bedolla2003; Andersen Reference Andersen1979; Cain, Kiewiet and Uhlaner Reference Cain, Kiewiet and Uhlaner1991; Hajnal and Lee Reference Hajnal and Lee2011). They argue that the unavailability stems from the unique immigrant socialization process, in addition to the absence of the European model of the past, which relied on powerful outreach by political machines to push specific ethnic immigrant groups into party politics upon entry to the United States (Cornwell Reference Cornwell1960; Reference Cornwell1964; Dahl Reference Dahl1961; Erie Reference Erie1998). However, beyond acknowledging that the offspring of immigrants are unlikely to comport with the traditional partisan acquisition story, no one to my knowledge has asked if and how that partisan void is filled, or how long the process takes. This paper seeks to answer these questions.
Using the knowledge that the traditional partisan attainment story is unavailable to many second-generation Americans, an explanation will be presented below as to why institutions are more influential in shaping their partisan development. To test the impact of institutional links, this paper will analyze and compare the associations of second-generation Americans with institutions to determine to what degree these associations increase the likelihood of partisan acquisition for second-generation Americans as compared with both first-generation Americans and those with at least third-generation status. This paper distinguishes itself from past scholarship by directly testing the impact of institutional associations on the partisanship attainment of those who choose to identify with a party and, more importantly, for those who have chosen not to, within each generational iteration. Individuals claiming no partisan preference are customarily dropped from the analysis and subsequently ignored. By overlooking this vital subgroup, which often makes up the plurality of the contemporary immigrant population, as is the case with Latinos and Asians, we fail to get a realistic picture of the overall population's behavior (Hajnal and Lee Reference Hajnal and Lee2011).
Additionally, this paper contends that despite the critical contribution Hajnal and Lee (Reference Hajnal and Lee2011) make by focusing on individuals claiming no partisan preference, their claims that these individuals are actively choosing to opt out of party politics because the parties are not speaking to their issues may be premature. This paper instead argues that because of the lack of direct parental partisan transmission, many of these children of immigrants will eventually opt into party politics, but will do so much later when they are finally exposed to such politics as they age. Finally, much of the research has treated generational status and institutional association as separate independent variables within a single model. This obscures any reliable measure of the effect and direction of the relationship between institutional associations and each generational status. The expectation is that by accounting for those with no partisan connections and treating each generational iteration as a unique population, we can get a better measure of the partisan acquisition of each generational subgroup.
MAKING SENSE OF LOW LEVELS OF PARTICIPATION
This paper finds that temporal considerations and institutional associations are playing a larger role in shaping the partisan outlook of second-generation Americans than was previously understood. These results provide persuasive evidence supporting the claim that the partisan attainment of a large and rapidly expanding segment of the U.S. population is not captured by the wide net cast by the Michigan representation. This is not meant to suggest that the Michigan Model's conception of partisan attainment is no longer applicable. The aim here is a bit more modest: the goal of this paper is instead to emphasize that the Michigan Model could benefit from a critical modification that would make it more restrictive when dealing with this ever-expanding populace. Without this provisional but critical modification to the partisan attainment story, we will continue to be perplexed about what appears to be seemingly apolitical behavior from sizable segments of the U.S. population.
While partisan identity is not a prerequisite for political participation, its heuristic nature helps to minimize the cost of entry into politics by supplying cues that allow individuals to assess their political environment. As The American Voter states, “most elements of national politics are far removed from the world of the common citizen [and this] forces the individual to depend on sources of information from which he may learn indirectly what he cannot know as a matter of direct experience” (Campbell et al. Reference Campbell, Converse, Miller and Stokes1960, 128). We also know that when the cost of information is lowered by the party heuristic, the effort it takes to process information about a party position relative to an individual's core values becomes much less burdensome (Alvarez and Brehm Reference Alvarez and Brehm2002; Feldman Reference Feldman1998). Therefore, once these costs are reduced, individuals are more likely to comport with traditional political activity related to political participation.
The lack of understanding concerning second-generation Americans and their partisan development is not without serious consequences. According to Pew Research (Taylor et al. Reference Taylor, Cohn, Livingston, Funk and Morin2013), there are about 20 million second-generation Americans in Latino and Asian communities alone (with an additional 16 million still under the age of 18). These staggering demographics from Pew fail to report the numbers of the 1.5 Generation. Additionally, it should be noted that the rates of immigration continue to grow, particularly from Asian and Latin American countries. In the fiscal year of 2017, the federal government reported that more than 40% of immigrants are coming from six countries: Mexico, China, India, Cuba, the Dominican Republic, and the Philippines (U.S. Department of Homeland Security 2017). These numbers do not give any indication of how many undocumented individuals continue to come into the United States, but based on data from the Department of Homeland Security it appears that these migrants continue to come primarily from Mexico, El Salvador, Guatemala, and Honduras (Baker and Williams Reference Baker and Williams2014). However, the biggest reason we should be paying closer attention to the importance of second-generation Americans and their political socialization comes from their projected population growth. If immigration flows and birth rates stay on their current pace, second-generation Americans will make up 18% of the total U.S. population by 2050 (Taylor et al. Reference Taylor, Cohn, Livingston, Funk and Morin2013). If these 2050 population projections are correct, the total population of second-generation Americans will be the highest in modern history. These figures should add to the sense of urgency we should have concerning these under-examined populations and their political development in the political socialization literature.
