The Impact of State Immigration

“In the winter of 1910 Congress received the longest report ever submitted by a government

investigating body… Buried in that statistical mountain was at least one table of figures

which was to provide peculiarly prophetic for our own times. This table showed that a

majority of the children in the schools of thirty-seven of the nation’s leading cities had

foreign-born fathers…Viewed in today’s perspective, it is clear that hose figures

forecast a major political upheaval sometime between 1930 and 1940. By then all of these

children, plus baby brothers and sisters not enrolled in school, would have grown to voting

age. Massed as they were in the states commanding the largest electoral vote,

their sheet numbers would topple any prevailing political balance” -Lublin 1965 (43)

Immigration since the 1960s has once again changed the face of American society.

Much of the current racial/ethnic diversity in the U.S. can be traced to changes in U.S.

immigration policy and the resulting changes in immigration patterns over the past six

decades. There is an increase in the number and proportion of immigrants, the U.S.-born

children of immigrants have become more numerous in public schools, populations of color

are increasing such that the majority of births to women in the U.S. are children of color, and

the Census Bureau predicts that the U.S. will be a ‘majority-minority’ nation by 2050. That

immigrants are more numerous and more racially/ethnically diverse is clear – less clear is the

degree to which new immigrants and their families are being successfully incorporated into

U.S. society and the polity.

This dissertation is set in the context of a 21st-century America undergoing rapid

immigration-driven demographic change accompanied by highly polarized debates about

immigrants and immigration policy among the general public and policymakers. Just as in

the early part of the twentieth century when immigrants and their children contributed to the

massive political changes of the 1930s and 1940s, immigrants and their children today are

poised to potentially exercise substantial political influence as their numbers grow. This

dissertation is concerned with the ways in which the immigration policymaking environment

of the 2000s and beyond is impacting immigrants and the second generation, specifically

how it is impacting political participation and well-being in these communities.

Immigration-Driven Population Changes

At approximately 70 million, the combined population of foreign-born and second

generation in the U.S. is at an all-time high. As illustrated in Figure 1.1 below, taken together

the foreign-born and second generation make up approximately 23% of the total U.S.

population, a proportion surpassed in modern history only in the early decades of the 20th

century.

The roots of contemporary immigration-driven population increases and

accompanying racial/ethnic diversity lie in U.S. immigration policies of the 1960s. The 1965

Immigration and Nationality Act eliminated racial exclusion in U.S. immigration law,

allowing immigrants from previously-excluded Asian countries to immigrate in large

numbers legally. The end of the Bracero program in 1964 and increases in border

enforcement beginning with the Immigration Reform and Control Act of 1986 together

transformed what had previously been primarily circular seasonal/temporary labor migration

from Mexico to increasingly long-term settlement (Massey and Pren 2012). In addition,

foreign intervention around the world has yielded a continuing flow of immigrants able to

gain legal status as refugees/asylees from a diverse array of countries such as Cuba, Vietnam,

El Salvador, Haiti, and others (Fuchs 1993).

These trends have created a population dynamic whereby those Americans who have

a connection to the immigrant experience - immigrants and their children - make up a large

portion of the very young in the U.S. but a much smaller portion of the age groups most

likely to vote, middle-age and older U.S. residents. The immigrant age structure dynamic is

detailed in Figure 1.2 below where we see that in 2010 the naturalized foreign-born

population is most concentrated in the 35-49 age bands, noncitizen immigrants in the 25-44

age bands, and the U.S.-born children of immigrants in the 0-29 age bands. Also notable is

the diversity of the immigrant-connected population, with the top trend line illustrating the

portion of the foreign-born and second generation in each age band who are either

Hispanic/Latino or Asian – over 70% in the 0-19 age bands, dropping to less than 20% in the

80-85+ age bands. What is also clear is that the leading edge of the huge growth in a

population of Americans with connection to the immigration experience and eligible to vote,

the U.S.-born second generation, have not yet reached voting age. As these Americans age

and, along with adult naturalized immigrants, make up a greater portion of the U.S. electorate

in the decades to come, the extent to which they participate politically will become evermore

relevant to American politics and public policy.

Scope and Purpose of Dissertation

This dissertation seeks to answer questions related to the impact of contemporary

immigration policymaking in the U.S. states on immigrant communities nationally,

specifically in terms of the political participation of adults and the well-being of children. I

am most interested in those close to the immigrant experience who are or will be eligible to

vote – thus in the empirical chapters to follow I focus on the impact of immigration policy on

the outcomes of interest among legal immigrants, naturalized immigrants, and U.S.-born

children of immigrants. This dissertation is one of the first studies to examine the relationship

between state immigration policymaking and the outcomes of political engagement and

wellbeing.

Political incorporation of legal immigrants is one indication of the overall

incorporation of immigrants into U.S. society. While studies of immigrant incorporation

often focus on economic incorporation (Donato et al. 2005; van Tubergen et al. 2004) or

assimilation (Alba and Nee 1997; Zhou 1997), political incorporation is an additional

indicator of how well U.S. society embraces those who we choose to admit as immigrants. In

contrast to other western, English-speaking countries made up largely of immigrants and

their descendants (such as Canada, Australia, and New Zealand), the U.S. has few formal

federal integration policies targeting immigrants and allocates relatively few resources to this

endeavor (Fix and Zimmerman 1994; Bloemraad 2006). This fact combined with the large

increase in sub-national anti-immigrant legislation during the 2000s leaves us to question

whether the failure to support full integration of immigrants into U.S. society will lead to

continued or worsening societal divisions.

Perhaps even more importantly, citizen participation in government is necessary to

ensure representation and legitimacy in a democratic society, “Citizen participation is at the

heart of democracy” (Verba et al. 1995, 1). The immigrant populations examined in this

dissertation are either citizens by naturalization or birth or those will be eligible for

citizenship based on legal immigration admission. To the extent that the immigrants

examined here are future voters, and to the extent that there are real options for them to

engage politically in U.S. society before they attain the right to vote through citizenship or

age, the questions of why, how, and under what circumstances immigrants and their children

engage constitute legitimate and interesting political science scholarship. When we consider

the great social impacts that high levels of immigration are having on contemporary society

and the evidence of low levels of political engagement among some groups of naturalized

citizens these questions become even more consequential.

This research is particularly relevant during this dynamic time of both increased

subnational immigration policymaking activity and increases in immigrant and second

generation populations. Given this contemporary social context, this dissertation asks

questions regarding the impact of immigration policymaking in the U.S. states on outcomes

in immigrant communities in terms of adult political participation and child well-being. How

do contemporary state immigration policies influence immigrant political incorporation in

the U.S.? Are the effects of such policies observable across a range of immigrant settlement -

including new legal immigrants, naturalized citizens, and U.S.-born second generation

citizens? How are such policies influencing child well-being in immigrant families? And how

do the effects of such policies differ across immigrants identifying with the four largest U.S.

racial/ethnic groups?

In an effort to gain empirical leverage on the multi-generational processes examined

in this dissertation, I make use of secondary data from four large nationally representative

survey datasets to interrogate my research questions from the vantage point of different

subpopulations within the immigrant community: First among newly-legalized immigrants,

then naturalized immigrants, and finally U.S.-born children of immigrants. Examination of

the role of immigration policy in influencing outcomes in each of these sub-populations thus

approximates a view to the larger multi-generational process that is political and social

incorporation of immigrants and their descendants in the contemporary U.S. context.

State Immigration Policymaking 2003-2012

The four empirical chapters that follow make use of state immigration policy as the

primary predictor of outcomes of interest; thus I discuss the environment that characterizes

this period and describe the construction of my primary predictor variable here. The

federal government has clear responsibility for immigration policy in the U.S.; however,

policymaking by states has increased substantially in the contemporary period and appears

to represent a “steam-valve” of sorts for public opinion (Spiro 1997). In the absence of

passage of any major federal immigration policy by Congress, the 2000s witnessed the start

of an especially active period of immigration policymaking by the U.S. states. While

Arizona’s SB 1070 (enacted in 2010) is perhaps the most well-known among anti-immigrant

state laws, Alabama, Georgia, South Carolina, and Utah have enacted similarly broad

restrictive omnibus legislation with provisions that range from identifying how local law

enforcement will cooperate with federal immigration enforcement agencies to restricting

undocumented immigrant’s access to public benefits, employment, housing, and education

(Sinema 2012; Wallace 2014). In addition, the National Conference of State Legislatures

(NCSL) Immigration Policy Project has documented immigration-related policymaking in all

50-states since 2005 (NCSL 2015). Scholars examining this trend in state immigration

policymaking have found various contributing factors, including state legislative partisanship

(Zingher 2014), citizen ideology (Monogon 2013; Chavez and Provine 2009), Hispanic

population growth (Marquez and Schraufnagel 2013), special interest groups

(NicholsonCrotty and Nicholson-Crotty 2011), foreign-born population growth (Boushey and

Luedtke 2011), and the interaction between state budget pressures and growing racialized

immigrant populations (Ybarra, Sanchez, and Sanchez 2014). Whatever the drivers of this

observed increase in immigration policymaking by the U.S. states, this policy environment

constitutes an important component of the political socialization context for the immigrants

and their children examined for this dissertation.

To operationalize state immigration policies, I have created indices of state

immigration policies that include both punitive (exclusionary) and beneficial (inclusionary)

policies. Consistent with other scholarship of contemporary state immigration policy (Chavez

and Provine 2009; Monogon 2013; Marquez and Schraufnagel 2013; Zingher 2014), I make

use of the NCSL dataset on state immigration policy to identify enacted state immigration

legislation from 2005-2012. As a team of scholars (as described in Ybarra, Sanchez, and

Sanchez 2014), we coded each piece of enacted legislation in the NCLS dataset 2005-2012 as

to content and direction – neutral, beneficial to immigrants, or punitive to immigrants - and

we disaggregated omnibus legislation into separate policies so that each provision could be

coded accurately as to content and direction (see Appendix 1.A for NCLS 2005-2012 coding

detail). This variable is limited to enacted policies passed by the legislature, so it excludes

bills introduced and not passed and policy action taken independently by the executive or

judicial branch. Since NCSL reports no data on enacted state-level immigration policies prior

to 2005, for 2003 and 2004 I make use of a report from the Progressive States Network

(2008). This report identifies beneficial and punitive policies enacted by each of the 50 U.S.

states from 1997-2008, however I make use only of their 2003-2004 data. Very few of the

policies enacted during this period occur before 2003, and my interest is in examining the

period of heightened policymaking activity. For these policies I rely on policy direction

identified in the publication itself for directional coding.

I agree with other authors using NCSL data (for example see Nicholson-Crotty and

Nicholson-Crotty 2011) that combining punitive and beneficial state immigration policy into

a single index makes sense both theoretically and empirically. Theoretically, research on state

welfare and economic development policies demonstrates that in the absence of strict control

over who may enter and leave their local jurisdictions, state and local policymakers make use

of a mix of policy burdens and benefits to either attract or repel certain populations (Bailey

and Rom 2004; Eisinger 2000). Empirically, many states seem to be passing both punitive

and beneficial policies in the same years. As detailed in Figure 1.3 below, between

2003 and 2012 states enacted 1,179 either punitive or beneficial immigration-related policies

(author tabulation), and of these all states enacted both punitive and beneficial policies. Even

the state enacting the least immigration legislation, Wisconsin, enacted one policy coded as

punitive and one coded as beneficial during this period.

To represent the cumulative burden/benefit created by immigration policies enacted

by state legislatures from 2003 to 2012, I construct indices that include all state immigration

policies enacted from 2003 through the year of relevant outcomes in each of the empirical

chapters to follow. Policies are individually counted -1 if punitive, +1 if beneficial and 0 if

neutral if they deal with a single issue. Since many states have enacted large omnibus

legislation addressing immigration containing multiple policy provisions, these laws have

been disaggregated into separate policies with each policy separately coded as to direction. I

sum all policies enacted from 2003 to the outcome year (through 2012) to create a single

cumulative additive index of immigration policy climate for each state. I chose a cumulative

policy index rather than a discrete year-to-year policy index because of the political

socialization processes which are believed to precede politicization of both young people and

adults (McIntosh and Youniss 2010; Sapiro 1996). A cumulative measure represents an

advantage over year-to-year measure in that it contains in one year the social constructions,

both positive and negative, expressed in state level immigration policies and experienced by

the target population for the previous years since 2003 combined. To ease interpretation, in

most of the empirical chapters to follow I convert the additive index to a z-score to

standardize the variable across all states with the resulting distribution of index scores across

states having a mean of zero and standard deviation of 1.0. Although the process of

constructing this variable is common across all four empirical chapters, the actual range

varies because the years in which the outcomes of interest are measured varies.

Outline of Dissertation

This dissertation focuses on the impact of state immigration policy on immigrant

communities. Chapter Two sets the stage with an in-depth review of literature and theory.

Here I join previous scholarship on immigrant political incorporation with social

construction-policy design theory and identify gaps in both literatures that this dissertation

proposes to fill. I contend that social construction-policy design theory is a useful public

policy theoretical lens to apply to the examination of the feed-forward effects of

contemporary U.S. immigration policy. I also provide a brief overview of my approach to testing

the theory in the empirical chapters to follow.

In Chapters Three, Four, and Five I conduct empirical tests on the impacts of

contemporary immigration policy in the U.S. states on the political engagement of adult

immigrants and the second generation. I do this in Chapter Three as concerns new legal

immigrants, examining the outcomes of political knowledge and intent to naturalize. I find

here that state immigration policy has significant impacts that differ both in relation to the

outcomes examined and by race/ethnicity of immigrants. In Chapter Four I test the effects of

state immigration policy on voting behavior among naturalized immigrants and find a

positive relationship in that beneficial (inclusive) policies result in higher voting among

Hispanic/Latino and non-Hispanic White naturalized immigrants and punitive (exclusive)

policies result in lower voting. In Chapter Five I examine the effects of state immigration

policy on non-voting political behavior of U.S.-born children of immigrants and find a

negative effect only among the Hispanic/Latino children of immigrants. Since each chapter

makes use of a different nationally representative survey dataset, I provide detail on the

datasets in each relevant chapter.

