Abstract
The current US undocumented population is large and settled. As a result, millions of US-born citizens are growing up with undocumented parents or siblings. In this paper, we use original survey data to study the politics of the US-citizen offspring of undocumented migrants. We test theories of parental political socialization, which imply that having undocumented parents may have chilling effects on political engagement. We also test theories of social activism, which predict that the offspring of the undocumented may be motivated to make use of their rights as US citizens by protesting on behalf of their parents. We find no evidence of lower political engagement among those with undocumented parents. Instead, we find that the offspring of the undocumented are more likely to protest on immigration issues, and more optimistic that popular protest can induce political change. We use an instrumental variables design to test whether these differences warrant a causal interpretation, and find tentative evidence that having undocumented parents does indeed have mobilizing political effects.
The United States is home to around eleven million undocumented immigrants. Even though they are legal outsiders, most have close ties to the country. Many have US-citizen relatives, including partners and children. A great deal of research has addressed the causes and consequences of anti-immigrant attitudes in this era of large-scale “illegal” migration (e.g., Hainmueller and Hopkins 2014; Wright, Levy, and Citrin 2016). And yet, we still know very little about the politics of undocumented immigrants themselves, or the political implications of this regime of mass illegality for the millions of American citizens with undocumented family members (McCann and Jones-Correa 2016).
Until scholars address these topics, we will not understand the contemporary immigrant experience. While most of the country’s immigrants are set on a path to permanent residence and citizenship, about one in four are marked as “illegals.” Because so many undocumented migrants are Latinos, without research on the politics of mass illegality, we will struggle to understand Latino politics, or indeed the factors that—under some circumstances—make “Latino” a salient political identity (Lee 2008).
Research in this area also promises to provide new ways to test and build upon theories of political learning and activism. How do young US citizens whose parents have few (political) rights think about and engage with the state? In this paper, we provide some of the first evidence on the political effects of having undocumented parents. As undocumented migrants are wary of attention, and they and their children only make up a small share of all US residents, it is hard to gather representative data on this population. We use an original survey of young US-born Latinos, some of whose parents are undocumented, to start to fill the gap in our knowledge of these issues.
To frame the political effects of mass undocumented migration on the US-born second generation, we draw together theories and findings from several literatures. One prediction, following from research on parental political socialization and on immigrant incorporation, is that having undocumented parents has chilling effects on civic and political engagement. However, other research on social movements and on how people respond to political threats implies that those with undocumented parents might be pushed toward activism on immigration issues. In fact, our data provide scant evidence of chilling effects, but more support for the idea that family exposure to the risk of deportation serves to mobilize the US-born offspring of the undocumented to demand immigrants’ rights. These results imply that far from being inert, many undocumented migrants are politically engaged. The findings also reaffirm the insight that political behavior is shaped by social and, especially, by family ties.
Immigrants in the Shadows
Until 1965, Congress placed no formal limits on migration to the United States from the Western hemisphere (Ngai 2004). The new quotas, when they arrived, were unrealistic given the history of the southern border regions (Massey, Durand, and Malone 2003). Seasonal migrant workers had long supplied much of the labor in the Southwest, and, at times, manual workers or “braceros” were actively recruited by the US government (Zolberg 2006, 310). Because the border was long, remote, and hard to patrol, and checks on work permits were lax, workers continued to come and go without papers even after the 1965 reforms. The Immigration Reform and Control Act (IRCA) of 1986 promised to change this. Amnesties were passed for undocumented migrants who had been in the country since before 1982, and for seasonal laborers who worked in agriculture in 1986. A total of 1.8 million people regularized their status via the former provision, and a further 1.3 million via the latter. IRCA also promised to stop unauthorized migration by sharply increasing border enforcement and by imposing new sanctions on employers who hired workers without papers.
In fact, however, border enforcement remained difficult, and the sanctions regime was lax due to opposition from employers and executive branch skepticism of regulation (Zolberg 2006, 373). As a result, undocumented migration continued apace. This became more politically contentious over the course of the 1990s. The federal government responded with “symbolic” shows of force at the US–Mexico border (Andreas 2009). The fees charged by border smugglers rose but, at first, the number of migrants crossing was little affected. Gradually, though, as the danger and cost of unauthorized migration rose, as demographic changes reduced the number of young Mexicans in need of work, and especially after the US economy stalled in 2008, the flow of new immigrants slowed (Massey 2013). Yet, the stock of undocumented migrants has remained stable. Harsh border enforcement has made return migration too risky for most, and the median length of residence has risen to fourteen years (Passel and Cohn 2016, 4). Undocumented migrants have become ever more integrated into US society (Donato and Armenta 2011).