Much of the prior research either ignores the inability of immigrants to transmit partisanship or it assumes that they are able, ready, and willing to provide their offspring with a partisan model. The consequences of this approach are precarious because scholars continue to conflate minority groups directly connected to the immigrant experience with groups who are generations removed from it and are expecting the same level of political activity, including partisan identification, from both populations. There is a strong possibility that the reality may be more of a story of these second-generation Americans experiencing a prolonged partisan socialization process. In other words, it is not that many with contemporary immigrant ties have opted out of party politics or are completely uninterested, it is that many have yet to opt in.
PREVIOUS RESEARCH
Social scientists have long contended that parents or guardians play a prominent role in the political socialization of children. Socrates expressed this belief when he asked his interlocutors in Plato's The Republic, “Don't you know that the beginning is the most important part of every work and that this is especially so with anything young and tender?” (Bloom Reference Bloom1968, 377a). Socrates reasons that if anyone wants to understand the adult before us, we need to understand what was instilled in the child at an early age. Subsequent research in the social sciences has gone on to echo Socrates’ sentiment. The opinion is widely held that values learned early in childhood work to build a strong foundation in the formation of the political attitudes of adults (Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Dawson and Prewitt Reference Dawson and Prewitt1977; Easton and Dennis Reference Easton and Dennis1969; Greenstein Reference Greenstein1960; Reference Greenstein1965; Hess and Torney Reference Hess and Torney1967; Hyman Reference Hyman1959; Jennings and Markus Reference Jennings and Markus1977; Jennings and Niemi Reference Jennings and Niemi1968, Reference Jennings and Niemi1974; Jennings, Stoker and Bowers Reference Jennings, Stoker and Bowers2009; Niemi and Sobieszek Reference Niemi and Sobieszek1977).
Partisanship falls within this purview; it is widely accepted that parents are the driving force behind the initial development of an individual's partisanship (Beck and Jennings Reference Beck and Jennings1991; Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Jennings and Niemi Reference Jennings and Niemi1968; Reference Jennings and Niemi1974; Lewis-Beck et al. Reference Lewis-Beck, Jacoby, Norpoth and Weisbert2008; Niemi and Jennings Reference Niemi and Jennings1991; Niemi and Sobieszek Reference Niemi and Sobieszek1977). The traditional understanding of partisan transmission paints a story of partisan loyalties gained during childhood, often in pre-adolescent children, where they frequently model parental partisan behavior. Children, as early as fourth grade, are known to mimic parental party attachment (Greenstein Reference Greenstein1965). While children in the fourth grade are unable to grasp the significance of what it means to identify as a partisan, the implication is that partisan transmission starts early. Additionally, scholars argued that the similarity between parents and their progeny's partisan outlook emphasizes a possible causal link when it comes to the transmission of party preferences (Jennings and Niemi Reference Jennings and Niemi1974).
However, what if children have not received any direct parental partisan transmission? Second-generation Americans are often born to parents who may simply not be ready or able to transmit a partisan identity. A significant number of Latino and Asian immigrants do not arrive to the United States with an intimate knowledge of the political system or its political parties in general (Alvarez and García Bedolla Reference Alvarez and Bedolla2003; Andersen Reference Andersen1979, Reference Andersen2010; Cain, Kiewiet and Uhlaner Reference Cain, Kiewiet and Uhlaner1991; García Bedolla Reference García Bedolla2005; Levin Reference Levin2013; Portes and Rumbaut Reference Portes and Rumbaut2006; Uhlaner and Garcia Reference Uhlaner and Garcia2005; Wong Reference Wong2000). We also know that the partisan learning curve for immigrants is steep (Andersen Reference Andersen1979; Cain, Kiewiet and Uhlaner Reference Cain, Kiewiet and Uhlaner1991; Converse Reference Converse1969; Wong Reference Wong2000). Even after almost two decades in the United States, many eligible immigrants do not become citizens and still express strong desires to return to their country of origin. The politics of immigrants in the United States is often characterized by ongoing concerns about the politics of their country of origin (Portes and Rumbaut Reference Portes and Rumbaut2006). According to Portes and Rumbaut, “old loyalties die hard because individuals socialized in another language and culture have great difficulty giving them up as a primary source of identity” (Reference Portes and Rumbaut2006, 120).