In Chapter Six I extend the examination of the impact of state immigration policy to

the outcome of child health and well-being within the context of the erosion of the immigrant

health advantage. I contend that immigration policy may be an important social determinant

of health for immigrants and their children in that such policies define belongingness and

may racialize new immigrants (Nevins 2002; Viruell-Fuentes et al. 2012). Here I find that

state immigration policy exerts a significant impact among the Asian, Hispanic/Latino, and

White children of immigrants, even after controlling for relevant community, family, and

individual-level factors; the impact is greatest among Asian children. The effect is in the

expected direction in that children of immigrants from these three racial/ethnic groups living

in states with more beneficial immigration policies have fewer expected adverse child/family

experiences.

Chapter Seven is the concluding chapter, and here I return to the core research

questions that provided the impetus for this study focused on the ways that contemporary

state immigration policy is acting on outcomes, primarily political engagement but also child

well-being, in the various immigrant communities present in the United States. I review the

main findings from the empirical chapters above with an eye toward interpreting the

multigenerational processes at work, and suggest explanations for some of the differences

found between the subpopulations examined. Here I also draw on the collective results to

point to future scholarship in this area. Lastly, I discuss the applied implications of this

research for public policy and political mobilization efforts.

CHAPTER 2

LITERATURE AND THEORY:

Social Construction-Policy Feedback Theory and the Political Incorporation of

Immigrants and their U.S.-Born Children

How do contemporary immigration policies influence immigrant political

incorporation in the U.S.? Are the effects of such policies observable across a range of

immigrant settlement - including new legal immigrants, naturalized citizens, and U.S.-born

second generation citizens? And how do the effects of such policies differ across immigrants

identifying with the four largest U.S. racial/ethnic groups? In approaching these research

questions I seek to join two major and distinct theoretical threads in political science: first

theories of immigrant political incorporation that take a multi-generational view of how

immigrants and their descendants become a part of the polity in their destination country, and

second policy design-social construction theory that posits the active role U.S. public policy

plays in encouraging (discouraging) civic and political engagement among those who are the

target.

In this chapter I first review the literature on political incorporation of immigrants

broadly, including a review of the contribution of studies in political behavior scholarship.

After laying this groundwork, I move to an extended discussion of my primary explanatory

theory – policy design-social construction – and its application to the study of political

incorporation of immigrants in the U.S. in the contemporary period. Lastly I provide a brief

overview of my approach to testing the theory in the empirical chapters to follow.

Immigrant Political Incorporation

Immigrant political incorporation describes the processes by which immigrants

“accumulate political experience in the U.S. polity” (Marrow 2005, 785), “gain political

voice in the U.S. or find their political opportunities constrained” (Lee et al. 2006, 15), and

develop “the capacity for sustained claims making about the allocation of symbolic or

material public goods” (Hochschild et al. 2013, 16). Briggs (2013) conceptualizes two “core

dimensions” of immigrant political incorporation encompassing both “membership,

connoting recognition and belonging, felt by the immigrant ‘outsiders’ and also host country

insiders” and “capacity for influence, that is, to successfully make claims in the polity” (323).

While engagement represents activity, political incorporation is a process that occurs over

the course of the immigrants’ time in the host country and extends to their children born in

the U.S. following immigration (Brown and Bean 2011; Hochschild and Mollenkopf 2009).

Ramakrishnan (2013) takes care to draw a conceptual differentiation between

immigrant political assimilation from immigrant political incorporation, with the former

focused primarily on the role of individual-level behavior and the latter focused more on the

role of institutions and groups. He states that “…The central distinction in the definitions on

either side of the conceptual divide involves the focus on individual-level outcomes in the

case of assimilation and the institutional-level processes in the case of incorporation” (34).

The empirical studies that make up the chapters to follow in this dissertation all use

individual-level survey data that measure the behavior of individuals with a focus on the role

of institutions (policies) in structuring and shaping that behavior, a conceptual space that

Ramakrishnan would refer to as a blurred boundary. Because my interest is clearly focused

on examining the central role of institutions (public policies) in shaping behavior in groups

of individuals, in examining the differences among those groups, and most importantly

examining those group impacts across the progression of immigrant generations, I place this

study in the theoretical tradition of political incorporation studies.

Comparative political incorporation literature is theoretically rich and situates

immigrant political incorporation within the context of immigrant assimilation and

incorporation writ large – inclusive of social, economic, cultural, and political assimilation

and incorporation – with the recognition that political incorporation often occurs after other

types of incorporation (Mollenkopf and Hoschschild 2009). Studies in this vein are

concerned with the social and structural determinants of political incorporation including the

demographic and social contexts of reception (McDermott 2013), the political environment

of the receiving countries (Minkenberg 2013), the role and density of nonprofit organizations

(deGraauw 2008), and the interplay between agency of the immigrant and structures in

receiving countries (Briggs 2013).

Most authors in this comparative tradition acknowledge the important role that

political institutions in the host country play in providing the political opportunity structure

within which immigrants are incorporated. Thus conceptualized, political institutions include

laws governing citizenship and voting, electoral structures, operation of parties, variation of

local elected offices, freedom (lack of freedom) for free speech and political protests, and

judicial systems (Mollenkopf and Hoschschild 2009). Empirical studies in this tradition then

typically incorporate public policy as a part of an examination of laws in the host country

structuring immigrant political engagement and most often take an historical-institutional

approach. For example, Joppke’s (2009) comparative examination of political integration of

Muslims in France and Germany recounts the history of Muslim integration in both

countries, their integration in terms of religious education and cultural practices, and the

socioeconomic marginalization of Muslims in both countries; differences he contends

contribute to variation in political incorporation.

Two studies in the comparative political incorporation literature are of particular

relevance to this dissertation in that they have focused on the role of public policy and

include the U.S. in their set of cases (Bean et al. 2012; Bloemraad 2006a; 2006b; 2006c). In

both studies the authors are explicitly interested in public policies which positively impact

incorporation of immigrants, thus they operationalize policy as either the presence or absence

of such incorporative policies. In their examination of two U.S. immigrant-receiving cities,

New York and Los Angeles, Bean et al. (2012) found that the direction of the relationship is

positive: city-level inclusionary policy is found to have a positive impact on political

incorporation of young immigrants after controlling for a vector of individual-level

covariates (age, gender, parental education, citizen parent, two-parent household, living

abroad, and number of siblings).

In a much more expansive study spanning several publications, Bloemraad (2006a;

2006b; 2006c) posits a new theory of “structured mobilization” to explain differences in

immigrant political incorporation (naturalization rates, civic engagement, seeking elected

office) in Canada and the U.S., where the political incorporation of immigrants is dependent

not only on individual and sending country characteristics but also on national-level policies

that structure instrumental and interpretive/symbolic contexts of reception and incorporation.

In her empirical examination comparing immigrants to Canada and the U.S. from Portugal,

Vietnam, and other countries, Bloemraad finds that although these immigrant groups

naturalize and participate at different rates in either country, they each naturalize and

participate at significantly higher rates in Canada than in the U.S. even after controlling for

standard individual-level covariates (Bloemraad 2006b; 2006c). She argues that these

persistent differences are due to the differences in instrumental/material and

interpretive/symbolic support provided by the nation-state and identifies specific national

policies in Canada that work to structure immigrant incorporation. These include the

Canadian federal immigration bureaucracy’s emphasis on incorporation rather than law

enforcement; Canada’s financial assistance for newcomer settlement in the form of language

training, employment counseling, and social assistance; and differences in policies on

ethnoracial diversity (Bloemraad 2006a). Bloemraad’s comparative research shines a light on

the differences between nation-states in the policies they may or may not adopt to facilitate

positive integration of immigrant groups and reminds us of the work of other scholars who

describe the relative lack of such incorporative policies in the U.S. (Fix 2007).

Opportunity gaps in the comparative political incorporation literature are primarily

around empirical applications. Scholars have critiqued the field broadly for the lack of use of

representative data (Brown and Bean 2011), and as noted above most of the empirics in this

field are in the historical-institutionalist tradition rather than rigorous quantitative methods.

With the single exception noted above (Bean et al. 2012), this field fails to address

subnational U.S. contexts; when the U.S. is examined it is generally as one among a number

of nation-states and taken as a whole. Finally, with the exception of the rich work of

Bloemraad (2006a; 2006b; 2006c) noted above, theorizing around the impact of public

policies on immigrant political incorporation is limited in that public policy is viewed as one

of a number of contributions to the overall political institutional context of the receiving

country.

Immigrant Political Behavior

Most of the scholarship studying immigrant political participation is the U.S. is in the

political behavior literature. This body of scholarship is largely influenced by mainstream

theories of political participation, most notably the resource model of political participation

developed by Verba, Schlozman, and Brady (1995; with Nie 1993). They found that three

individual-level resources are key to political participation – time, money, and civic skills.

They further posited that social institutions not associated with economic position -

nonpolitical voluntary associations and religious institutions - provide resources that

facilitate political participation. They consider resources to be as broad as education, time,

money, English proficiency, and resources derived from involvement in non-political

institutions (for example organizational skills and interpersonal networks). When introduced

in the mid1990s, its authors contended that the resource model of political participation was

an improvement over previous that relied most heavily on socio-economic status or others

that relied on feelings of efficacy and motivation.

“A model that includes resources has several advantages in explaining political

activity. Resources can be measured more reliably than is possible with the

motivations (e.g., efficacy or political interest) that often are used to explain

activity. Furthermore, they are causally prior to political activity, deriving from

home and school, choices about jobs and family, and involvements in

nonpolitical organizations and churches” (Brady, Verba, and Schlozman 1995,

285).

The authors further demonstrated that the resource model did in fact explain much of the

observed variation in American political participation outcomes (Brady, Verb, and

Schlozman 1995).

However, while the resource model may have represented an advance over previous

dominant models of political participation, the resource model was found by scholars of race

politics to be inadequate to explain voting among racial/ethnic groups. Scholars of political

participation among non-white racial/ethnic groups began to focus on the importance of

contextual and community factors to explain voting and other political behaviors among

marginalized groups. Jan Leighley, for example, critiques mainstream models and their

assumption that “…that individuals think and act politically independent of their social and

political context” (Leighley 2001, 172). Perhaps more importantly, scholars in of

racial/ethnic participation searched for explanations beyond traditional models to explain

persistent differences in participation observed between racial/ethnic groups. Scholarship in

this vein demonstrated a variety of other factors beyond individual-level resources that were

important in explaining participation among racial/ethnic groups – including the importance

of co-ethnic candidates for Latino and African American voters (Barreto 2007; Bobo and

Gilliam 1990); residential mobility for Latinos (Ramirez 2007); co-ethnic mobilization

among Latino and African-American voters (Leighly 2001); linked fate for AfricanAmerican

voters (Dawson 1994); and group consciousness for Latino voters (Sanchez 2006).

Similarly, scholars of immigrant political participation have found traditional models of

voting and non-voting political participation, with their focus on individual-level resources,

to be of limited use in explaining voting among immigrants and their descendants. For

example, using the 1994-2000 Current Population Survey, Ramakrishnan (2005) found that

while age, education, employment status, residential stability, and marital status are all

significantly predictive of voting among naturalized immigrants, the effects of education are

weakest among the first generation of immigrants in each of the four largest racial/ethnic

groupings (non-Hispanic Asian, non-Hispanic Black, Hispanic/Latino and non-Hispanic

White). Additionally he found that income, a usually strong predictor of political

participation among the U.S.-born, is not significantly predictive of voting among naturalized

Latino, Asian, or Black immigrants.

Thus scholars of immigrant political participation have increasingly explored the

impacts of structural, contextual, and community factors on political participation in

immigrant communities. Lee et al. (2006) highlight the importance of moving beyond

individual-level factors to consideration of contextual factors, “…future research of

immigrant politics needs to give a more thorough consideration to the effects of social and

political contexts on subsequent participation. Factors such as visa status and age of entry

into the United States influence not only the socioeconomic adaptation of immigrants, but

also their political and civic incorporation” (268). Extant research has verified that these

factors are significant and important in predicting political participation among immigrants.

For example, co-ethnic immigrant concentration at the neighborhood level increases voting

among Asian registered voting in California (Tam Cho et al. 2006); state policies providing

greater access to the social safety net for immigrants is associated with increased voting

among Latinos and Asians and increased registration among Asians (Logan et al. 2012); dual

nationality policies and other sending-country characteristics effect both naturalization and

voting among immigrants (Bueker 2005; Jones-Correa 2001); gendered social networks

impact opportunities for immigrants to adjust to legal status in Houston (Hagan 1998); and

nonpartisan voter contact effects voting among Latino and Asian immigrants in Southern

California (Ramirez and Wong 2006).

Importantly, political behavior research on immigrant political participation has

revealed significant and substantive differences in predicted rates of voting that generally

persist among immigrants of differing racial/ethnic groups even after controlling for

individual-level and contextual covariates. On average, Black, Latino, and Asian naturalized

immigrants and second generation citizens (U.S.-born children of immigrant parents)

demonstrate significantly lower levels of voting than their non-Hispanic White counterparts

(Ramakrishnan 2005). While predicted probabilities of voting increase from the immigrant

generation to the second generation and again to the third-plus generations among Blacks and

Asians, among Latino second and third-plus generations voting remains at the same low rate

as observed among naturalized citizens and all lag behind the rates of comparable

nonHispanic White voters (Ramakrishnan 2005). Scholars have noted similar trends for

nonvoting political participation such as volunteering for campaigns and writing

representatives (Pearson and Citrin 2006).