Strikingly, since 1980, around seven million people have been born in the United States with at least one undocumented parent. The number peaked at about 370,000 per year in the mid-2000s, but is now trending down in a pattern that lags the fall in undocumented migration (Passel and Cohn 2015). These young people are all US citizens, thanks to the Fourteenth Amendment—which was originally intended to turn slaves into citizens but is now arguably the most important single factor for immigrant integration in America. Even if undocumented migrants leave the country, or never become citizens, their US-born children will remain and are coming of age to influence US politics and society.
How does having undocumented parents affect American citizens? Most US-citizen children of the undocumented were born since the 1980s, as undocumented migration increased around that time (Passel and Cohn 2015). Given the young age profile of the group, scholars in education, psychology, and public health have been among the first to study the offspring of the undocumented (Suárez-Orozco et al. 2011; Suro, Suárez-Orozco, and Canizales 2015). For example, Dreby (2012) describes emotional problems among US-born children who fear their parents may be deported. For their part, undocumented parents are more likely to have concerns about the developmental progress of their children (Ortega et al. 2009). Yoshikawa (2012) finds that the US-born children of undocumented parents are less likely to enroll in the educational and welfare programs for which they are eligible, because their parents fear contact with state agencies (but see Terriquez 2012). Correspondingly, Bean et al. (2011) find that undocumented parents pass on a legacy of lower educational attainment to their children, compared with the children of legal immigrants. In a recent overview, leading demographer Douglas Massey (2013, 13) goes so far as to argue that “lack of legal status constitutes an insurmountable barrier to social and economic mobility, not only for the undocumented immigrants themselves, but also for their citizen family members.” Political scientists have been slower to study the implications of mass illegality, whether for the immigrants themselves or their US-citizen offspring.
Politics Overshadowed?
We can start to fill the gaps in our knowledge by gathering new data and by applying core theories in political science to this new situation. In particular, there is a rich literature on the process of “political socialization” whereby young people learn about politics and pick up habits of political (non)participation. We can also draw on relevant research on contextual “cohort effects” and on collective action among marginalized groups.
Undocumented Parents as Role Models?
Much of the existing research on political socialization is based on social learning theory (Bandura 1969, 1971), which posits that people learn not only from their own experiences but also by modeling the behavior of others, especially their parents (e.g., Jennings and Niemi 1981). In a recent review, Jennings, Stoker, and Bowers (2009) reaffirm that parents serve as role models to shape the partisanship of their children, and, to a somewhat lesser extent, also affect their offspring’s proclivity to political engagement.
It is an open question whether undocumented parents serve as political role models and thereby help to socialize their US-born children into American politics. Some of the political rights secured in the US Constitution—such as freedom of expression or assembly—extend to all residents, not just citizens. Undocumented migrants can engage in many of the civic activities, such as joining clubs or going to public meetings, that tend to promote more explicitly political behavior (see Schlozman, Verba, and Brady 2012). But, of course, undocumented migrants cannot vote. And parental socialization usually means not only showing children how to exercise their political rights, but also explaining the ideas and groups that structure political life. Research suggests that migrants may struggle at this, because they themselves often take years to learn about the politics of the new homeland (Fraga et al. 2012; Hajnal and Lee 2011; but see also Sears, Danbold, and Zavala 2016).
Terriquez and Kwon (2015, 425) find that “barriers to immigrant parents’ political engagement suppress the civic and political participation” of the US-born second generation. Legal status may be just such a barrier. Along these lines, S. K. Brown and Bean (2016) find that Latino immigrants, including legal residents who do not yet hold voting rights, pick up more political knowledge the longer they live in the country. But the authors find that this does not apply to the undocumented. Based on this literature, one might expect that undocumented migrants are not only unable to serve as role models when it comes to voting but are also less likely to engage in the kinds of civic and political behavior that are open even to noncitizens. For these reasons, undocumented parents may be less likely to socialize their children into habits of political participation. 1
Undocumented parents might even serve as models of political distancing. Beyond the US border zone, undocumented migrants are most likely to be deported if they come into contact with the police after committing a crime, misdemeanor, or traffic offense (Meissner et al. 2013, 105). In some parts of the country, police officers have great discretion to ask for evidence of immigration status (Farris and Holman 2017). In this context, undocumented migrants even fear that they will be deported for reporting a crime (e.g., Menjívar and Bejarano 2004, 137). The result is that they avoid public places and state agencies, and fail to claim even the rights to which they are entitled (Gleeson 2010; Yoshikawa 2012). One might expect the children of undocumented parents to learn by example that it is best to avoid state agencies and the police. This would fit with earlier findings that, when it comes to views of the state, parental socialization carries a less trusting message for children in racialized minority groups (Abramson 1972; Garcia 1973; Michelson 2003).