Beyond identity, there are other hurdles to entry into partisan politics for contemporary immigrants. The language barrier makes it difficult for immigrants to find a party that meets their needs even if they were to become interested in seeking one out upon arrival (Andersen Reference Andersen2010; Cain, Kiewiet and Uhlaner Reference Cain, Kiewiet and Uhlaner1991; Rosenstone and Hansen Reference Rosenstone and Hansen1993). Non-naturalized immigrants sometimes stay away from politics because they feel it is not appropriate for immigrants to have a say in American partisan politics (Levin Reference Levin2013). Moreover, undocumented immigrants are hesitant to participate in politics in general because they fear federal agencies and the real threat of detainment or deportation (Andersen Reference Andersen2010; Levin Reference Levin2013). There are also economic considerations. Many newly arrived immigrants may be largely focused on economic survival and have little time or energy to dedicate to learning a new brand of partisan politics.
Adding to the difficulty of party incorporation is the fact that party outreach to immigrants is sporadic and in some cases even antagonistic at both the local and national level (Andersen Reference Andersen2010; Hajnal and Lee Reference Hajnal and Lee2011; Jones-Correa Reference Jones-Correa1998; Wong Reference Wong2006). Another issue is that parties still appear to behave in the strategic manner that Rosenstone and Hansen (Reference Rosenstone and Hansen1993) emphasized, which suggested that people who were unlikely to vote would not be targeted for outreach by parties. As Andersen (Reference Andersen2010) highlights, parties are still chiefly mobilizing participants whose behavior is predictable and, because of advances in technology in the era of “big data,” the processes of strategic targeting have become more efficient than ever. The consequence of this efficiency is that parties and candidates can survive and in some cases, thrive in low-turnout political environments because parties and candidates can fine-tune appeals to target supporters “while ignoring large blocs of nonregistered, nonvoting, or unpredictable voters” (Andersen Reference Andersen2010, 56). The result of this type of strategic party behavior is to marginalize those who may otherwise be incorporated into partisan politics (Andersen Reference Andersen2010; Wong Reference Wong2006).
The primary takeaway from the literature is that a majority of first-generation contemporary immigrants have been either reluctant or in some cases unable to become politically involved in American partisan politics until much later.Footnote 3 This paper is not in any way implying that immigrant parents are not transmitting social values to their children that work to socialize the children in important ways. As García-Castañon (Reference García-Castañon2013) discusses, the resources and values immigrants acquire before migration play an important role in the way their children work to stake their political identity in the United States. The point this paper is making is that many immigrant parents are unable to map those values onto the appropriate political party in the United States early enough to influence their child's partisanship as the canonical socialization literature has contended. Plainly parental nurturing, in terms of partisan attainment, cannot account for everyone. Therefore, exploring how and when second-generation Americans orient themselves in the political world when they are missing this key partisan heuristic, which is conventionally understood to come from parents, becomes vital to understanding their seemingly apolitical behavior.
LATINOS AND ASIANS AS A TEST CASE
The 2006 Latino National Survey (LNS) (Fraga et al. Reference Fraga, Garcia, Hero, Jones-Correa, Martinez-Ebers and Segura2006) and the 2008 National Asian American Survey (NAAS) (Ramakrishnan et al. Reference Ramakrishnan, Junn, Lee and Wong2008) will both be analyzed to examine the extent to which the socialization literature on partisan acquisition holds up to scrutiny once the scope is broadened to include the generational status of the respondents and the potential spaces that could lead to partisan exposure for second-generation Latinos and Asians as they age. The 2006 LNS and the 2008 NAAS are the largest and most detailed accounts of political attitudes of Latinos and Asians in the United States.Footnote 4 The expectation is that the results will prove to be generalizable once other groups with contemporary immigrant ties are examined.
The descriptive statistics from the LNS buttress the claim that second-generation Latinos are not acquiring their partisan identification as the leading theories contend. The leading theories of partisan socialization imply that we should expect all second-generation Americans to identify with a partisan identity at roughly the same rate. We find that 54% of first-generation Latinos (regardless of citizenship status) in the 2006 LNS are not identifying as Democrats, Republicans, or Independents.Footnote 5 These results are consistent with the scholarship on immigrant partisan attainment. With that said, we should expect to see the rate of partisan identification for the remaining second- and third-generation Latinos in the sample to be analogous. This is not the case. We find only 21% of Latinos with third-generation status not identifying as partisans. At the same time, the data indicate that 34% of second-generation Latinos are not identifying with either party or as Independents. This difference between third- and second-generation Latinos translates to a 47% drop-off in the likelihood of identifying as a partisan. The difference of means between generational status as related to partisan identification is substantive and credibly non-zero at every iteration of generational status (see Tables A1–A3).