Two studies in the political behavior tradition are of particular relevance for this

dissertation in that they have focused on the role of policy and political environments in U.S.

states and attempted to measure the impact of such environments on political engagement of

immigrants (Pantoja et al. 2001; Ramakrishnan 2005). Both are placed in the anti-immigrant

environment of California in the mid-1990s which saw the passage of Proposition 187 in

1994 to restrict public services for illegal immigrants (Campbell et al. 2006). In both studies

the primary independent variable of interest is framed as the hostile or threatening political

environment. In terms of theory, Pantoja et al. (2001) put forward a novel theory of

naturalized voter participation which posits that immigrants to some degree self-select when

to naturalize and that “subsequent levels of political participation are endogenous to the

selfselective characteristics of the naturalization process and the raised expectations that

accompany such life choice” (735). Ramakrishnan (2005) builds on and extends this

scholarship using a theoretical lens of political threat which has been shown in other

populations to stimulate increased voting. The authors in both studies found that the political

environment related to immigrants in California in the mid-1990s did impact immigrant

political behavior negatively, meaning that the political environment that was negative

(threatening) toward immigrants contributed significantly to in increased political

engagement among naturalized immigrants. The main interest of the scholars in both these

studies extended beyond the public policy itself to the charged political environment in the

years both before and after passage of California’s Proposition 187. This interest is reflected

in how they each operationalized the political environment predictor variable; neither

operationalized the policy itself. Pantoja et al. (2001) operationalized the political

environment predictor by running separate models for the three states in their study

(California, Texas, and Florida), comparing rates of voting among a select group of

naturalized Latino immigrants across the three states and attributing those differences to the

political environment. Ramakrishnan (2005) operationalized the threatening anti-immigrant

political environment by using newspaper headlines to create an index of threat environment

for California and comparison states (Texas and New York) during the mid-1990s. Rather

than claiming to test the impact of the policy itself, Pantoja et al. (2001) were testing what

they referred to as a politically hostile state environment surrounding Proposition 187, and

Ramakrishnan (2005) was testing a political environment that was threatening to immigrants,

inclusive of Proposition 187 but also including the Republican takeover of Congress in 1994

and both welfare and immigration reform legislation enacted by Congress in 1996.

In a related study, Barreto et al. (2009) similarly found that an increasing number of

Latinos, both naturalized and native-born, entered the electorate during the highly

antiimmigrant political environment of California of the mid-1990s and drove the large

growth in Latino voting observed in California between 1996 and 2000. Although these

authors did not operationalize public policy, their findings generally support the findings of

Pantoja et al. (2001) and Ramkrishnan (2005) that point to increased mobilization as a result

of an antiimmigrant policy environment.

Based on this review, I conclude that a substantial and meaningful opportunity gap in

the political behavior literature exists around the need to more adequately operationalize and

test the impact of public policy on the political behavior of immigrants, along with the need

for more robust theorizing in this area.

Policy Design and Social Construction Theory

In contrast to positivist/rational theories of public policy which assume public policy only as

a tool to address the problems of society, Schneider and Ingram’s policy design-social

construction theory (1997) represents a post-positivist approach in policy research taking us

back to Easton’s definition of politics – the authoritative allocation of values (Easton 1953).

If public policy is viewed as the way such allocation occurs, then the policy design-social

construction theory asks us to focus on just whose values are being reflected, and thus it

places the power relationships in society at the center of our study of policy.

The policy design-social construction framework emerged in the late 1980s, developed by its

primary authors, Anne Larason Schneider and Helen Ingram (1993), to help explain the

enduring dilemmas represented by policy targeting marginalized groups, dilemmas not

adequately addressed by rationalist frameworks. The policy design-social construction theory

contends that because target populations vary in the level of power they have in the political

realm, policymakers socially construct them in positive or negative terms and distribute

benefits and burdens that reflect and reinforce these constructions (Schneider and Sidney

2009; Ingram, Schneider and deLeon 2007). This theory emphasizes the ways in which

policy affects politics for the target populations – thus conceptualizing policy design serving

at once as both a dependent variable dependent, in large part on social constructions of

knowledge and of the target population, and an independent variable structuring and

conditioning future policy/political engagement on the part of the target populations

(Schneider and Ingram 2005).

“Policy design” is the unit of analysis in much policy research utilizing the policy

design-social construction theory, referring to the actual content and substance of public

policy (Schneider and Sidney 2009). This approach assumes that policy design contains

elements and characteristics that can be observed, such as target populations, goals,

problems, rules, rationales, or assumptions. “Social constructions” are key in this theoretical

approach and refer to the “underlying understanding of the social world that places

meaningmaking at the center. That is humans’ interpretations of the world produce social

reality; shared understandings among people give rise to rules, norms identities, concepts and

institutions” (Schneider and Sidney 2009, 106). The theory then is interested not just in the

instrumental means of policy design that are directed at the policy goals but also in the

symbolic and interpretive dimensions that reflect the social constructions of policymakers.

The theory assumes that public policy is purposive and normative and that the elements and

characteristics of the policy design are intentionally arranged to serve particular values and

interests (Schneider and Ingram 1997).

Schneider and Ingram’s policy design-social construction theory has roots in the critics of the

rationalist/positivist approach. Lowi was among one of the first such critics – he argued that

the ideal of pluralism, in which groups competed with one another for benefits from

government, no longer existed in the U.S. Instead, he argued that “interest group liberalism”

had replaced the ideal of pluralism and that powerful interest groups had captured

policymaking and the implementation process (Lowi 1964). Lowi further blamed, at least in

part, the policy itself – describing “arenas of power” in which certain types of policy

structure the patterns of politics, enabling policy to engender politics (Lowi 1979; 2009).

Similarly, Fischer (1980) wrote early in his career of how value-laden the entire policy

process is, and he remained critical of the policy sciences positivist approach which he

assessed as a failure at providing “solutions to the problems facing modern societies”

(Fischer 1998, 130). In building on the work of the post-positivists then, Schneider and

Ingram reject the rational approach to policy based on its technical orientation and its lack of

attention to normative matters and principles of justice. Instead they propose an alignment

with critical theories, which focus on macrostructural and institutional explanations for what

ails society (Schneider and Ingram 1997).

In their recent review of the published empirical applications of policy design-social

construction theory between 1993 and 2012, Pierce et al. (2014) distill the eight basic

assumptions of the theory into three topics - the model of the individual (containing four

assumptions), power (containing one assumption), and the political environment (containing

three assumptions). As concerns the model of the individual, the theory assumes that actors

rely on mental heuristics, that mental heuristics filter information in a biased manner, that

people use social constructions in a subjective manner, and that social reality is boundedly

relative. As concerns power, the theory assumes that power is unequally distributed among

individuals in the political environment. And finally, as concerns the political environment,

the theory assumes that policy creates future politics, that policies send messages to citizens

capable of affecting orientations and participation, and that those policies are created in an

environment of political uncertainty. These eight assumptions are detailed in Table 2.1 below,

reproduced from Pierce et al. (2014, 5).

Table 2.1. Assumptions of the Theory of Social Construction and Policy Design

Model of the Individual

1. Actors cannot process all of the information relevant to make a decision, and therefore rely on

mental heuristics to decide what information to retain.

2. Mental heuristics filter information in a biased manner, thereby resulting in a tendency for

individuals to confirm new information that is consistent with preexisting beliefs and reject

information that is not.

3. People use social constructions in a subjective manner that is evaluative.

4. Social reality is boundedly relative where individuals perceive generalizable patterns of social

constructions within objective conditions.

Power 5. Power is not equally distributed among individuals within a political

environment.

Political Environment

6. Policy creates future politics that feeds forward to create new policy and politics.

7. Policies send messages to citizens that affect their orientations and participation patterns.

8. Policies are created in an environment of political uncertainty.

Reproduced from Pierce et al. (2014) Table 1

These assumptions then support the two core analytic propositions of the theory, illustrated in

Figure 2.1 below, that form the bases for empirical applications - the “target population

proposition” and the “feed-forward proposition” (Pierce et al. 2014). Based largely on

assumptions of the individual and power noted above, the target population proposition

posits that policymakers embed social constructions of target groups, or groups that are the

intended target of public policy, in the public policy itself. And further, that the allocation of

benefits and burdens contained in the policy depends both on the relative political power of

the target group and the positive or negative social construction of the group (Pierce et al.

2014). To elaborate this proposition, Schneider and Ingram illustrate a typology of target

groups containing two axes - political power and social inclusion or “deservingness” – and

they identify four target group types: advantaged, dependents, contenders, and deviants

(Ingram, Schneider and deLeon 2007). The authors acknowledge that not all groups are

socially constructed in public policy and that some constructions are more contested than

others (Schneider and Ingram 1993).

The second and more relevant proposition for the purposes of this dissertation is the

feed-forward proposition. Based largely on the assumptions of the political environment

noted above, this proposition posits that policy leads to politics, or more specifically that the

design of the policy conditions the opportunity structures, political participation, and

mobilization of the target group (Ingram, Schneider and deLeon 2007; Pierce et al. 2014).

The design elements of the policy itself are conceptualized as active and consequential and

are tools that transmit both instrumental and interpretive benefits and burdens (Schneider and

Ingram 1997). Policy design contains “implicit ideas, values, and broader meaning within

society” that bring about “patterns of political voice, power and democratic responsiveness”

(Schneider and Sidney 2009, 112). “The net result is a powerful influence on the behavior

and understanding of self by target populations” (Pierce et al. 2014, 6). It is this second

proposition of the theory, that of feed-forward effects, that I will test in the empirical chapters

of this dissertation.

As illustrated in Figure 2.1 below, the two propositions may be viewed as working

together as a feedback loop so that those target groups positively constructed in public policy

receive benefits that structure political opportunities and encourage civic/political

engagement. Such engagement of positively constructed target groups encourages policy

responsiveness on the part of political elites. The feedback loop also operates in the negative

so that groups negatively constructed in public policy are recipients of the burdens of policy

that lead to limited opportunity structures and a lower likelihood of civic/political

engagement. Groups with low political engagement generally elicit low levels of policy

responsiveness from political elites. This feedback loop may then lead to path dependence in

the form of either a virtuous (positive) cycle or a vicious (negative) cycle, depending on how

target groups are constructed in public policy.

Figure 2.1. Policy Design-Social Construction Propositions

The authors of this model theorize that the mechanisms that underlie the target

population proposition have to do with the interaction between the relative position of the

group being constructed and their political power. The four-quadrant ideal type model of

group construction as either deserving (including advantaged and dependent groups) or

undeserving (including either contenders or deviants) posits that those groups constructed as

deserving and with the most political power will receive the most benefits in public policy,

while those groups constructed as undeserving and with the least political power will receive

the most burdens (Schneider and Ingram 1997). Such constructions are translated into public

policy in this way because the “Policymakers, especially elected politicians, respond to,

Social

Constructions

Embedded

Public

Policy

Civic/Political

Engagement

Target

Populations

Target

Population

Proposition

Feed

‐

Forward

Proposition

perpetuate, and help create social constructions of target groups in anticipation of public

approval or approbation” (Ingram, Schneider and deLeon 2007, 106). This mechanism is

thought to be especially powerful as it concerns negatively-constructed target groups with

little political power (Nicholson-Crotty and Meier 2005).

We may also gain insight into possible mechanisms behind the target population

proposition from the scholarship on the symbolic language contained in elite political

rhetoric. Interpretive effects are “the impact of policies on the cognitive processes of social

actors” (Pierson 1993, 610) and occur primarily through the role of policy in transmitting

information and meaning. Scholars have long recognized the interpretive use of symbolic

language in meaning-making, the power of symbolic language to mobilize and demobilize

groups (Edelman 1964), and the capacity of political language and rhetoric to create and

transmit social constructions of groups as powerful or power-less. In his seminal work on

symbolic power Bourdieu (1977) theorized that “linguistic exchange,” communication

between sender and receiver, transmits symbols of authority and that groups are constructed

as more or less powerful through linguistic exchanges.

“It is in the struggles which shape the history of the social world that the

categories of perception of the social, and the groups produced according to

these categories, are simultaneously constructed” (134).

As one set of privileged social actors, political elites seek to impose their representation of

the social world through their use of authoritative language, they “undertake to transform the

social world in accordance with their interests – by producing, reproducing and destroying

the representation that make groups visible for themselves and for others” (Bourdieu 1977,

127). Because the language contained in public policy necessarily comes from the ideas and

concepts held by policymakers, political elite rhetoric may be one mechanism by which

social constructions of target populations become embedded in public policy.

Schneider and Ingram theorize that the mechanisms behind the feed-forward

proposition are both material and interpretive. “Policy designs contain both instrumental and

symbolic messages that teach lessons about democracy, justice, citizenship, and the capacity

of the society to solve collective problems” (Schneider and Ingram 1997, 104). Empirical

studies testing the feed-forward proposition in policy areas such as veteran’s benefits and

social welfare policy have found evidence of such policy learning. In her examination of

veterans who received educational benefits from the G.I. Bill after World War II, Mettler

(2002) found evidence that both material and interpretive benefits of the program resulted in

greater civic engagement among participants. She found that the interpretive benefits derived

largely from the inclusiveness of the program and the ease with which participants received

their benefits. Further, she found that these interpretive benefits accrued primarily to veterans

who had grown up in lower income families.

“The resources the program extended were likely to have been especially

instrumental in enhancing the well-being of such individuals, ameliorating the

deterrents to civic activity they experienced in childhood and thus enhancing

their civic capacity most dramatically. In addition, the interpretive effects of the

G.I. Bill were especially powerful for such veterans, conveying to them a sense

of an elevated status in the polity” (361).

In his comparative investigation of the impact of policy design on political

participation among AFDC (Aid for Families with Dependent Children; a.k.a. welfare) and

social security disability recipients, Soss (1999; 2005) found evidence of individual-level

policy learning which he posits stemmed from differences in the benefit amounts,

administration procedures, and perceived messages in the two programs. He finds that

participants of both programs, who otherwise are very similar, learn different lessons from

participating in the programs that impact their general views of government and level of

political efficacy. He concludes that these differences are sufficiently powerful to produce the

significant differences observed in political engagement between the two groups. Soss (2005)

further suggests a group identity mechanism at play in which social constructions contained

in public policy may impact collective action. “High levels of stigma consciousness tend to

undermine group identification, perceptions of shared interest, expectations of effective

agency, and feelings of collective injustice” (320).

The Two Propositions and Immigrants in U.S. Public Policy

A number of scholars have examined the presence of social constructions of

immigrants contained either in U.S. public policies or in the political rhetoric leading to the

passage of legislation that targets immigrants. Although some of these scholars do not make

explicit use of the policy design-social construction theory, they all may be viewed as testing

the first or target population proposition for its application to immigrants and immigration

policy. Taken together, this line of scholarship finds that social constructions of immigrants

are contained in political rhetoric leading to public policies that target immigrants and are

embedded in U.S. public policy. Further, the valence of these constructions varies with some

carrying negative and others positive valence and variously communicating the targeted

immigrants as either deserving or undeserving of the benefits of public policy. Newton

(2005) contends that the national immigration policy debate as a whole has been effectively

recast as one of deserving vs. underserving policy targets with origins in the rhetoric of

political elites:

“These social constructions of immigrants, both positive and negative, have

become a part of our national discourse on the issue. They also provide

important political currency for elected officials, who have the skill and access

to avenues of communication through which they can advance these

constructions to serve their policy agendas” (Newton 2005, 141).