Taken together, then, past research on political socialization and immigrant political incorporation suggests chilling effects on US citizens with undocumented parents. In this paper, we test two ways in which these effects could work. Our first two hypotheses refer to the idea that undocumented parents are less likely to model political engagement for their children, while the third refers to the idea that the children of the undocumented may (implicitly) learn from their parents that the state should be avoided.
A Generation of Immigration Activists?
In contrast, other theories imply that having undocumented parents may make second-generation Latinos more politically active, at least on the issue of immigrants’ rights. Unauthorized immigration has provoked a fierce political response, and this in turn has politicized many immigrants. An example is Proposition 187 in California, a ballot initiative from 1994 that would have denied people without papers (even minors) services such as schooling and health care. The proposition passed by a large margin but was blocked by the courts, as it trespassed on federal power over immigration laws. The anti-immigrant backlash, in turn, made legal immigrants in the state more likely to naturalize and pushed Latinos to see themselves as having common political interests (Pantoja, Ramirez, and Segura 2001; Ramakrishnan and Espenshade 2001). Proposals to criminalize the undocumented and their family members also prompted large pro-immigrant marches in 2006, which brought millions of undocumented migrants and their US-born relatives onto the streets (Voss and Bloemraad 2011; Zepeda-Millán 2014). Since this wave of protest, young activists have continued to push for immigration reform using innovative tactics such as occupying congressional offices and getting arrested in graduation garb, or self-deporting and then applying for asylum (Carrasco and Seif 2014).
Research on contextual socialization implies that the activism around this issue may have made young US-born Latinos more likely to protest and to see protest as a viable political tool. Theories of “cohort effects” build on Mannheim’s (1952) observation that generations are marked by the politics they experience in the “formative years” of adolescence and early adulthood (Bartels and Jackman 2014; Carlsson and Karlsson 1970). Cohort effects are clearest in times of upheaval such as the Great Depression, but can also reflect narrower events like the Watergate scandal (Dinas 2013). There is also evidence that cohort effects differ for ethnic and racial minorities (Abrajano and Lundgren 2014), especially when racial politics is at stake, as in the civil rights era (Schuman and Scott 1989).
Furthermore, research on social movements and on political threats implies that this context of activism may have had even stronger effects on young Latinos with undocumented parents. Following Durkheim ([1897] 1997), social movements were once thought to result from anomie and the breakdown of social order. This view was shaken by scholars who pointed out that activists are often well-connected rather than isolated (Tarrow 1994). But recent research provides a more specific account of how movements arise from the disruption of norms and routines. Snow et al. (1998, 9) argue that “actual or threatened intrusions into culturally defined zones of privacy” such as the family can inspire people to take mass action. People are strongly motivated by threats or losses (Marcus 2000; Tversky and Kahneman 1991). Intense fear of threats to “culturally inviolable zones” such as the family can help people overcome collective action problems (Snow et al. 1998, 17). Because concerns for family members are known to be shared by many Latinos, this should increase confidence that protest will arise and may succeed. In line with these ideas, messages such as “Stop Separating Families” are ubiquitous at immigration protests (Pallares 2014).
The number of true activists may be small, compared with the cohort of Latinos that has grown up in this era. But an activist political climate can have wider effects thanks to media coverage (Merolla et al. 2013) or social exposure. As Cohen-Marks, Nuño, and Sanchez (2009, 713) wrote in the wake of the 2006 protests, the strongest effects of immigrant activism may yet be found among “the countless children who will grow up hearing tales of the day millions emerged from the shadows to declare their determination to pursue the American dream.” Even people who did not join marches on this issue may come to think that, given their weak legal status, public protest is an effective political tool for undocumented migrants and their allies. Again, these effects may be stronger among people more directly exposed to the issue because of their undocumented parents.
Based on these ideas, we test two further hypotheses. The first is that the combination of a context of activism plus exposure via family members makes US-born Latinos with undocumented parents more likely to become activists. The second is that even if this combination does not turn people into activists, it makes them more likely to believe that protest is an effective way to pursue policy change.
To summarize, theories of parental socialization and immigrant incorporation predict chilling political effects for US-born children who have undocumented parents. In contrast, theories of cohort effects and social movements imply that young Latinos with undocumented parents who have grown up in an era of immigrant activism are especially likely to mobilize and protest on immigration issues. Readers should note that the two sets of hypotheses are not incompatible. We may see evidence only of chilling effects, or only of mobilization. But it is also possible that the offspring of the undocumented are generally less politically engaged, yet active on immigration issues.