The descriptive statistics from the NAAS are not as clear as those of the LNS when it comes to generational partisan identification. This limitation is likely related to the fact that 79% of the NAAS sample is of first-generation status, and third-generation respondents make up only 4%. Even still, moving forward with the limited generational data available, we find that 43% of first-generation Asians (regardless of citizenship status) are not identifying with a political party or as Independents. The data also show that 31% of second-generation Asians are not identifying with either party or as Independents. Finally, for the third-generation Asian sample, only 26% of them are not identifying as Republicans, Democrats, or Independents. While the drop in those choosing to identify with a political party translates to an 18% difference between the second and third generation, we should be careful not to read too much into these findings. As mentioned, the limited number of observations in the NAAS for the third-generation category (196 total N) make it difficult to have too much confidence in the results produced from this generational cohort. While the difference of means between generational status as related to partisan identification is substantive and credibly non-zero when we compare the first-generation and second-generation cohorts, and the first- and third-generation cohorts, the difference in means between the second- and third-generation subgroupings is not significant (see Tables A4–A6). Therefore, because of data limitations, much of the analysis of the NAAS will focus on the first- and second-generation Asian cohorts.
Hajnal and Lee (Reference Hajnal and Lee2011) highlight the importance of accounting for those who report that they “do not know,” “do not care,” “refuse,” or “none of the above,” when asked about their partisan affiliation. Hajnal and Lee labeled those that fall into one of these aforementioned categories as “Nonidentifiers” (Reference Hajnal and Lee2011, 5). As Hajnal and Lee point out, a major reason that much of the analysis of partisanship attainment overlooks the unique socialization processes of those with direct immigrant ties is that these Nonidentifiers, who principally come from immigrant communities, are treated as missing data and subsequently dropped from the analysis. It is important instead to account for Nonidentifiers, because dropping them from the analysis presupposes that these respondents are randomly opting not to identify as partisans and are therefore not different from those respondents who identified as partisans. Considering that Nonidentifiers make up the plurality of the respondents in the LNS (45%) and the NAAS (40%), it seems implausible that this is the case. As Brehm (Reference Brehm1993) and Berinsky (Reference Berinsky2004) emphasize, those who opt out are likely substantively different from those who agree to answer specific surveys or survey questions. Again, while partisanship is not a required prerequisite for political participation, the data from both the LNS and NAAS reveal it is critical, because individuals with no partisan identity are significantly less likely to be politically active. Specifically, the data show that Latino and Asian Nonidentifiers are both substantively different from those who identify with a party and are significantly less likely to either be interested in politics, registered to vote, turn out to vote, contact public officials, or discuss politics with friends or family (see Tables A7 through A15).
To continue to disregard Nonidentifiers is to continue to overlook the largest segment of these populations.Footnote 6 The strategy this paper employs is to treat Nonidentifiers as a category unto themselves. In sum, when examining the data, we find that the transmission of partisanship is not happening for second-generation Americans because the numbers tell us that for a large plurality of first-generation Latinos and Asians, there is no partisanship to transmit. This leads us to ask where and when it is that the children of immigrants are potentially acquiring their partisan outlooks.Footnote 7
THE PROMINENCE OF INSTITUTIONAL ASSOCIATION IN THE PARTISAN FORMATION
The power of institutions and the social interactions within them to politically socialize those who interact with them, either via informational or social mechanisms, has generated a lot of discussion (Berelson, Lazarsfeld and McPhee Reference Berelson, Lazarsfeld and McPhee1954; Cain, Kiewiet and Uhlaner Reference Cain, Kiewiet and Uhlaner1991; Huckfeldt, Plutzer and Sprague Reference Huckfeldt and Sprague1993; Jennings and Markus Reference Jennings and Markus1977; Leal Reference Leal1999; Mutz and Mondak Reference Mutz and Mondak2006; Rosenstone and Hansen Reference Rosenstone and Hansen1993; Schlozman, Verba and Brady Reference Schlozman, Verba and Brady2012; Sinclair Reference Sinclair2012; Verba, Schlozman and Brady Reference Verba, Schlozman and Brady1995; Wolfinger and Rosenstone Reference Wolfinger and Rosenstone1980). Although this paper is agnostic about which of these mechanisms ultimately has the larger impact on socializing its members, the belief is that these mechanisms likely work in tandem for partisan novices. The key point remains that institutions provide the context for such alternative partisan socialization to occur. However, as mentioned above, institutional effects are thought to be subordinate to parental socialization when it comes to partisan attainment.
In contrast, this paper asserts that the role of parental influence in developing the partisan identities of second-generation Americans is suspect and that institutions are likely to fill this partisan void over time. Institutions are known to provide informational resources, psychological engagement, the opportunity to acquire civic skills, and the opportunity to become politically active via recruitment and engagement (Verba, Schlozman and Brady Reference Verba, Schlozman and Brady1995). Additionally, this paper argues that exposure and reinforcement of particular partisan norms in an institutional context can provide those who have received no direct parental partisan transmission the ability to begin to shape a partisan identity by using the partisan cues from institutional members as informational guideposts. These partisan novices are unlikely to resist the partisan cues coming from these familiar institutions because these cues work to fill in the gap left by the lack of direct parental transmission. Additionally, the institutional setting allows for the message to be delivered in such a way that it reduces the effort it takes to process the information about a party the institution appears to support (Shah and Oppenheimer Reference Shah and Oppenheimer2008).