Some scholars outside the policy design-social construction framework have

examined the social construction of the “illegal immigrant” contained in immigration policy

from elite rhetoric. For example, Ngai (2004) employed a historical/sociolegal framework

using court documents and elite discourse to examine U.S. national immigration policy from

1924-1965, and demonstrated that the racialized illegal alien was constructed from

exclusionary immigration policy in the early 1900s. Similarly, Nevins (2002) details the

construction of the illegal immigrant in contemporary public discourse, which he contends is

rooted in the U.S. history of “largely race-based anti-immigrant sentiment” (96). And further

how political actors during the last half of the twentieth century used “discursive devices”

(121) to capitalize on rising concerns among the public about illegal immigration and justify

increased border enforcement efforts by the nation-state. Discourse is institutionalized in

laws, and laws related to immigration construct categories of belongingness and membership

that often become a part of the social identities of those residing within U.S. borders (Nevins

2002).

The Immigration Reform and Control Act (IRCA) was passed by Congress in 1986

on a bipartisan basis and contained three main provisions to address the problem of

undocumented immigration as it was viewed at the time: increased border enforcement,

employer penalties for knowingly hiring undocumented immigrants, and legalization for 2.7

million previously undocumented workers (Chishti et al. 2011). Using narrative analysis of

text of congressional debates leading up to passage of IRCA1986 and applying an explicit

policy design-social construction framework, Newton (2008) found that illegal immigrants

who would be eligible for IRCA’s amnesty provisions were primarily constructed as

undeserving by opponents of the legalization provisions contained in the proposed

legislation. “Statements to this effect juxtaposed the illegal alien with the legal alien and

reminded members of Congress that the policy was an injustice to those who had chosen to

abide by U.S. immigration law” (Newton 2008, 86). She also found evidence of contestation

of this primary undeserving narrative with a deserving immigrant construction put forward

by sponsors of the legislation, “…supporters of amnesty could recast immigrants in terms of

national mythology, representing them as the kind of people who made this country great, the

kind of folks we should embrace for what they have to offer” (Newton 2008, 90).

The next major federal immigration legislation was the Illegal Immigration Reform and

Immigrant Responsibility Act (IIRIRA) enacted in 1996 which enhanced border patrol,

increased enforcement and penalties for alien smuggling and false documents, strengthened

deportation laws, provided for electronic verification of required employment

documentation, and changed the processes for accepting refugee and asylum applications

(Fragomen 1997). Using a citizenship lens, Schneider (2000) examined the congressional

debates leading up to IIRIRA 1996 and found that members of Congress relied heavily on

symbolic images of citizenship based on status as responsible taxpayers, workers, and family

members in order to contrast characterizations of undocumented immigrants as irresponsible.

As one example, “In a number of speeches Representatives noted the high fertility rate,

especially among illegal aliens, and hinted toward these immigrants ’ cunning manipulation

of benefit provisions in order to support their large families” (Schneider 2000, 266). As she

had done with the 1986 legislation, Newton examined the congressional debates leading to

passage of IIRIRA 1996 and similarly found that both undocumented and legal immigrants

were negatively constructed as consumers of welfare and public goods and that

undocumented immigrants were further constructed as criminals. In contrast to the 1986

legislation, she found no contestation to the negative constructions in the debates leading up

to the 1996 legislation (Newton 2008).

Also passed by Congress in 1996, the Personal Responsibility and Work Opportunity

Reconciliation Act (PRWORA) reformed the nation’s social welfare system and “redefined

immigrants’ access to public benefits” by excluding all undocumented and most legal

immigrants from receiving many public benefits (Fix and Passel 2002, 1). A number of

scholars have examined the elite discourse leading to PRWORA 1996 for symbolic language

and social constructions of immigrants. Using discourse analysis, Agrawal (2008) found that

members of Congress relied on symbolic language related to self-sufficiency and the value of

citizenship to support provisions excluding immigrants from public benefits. Yoo (2008)

makes use of social construction theory, framing theory and content analysis to examine the

social constructions of elderly immigrants at the time of the passage of this legislation

excluding all undocumented and most legal immigrants from receiving disability benefits

under the Social Security Administration’s disability programs. She found that social

constructions of elderly immigrants by political elites were dominated by characterizations as

underserving, fraudulent, benefit-seeking, rule breakers, and non-contributing members of

society, “…immigrants were constructed as fraudulent and immigrating to the U.S. to receive

public assistance” (Yoo 2008, 490).

In addition to high-profile congressional debates, scholars have found social

constructions of immigrants as well in symbolic language used by state legislative elites

surrounding state public policy. Using a federalism lens and discourse analysis to examine

legislative resolutions passed in states along the U.S.-Mexico border from 1993-2007,

Filindra and Kovacs (2011) found substantial variation in how immigrants were constructed.

For example, elites in Arizona relied heavily on negative symbolism toward undocumented

immigrants while those in California and New Mexico relied more heavily on language

emphasizing the importance of immigrant integration. “For Arizona, the solution to

immigration problems is enforcement, not improvement of public services to immigrants. On

the other hand, New Mexico and California are more concerned with integration issues,

seeking to study immigration-related problems, reward programmes that help immigrants,

and provide additional services” (Filindra and Kovacs 2011, 20). In a case study on Arizona

legislation using policy design-social construction theory, Magana (2013) found that elite

characterizations of undocumented immigrants as criminals and as “out-of-control”

contributed to the passage of a series of anti-immigrant state laws in Arizona between 1996

and 2012. Also using policy design theory, Reich and Barth (2010) examined policy

outcomes by examining elite (state legislator) rhetoric and resulting legislator vote on state

tuition policies for undocumented students in Arkansas and Kansas. They concluded that elite

social constructions of undocumented students contained in political rhetoric surrounding the

proposed legislation in Kansas constructed them as “proto-citizens”, thus contributing to the

passage of legislation favorable to those students.

Although these examples of scholarship finding social constructions of immigrants in

the political rhetoric leading up to and embedded in U.S. public policy focus on the

legislative branch, we know that policy is made in the executive and judicial branches, and

scholars have found social constructions of immigrants here as well. Using framing theory

and a survey experiment, Knoll et al. (2011) find that positive and negative symbols of

immigrants used by presidential candidates of both major parties during the 2008 Iowa

primary campaigns were consequential for policy preferences among voters. DiAlto (2005)

finds that the courts played an important policymaking role in constructing Japanese

Americans as non-White, ineligible for citizenship, and “enemy aliens” during the early part

of the twentieth century. “The significance of the courts in socially constructing group

identities is especially evident when it comes to issues of race, as courts have been at the

forefront of racially constructing groups for more than a century (DiAlto 2005, 85). In

terms of the second or feed-forward proposition of policy design-social construction theory,

that social constructions contained in public policy are sufficiently active to impact the civic

and political engagement of the target population, I located just one scholar who empirically

tests this proposition as it relates to immigrants in the U.S. using an explicit policy design-

social construction theoretical lens. Coutin (1998) conducts a historical-institutional

examination of the legalization strategies and experiences of people coming to the U.S. from

El Salvador starting in the 1980s. Most Salvadorans fleeing civil war in their country in the

1980s arrived as undocumented migrants, and like other undocumented non-citizens their

lives became more difficult following implementation of IRCA 1986 which required proof of

legal residency in order to work legally. Although they were fleeing war and potential

persecution, most were considered undocumented immigrants barred from working and

subject to deportation if apprehended. During this time, the Reagan administration was

supporting the Salvadoran government, and claimed most of the

Salvadorans in the U.S. were “economic immigrants who did not deserve asylum” (Coutin

1998, 906) under existing U.S. refugee/asylee provisions. Sanctuary advocates, Salvadorans,

and allies in the U.S. during the 1980s mobilized to successfully challenge and change the

construction in immigration law from undocumented immigrants to refugees/asylees and

ultimately to legal immigrants eligible to pursue naturalization (Coutin 1998).

Target Populations - Contingency, Contestation and Adjacency

Although the review above demonstrates that both legal and undocumented

immigrants have been target populations in contemporary immigration policymaking at both

the federal and state levels, the primary negative constructions have been of undocumented

immigrants portrayed as an undeserving threat. In considering an examination of the

feedforward proposition of policy design-social construction theory it is important to

understand that the social constructions of immigrants in U.S. public policy are complex -

immigration status is contingent and contested, and spillover effects occur by which

individuals and groups adjacent to the explicit policy targets may be affected by policy.

Cook (2013) details the contingent nature of legal status among even undocumented

immigrants in the U.S., many of whom have deep community roots. This group of the most

excluded immigrants in the U.S. has at times become eligible for legalization, as with the

Immigration Reform and Control Act (IRCA) 1986 in which 2.7 million

previouslyundocumented workers were offered a path to citizenship (Chishti et al. 2011).

They also have become eligible for protected status, as with the more recent executive

actions such as the Deferred Action for Childhood Arrivals (DACA) of 2012 and the

Deferred Action for Parental Accountability (DAPA) of 2015 in which an estimated 2.1

million undocumented youth and another 3.7 million undocumented parents of citizen and

permanent resident children have opportunity to temporarily adjust their status to avoid

deportation (Hooker et al. 2015; MPI 2014). Because the status of individual immigrants may

change - from undocumented to legalized or protected, from temporary status to permanent,

and from resident to naturalized citizen - someone who may be clearly defined as a member

of a target population for a particular policy one day may not be the next. The contingent

nature of immigration status suggests that the feed-forward proposition of social

constructions of immigrants contained in public policy may affect those who previously held

the status targeted, even though they may not currently hold that status.

In the contemporary immigration policymaking space we see active attempts to

reconstruct immigrants from undeserving to deserving. Segura (2013) reminds us that

incorporation of immigrants is a contested political process. Contemporary contestation

efforts may be observed most vividly as they concern undocumented immigrant youth, now

known as DREAMers, who were brought to the U.S. by their undocumented immigrant

parents as minors. Early scholarly work finds that in the wake of DACA 2012 the

DREAMers and their allies have successfully contested a negative construction of them in

the popular media (Garcia Rodriguez 2014). The contestation by Salvadorans of their

classification of undocumented immigrants cited above (Coutin 1998) is another example of

contestation resulting in increased rather than decreased political engagement. These findings

are consistent with the prediction of policy design-social construction theory that

negativelyconstructed groups may be capable of using the very policies in which they find

themselves characterized as undeserving as a point of contention in mobilizing to resist and

challenge the negative constructions (Ingram and Schneider 2005). In their

specification of social construction-policy design theory, Schneider and Ingram describe that

the expectation for a positive relationship between public policy and political engagement is

due in large part to the importance of social identity in political engagement and mobilization

and the variation in the messages about the social identity of target populations contained in

public policies. Furthermore, the path by which negatively-constructed groups use the very

policies in which they find themselves characterized as undeserving as a point of contention

to mobilize and resist the negative constructions is both difficult and unusual (Ingram and

Schneider 2005). While the mechanisms at work that (in some cases) enable negatively

constructed immigrant target populations to contest the construction is not entirely clear,

related scholarship suggests a possible group consciousness mechanism by which anti-

immigrant policymaking may be triggering pan-ethnic solidarity and increased political

organizing among target groups (Schmidt et al. 2010). This suggests that in empirical

examination we should be alert to the possibility that contestation may moderate or mediate

the relationship between social constructions contained in policy and civic/political

engagement of target population observed in the feed-forward proposition. We perhaps might

not expect the same relationship between social constructions in policy and the outcome of

civic/political engagement as concerns a target population that is actively contesting the

constructions compared with one that is not.

In the contemporary period an estimated 85% of families with at least one immigrant

parent are considered “mixed-status” families, meaning they contain both citizen and

noncitizen family members (Fix and Zimmerman 2006). An estimated 5.5 million children in

the U.S. have at least one undocumented parent, and 4.5 million of these children are U.S.

citizens (Suarez-Orozco et al. 2011). This type of family complexity suggests that even when

family members may not be the intended target of public policy, because the lived

experiences of family members (especially children) are so dependent on one another policy

effects on one family member may in turn affect others. In fact, empirical findings support

this suggestion at least as concern health and well-being outcomes for children of

immigrants. This type of adjacency effect may be most apparent in the case of increases in

restrictive/exclusionary policies enacted at the state and local level in the 2000s in which

“enforcement increasingly intervenes in everyday life spaces, both public and private,

whether one lacks or possesses legal status” (Varsanyi 2010). Children of immigrants

demonstrate higher high school graduation rates in states with welfare policies that are more

inclusive of immigrants (Filindra et al. 2011), and immigrant families with eligible citizen

children have lower enrollment in the federal Supplemental Nutrition Assistance Program

(SNAP) in states with more restrictive immigration policy (Skinner 2012). Children whose

parents benefited from legalization, allowing them to move from undocumented to legalized

status, experience improved educational outcomes (Bean et al. 2006). The nationwide move

in public policy toward criminalization, detention, and deportation of undocumented

immigrants threatens the health and well-being of children of immigrants (Androff et al.

2011). A household survey in Arizona following the passage of the punitive SB 1070 in 2010

found that households in which at least one member was undocumented reported planning to

avoid accessing health insurance and healthcare because of fear related to SB 1070 (O’Leary

and Sanchez 2011).

Finally, the racialization of immigration policy in the U.S. has implications for the

feed-forward proposition. With immigration since the 1970s dominated by immigration from

Latin America and Asia, such that the majority of the current foreign-born population is from

those regions (Grieco et al. 2012), contemporary immigration policy contexts are often

racialized. Substantial scholarship speaks to the continued racialization of immigrants from

Asia and Latin America as non-White in contemporary America (Martin and Duignan 2003;

Rumbaut 2009; Schmidt et al. 2010; Bean et al. 2013). The racialization of some immigrants

has two potential implications for the research presented here that examines the feed-forward

proposition. First, because immigrants racialized as non-White experience a racialized

socialization or assimilation process, one which impacts social incorporation (Golash-Boza

and Darity 2008) and economic incorporation (Hersch 2003), I reason that racialization may

also impact political incorporation and the relationship between social constructions

contained in public policy and political engagement of the target population. Ramakrishnan

(2005) finds evidence of a “persistent racial gap” (85) in which non-White immigrants

demonstrate significantly lower rates of voting than White immigrants, a difference that

persists into the second and third generations. Second, because the operationalization of

racism in the U.S. relies so heavily on identifiable phenotype, residents of the U.S. who are

phenotypically similar to intended racialized immigrant targets of public policy may

experience adjacency effects. Research on contemporary immigration enforcement has found

such adjacency effects in which legal immigrants and U.S. citizens who are Latino have been

swept up in enforcement activities targeting undocumented immigrants (Romero 2006;

Stevens 2005; Preston 2011). Short and Magana (2002) suggest that phenotypic similarity

and adjacency may elicit higher levels of political engagement in the form of contestation,

“as a theoretical addition to Schneider and Ingram’s model, it would be a useful avenue for

future researchers to investigate the political impact of a phenotype shared with a group that

is socially constructed negatively as a motivating force for greater political

empowerment….” (Short and Magana 2002, 709).