Data and Measures
One reason there has been little research on the political implications of having so many undocumented migrants living in the United States for so long is that representative data on this group are scarce, as are data on their US-citizen family members. Undocumented migrants themselves may be reluctant to participate in surveys (but see Bean et al. 2011; Terriquez 2012), although this is less of a concern for their US-citizen offspring, who have much less to fear. Arguably, the bigger problem for gathering representative data is that the offspring of undocumented migrants make up a small share of the total US population, so that only a few are included in standard surveys.
To recruit a sample large enough for statistical analysis, we commissioned a new survey in the summer of 2013. Because the undocumented population is about 80 percent Latino, we focused on this group (Passel and Cohn 2014). Unequal enforcement means the risk of deportation is even more concentrated on Latinos: in recent years, around 95 percent of those deported were from Latin America (Immigrations and Customs Enforcement 2013). In this sense, undocumented migration is a racialized phenomenon (Chavez 2013; Zepeda-Millán 2014). To make comparisons, we aimed to survey the US-born children of both authorized and unauthorized migrants. As noted above, most of the people who have grown up in mixed-status households were born since the 1980s. In an effort to include some people whose parents were unable to regularize through IRCA, we opted to recruit Latinos born in the United States between 1982 and 1995. That is, they were born after the deadline for the first of the IRCA regularization programs. This does not guarantee that their parents arrived in the United States only after that deadline, but it helps to ensure that some did so. The survey participants also had to be adults at the time of the survey. In early 2013, we used Current Population Survey (CPS) data to estimate the population of interest at 3.1 million Americans born in this time period to immigrant parents, with at least one parent from Latin America. That is about 1 percent of the US population.
We commissioned the survey through the research firm GfK, formerly Knowledge Network. This allowed us to combine two sources to reach the desired sample size. First, we drew from the GfK Latino panel, which is recruited as a probability sample based on mail addresses and random digit dialing, and is then periodically surveyed online. In return for participation, GfK provides Internet access where needed, which helps to ensure a representative sample. Although sizable, the panel contained only a few hundred people who met our criteria. GfK, therefore, arranged to recruit extra participants online via English and Spanish-language websites. 2 Combining the probability and the opt-in samples is much more costly than an Internet-only sample, but has several advantages. Including the probability sample increases our confidence about representativeness, allows for comparisons across sample types, and also allows for more credible weighting (DiSogra et al. 2011). The survey was fielded online in July and August 2013. In the final dataset, one-third of the subjects are from the GfK probability sample, and the rest are opt-in recruits, for a total of 1,050 people.
Table 1 describes our sample. The unweighted sample is more educated and more female than the CPS data imply. Higher response rates from more educated people are common in survey research and suggest that we may be overestimating levels of political engagement, which tend to be higher among the more educated. However, it is encouraging that the probability and opt-in samples are very similar, allaying some concerns about the representativeness of data from people recruited online. Given the paucity of data for research on these issues, our survey is a big step forward. Of course, further research will be needed to corroborate our findings. In this paper, we present our results with design and poststratification weights, based on the sampling procedures and CPS demographic data. Missing data were replaced by multiple imputation; we find stable results over five sets of imputations.
Comparing Demographics from CPS Data with Our Probability and Opt-In Samples of Young, Second-Generation Latinos.
Source. The first source is the March 2013 CPS, while the second source is the Latino Second Generation Study–ICPSR Number 36625.
CPS = Current Population Survey; ICPSR = Inter-university Consortium for Political and Social Research.
Survey Items
We asked our second-generation Latino survey participants a series of questions about the immigration history, civic, and political behavior of their parents. Of course, it would be preferable to ask the parents themselves, to avoid the risk of children projecting from their own experiences onto their parents or simply misremembering their parents’ habits. But surveying both undocumented residents and their children presents great logistical difficulties. In an effort to limit the risk of respondents projecting their own values, we mostly asked about specific actions rather than parental attitudes. We found that 45 percent of our respondents have at least one parent who lived for a period as an undocumented migrant in the United States, and that one-third of those parents were still undocumented at the time of the survey (we did not ask about undocumented status directly, but inferred it from the questions on migration history; see online appendix for full details). 3 As in prior research (e.g., Bean et al. 2011), we find some signs of socioeconomic disadvantage among the US-born offspring of the undocumented, compared with the offspring of other immigrants. Those with at least one parent who is still undocumented are more likely to have a high school education or lower, and are less likely to hold a college degree (see Figure S1 in the online appendix).