Furthermore, research tells us that people are influenced by the social networks outside of their familial spaces (Abrajano and Alvarez Reference Abrajano and Alvarez2010; Asch Reference Asch1955; Berelson, Lazarsfeld and McPhee Reference Berelson, Lazarsfeld and McPhee1954; Cain, Kiewiet and Uhlaner Reference Cain, Kiewiet and Uhlaner1991; Conover, Searing and Crewe Reference Conover, Searing and Crewe2002; Dawson and Prewitt Reference Dawson and Prewitt1977; DeSipio Reference DeSipio, Gerstle and Mollenkopf2005; García Bedolla Reference García Bedolla2005; Huckfeldt, Plutzer and Sprague Reference Huckfeldt and Sprague1993; McPhee Reference McPhee1963; Rosenstone and Hansen Reference Rosenstone and Hansen1993; Schlozman, Verba and Brady Reference Schlozman, Verba and Brady2012; Sinclair Reference Sinclair2012). In many of these cases, people want to be accepted as part of the group. Therefore, there is a tendency for individuals to adopt the opinion of the majority of the group in order to fit in. One way to do this within institutional settings that happen to signal partisan cues is to take on the partisan identity of those within the institution, even if initially it is only superficially so. Once acceptance occurs, we can expect that partisan attachment, no matter how nascent, to start a self-reinforcing process (Lewis-Beck et al. Reference Lewis-Beck, Jacoby, Norpoth and Weisbert2008). Not everyone will give up their preexisting partisan identification on a whim to be a part of the group, but if an individual has not received any direct partisan transmission from a parent, there is unlikely to be much resistance to adopting the majority opinion of institutional members.
Examples of institutional spaces that this paper will focus on include the university system via having obtained a college degree, unions, the church through attendance, the military, and civic organizations. It is not uncommon that, as people reach adulthood, they experience significant and frequent changes in their social milieu and as they make these changes they can find themselves interacting with institutions that have a partisan element (Green, Palmquist and Schickler, Reference Green, Palmquist and Schickler2002). As Verba, Schlozman and Brady, contend, “ordinary and routine activity on the job, at the church, or in an organization, activity that has nothing to do with politics or public issues, can develop organizational and communication skills that are relevant for politics and thus can facilitate political activity” (Reference Verba, Schlozman and Brady1995, 17–18). Many teens out of high school will go to college, join a new church, a civic organization, the military, or the workforce. All of these institutions can be partisan in nature. The overall contention is that, in the desire to fit in within the institutional context, those who have not received any direct parental partisan transmission will be more open to receiving and accepting partisan prompts when they find themselves interacting with these institutions, before they move on to costlier forms of political participation.
To test the partisan effects that an association with an institution can have on second-generation Latinos and Asians, the link between their association and their probability of identifying as either a Republican, Democrat, Independent (all three choices make the respondent an identifier) or as a Nonidentifier will be examined and then compared with both first- and third-generation cohorts. The expectation is that second-generation Americans’ institutional interactions will reduce the probability that they will be Nonidentifiers at a substantively higher rate than those second-generation Americans who do not have such interactions. Additionally, we expect to find that the impact will also be greater among second-generation Americans who have these institutional associations when compared with first- and third-generation Americans. The logic is that an affiliation with these institutions will have the greatest impact on the second generation because their parents were largely unable to transmit their partisanship preferences when they were children, thus leaving the second generation most susceptible to institutional socialization effects, ceteris paribus. On the other hand, the partisan formation of third-generation Americans is expected to develop largely through parental transmission, as the leading theories have contended, and therefore they should be the least impacted by their institutional links.
METHOD
A logit regression model will be employed to measure the effect that age and institutional associations have on the partisan attainment of each generational subgroup of the Latino and Asian samples. Logit regression is used because the dependent variable of Nonidentifier is dichotomous. The dependent variable used was created to account for the choice in which the respondent derives utility by either choosing to identify as a partisan or not. The dependent variables for both the Latino and Asian samples are created from questions analogous to the language used in the LNS, which asks, “Generally speaking, do you usually consider yourself a Democrat, a Republican, an Independent, some other party, or what”?