The paucity of studies empirically examining the feed-forward proposition of the

policy design-social construction theory as concerns immigrants as target populations of

public policy represents a substantial opportunity gap in the literature, especially given the

rich value-laden U.S. immigration policymaking environment in the contemporary period.

Overview of Empirical Approach

The empirical analyses contained in the chapters to follow build on the theory and

previous empirical research described here in these various traditions – comparative political

incorporation, political behavior, and policy design-social construction theory. In addition,

these analyses are an attempt by the author to address opportunity gaps identified in each of

these three bodies of literature – including the gap in use of representative data, rigorous

quantitative methods, and application to sub-national U.S. contexts in the comparative

immigrant political incorporation literature; the gap in operationalization of public policy and

general lack of theorizing public policy in the political behavior literature; and the scarcity of

studies empirically examining the feed-forward proposition of the policy design-social

construction theory as it concerns immigrants as target populations.

The decade under study for this dissertation, 2003-2012, witnessed high levels of

policymaking around immigration in the U.S. states. Both this general increase as well as the

variation in policymaking among the states make state immigration policy during this period

of time an opportune application of the feed-forward proposition theorized by policy

designsocial construction theory. Although I conceptualize political incorporation as

occurring over multiple generations, in the absence of multi-generational longitudinal data I

measure engagement/behavior at the level of the individual. Thus I examine individual-level

political behavior in the empirical chapters to follow. I make use of a variety of

nationallyrepresentative survey datasets to seek answers to my larger research questions by

examining subpopulations – first newly-legalized immigrants, then naturalized immigrants,

and finally U.S.-born children of immigrants. Examination of the role of immigration policy

in influencing the political engagement of each of these sub-populations should approximate

a view to the larger multi-generational process that is political incorporation of immigrants

and their descendants in the contemporary U.S. context.

The concept of adjacency is particularly important for this dissertation in that the

populations I examine empirically are not themselves undocumented immigrants, the explicit

target of many of the policies I test. Yet the populations I examine occupy social spaces

adjacent to undocumented immigrants, as described above. My reasons for using adjacent

populations rather than undocumented immigrants themselves are twofold: First, adjacent

populations with some legal status – here either new legal immigrants, naturalized

immigrants, and U.S.-born children of immigrants – have the potential for a greater political

(electoral) impact in the U.S. polity into the future due to their legal status. Because of my

interest in the historical patterns of immigrant political incorporation, taking this long view is

of interest to me. Second is the very practical issue of the difficulty in finding or generating

nationally representative datasets of undocumented immigrants sufficiently large so as to be

suitable to test my quantitative models.

CHAPTER 3

State Immigration Policy and Political Incorporation of New Legal Immigrants, 2003-

2007

Legal immigrants are future citizens, and the political incorporation of legal

immigrants is an indication of their overall incorporation into U.S. society; yet the

mechanisms by which legal immigrants experience political integration in the U.S. are less

than clear. While studies of immigrant incorporation often focus on economic incorporation

(Donato et al. 2005; van Tubergen et al. 2004) or assimilation (Alba and Nee 1997; Zhou

1997), political incorporation of legal immigrants is an additional indicator of how well U.S.

society embraces those we choose to admit as immigrants. A number of scholars note that in

contrast to other Western, English-speaking countries made up largely of immigrants and

their descendants (such as Canada, Australia, and New Zealand), the U.S. has few formal

federal integration policies targeting immigrants and allocates relatively few resources to this

endeavor (Fix and Zimmerman 1994; Bloemraad 2006). The 2000s witnessed spikes in

antiimmigrant sentiment evidenced by record numbers of anti-immigrant policies passed at

the state and local levels (Chavez and Provine 2009; Hopkins 2010; Monogon 2013), thus we

are left to question whether the failure to support full integration of immigrants into U.S.

society will contribute to continued or worsening societal divisions.

Perhaps even more importantly, citizen participation in government is necessary to

ensure representation and legitimacy in a democratic society, “Citizen participation is at the

heart of democracy” (Verba et al. 1995, 1). Legal immigrants are future citizens – they have

met U.S. government requirements for admission as legal permanent residents and are on a

legal pathway to citizenship. While we know that not all legally admitted immigrants

ultimately access citizenship, given their situation, we must assume at the outset that they are

all potential future citizens and thus potential participant in the political system. To the extent

legal immigrants are future citizens and to the extent that there are real options for these

future citizens to engage politically and civically in U.S. society before they attain the right

to vote through citizenship, the questions of why and how and under what circumstances

legal immigrants engage constitute legitimate and interesting political science scholarship.

When we consider the great social impacts that high levels of immigration are having on

contemporary society and the evidence of low levels of political engagement among some

groups of naturalized citizens these questions become even more consequential. This study

seeks to further our understanding of the factors that contribute to contemporary immigrant

political incorporation by assessing the ways in which factors known to contribute to voting

behavior among naturalized immigrants affect the non-voting political incorporation of new

adult legal immigrants. I am particularly interested in the role that immigration policy itself

plays in shaping the political incorporation of new adult legal immigrants, after taking into

account other factors such as characteristics of immigrants themselves, immigrant social

integration, political experiences in the home country, and household composition. This

single study is the first empirical chapter in a dissertation that examines the impact of

immigration policy on political incorporation of immigrants and their children from different

perspectives. To some extent this chapter serves to test the feasibility of my overall theory

and approach.

In this chapter I first review the relevant literature on immigrant political behavior.

After describing the data and methods used in the study, I present the analyses. In the

concluding section I discuss the implications of the findings on future research. My findings

largely confirm that individual factors, social factors, and home country factors affect

political knowledge of new legal immigrants in much the same ways as they do voting

behavior among naturalized immigrants. I also find that state immigration policy is

consequential for political incorporation. These findings suggest that the effects of

immigration policy on political engagement of immigrants may be a fruitful area of further

research.

Immigrant Political Behavior

For this study I draw on what is known about predictors of voting and non-voting

political engagement and incorporation among adult immigrants in the United States. The

research on political engagement among adult immigrants is sparse, and what has been

published has focused most heavily on the voting behavior of naturalized immigrants.

Political engagement or participation is “activity that has the intent or effect of influencing

government action – either directly by affecting the making or implementation of public

policy or indirectly by influencing the selection of people who make those policies” (Verba et

al. 1995, 38).

Of course legal immigrants who are not yet naturalized do not have the option of

voting, yet these proto-citizens do have options for political engagement. Although relatively

few studies have examined predictors of non-voting political engagement among immigrants

to the U.S., the few that have been completed suggest that resources, race, time in the U.S.,

generational status, involvement in transnational politics, ethnic residential concentration,

and the family all may play a role (Pearson and Citrin 2006; Merelman 1980; Keefe et al.

1979; Nee and Sanders 2001; Bloemraad and Trost 2008; Tam Cho 1999; Ramakrishnan and

Viramontes 2010; Ramakrishnan and Espenshade 2001).

In their examination of Public Policy Institute of California (PPIC) surveys from

1998 to 2002, Pearson and Citrin (2006) find variation in non-voting political participation

among non-citizen immigrants in California by race/ethnicity, with non-citizen Latinos

scoring highest (.49) on their political participation index compared to non-citizen whites

(.33) and non-citizen Asians (.30). Ramakrishnan (2006) notes that non-political civic

volunteerism among immigrant non-citizens has important implications for political

incorporation. In his analysis of the Current Population Survey Volunteer Supplement 2002,

he finds that civic volunteerism (volunteering and belonging to civic organizations) increases

with time in the U.S. and immigrant generational status and that increased ethnic residential

concentration is associated with some increase in civic volunteerism for Latinos and Asians.

Scholars have also posited that family composition is an important support in social

incorporation of immigrants that may also affect political and civic incorporation - including

extended family, non-family household members, and minor children (Merelman 1980).

Immigrants to the U.S. frequently experience households with extended families or

nonfamily household members. Often these function as a supportive factor in immigrant

family societal integration (Keefe et al. 1979). In their studies of the labor-market aspects of

immigrant incorporation, Nee and Sanders (2001) emphasize the central role that extended

family members play “as a repository of the different forms of capital that immigrants bring

with them and accumulate…The family is viewed as a key social institution providing the

basis for trust and collective action” (388). In addition to extended family members,

Bloemraad and Trost (2008), in their qualitative examination of family and household

participation in the Spring 2006 immigration protests, find evidence that the children in

immigrant households are mobilizing parents. Thus they posit a model of “dual political

socialization” whereby a bi-directional process of socialization is occurring - both the

expected parent-to-child socialization and in an unexpected child-to-parent socialization.

Hypotheses

The conceptual model I use to guide the proposed analyses relies on the previous

review of theoretical and empirical research on political engagement and incorporation of

adult immigrants as well as the role of policy in shaping political participation among policy

targets. In this study I test whether or not social constructions of immigrants present in

immigration policy have consequences for their political incorporation. The model of factors

impacting adult immigration political and civic engagement (illustrated in Figure 3.1 below)

incorporates previous empirical research starting from the broadest level and ending on the

individual level, including policy predictors, home country factors, household factors, social

incorporation, and individual factors.

Figure 3.1. Factors Impacting Adult Immigrant Political Incorporation

Immigration Policy

State Immigration Policy Visa Admission Category

I propose that influences known to predict voting behavior among naturalized

immigrants are also at work in pre-voting political incorporation of new legal immigrants.

Furthermore, I propose that the relationship between state immigration policy and

immigrants residing in states is such that immigrant residents receive interpretive messages

from the enactment of state immigration policy and the political rhetoric surrounding such

passage, whether positive or negative. I posit that such interpretive messages have the ability

to influence the non-voting political incorporation among new legal immigrants. I further

posit that state immigration policy often has a racialized component such that Asian and

Hispanic/Latino immigrants may experience higher levels of interpretive benefits and

burdens than immigrants who identify as non-Hispanic White.

Based on these expectations, I propose the following three hypotheses:

H1: Factors known to predict voting behavior of naturalized immigrants to the U.S.

will similarly predict non-voting political incorporation among recent adult legal

immigrants to the U.S.

H2: State immigration policy indices will be positively associated with non-voting

political incorporation among new legal immigrants such that higher policy indices

lead to higher levels on the outcomes and lower policy indices lead to lower levels.

Home Country Factors

Political engagement in country of origin

Civic engagement in country of origin

Voting in country of origin

Household Factors

Size of household

Children present in household

Non-family members present in households

Social Incorporation Factors

Years in US

English-speaking ability

Religious attendance in US

Employed at second interview

Individual Demographic Factors

Gender

Age

Marital Status

Education

Race/Ethnicity

H3: Among non-Hispanic Asian and Hispanic new legal immigrants, the impact of

state immigration policy indices on non-voting political incorporation will be greater

than among non-Hispanic White new legal immigrants.

Data and Methods

In this study I examine the predictors of political incorporation of recent adult legal

immigrants to the U.S. using the New Immigrant Survey 2003 (NIS-2003), a nationally

representative, longitudinal survey of immigrants recently granted legal permanent residence

(LPR) status. The baseline round of data collection from the 2003 cohort (NIS-2003-1) was

conducted in 2003, and a follow up round of data was collected from the same cohort in

2007. The NIS-2003 sample was drawn from the administrative records of new immigrants

admitted to LPR status in the seven months from May to November of 2003. The NIS-2003

adult sample, upon which this study is based, includes all immigrants 18 years and older at

admission to LPR status who have visas as principals or as accompanying spouses, and

excludes other accompanying categories such as adult children and other accompanying

relatives. The NIS-2003 designers used geographic cluster sampling, drawing a random

selection of 10 Metropolitan Statistical Areas (MSAs) and 15 remaining counties from a

sampling frame of 85 MSAs and 38 counties where 89% of all LPR immigrants entering the

U.S. in 1996-2000 resided. From 12,500 immigrants randomly selected to make up the

NIS2003 adult sample, the researchers obtained a 68.6% response rate yielding 8,573

completed adult interviews. All respondents were interviewed in their preferred language.

The NIS2003 undersamples spouses of U.S. citizens (about half of their occurrence) and

oversamples employment principals (about twice their occurrence), so the overall sample is

not necessarily reflective of the proportion of each visa subgroup in the larger population of

LPRs. These and further details of the NIS 2003 design and sampling are described in Jasso

et al. (2005).

Of the 8,573 completed adult interviews represented in the NIS-2003 data, 4,363

completed the second round of interviews in 2007. The analytical approach for this study

consists of using the publicly available NIS-2003 dataset. Because of the need for state of

residence to match the policy predictor, the analytic dataset is further limited to the subset of

respondents indicating residence in one of the six states identified in the public dataset,

2,788 in all. My unit of analysis is individual new immigrant respondents to the NIS-2003

baseline and follow up surveys, with predictor variables drawn primarily from the baseline

survey and dependent variables drawn from the follow up survey. This allows me to separate

temporally those factors conceptualized as causal from the behavior they are hypothesized to

impact. I conduct two multivariate ordered logit regressions using the ologit command in

Stata 13.1 with clustering on states. This method allows for clustering of respondents within

states, accounts for unobserved differences between states, and appropriately adjusts standard

errors producing robust standard errors. I also use sampling weights in all estimations.

Outcome Variables

The two political outcomes from the 2007 follow-up survey of 2003 NIS respondents

are Political Knowledge and Intent to Naturalize. Table 3.1 contains summary descriptive

statistics for variables and cases included in all analyses.