We used a standard set of questions on civic behavior to test for evidence of a “chilling” effect of undocumented status on parents or their US-born offspring (H1 and H2). Specifically, we asked about the following: attending Parent Teacher Association (PTA) or school group meetings; attending community meetings; giving blood; donating money to charity; volunteering for a charity or church; working with others in the community to solve a problem; organizing an event in the community; participating in an ethnic organization, or in an organization linked to the country of origin; participating in a sports league, or in a labor union, or a professional association, or in an organization that supports candidates in elections. In answering these questions about their parents, we told the second-generation Latinos in our survey to think back to the time when they were 16 years old, so as to focus on the “formative years” of adolescence and to avoid casting too far back into the past, when memories are less reliable. We also asked survey participants to answer the same questions about themselves at the time of the survey. In addition, we used three measures of offspring political engagement. The first reports the strength of political interest, the second the frequency of political discussions in a range of settings, and the third reports consumption of political news from various media (see online appendix for details of question wording and coding).
To test whether the offspring of undocumented migrants avoid state actors (H3), we used questions about trusting, or avoiding contact with, the government and the police. To test for mobilizing effects of having undocumented parents in an era of Latino activism (H4), we asked whether survey respondents had participated in rallies or marches about immigration (among other issues). Finally, to test for broader effects on the perceived efficacy of protest (H5), we asked survey participants, “What kind of activities do you think are effective in pushing for political change?” The list of options is voting, participating in legal protests, participating in illegal protests or damaging property, taking issues to the courts, and lobbying politicians such as the President, Members of Congress, or state representatives.
When comparing survey participants by parental legal status, we use statistical models to control for other factors that may be related both to legal status and political behavior. This is important because undocumented migrants tend to be poorer, less educated, and concentrated in certain regions, and these differences may also curtail political engagement. It would be wrong to construe all differences due to demographic factors as reflecting the effects of parental legal status. Specifically, we control for the education of the survey participant and the parents, the survey participant’s age and gender, whether the parent(s) come from Mexico, whether the survey participant is fluent in Spanish, whether the parents held manual versus office jobs when the survey respondent was young, and region of residence in the United States (see online appendix for more on these items).
Results
We begin with the findings on parental civic behavior. Counter to our prediction (H1, chilled parents), we find similar levels of civic and political activity among documented and undocumented parents. This is illustrated in Figure 1, where the points show the share of parents said to have engaged in each activity, by legal status at the time of our survey. Horizontal lines through each point show 95 percent confidence intervals, based on robust standard errors. Overall, Figure 1 reveals few differences by legal status. Furthermore, summing civic activity across all domains, we find no significant difference by parental legal status in either a bivariate model (p = .49), or a model with demographic controls (p = .64). 4 We also asked survey participants about the frequency of political discussions at home when they were aged 16, and find no significant difference by parental legal status (p = .94).
Parental civic behavior, by legal status.
Admittedly, these results are indirect, because they are based on the answers given about their parents by the US-born offspring. The results about chilling effects on the US-citizen offspring themselves should be clearer (H2). In fact, however, we see no evidence to support the prediction of a chilling effect on the civic and political behavior of the offspring of undocumented migrants. These results are shown in Figure 2. If anything, the children of the undocumented appear slightly more active in the civic and political domain, and statistical models of the total amount of activity across domains provide some support for this view (p = .08 in bivariate comparison, and p = .1 with controls). Although this runs counter to our original hypothesis (H2, chilled children), it is consistent with theories of parental socialization given the evidence from Figure 1 that the parents themselves did not show any chilling effects. Some of the percentages in Figures 1 and 2 are unrealistically high, a common problem with self-reported measures of socially desirable behavior. This problem may be even worse when the reports are indirect, as in our case when we asked the offspring to report on their parents. Note, however, that the key concern for our comparisons is differential rates of overreporting among respondents with versus without undocumented parents. We think this is unlikely.
Offspring civic behavior, by parental legal status.
In line with these findings, it does not appear that the US-born children of undocumented migrants are less engaged with US politics. Table 2 shows results from models with controls for background characteristics. The first three outcomes in Table 2 are interest in politics, following politics through the media, and frequency of political discussions. The estimated coefficient on the measure of parental legal status is consistently close to zero and nonsignificant. The last two outcomes, in the columns on the right side of Table 2, are indices of distrusting the federal government and police, and a desire to avoid the government and police. Again, we see no significant association with parental legal status. Overall, our results so far provide no support for H1 (chilled parents), H2 (chilled children), or H3 (socialized avoidance).