The advantage of the LNS data is that it allows for the testing of a large number of possible explanations when it comes to understanding the characteristics of those who identified either as Democrats, Republicans, Independents (Identifiers) or Nonidentifiers. In efforts to try to increase the explanatory power of the model, a total of 17 explanatory variables are introduced from the LNS. These variables include six main independent variables of interest, with five related to institutional associations. Institutional variables include whether the respondent has any association with the following institutions: the university system via having obtained a college degree, civic organizations, the military, the church through attendance, and unions. The sixth independent variable of interest is age. Again, the role of age is important because this paper is arguing that temporal considerations play a critical role in the accounting of the partisan identification of second-generation Americans. It should be noted that this paper focuses on time spent in the United States instead of age for the first-generation cohort in the Latino and Asian samples to provide a more accurate picture of the effect of time.Footnote 8
The remaining independent variables consist of covariates associated with political behavior believed to influence partisanship. These variables include interest in politics, employment status, income, gender, whether at least one parent has a college degree, whether the respondent has children and geographic location.Footnote 9 Additional variables were added to include whether the respondent receives government assistance, whether the respondent believes Latinos can get ahead in the United States, and finally whether the Latino respondent identifies as Mexican, Puerto Rican, or Cuban (the three largest ethnic groups in the sample). These additional controls are meant to account for some of the alternative explanations that have been presented as to why those with ties to contemporary immigrant communities do not join a political party. By looking to whether a respondent receives government assistance, we attempt to see whether those who received assistance know which party is helping provide these benefits and thus are therefore more likely to support that party. The variable concerning whether Latinos believe they can get ahead is included to account for the alternative explanation that positive affect group attachment affects participation in the political process before Latinos decide to engage (García Bedolla Reference García Bedolla2005). Finally, we add ethnic identifiers to account for the fact that Latinos cannot be easily explained as a homogenous group (Abrajano and Alvarez Reference Abrajano and Alvarez2010).Footnote 10
When turning to the NAAS survey to test the effects of age and institutional associations, there is an attempt to mirror the model created using the LNS when it comes to understanding the characteristics of those who identified either as Democrats, Republicans, Independents or Nonidentifiers. However, because the LNS and NAAS surveys do not always ask the same type of questions, or even in the same way, there were limitations on what could be included. Therefore, when concentrating exclusively on the NAAS data, the main independent variables of interest were reduced to five, with only four related to institutional associations. Institutional variables for the NAAS model included whether the respondent has any association with the following institutions: the university system via having obtained a college degree, civic organizations, the church through attendance, and unions. The fifth primary independent variable of interest is age or time spent in the United States. The remaining independent variables are meant to mirror some of the alternative explanations included in the LNS model; these variables include interest in politics, employment status, income, gender, trust in government, and ethnic identity.Footnote 11 The ethnic identities used in the model include Indian, Chinese, and Vietnamese respondents, the three largest groups in the NAAS sample.Footnote 12
RESULTS
The results presented in Table 1 and Figure 1 focus on the LNS model. As Table 1 denotes, many of the predictor variables in the logit model register some level of statistical significance within at least one of the Latino generational subgroups. To get a better idea of the size of the effect each independent variable has on the probability of respondents choosing to be Nonidentifiers, this paper will employ the use of predicted probabilities. Of course, in a logit model, probabilities are conditional on the values of the other variables. Here, we can demonstrate the effect of each specific dichotomous variable in predicting the probability of a respondent being a Nonidentifier by varying whether a respondent is affiliated or non-affiliated with the variable of interest, such as institutional affiliation, while holding all the other variables in the model at their means. With respect to the effect of non-dichotomous variables in predicting the probability of a respondent being a Nonidentifier, this paper varies the specific variable to one standard deviation above the mean (which will be referred to as the “maximum”) while holding all the other variables in the model at their means.
Figure 1. The predicted effect of time on Latino respondent's probability of being a nonidentifier.
Table 1. Probability of being a nonidentifier by generational status and marginal effects for the 2006 Latino National Survey
Note: *p ≤ .05; **p ≤ .01; ***p ≤ .001; the blank cells under marginal effects indicate the coefficients are non-significant. Coefficients in bold maintained their statistical significance after running robust standard errors testing.
Turning to predicted probabilities, we see that the probability of a first-generation Latino being a Nonidentifier is 54%, all things being equal. The results also show that first-generation Latinos are only significantly affected by one of the five institutional association measures when it comes to the chances that they will be Nonidentifiers. Consequently, since first-generation Latinos who have a connection with the military are predicted to have a 50% probability of being an identifier, they are 4% less likely to be a Nonidentifier. However, there are other factors in the model that are just as likely to impact Nonidentifier status in either direction.
The model, unsurprisingly, reveals that interest in politics has considerable influence. Those who have a maximum interest in politics are 9% less likely to be Nonidentifiers compared with the average person in the sample. Additionally, as mentioned above, past research has revealed that time in the United States has a significant influence on whether first-generation Latinos will identify as partisans. The model here validates that claim, as those who have spent the maximum number of years in the United States (31 years) are 11% less likely to be Nonidentifiers compared with the average first-generation respondent, who has only spent 18 years in the country.