For the Political Knowledge outcome variable, the 2007 follow up survey asks two

political knowledge questions from which I created an additive index of political knowledge

– “Do you happen to know the name of the person who holds the following positions in U.S.

government:”… “Secretary of State?”… “Speaker of the United States House of

Representatives?”. In the U.S., individual increases in political knowledge are known to be

associated with increased political participation (Delli Carpini and Ketter 1996; Popkin and

Dimock 1999). A simple additive index then makes up the first dependent variable, treated as

an ordinal variable ranging from 0-2 based on how many of these two people the respondent

could correctly identify. Overall NIS respondents included in the Political Knowledge

analysis demonstrate a mean of 0.532 on this additive scale of political knowledge. For

the Intent to Naturalize outcome variable, the 2007 follow up survey asks a multiple choice

question from which I create an index reflecting strength of intent to naturalize – “Do you

intend to file to become a citizen of the U.S.”…“Yes, intend to file.”...

“No, do not intend to file”…“Have already filed, but not a citizen”… “Already a citizen”…

“Don’t know.” I have recoded so that 0=”No, do not intend to file” and “Don’t know”;

1=”Yes, intend to file”; and 2=”Have already filed, but not a citizen” and “Already a citizen”.

Other immigrant scholars have posited that naturalization may be viewed as a political act for

those who are not yet able to vote (Bueker 2005). Overall NIS respondents in 2007 included

in the Intent to Naturalize analysis demonstrate a mean of 0.961 on this 0-2 scale of intent to

naturalize.

Table 3.1. Descriptive Statistics

New Legal Immigrants

Variable N Mean

Min

Max

Political Knowledge Base Model

Political Knowledge

1737

0.474

0.699

State Policy Index

1737

0.302

0.840

-1.225

1.225

Family-sponsored Visa

1737

0.612

0.4861

Employer-sponsored Visa

1737

0.087

0.282

Other Visa (omitted)

1737

0.310

0.463

Political Engagement Home Country

1737

0.410

1.030

Vote in Home Country

1737

0.042

0.200

Number in Household (mean)

1737

3.985

2.001

Children Present in Household

1737

0.642

0.480

Non-Family Members Present in HH

1737

0.189

0.391

Years in US

1737

3.983

4.637

English speaking (well/very well)

1737

0.412

0.492

Religious Attendance in US

1737

0.609

0.488

Employed at second interview

1737

0.681

0.466

Female

1737

0.598

0.490

Age (mean)

1737

38.574

12.268

Married

1737

0.818

0.386

Education at US entry (mean years)

1737

11.835

4.861

Asian (non-Hispanic)

1737

0.264

0.441

Black (non-Hispanic)

1737

0.039

0.193

Hispanic

1737

0.519

0.500

White (non-Hispanic) (omitted)

1737

0.196

0.697

Variable

Mean

Min

Max

Intent to Naturalize Base Model

Intent to Naturalize

1684

0.961

0.455

State Policy Index

1684

0.291

0.833

-1.225

1.225

Family-sponsored Visa

1684

0.567

0.496

Employer-sponsored Visa

1684

0.101

0.302

Other Visa (omitted)

1684

0.326

0.469

Political Engagement Home Country

1684

0.392

1.001

Vote in Home Country

1684

0.045

0.207

Number in Household (mean)

1684

4.108

1.998

Children Present in Household

1684

0.664

0.473

Non-Family Members Present in HH

1684

0.208

0.406

Years in US

1684

4.282

4.885

English speaking (well/very well)

1684

0.382

0.486

Religious Attendance in US

1684

0.604

0.489

Employed at second interview

1684

0.674

0.469

Female

1684

0.579

0.494

Age (mean)

1684

39.677

12.709

Married

1684

0.788

0.409

Education at US entry (mean years)

1684

11.467

4.927

Asian (non-Hispanic)

1684

0.268

0.443

Black (non-Hispanic)

1684

0.042

0.201

Hispanic

1684

0.523

0.500

White (non-Hispanic) (omitted)

1684

0.194

0.396

Predictor Variables

I chose predictor variables that represent the five broad categories of predictors of

political engagement reviewed previously in this paper and reflected in Figure 1 –

immigration policy, home country, household makeup, social incorporation, and individual

demographics. Summary statistics from the NIS-2003 for each of the predictor variables are

detailed in Appendix 3.A.

Table 3.1. Descriptive Statistics (Cont’d

)

New Legal Immigrants

STATE IMMIGRATION POLICY: I describe in detail the construction of the state

immigration policy predictor in the Introduction of this dissertation. For these analyses I

make use of the same positive-to-negative cumulative state policy index; however, because

the dataset used in this chapter has baseline data collected in 2003 and outcomes collected in

2007, I re-calculate the state policy index to include only state policies enacted 2003-2007.

Additionally, because (as noted above) my analysis includes only new legal immigrants

residing in the six largest immigrant-receiving states, in standardizing the state policy index I

include only those six states. The final standardized 2003-2007 cumulative policy index used

for this analysis has a mean of 0 and standard deviation of 1, and a range from -1.225 to

1.225. To ensure that this variable is capturing actual exposure to the state immigration

policy to the greatest extent possible, analysis is limited to those respondents reporting the

same state of residence in both 2003 and 2007; overall 91% of 2007 respondents reported

living in the same state as they had in 2003.

VISA CATEGORY: Visa category is a conceptually and empirically meaningful

variable. Federal immigration policy creates “selection rules” which may be important in

understanding the attitudes and behavior of immigrants (Jasso et al. 1997). The

employersponsored visa may generally be viewed as providing greater material and

interpretive resources to the immigrant, whether principal or accompanying spouse. These

immigrants generally have higher levels of education and are most often immigrating

through a highlyskilled visa category, while family-sponsored visa categories may generally

be viewed as providing a lower level of resources (Woodrow-Lafield 2004; Kanjanapan

1995). In contrast, refugees and asylees are seeking refuge from persecution, and are

provided higher levels of material benefits through government-sponsored resettlement

programs (Ralson 2012). In both situations the visa category is a social space that both

“selects” certain immigrant characteristics such as education and political history, and

confers additional benefits or burdens based on the category. My intention in considering

visa category is not to

disentangle which effects are based on selected versus conferred characteristics, but rather to

control for both so as to avoid confounding potential visa category effects with those of my

main predictor of interest.

I construct three mutually exclusive predictor variables to measure the potential effect

of visa category at entry to LPR status on political incorporation of new immigrants. The

measures included here are based on respondent self-report during the NIS interview. The

first visa category measure in this category is whether or not the respondent’s visa is an

employer-sponsored visa, with yes=1 and no=0, this includes principal and accompanying

family members entering on an employer-sponsored visa. The second visa category measure

in this category is whether or not the respondent’s visa is a family-sponsored visa, with yes=1

and no=0, this includes principal and accompanying family members entering on a family-

sponsored visa. The third and final measure in this category is “other,” indicating that the

respondent’s visa is neither employer-sponsored nor family-sponsored, coded with yes=1 and

no=0. Respondents fitting into this final “other” category are admitted through a collection of

non-employer and non-family based visa types including diversity immigrants, refugees, and

other special categories not otherwise specified. I enter family-based and employer-based

visa types in the models so that “other” visa type is the excluded category. So the effects of

the family-based and employer-based visa categories are measured against the “other” visa

category. Since previous scholarship has found that employer-based immigration both

indicates and confers resources helpful to political incorporation

(Enchautegui 2013) and refugees demonstrate higher rates of naturalization (Bloemraad

2006a), I anticipate that those respondents entering through the family-based visa category

will demonstrate lower levels of political incorporation than those entering through

employer-based and “other” visa categories.

HOME COUNTRY PREDICTORS: Three predictor variables represent home

country political factors that may affect political engagement in the United States. I created

an additive scale to measure for political engagement in home country from a series of seven

NIS survey questions about political activity, “While living outside the United States did you

ever”… “Talk to anyone about a local or national problem?”… “Sign a petition regarding an

issue that concerns you?”… “Contact a public official about an issue that concerns you?”…

“Attend a public meeting about an issue that concerns you?”… “Go to any meetings, rallies,

speeches, or dinners in support of a political candidate?”… “Work for a party or political

candidate, either as a volunteer or for pay?”… “Contribute money to an individual political

party, candidate, or organization?”. I include the resulting additive scale as a continuous

measure with possible values from 0-7. Based on previous research on the effect of home

country on political engagement in the U.S. (Ramirez and Felix 2011), I expect that those

respondents reporting more political engagement in their home country will demonstrate

higher levels of political engagement in the U.S.

I also include here a measure of voting in home country since moving to the United

States. The survey asks “While living in the United States, have you voted in any election

held in your country of origin?”. I code this variable as binary, 1=Yes and 0=No, excluding

Refused and Don’t Know responses. Previous research on dual citizenship and voting in

home country is mixed (Cain and Doherty 2006; DeSipio 2006), thus I expect that voting in

home country following moving to the U.S. will be negatively associated with political

incorporation.

HOUSEHOLD PREDICTORS: I choose three predictor variables to measure

household factors. Emerging research is pointing to the role that household members, both

family and non-family, may play in the political incorporation of immigrants (Keefe et al.

1979, Bloemraad and Trost 2008). The NIS survey contains a series of questions asking

about each member of the household, including relationship to respondent. From these

questions I am able to construct a series of measures to indicate members of the household. I

include total number in household (including respondent) as a continuous measure, and two

0/1 binary variables to indicate the presence of child/children (including adopted, step, and

foster children) and non-family household members. Based on the limited previous research

in this area (Merelman 1980; Bloemraad and Trost 2008), I expect that respondents with

more household members, as well as those with children and non-family members in the

household, will demonstrate higher levels of political incorporation than those with fewer

household members and no children or non-family household members.

SOCIAL INCORPORATION PREDICTORS: Four predictor variables represent

social incorporation in the U.S. I include years in the U.S. as a continuous measure. Ability to

speak English is taken from an NIS baseline survey question in which the respondent is

asked to rate their level of English language ability, “How well would you say you speak

English?” on a four point scale from “very well” to “not at all.” I include as a 0/1 binary

variable where well/very well=1 and not well/not at all =0. Religious attendance in the U.S.

is taken from an NIS baseline survey question in which the respondent is asked, “Since

becoming a permanent resident, how many times have you attended religious services?”. I

include as a 0/1 binary variable where any report of attending religious services in the U.S.=1

and no attendance=0. Employment status is taken from an NIS 2007 follow up survey

question in which the respondent is asked, “Are you working now, temporarily laid off,

unemployed and looking for work, disabled and unable to work, retired, a homemaker, or

what?” I include as a 0/1 binary variable where report of working now=1 and all other

response options=0. Note that this employment variable is the only predictor variable I have

taken from the 2007 follow up survey; all others are from the baseline survey. I reason that

because in 2003 respondents were new immigrants, many had not yet had the opportunity to

gain the social integration benefit of employment that I seek to capture with this variable.

This is borne out by examining the baseline and follow up survey data, with 58.3% of

respondents in the entire baseline survey indicating that they were currently working

compared with 72.8% at the follow up survey.

Based on previous research (Ramakrishnan and Espenshade 2001; Ramakrishnan

2005), I anticipate that those respondents with more years in the U.S., those with higher

levels of English ability, those who attend religious services in the U.S., and those currently

in the workforce to demonstrate higher levels of political incorporation than those with fewer

years in the U.S., those with lower levels of English ability, those who do not attend religious

services in the U.S., and those not currently employed.

DEMOGRAPHIC PREDICTORS: Five main predictor variables fall under the broad

heading of demographic factors, all of which have been used as standard measures in

previous studies of political engagement. I measure gender as a 0/1 binary variable where

female=1 and male=0, age in years as a continuous variable, marital status as a 0/1 binary

variable where married or living with partner as married=1 and any other status=0, and

education in years as a continuous variable. I expect female respondents, those who are

married, and those with higher levels of education to demonstrate higher levels of political

incorporation than males, those who are unmarried, and those with lower levels of education.

I include self-reported race/ethnicity variables measured as a series of 0/1 binary variables,

one each for non-Hispanic Asian, non-Hispanic Black, Hispanic, and non-Hispanic White. I

code the four racial/ethnic categories so that they are mutually exclusive starting with

Hispanic ethnicity (all respondents who reported Hispanic ethnicity are included in this

category) so that Asian race indicates non-Hispanic Asian, Black race indicates non-Hispanic

Black, etc. I enter Asian, Black, and Hispanic in the models so that they are compared against

the excluded White category. Based on previous research that finds that immigration policy is

often racialized in the contemporary environment as Asian or Hispanic (Martin and Duignan

2003; Rumbaut 2009; Schmidt et al. 2010; Bean et al. 2013), I expect that the impact of state

policy on political incorporation among new legal immigrants will be stronger among Asians

and Hispanics.

Findings

To test my hypotheses I make use of all available NIS data with complete responses

on the dependent and predictor variables. I test two models for each of the two political

incorporation outcomes: first a base model using all predictor variables but no interactions,

and second a complete model in which I interact the state immigration policy index with

race/ethnicity. These estimation results are detailed in Tables 3.2 and 3.3. Here I interpret the

odds ratios and some substantive impacts in turn for each category of predictors on each of

the two dependent variables. Finally, I interpret the impacts of state immigrant policy by race

interaction terms.

The direct effect of my primary predictor of interest, state immigration policy index,

is not significant in predicting either of the political integration outcomes here in either base

model. However, in both Model 2 and Model 4 where I test the interactions between state

immigration policy index and race/ethnicity we see evidence of both direct and indirect

effects, we may interpret the interaction terms as the impact state policy is having on that

racial/ethnic population as compared with the excluded White population. Interestingly, these

effects are in opposite directions for political knowledge outcome and intent to naturalize.

State immigration policy is significantly and negatively predictive of political knowledge and

strongly positively predictive of intent to naturalize such that each single point increase on

the state immigration policy index is associated with a 25.8% reduction in the odds of having

the highest level of political

Table 3.2. NIS Political Knowledge Ologit Models

Odds Ratios & 95% Confidence Intervals

Model 1

Model 2

State Immigration Policy

0.922

0.742**

[0.773, 1.100]

[0.609, 0.904]

State Policy X Asian

2.662*

[1.006, 7.043]

State Policy X Black

4.099** [1.439,

11.68]

State Policy X Hispanic

1.623

[0.569, 4.632]

Admission Category

Family-Sponsored Visa

0.884

0.880

[0.780, 1.002]

[0.773, 1.001]

Employer-Sponsored Visa

1.219**

1.250***

[1.080, 1.375]

[1.124, 1.389]

Home Country Factors

Political Engagement in Home Country

1.175***

1.178***

[1.076, 1.283]

[1.088, 1.276]

Vote in Home Country

1.547

1.572

[0.948, 2.525]

[0.971, 2.542]

Household

Number in Household

0.985

0.982

[0.922, 1.052]

[0.920, 1.049]

Child/Children Present in HH

0.988

1.000

[0.626, 1.559]

[0.634, 1.577]

Non-Family Household Members

0.802***

0.801***

[0.742, 0.868]

[0.737, 0.871]

Social Incorporation

Years in U.S.