Parental Legal Status as a Predictor of Political Trust or Alienation.
Source. The source is the Latino Second Generation Study–ICPSR Number 36625.
Robust standard errors in parentheses. ICPSR = Inter-university Consortium for Political and Social Research.
p < .1. *p < .05. **p < .01.
We now turn to the possibility that young, US-born Latinos with undocumented parents are more involved in or optimistic about political activism on immigration issues, compared with other second-generation Latinos. We find that the offspring of undocumented migrants are indeed more likely to have attended an immigration march or rally in the past year (p = .03, with controls), and are more likely to have family members who joined the large 2006 immigrants’ rights marches (p = .05), although they are not significantly more likely to have marched themselves (p = .17; see Table S3 in the online appendix for detailed results). These results tend to support H4 (activism).
Finally, Figure 3 shows bivariate results for optimism about the efficacy of a range of activities in pushing for political change. We find that the children of undocumented migrants are more likely to see both legal protests, and “illegal protests or damaging property,” as politically effective (both at p = .01 or less, with controls). There are no significant differences by parental legal status for the questions on the efficacy of lobbying, working through the courts, or voting. This all supports H5 (protest efficacy).
Perceived efficacy of activism, by parental legal status.
The estimated effects are moderate in size. In models that hold other variables at typical values, people with one or more undocumented parents are 8 percentage points more likely to say they attended an immigration rally in the past year, and 4 percentage points more likely to have family members who marched in 2016. These results are in the same range as the estimated difference between people with Mexican origins and other Latinos; scholars have found that Mexican Americans have been especially active in immigration protests (e.g., Zepeda-Millán 2014). Holding other factors constant, people with undocumented parents are about 19 percentage points more likely to say that legal protest is politically effective, and about 2 percentage points to say this about illegal protests (from a low baseline; only about 9% see such risky methods as effective).
Checks for Internal Validity
Comparing people with and without undocumented parents is not enough to show a causal effect of parental legal status. In observational studies such as this, there is always the possibility of bias due to omitted variables. We, therefore, use another strategy to further test for a causal connection between parental legal status and the political behavior of the US-born offspring. This hinges on the fact that IRCA made it easier for migrants who arrived before the reform to regularize their status. The timing of the reform does not impose a sharp cutoff, since not everyone who arrived by 1986 was able to regularize. The fact that there were two regularization programs—with cutoffs at the end of 1981 and for agricultural workers in 1986—makes the comparison even fuzzier.
Although we only sampled people born in the United States from 1982 to 1995, some of their parents arrived in the country years earlier. So we can use the year of parental arrival, combined with the timing of the IRCA regularization programs, as an exogenous source of variation in parental legal status. For reasons largely beyond their own control, parents who arrived after IRCA are more likely to have remained stuck in undocumented status. Figure 4 implies that this approach is viable. The figure shows the percentage of parents who were reported to be still undocumented at the time of our survey, depending on their reported year of arrival in the United States. This percentage jumps upward after the mid-1980s when IRCA came into force. 5
Legal status of parents, by their year of arrival in the United States.
Although our survey participants whose parents arrived after IRCA are on average a bit younger, there is no reason to expect a sharp break that coincided with IRCA in the variables that might confound the effect of parental legal status. 6 We, thus, include time trends to control for smoothly changing characteristics (e.g., a gradual process of parental assimilation), and use indicators of whether each parent arrived after 1986 as instrumental variables to estimate the effect of having undocumented parent(s). Following the logic of two-stage least squares, in our case the first-stage regression is a model of having one or more undocumented parents, as predicted by demographic controls, linear controls for parental year of arrival, and the indicators for parents arriving in the United States after 1986. The second-stage regression is a model of offspring politics using the predicted values for parental legal status from the first-stage model, plus the same set of controls. 7
Our data suggest that many people were deterred from regularizing by having missed the IRCA deadlines; we reject the null in a test for weak instruments (p < .01). Table 3 presents instrumental variables (IV) results for two outcome measures. The first is an index (scaled from 0 to 1) that sums participation in the 2006 marches, having a close family member who joined the marches, and having rallied on immigration issues within the past year. The second is an index (scaled from 0 to 1) that sums the responses “legal protest” and “illegal protest or damaging property” to the question about effective measures for political advocacy. We estimate positive effects of having undocumented parents on participation in immigration rallies and on optimism about the effects of such activism, although the estimates are not quite statistically significant at conventional levels (respectively, p = .09 and p = .1).