Additionally, the results reveal that employment, increases in income, parental education, and Puerto Rican ethnicity also significantly reduce the chances of being a Nonidentifier for first-generation Latinos. While employment and an increase in income only account for a 1–2% reduction in the probability of being a Nonidentifier, parental education, and Puerto Rican ethnicity are more impactful. Those first-generation Latino respondents who reported having at least one parent who was college educated were 7% less likely to be Nonidentifiers when compared with the average first-generation Latino respondent. The fact that first-generation Puerto Ricans are 9% less likely to be Nonidentifiers compared with other Latino immigrants is no doubt related to the fact that Puerto Ricans are U.S. citizens and are by default connected to the American political process in some capacity.
The results for the first-generation subgroup are relatively consistent with the past literature. Nonetheless, the conventional approach suggests that the uniqueness of these results regarding the role of time should disappear for the children of immigrants. As Table 1 and Figure 2 make explicit, temporal considerations continue to play a significant role among second-generation Latinos. Not only is age, as it relates to getting older, a vital component when it comes to second-generation Latinos and their chances of being Nonidentifiers, but its impact is just as pronounced as it was for those in the first-generation subgroup. Age accounts for a 9% difference in whether second-generation Latinos are Nonidentifiers compared with the average second-generation Latino respondent, who is 37 years old. The type of substantive impact age is having on second-generation Latinos has not been accounted for by the previous literature. The traditional approach would have us suppose that these second-generation Latinos have come to a partisan choice early in their childhood.
Figure 2. The predicted effect of time on Asian respondent's probability of being a nonidentifier.
Table 1 also demonstrates that for second-generation Latinos, three out of the five institutional variables examined have a substantive effect on their Nonidentifier status. Second-generation Latinos with a college degree are 4 percentage points less likely to be Nonidentifiers compared with the average second-generation respondent. Second-generation Latinos are also 5 percentage points less likely to belong to the Nonidentifier category when they are associated with at least one civic organization. Union association is even more impactful: those who are linked to a union are 10 percentage points less likely to be Nonidentifiers compared with the typical respondent. The model highlights that institutional connections are meaningfully impacting the probability of individuals moving out of Nonidentifier status in ways not previously accounted for. While a 4–10 percentage point decrease may not appear to be much on the surface, this type of movement from a high Nonidentifier population is far-reaching. It seems age and institutional associations are helping one move from having no partisan identification at all into claiming one. This process has largely been ignored by the orthodox understanding of partisan acquisition. This type of shift away from the Nonidentifier category suggests that we should also start to consider seriously the kind of impact institutions can have on the partisan attainment of second-generation Americans much later in their adult lives.
For the third-generation subgroup, we find that only one institutional association is significantly affecting the probability of a person moving out of Nonidentifier status by 1%. Age, it should be noted, still proves to be impactful for the third-generation subgroup. As one gets older, we see a 4% drop in the prospect of someone being a Nonidentifier when compared with the average third-generation respondent. This impact suggests that the prolonged partisan socialization process may extend into the third generation, even if the effect has attenuated. In sum, it seems that those Latinos with third-generation status by and large comport with the traditional partisan attainment story.Footnote 13
Turning to Table 2 and Figure 2 to examine the NAAS sample, we continue to see support for the claims this paper makes regarding the impact of time on second-generation Americans. The results reveal that the average first-generation Asian respondent who has spent 21 years in the United States has a 42% chance of being a Nonidentifier, while those in the sample who have spent the maximum time in the United States (33 years) only have a 37% chance of being a Nonidentifier, a 5% decrease. When examining the second-generation cohort in the NAAS sample, we again see the significant role age plays in reducing the probability of being a Nonidentifier. In comparison, a second-generation Asian who is 60 years old would be predicted to have a 25% probability of being a Nonidentifier, or a 4% difference from the mean. Age plays no significant role for the third-generation cohort.
Table 2. Probability of being a nonidentifier by generational status and marginal effects for the 2008 National Asian American Survey
Note: *p ≤ .05; **p ≤ .01; ***p ≤ .001; the blank cells under marginal effects indicate the coefficients are non-significant. Coefficients in bold maintained their statistical significance after running robust standard errors testing.
While the results concerning the Asian sample are a bit muted compared with those found among Latinos when examining institutional considerations, the results indicate that institutions can still play a major role in the partisan attainment of second-generation Asians. While institutional associations play no role for first-generation Asians, church attendance and civic associations substantively affect the probability of second-generation Asians becoming Nonidentifiers. Those second-generation Asian respondents associated with at least one civic organization are 8% less likely to be Nonidentifiers compared with the typical second-generation respondent. However, not all institutional associations worked to lower the probability of becoming a Nonidentifier. Increased church attendance can increase the chances of second-generation Asians becoming Nonidentifiers. Those second-generation respondents in the NAAS sample who attend church every week are 6% more likely to be Nonidentifiers compared with the average respondent who only attend church a few times a month.