1.012

1.014

[0.993, 1.031]

[[0.997, 1.032]

English speaking ability

1.332**

1.327**

[1.089, 1.628]

[1.096, 1.607]

Religious Attendance in U.S

0.784***

0.806***

[0.682, 0.902]

[0.710, 0.915]

Employed at second interview

0.964

0.956

[0.757, 1.227]

[0.748, 1.222]

Demographics

Female

0.469***

0.465***

[0.418, 0.525]

[0.416, 0.520]

Age

1.013***

[1.007, 1.020]

Married

1.644**

1.637**

]1.178, 2.296]

[1.173, 2.284]

Education

1.119***

1.118***

[1.093, 1.145]

[1.090, 1.147]

Asian

0.998

0.505*

[0.618, 1.612]

[0.293, 0.870]

Black

3.258***

1.570

[1.840, 5.768]

[0.843, 2.921]

Hispanic

0.500***

0.356**

[0.375, 0.665]

[0.165, 0.772]

1737

*p<.05, **p<.01, ***p<.001; pweights applied

Table 3.3. NIS Intent to Naturalize Ologit Models

Odds Ratios & 95% Confidence Intervals

Model 3

Model 4

State Immigration Policy

1.064

1.738**

[0.910, 1.244]

[1.201, 2.514]

State Policy X Asian

0.268*

[0.0814, 0.884]

State Policy X Black

0.471

[0.0307, 7.223]

State Policy X Hispanic

0.190*** [0.108,

0.333]

Admission Category

Family-Sponsored Visa

0.515***

0.522***

[0.357, 0.745]

[0.367, 0.744]

Employer-Sponsored Visa

0.484***

0.487***

[0.363, 0.647]

[0.348, 0.681]

Home Country Factors

Political Engagement in Home Country

0.746***

0.743***

[0.676, 0.825]

[0.671, 0.823]

Vote in Home Country

0.864

0.863

[0.488, 1.530]

[0.484, 1.541]

Household

Number in Household

0.983

0.981

[0.931, 1.037]

[0.930, 1.035]

Child/Children Present in HH

1.208

1.230

[0.848, 1.721]

[0.880, 1.720]

Non-Family Household Members

0.848

0.853

[0.624, 1.154]

[0.618, 1.179]

Social Incorporation

Years in U.S.

1.018

1.019

[0.994, 1.044]

English speaking ability

1.071

1.109

[0.906, 1.266]

[0.891, 1.380]

Religious Attendance in U.S

1.071

1.048

[0.888, 1.291]

[0.846, 1.299]

Employed at second interview

1.318

1.308

[0.953, 1.823]

[0.952, 1.797]

Demographics

Female

1.040

1.035

[0.649, 1.669]

[0.603, 1.642]

Age

0.992

0.991

[0.981, 1.004]

[0.980, 1.003]

Married

0.702***

0.715**

[0.579, 0.851]

[0.577, 0.885]

Education

1.085***

1.082***

[1.050,1.121]

[1.046, 1.118]

Asian

1.407

3.475*

[0.714, 2.772]

[1.189, 10.15]

Black

1.463

2.802*

[0.725, 2.951]

[1.011, 7.769]

Hispanic

1.604

4.977***

[0.803, 3.203]

[2.845, 8.707]

1684

*p<.05, **p<.01, ***p<.001; pweights applied

knowledge (Model 2) and a 73.8% increase in odds of having the highest level of intent to

naturalize (Model 4). In terms of the indirect effect of state immigration policy via

race/ethnicity, we see that state policy is strongly positively associated with greater political

knowledge among Asian and Black respondents (2.6 times higher odds of having the highest

level of political knowledge for Asians and 4.1 times for Blacks than for Whites). For the

intent to naturalize outcome we see the policy and race interaction terms are strongly

negative, indicating that state policy produces a 73.2% reduction in the odds of having the

highest level of intent to naturalize among Asian respondents and a 81.0% reduction among

Hispanic respondents, both compared with the excluded White category.

For visa category, immigrants with family-sponsored visa category show no

significant difference on political knowledge, but significantly reduces the odds of having the

highest level of intent to naturalize, with immigrants with family visas demonstrating 48.5%

lower odds in Model 3 (base model) and 47.8% lower odds of having the highest level of

intent to naturalize in Model 4 (with interactions), when compared with the excluded “other”

visa category. Those immigrants with employer-sponsored visas demonstrate significantly

higher odds of having the highest level of political knowledge scores significantly lower odds

of having the highest level of intent to naturalize than those with the “other” visa types.

Those with employer-sponsored visas 21.9% and 25.0% higher odds of having the highest

level of political knowledge (Model 1 and Model 2) and 51.6% and 51.3% lower odds of

having the highest level of intent to naturalize (Model 3 and Model 4) compared with the

excluded “other” visa type that includes refugees.

Among home country factors, political engagement in home country is significantly

and positively associated with political knowledge, and significantly and negatively

associated with intent to naturalize. Political engagement in the home country prior to

coming to the U.S. is significant and positively associated with political knowledge,

increasing the odds of having the highest level of political knowledge by 17.5% (Model 1)

and 17.8% (Model 2) at each level of political engagement in the home country. In contrast,

political engagement in home country prior to coming to the U.S. is significantly and

negatively associated with intent to naturalize, decreasing the odds of having the highest

level of intent to naturalize by 25.4% (Model 3) and 25.7% (Model 4) at each level of

political engagement in the home country.

None of the household factors are significantly predictive of intent to naturalize, and

the presence of non-family household members is the only household factors significantly

predictive of political knowledge. Respondents with non-family household members have

approximately 20% lower odds of having the highest political knowledge score than those

who do not, and this relationship is the same in Model 1 and Model 2 specifications.

Considering next the category of social incorporation predictors, none are significantly

predictive of intent to naturalize, while English-speaking ability and religious attendance are

significantly predictive of political knowledge. As expected English-speaking ability is

significant and positively associated with political knowledge, such that those immigrants

reporting higher levels of English ability are about 1.3 times more likely to have the highest

level of political knowledge than are respondents reporting lower levels of

English ability, this relationship is the same in Model 1 and Model 2 specifications.

Additionally, those immigrants reporting regular religious attendance in the U.S. have 17.6%

and 19.4% lower odds (Model 1 and Model 2) of having the highest level of political

knowledge compared with immigrants who do not report regular religious attendance.

As concerns demographic predictors, all variables in this category are significantly

associated with political knowledge, while only education and race/ethnicity (Asian, Black

and Hispanic) are significantly associated with intent to naturalize. Being female decreases

the odds of having the highest level of political knowledge by 53.1% and 53.5% (Model 1

and Model 2), each year of age increases these odds by 1.0% (Model 1 and Model 2), being

married increases these odds by about 64%, and each year of education attained prior to

entering LPR status increases these odds by 12%. Similarly, each year of education increases

the odds of having the highest level of intent to naturalize by about 8% (Model 3 and Model

4).

After controlling for the indirect effect of state immigration policy by race (Models 2

and 4), Asian immigrants have 49.5% lower odds of having the highest level of political

knowledge and 3.5 times higher odds of having the highest level of intent to naturalize than

White immigrants. Hispanic immigrants demonstrate a similar pattern with 64.4% lower

odds of having the highest level of political knowledge and 4.98 times higher odds of having

the highest level of intent to naturalize than White immigrants. Black immigrants

demonstrate 3.2 times the odds of having the highest level of political knowledge in Model 1,

but this effect drops from significance in Model 2 once the indirect effects of state

immigration policy and race are accounted for. In terms of intent to naturalize, Black

immigrants demonstrate 2.8 times higher odds of having the highest level of intent to

naturalize than White immigrants (Model 4). Race/ethnicity are significantly associated with

political knowledge and intent to naturalize in the presence of other predictor and control

variables.

Discussion

I find partial support for my first hypothesis regarding factors known to predict

voting behavior of naturalized immigrants to the U.S. as concerns pre-voting political

incorporation of new legal immigrants, specifically as concerns the political knowledge

outcome. Political engagement in home country, household factors, social incorporation

indicators and demographic factors appear in both Models 1 and 2 to have similar predictive

power and direction that we see in previous immigrant incorporation literature

(Ramakrishnan and Espenshade 2001; Cain and Doherty 2006; Ramakrishnan 2005). This

appears to indicate that when examining pre-voting political behavior among immigrants not

yet eligible to vote, political knowledge may be a reasonable proxy for political behavior.

The findings point to a different story as concerns the intent to naturalize outcome, seeming

to indicate that different processes are at work here. Here we see that none of the household

or social incorporation factors are significantly associated with the outcomes, and among

demographic factors only education and race are significant. Political engagement in country

of origin is positively associated with political knowledge, consistent with previous literature

suggesting that political engagement as a learned skill is transferrable for immigrants from

home country to the U.S. (Portes, Escobar and Arana 2009; Ramakrishnan and Viramontes

2010; Ramirez and Felix 2011). However, it is negatively associated with intent to naturalize,

suggesting that that high levels of political engagement in the home country prior to entering

LPR status in the U.S. are a proxy for high levels of attachment to the home country that are

then reflected in a lower intent to naturalize.

Visa category behaves largely as expected as concerns political knowledge, indicating

that employer-sponsored visa contains resources consistent with greater political

incorporation. The negative relationship between both family- and employer-sponsored visa

with intent to naturalize indicates that those in the “other” visa category, which includes

refugees, are much more likely to report higher scores on the intent to naturalize scale. This

makes some sense if we consider that refugees are often fleeing political persecution and

violence and often will not have the option to return to their home country (Bloemraad

2006a).

I find partial support also for my second hypothesis regarding the association between

state immigration policy and the political incorporation outcomes examined here. State

immigration policy is strongly and positively associated with political knowledge among

Asian and Black immigrants, suggesting that beneficial (inclusive) policy is politically

mobilizing among these two groups. While the same policy is strongly and negatively

associated with intent to naturalize among Asian and Hispanic immigrants, suggesting that

punitive (exclusive) policy is impacting intent to naturalize among these two groups.

Finally, I find partial support for my third hypothesis regarding the impact of state

immigration policy among Asian and Hispanic immigrants being larger than among White

immigrants. Clearly in Model 2 we see the positive association between state immigration

policy among Asians is significantly larger than among Whites, and this relationship is true

among Black immigrants as well although not among Hispanic immigrants. While in Model

4 we see the relationship between state immigration policy and the outcome is stronger

among both Asians and Hispanics, as hypothesized, than among White immigrants.

Legal immigrants are now entering the U.S. at numbers greater than 1.0 million each year,

and immigrants are gaining citizenship at greater than half a million each year (Rytina 2013).

This population represents a numerically and substantively important portion of potential

future voters, and understanding the ways in which they are incorporated into the

U.S. polity is important to understanding their potential future overall impact on U.S. society.

If we agree that citizen participation in government is necessary to ensure representation and

legitimacy in a democratic society, then we should be concerned with the outcome of

political incorporation of legal immigrants.

This study is one of only a handful that examine factors in addition to individual-level

factors that contribute to or inhibit political incorporation of new legal adult immigrants, and

one of the only studies to consider how immigration policy itself contributes to or inhibits

such political incorporation. In addition to integrating the current major theories in the field

by incorporating various predictors in the empirical analysis, I have extended existing

knowledge by explicitly theorizing and modeling the role of immigration policy as a causal

factor in the political incorporation of new adult immigrants. Taken as a whole, the results

point to the need for further research to examine the independent role that immigration policy

plays in impacting the political incorporation of legal adult immigrants.

CHAPTER 4

The Impact of State Immigration Policy on Voting among Naturalized Immigrants

Naturalized citizens make up a statistically and substantively important part of the

present and future U.S. electorate, yet many questions remain as to the characteristics and

circumstances that either contribute to or suppress their participation in voting. In 2000 there

were 12.5 million naturalized citizens present in the U.S., and by 2012 that number had

grown almost 60% to 17.5 million (Gibson and Jung 2006; Grieco et al. 2012). Naturalized

citizens represented an estimated 7% of the voting population in the 2012 presidential

election (U.S. Census Bureau 2012), and their number is only expected to grow in the future.

This study seeks to further our understanding of the factors that impact voting among

naturalized citizens, with a particular focus on the impact of state immigration policy.

During the 2000s, states in the U.S. enacted a record number of policies related to the

immigrants within their borders. Some of these policies had the stated objective of driving

undocumented immigrants from the state, while others were designed to integrate new

immigrants into the communities and civic life of the state (Immigration Policy Center 2012;

Progressive States Network 2008). Scholars examining the determinants of state immigration

policymaking during the 2000s have variously found that conservative ideology, Republican

partisanship, special interests, and economic pressures contribute to state anti-immigration

policymaking (Zungher 2014; Marquez and Schraufnagel 2013; Monogon 2013;

NicholsonCrotty and Nicholson-Crotty 2011; Chavez and Provine 2009). Of greater

relevance for this study, however, is the finding that the growth rates of Hispanic, Asian, and

foreign-born populations have also played a substantial role in state enactment of

immigration policy independent of the size of undocumented immigrant populations

(Marquez and Schraufnagel 2013; Boushey and Luedtke 2011; Ybarra, Sanchez and Sanchez

2014).

This chapter examines the relationship between state immigration policy and voting

among naturalized immigrants. My primary research question is whether state immigration

policymaking has been mobilizing or demobilizing for naturalized immigrants. This study

extends existing scholarship in three important respects. First, I move the examination of

immigration policy and immigrant mobilization beyond Latino voters to include

consideration of effects across four major racial/ethnic subgroups of naturalized immigrants.