Instrumenting the Effect of Parental Legal Status by Arrival after 1986.
Source. The source is the Latino Second Generation Study–ICPSR Number 36625.
Robust standard errors in parentheses. IV = instrumental variables; ICPSR = Inter-university Consortium for Political and Social Research.
p < .1. *p < .05. **p < .01.
One can think of the estimates in Table 3 as the Local Average Treatment Effects for people whose parents were induced to remain undocumented by the fact that they missed the IRCA deadlines. The estimates are “local” in the sense that they only apply to people whose parents were indeed constrained by this context—as Angrist and Pischke (2009, 158) put it, these are the “compliers.” Overall, these results, though not conclusive, are consistent with a causal interpretation of our findings: having undocumented parents makes people more active on immigration issues.
Checks for External Validity
One concern for this project is whether the results from our sample can be expected to generalize to the broader population of interest. Relying in part on the Internet for our sampling frame may bias the survey toward including US-born Latinos who have more resources and are more assimilated. Of course, this problem does not apply to the one-third of respondents from the GfK panel. In addition, it is estimated that by 2013, around 95 percent of young Latinos had Internet access, so any such bias is likely to be modest (A. Brown, López, and Lopez 2016, 4–7).
The fact that our sample includes too many people with higher levels of education may lead us to overstate levels of political participation. This implies that one should not simply extrapolate the results to the wider population of young, US-born Latinos. But this is not necessarily a problem for the comparisons of those with versus without undocumented parents. We see no obvious reason to think that our key comparison would be biased by the fact that we have an overly educated sample, compared with the underlying population.
Another worry is that our aggregate results might mask varying effects across the country. Perhaps Latinos with undocumented parents, stigmatized by the association with illegality, are more likely to “fight” and become politically active in permissive climates (e.g., California), but more likely to “take flight” and withdraw from civic and political life in hostile contexts (e.g., Arizona). In that case, we should report the contrasting estimates rather than the overall effect. To test for varying effects, we interacted the indicator for having undocumented parent(s) with an index of “state-created immigration climate.” This index is based on factors such as whether state law enforcement officers are told to report undocumented migrants to federal officials (see Pham and Van 2013). States such as Illinois and California receive positive scores, while states such as Alabama and Arizona get negative scores. We tested for effects on parental and offspring civic behavior, offspring political engagement and trust, activism on immigration issues, and optimism about activism—twelve outcomes in all—and found only one significant interaction between parental legal status and the index of policy climate (the models use standard errors clustered by state). 8 As such, our data provide little reason to worry that the aggregate results might conceal varying state-level effects.
Discussion
Among our more striking findings is the lack of evidence for chilling effects on the civic and political behavior of the undocumented parents of US citizens. Our data suggest they engage in a range of activities that correlate with and can precede more formal engagement with the political system. In this, they are comparable with other immigrant parents. This implies that they can also model political engagement for their US-citizen children. Our paper may show a relatively strong case of political engagement among the undocumented, because the parents of US-born citizens are probably among the most socially integrated members of the undocumented population. Those who migrated illegally but did not have US-citizen children are probably less integrated, and indeed are more likely to have left the country by now. 9 However, our results arguably reflect the new reality. Even compared with the 1980s or 1990s, the undocumented population is now much more integrated into US society, as so many have lived in the country for so long.
Nevertheless, some queries might be raised about this finding. One possibility is that the “chilling” effects might be so deep that some young, US-born Latinos with undocumented parents refuse to participate in surveys such as ours. In that case, we would underestimate the chilling effects. We do not think this is likely, given that, as US citizens, they have little to fear from taking a survey. It is also worth considering how a distinct but related reluctance to discuss the immigration history of one’s parents could affect our results. The wary offspring of undocumented migrants might participate in surveys, but conceal the legal status of their parents. If this kind of wariness is also more common among people who are politically disengaged, that would lead us to overestimate political engagement among those whom we identify as having undocumented parents. While this is certainly possible, we doubt it is a big problem in our data. As we explain in detail in the online appendix, we infer undocumented status indirectly for parents who were not reported to have US citizenship, a green card, or a valid visa. We avoided asking a direct question about this sensitive issue. If our survey participants were trying not to expose their parents, one might also expect those recruited via the GfK panel to be more secretive, as GfK has their address, whereas the people we recruited online were not asked for theirs. And yet, as is clear from Table 1, the demographics of respondents from the two samples are similar, and comparable shares gave answers implying that they had one or more undocumented parents.