Income and interest in politics are also playing a substantive role for first- and second-generation Asians. Whereas a first-generation NAAS respondent with only average interest in politics has a 42% chance of being a Nonidentifier, the same respondent who has the maximum amount of interest in politics has a 34% probability of being a Nonidentifier, a difference of 8%. In contrast, second-generation respondents of the NAAS who have the maximum amount of interest in politics would be predicted to have a 25% probability of being a Nonidentifier, or a 4% difference from the mean. The effect of moving up to the maximum in the income scale translates to first-generation Asian respondents becoming 3% less likely to be Nonidentifiers compared with those who make the average of around $50,000. As to the effect of income on second-generation Asians, we see that maximum earners have a 23% chance of becoming Nonidentifiers compared with those who make the average of around $75,000, a difference of 6%.
The results presented here emphasize the importance of giving more thought to the socialization process of contemporary immigrant groups. However, they also work to underscore the idea that not all contemporary immigrant groups are impacted in the same way, even within their co-ethnic grouping. As the results indicate, second-generation Asians and Latinos are affected by temporal and institutional considerations, but not at the same level, not even within the same ethnic grouping, as is evidenced by the Vietnamese, Mexican, and Puerto Rican samples. Vietnamese respondents who have a historical connection to the United States that revolves around the United States’ response to Communism are on a different trajectory toward becoming partisans than those Asians who identify as Indian or Chinese. As the data show, identifying as a Vietnamese immigrant reduces a NAAS respondent's chances of being a Nonidentifier by 9%. The same scenario holds true for Latinos because Mexicans, Cubans, Puerto Ricans, and South Americans are not a monolithic political block.
We also see that Latinos are more likely to be impacted by temporal and institutional associations compared with Asians. This is likely due to group differences within the second generation. We know that when compared with Asians, Latinos in the second generation are younger, have lower levels of income, and initially have lower levels of access to institutional connections (Taylor et al. Reference Taylor, Cohn, Livingston, Funk and Morin2013). These differences do not, however, discount the importance of re-evaluating how we think about the contemporary immigrant experience and the need to push further to understand how these distinct ethnic groups, which this paper lumps together for the sake of parsimony, should be explored further. The learning curve for the children of immigrants, in many ways just as steep as the learning curve of their parents, still works to explain the perceived lack of political activity among these groups.
CONCLUSION AND IMPLICATIONS
The conventional treatment of partisan acquisition has largely elevated the presence of parental influence to account for the partisanship attainment of most Americans. By widening the scope of investigation to focus on the children of immigrants and Nonidentifiers, this paper finds substantial evidence for the inclusion of additional considerations such as temporal dynamics and institutional associations. In other words, it appears that institutional associations can significantly influence the partisan socialization process and that it takes time for the partisan socialization process to unfold for these Americans.
Taken together, the findings in this paper suggest that we do not have as clear a grasp on the political socialization story as was once thought. Unlike the traditional story that has been presented in the past, we must acknowledge and take care to make sure that those with direct contemporary immigrant ties, especially second-generation Americans, are not overlooked in the partisan socialization literature. Allowing for this temporal, yet critical, modification to our understanding of partisan socialization permits us to make sense out of the seemingly apolitical trends we see in many immigrant communities. If we continue to paint those with contemporary immigrant ties as apolitical without these additional considerations, we will continue to fail to realize that many Americans may eventually identify politically in terms of partisanship, which opens the door for other types of political participation. If this is the case, it makes it harder to understand certain political trends in behaviors, including why Latinos and Asians appear so disengaged from the political process. It also suggests that the idea of awakening the “sleeping giant” is still possibly decades away from becoming a reality, even in the age of Trump.
It is likely that in the foreseeable future, turnout among Latinos and Asians will not change in any significant way even as Donald Trump, who has been openly antagonistic toward immigrants, strengthens his anti-immigration rhetoric as president. This paper has presented evidence to suggest that the incongruity between reality and prediction that has persisted for over two decades is in large part due to the prolonged socialization process. With so many second-generation Americans within these contemporary immigrant communities, it will be some time before the prolonged partisan socialization of these groups makes its impact felt, allowing these groups to reach “partisan maturity.” We must be cautious of the litany of stories we will continue to encounter about the impact of the Latino and Asian vote. These expectations, which are set in place by our lack of understanding of the prolonged partisan socialization process, will only lead to more head scratching about the inability of these groups to turn out or be politically active. However, we should also take great care not to dismiss all of those who did not show up to the polls on November 8, 2016 as forever destined to be disengaged and therefore unworthy of examination.
ACKNOWLEDGMENTS
I want to thank the University of Chicago's Department of Political Science, in particular, I am grateful to John Brehm, Michael Dawson, John Mark Hansen, and Cathy J. Cohen for the helpful feedback I received from them. Additionally, I want to also thank David Leal, Ricardo Ramírez, and the anonymous reviewers for their insightful comments.
SUPPLEMENTARY MATERIAL
The supplementary material for this article can be found at https://doi.org/10.1017/rep.2018.21