Finally, our results might be due in part to “desirability” effects, where people give the answers they think are most socially acceptable. Young Americans with undocumented parents might feel they should say they are involved in immigrant activism. Again, this is possible but we doubt that it explains our findings, for three reasons. First, we did not directly ask about parental legal status, so we doubt that the survey sent a clear “desirability” signal on this issue. Second, not all of our outcomes are desirable. If anything, we suspect there is a taboo against admitting support for “illegal protests or damaging property.” And third, our survey participants did not give other socially desirable answers, although they had the chance. A classic example of a social desirability effect is saying you voted, even when you did not. Yet, our survey participants with one or more undocumented parents are not more likely to say they voted (p = .23).
Steps for Further Research
Our results align with prior research on cohort effects in political socialization, and with research showing that a sense of threat is a strong motive for political mobilization (Cho, Gimpel, and Wu 2006; Marcus 2000). The fact that so many Latino Americans have close ties to undocumented migrants, along with our finding that these ties make people more likely to protest on this issue, helps to explain some otherwise puzzling patterns. There has been a lot of activism for immigrants’ rights in recent years—even though undocumented migrants are in a tenuous legal position, they often lack the resources that fuel social movements, and the political context in much of the country is very hostile (Chavez 2013; Zepeda-Millán 2016). Scholars have argued that these activists are reacting to a sense of threat, but have emphasized threats to broad social identities such as “Latino” (e.g., Pantoja, Ramirez, and Segura 2001; Pérez 2015; White 2016). In contrast, we have shown a more intimate mechanism at work: the threats apply to specific people rather than an abstract group. It would be worth testing whether similar effects are seen for siblings of the undocumented, as many US-born Latinos have older brothers or sisters who arrived without papers or remained in the United States after their visas expired. More broadly, we think scholars should seek to understand the macro- and micro-level logics of solidarity and mobilization. Do they work separately? Or, for example, are macro-level identity appeals more effective for people who also have micro-level exposure to an issue via their family members? This is one way in which research on migration could link to broader theories about interpersonal dynamics in political behavior.
Notably, our findings are out of line with the dire predictions of some scholars on the long-run effects of having so many undocumented people living in the country, for example, with Massey’s (2013, 13) account of “insurmountable barriers” facing not only undocumented migrants but also their US-citizen children. When one looks a little closer at the existing research, however, the picture is more mixed. For example, Bean et al. (2011) find that the children of the undocumented are less educated than other second-generation children, but they also find evidence of upward mobility. Even the children of the undocumented are much more educated than their parents (see also Alba, Kasinitz, and Waters 2011). Rather than reaching for broad conclusions about “insurmountable” barriers, we think scholars should be more specific about the factors behind immigrant inclusion or exclusion. Our results are some of the first on the political consequences of a regime of mass illegality for the next generation. We should now ask why this legacy seems less damaging than in some of the research on education or mental health. One possibility is that the institution of birthright citizenship is such a strong equalizing force for the second generation that it largely offsets the political legacy of having one’s parents marked as “illegal.” Citizens can push back. Scholars should learn more about which rights are relevant in this case and who is most likely to use them.
Conclusion
A generation of Latino Americans is growing up with close ties to undocumented immigrants, often including siblings and parents. Theories of parental socialization imply that this might chill the political participation of the second generation, because undocumented parents are less likely to act as role models of political engagement. But our results suggest this is not the case. Instead, we have shown evidence that this is a generation shaped by a context of activism around immigration reform, and that Americans with undocumented parents are even more likely to join protests on this issue. A large body of previous research on political socialization shows that people tend to maintain the habits of political participation that they pick up in the “formative years” of early adulthood (e.g., Bartels and Jackman 2014; Jennings and Niemi 1981). For this reason, our results imply that one of the lasting legacies of this era of mass undocumented migration may actually be a new generation of political activists in America.
Footnotes
Acknowledgements
We thank the Russell Sage Foundation and the Cornell Institute for the Social Sciences for helpful reviewer feedback. We are also grateful for comments from Melissa Michelson, Robert C. Smith, Karen Schönwälder, and participants in the 2015 Western Political Science Association annual meeting.
Authors’ Note
A draft of this paper won the 2016 best paper award from the Latina/Latino Politics section of Western Political Science Association (WPSA). The data for the study are available as Inter-university Consortium for Political and Social Research (ICPSR) study 36625. Code for the results in this paper will be posted on the lead author’s Harvard Dataverse account within one month of publication.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We thank the Russell Sage Foundation and the Cornell Institute for the Social Sciences for funding this research.
Notes
Supplemental Material
The online appendix for this article is available with the manuscript on the Political Research Quarterly (PRQ) website.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