Second, I make use of a multilevel modeling technique that allows not only for nesting of

individuals within states but for estimation of the random effects of immigration policy at the

state level. Lastly, and most importantly, this study explicitly applies the theories of social

construction and policy feedback to the substantive issue of state immigration policy. These

public policy theories predict that public policy as an active institutional structure may

influence the political engagement of its target population via both material and interpretive

mechanisms. While previous scholarship has found that immigration policy often contains

both positive and negative social constructions of immigrants (Reich and Barth 2010;

Newton 2008; Coutin 1998), to date, none of the studies in this theoretical vein have

provided an empirical test of the impact of policies on voting among immigrants.

Immigrant Political Participation

Traditional models of political behavior focus on resources including income,

education, civic skills and social capital as the primary predictors of voting and other

political participation at the individual level. Brady, Verba and Schlozman (1995; Verba et al.

1993) found that an individual’s position in the economy generates resources available for

political participation. They consider resources to be as broad as education, time, money,

English proficiency, and resources derived from involvement in non-political institutions (for

example organizational skills and interpersonal networks).

As detailed further in Chapter 2 of this dissertation, one of the greatest limitations of

the dominant theories of political participation is their focus on individual-level factors as

predictors of political participation in the absence of contextual and structural factors.

Scholarship on political participation among immigrants, people of color, and other groups

with historically lower levels of participation has revealed the importance of contextual

factors. Jan Leighley critiques the standard socioeconomic model in its assumption that

“…that individuals think and act politically independent of their social and political context”

(Leighley 2001, 172). Scholarship on immigrant political participation has thus increasingly

incorporated structural and contextual factors as predictors including neighborhood level

coethnic population (Tam Cho et al. 2006; Ramakrishnan 2005; Ramakrishnan and

Espenshade 2001), descriptive representation and immigrant social safety net policies (Logan

et al. 2012), and voting regulations (Logan et al. 2012; Ramakrishnan 2005; Ramakrishnan

and

Espenshade 2001; Jones-Correa 2001).

Some research finds that anti-immigrant policymaking has been politically

mobilizing for some immigrants under some circumstances. This phenomenon has been most

studied relative to two policies in particular - California’s Proposition 187 and the federal

Sensenbrenner bill. Taken together these studies suggest that anti-immigrant policymaking,

or the threat of such policies even before implementation, can lead to increased political

participation among Latino immigrants. Participation outcomes studied include naturalization

(Pantoja et al. 2001), voting (Barreto et al. 2005), and mass protests (Rim 2009). Proposition

187 passed in California in 1994, barring undocumented immigrants from receiving social,

educational, and health services, but it was never implemented (Jacobson 2008). Pantoja et

al. (2001) found that the passage of Proposition 187, and the heated political environment

surrounding its passage, was significantly and positively associated with increased voting in

1996 among Latino immigrants in California who naturalized in that period, while they

found no such association for voting in 1996 among naturalized Latino immigrants in two

other comparison states, Florida and Texas, without extreme anti-immigrant ballot measures.

Barreto et al. (2005) found that an increasing number of Latinos, both naturalized and

nativeborn, entering the electorate during the highly anti-immigrant political environment of

California of the mid-1990s drove the large growth in Latino voting observed in California

between 1996-2000. H.R. 4437 (a.k.a. the Sensenbrenner bill) similarly did not ultimately

become law; however, this severely punitive anti-immigrant congressional legislation did

pass the House in December 2005 and provoked the mobilization of millions of immigrants

and their supporters in numerous large protest rallies across the country in the spring 2006

(Jonas 2006). Research on both of these instances of threatening policy reveals the key role

played by immigrant and ethnic mobilizing organizations in those geographic areas where

immigrant mobilization (voting and non-voting) occurred (Ramakrishnan 2005; Barreto et al.

2009; Rim 2009).

Racialized Immigration Policy

Finally, I incorporate the history of immigration policy as highly racialized into my

theorizing about the potential effects of state-level immigration policy on the political

behavior of naturalized immigrants. Federal immigration policy has played a central role

historically in racializing immigrants; this is most evident in immigration policies designed

to exclude immigrants from Asian countries such as the Chinese Exclusion Act of 1882, the

Gentlemen’s Agreement of 1907, and the McCarran-Walter Act of 1952. Although these

separate pieces of federal immigration policy spanned over a half-century, all were driven by

high levels of racialized anti-immigrant sentiment of their time (Tichenor 2002; Ngai 2004).

Changes in federal immigration policy in the mid-1960s led to waves of new immigrants

from previously-excluded Asian countries entering through highly-skilled categories, and

from Latin American countries entering through family-reunification categories and through

undocumented immigration (Tichenor 2002; Massey and Pren 2012). With contemporary

immigration since the 1970s dominated by immigration from Latin America and Asia, such

that the majority of the current foreign born population is from those regions (Grieco et al.

2012), contemporary immigration policy contexts are often racialized. Substantial

scholarship speaks to the continued racialization of immigrants from Asia and Latin America

as non-White in contemporary America (Martin and Duignan 2003; Rumbaut 2009; Schmidt

et al. 2010; Bean et al. 2013).

Hypotheses

This study is placed in the tradition of social construction and policy feedback

theories. I propose that the relationship between state immigration policy and immigrants

residing in states is such that immigrant residents receive interpretive messages from the

enactment of state immigration policy and the political rhetoric surrounding such passage,

whether positive or negative. I posit that such interpretive messages have the ability to

influence the probability of voting among naturalized immigrants. I further posit that state

immigration policy often has a racialized component such that non-White immigrants may

experience higher levels of interpretive benefits and burdens than immigrants who identify as

White.

Based on these expectations, I propose the following two hypotheses.

H1: State immigration policy will be positively associated with voting among naturalized

immigrants such that more beneficial policy lead to higher probability of voting and more

punitive policy leads to lower probability of voting.

H2: Among non-Hispanic Asian/PI, non-Hispanic Black, and Hispanic naturalized

immigrants, the impact of state immigration policy on voting will be greater than among non-

Hispanic White naturalized immigrants.

Data and Methods

In testing these hypotheses, I make use of the Current Population Surveys (CPS)

basic survey and the CPS Voting and Registration supplement for 2004-2012. The CPS basic

survey is a nationwide survey of the U.S. population age 15 and over administered monthly

by the Census Bureau with a focus on labor force participation. The CPS Voting and

Registration supplement is a supplemental survey administered in November of

evennumbered years to collect data on voting and registration in the general election among a

subset of those respondents to the basic CPS survey who are U.S. citizens and at least 18

years of age. Total respondents to the Voting and Registration supplement during the years of

interest are over 100,000 per year, of which approximately 1,000-5,200 per year are

naturalized immigrants. It is this naturalized immigrant subsample that forms the basis of the

dataset constructed for this research. To the pooled individual-level responses from the CPS

Voting and Registration supplement naturalized immigrant subsample for the general election

in 2004, 2006, 2008, 2010 and 2012, I have merged country-of-origin and state-level

variables (described below) to construct the complete analyzable dataset. After accounting

for missing data on any of the predictor variables, I am left with a complete dataset of 15,789

naturalized immigrant respondents identifying as one of the four major racial/ethnic groups

with which to conduct my analyses.

With these hypotheses I seek to understand the role of state-level immigration policy

on individual-level behavior among naturalized immigrants from the four major racial/ethnic

groups while controlling for both state-level, voting year, and individual-level factors. I can

best accomplish this by estimating variations in outcomes at multiple levels, thus I make use

of a series of multi-level mixed-effects with crossed random-effects logistic regression

models to account for the hierarchical relationship between respondents, the states where

they reside and presidential/non-presidential voting years. Mixed-effects multi-level models

with random effects appropriately allow me to estimate the significance and magnitude of the

effects of these hierarchical relationships (Hox 2010). I specify the crossed random-effects at

the second level, using both states and presidential/non-presidential voting years, as I am

assuming that the random effects at these levels are not nested within one another but instead

are crossed, meaning that the effect due to state is the same regardless of

presidential/nonpresidential year, and conversely the effect due to presidential/non-

presidential year is the same regardless of state. In order to test my specific hypotheses I

make use of four such models, one for each of the four racial/ethnic groups present in the

CPS Voting and registration supplement.

For all statistical modeling I make use of Stata/IC 13.1 using the multilevel mixed

effects logistic regression command with QR decomposition of the variance-components

matrix to aid convergence, meqrlogit. This estimation technique employs adaptive Gaussian

quadrature when more than one integration point is specified. I have followed the

modelbuilding procedures described by Rabe-Hesketh and Skrondal (2012) to achieve three

to four integration points for each multi-level crossed random-effects model reported.

While Stata does not support reporting intraclass correlation in a crossed-random effects

model, I may get a sense a priori of the proportion of variation accounted for at the individual

level as compared to higher-level predictors. A simple examination of the intraclass

correlation for intercept-only models of presidential and non-presidential years, aggregating

all four racial/ethnic groups together and partitioning the variance between individual and

state levels, reveals that in non-presidential years, approximately 4.0% of the total variation

in the probability of voting may be accounted for at the state level as compared to about

3.7% in presidential years. While small, this proportion of variance accounted for at higher

levels is consistent with other studies of individual behavior (Hox 2010). The likelihood ratio

tests for each of the two intercept-only models are significant at p<.0001, indicating that the

results are statistically significantly different than the same test done at using simple (single-

level) logistic regression. Taken together these results provide a priori justification for

utilizing multi-level models rather than simple (single-level) logistic regression models.

Dependent Variable

I model the dependent variable as self-reported voting in the general election on the

part of CPS respondents in 2004, 2006, 2008, 2010, and 2012. I chose these years to

maximize coverage and variation in the focal predictor of interest, state immigration policy,

as well as to maximize the number of voting years. The dependent variable is a simple binary

variable, coded as 1 if the respondent reported voting in the most recent general election and

0 if they reported not voting. Respondents providing all other response options (“don’t

know” and “refused” are excluded from the analyses.

Random-Effects Predictor Variables

In the crossed random-effects portion of the model I specify states as a primary

second-level variable, indicating that all first-level (individual-level) responses are grouped

within states. I estimate random slopes for the primary predictor variable, state immigration

policy index and for co-ethnic population concentration at the state level. In addition,

presidential year is designated as a crossed random-effects random variable to control for the

increased voter turnout associated with presidential years.

STATE IMMIGRATION POLICY: For the focal predictor variable, state

immigration policies, I have created an index of state immigration policies that includes both

punitive and beneficial policies. Consistent with other scholarship of contemporary state

immigration policy (Chavez and Provine 2009; Monogon 2013; Marquez and Schraufnagel

2013; Zingher 2014), I make use of the National Conference of State Legislature (NCSL)

dataset on state immigration policy to identify enacted state immigration legislation from

2005-2012. As detailed in Chapter 1, each piece of enacted legislation in the NCLS dataset

2005-2012 has been coded as to direction – neutral, beneficial to immigrants, or punitive to

immigrants - and omnibus legislation has been disaggregated into separate policies. This

variable is limited to enacted policies passed by the legislature, so it excludes bills introduced

and not passed and policy action taken independently by the executive or judicial branch.

Since NCSL reports no data on enacted state-level immigration policies prior to 2005, for

2003 and 2004 I make use of a report from the Progressive States Network (2008) that

identifies integrative and punitive policies enacted by each of the 50 U.S. states from 1997-

2008. For these policies I rely on policy direction identified in the publication itself.

I agree with other authors using these data (for example see Nicholson-Crotty and

Nicholson-Crotty 2011) that combining punitive and beneficial state immigration policy into

a single index makes sense both theoretically and empirically. Theoretically, research on state

welfare and economic development policies demonstrate that in the absence of strict control

over who may enter and leave their local jurisdictions, state and local policymakers make use

of a mix of policy burdens and benefits to either attract or repel certain populations (Bailey

and Rom 2004; Eisinger 2000). Empirically, many states seem to be passing both punitive

and beneficial policies in the same years. As detailed in Figure 1.3 in Chapter 1 of this

dissertation, in the years included in this study states enacted 1,179 either punitive or

beneficial immigration-related policies (author tabulation), and of these all states enacted

both punitive and beneficial policies. Even the state enacting the least immigration

legislation, Wisconsin, enacted one policy coded as punitive and one coded as beneficial

during this period.

For each state I have constructed an index that includes all state immigration policies

enacted from 2003 to the year of the CPS data collection. Policies are individually counted -1

if punitive, +1 if beneficial, and 0 if neutral. All policies are summed to create a single

additive index of immigration policy climate in each state for each year included in the

dataset. Since the CPS collects data only during even-numbered years, the years in which

congressional and presidential elections are held, the policy index variable contains

cumulative data from 2003 to each even-numbered year. The final index before

standardization ranges from -27 to +36, with most states falling in the middle in most years.

Among the 250 state-years represented in this dataset (50 states x five election years), 28.0%

(n=70) state-years have a value of zero, 36.8% (n=92) have negative values and 35.2% (88)

have positive values. To ease interpretation, I then standardize the policy index variable to

center on a mean of 0 with a standard deviation of 1.0 by subtracting the mean value (-0.42)

and dividing by the standard deviation (6.782596). As described in my hypothesis, given the

impact of social constructions of state immigration policy, I expect that more positive

immigration policy indices at the state level will be associated with higher probability of

immigrant voting.

STATE CO-ETHNIC CONCENTRATION: A number of scholars (Tam Cho et al.

2006; Ramakrishnan 2005; Ramakrishnan and Espenshade 2001) find that residential

coethnic concentration impacts voting in immigrant communities. I specify here as state-

level population percentage of each of the four major racial/ethnic groups – Asian/PI

(nonHispanic), Black (non-Hispanic), Hispanic, and White (non-Hispanic), averaged over

each two-year period ending with even-numbered years. I match this percentage of

racial/ethnic concentration to each respondent based on respondent self-identified

race/ethnicity for each specific year. For this variable I make use of data from the American

Community Survey (ACS) 1-year data for 2005-2012. Since ACS data for 50 states are not

available prior to

2005, I estimate 2003-2004 data by interpolating between the 2000 Census and the 2005

ACS. Based on previous scholarship, I expect that higher levels of co-ethnic concentration at

the state level will be associated with higher probability of immigrant voting.

PRESIDENTIAL YEAR: Since voting turnout among all groups is known to be

substantially higher in presidential election years than in non-presidential election years, I

include presidential year as one of the crossed random-effects. I expect presidential year to

be strongly and positively associated with voting, but include this variable primarily as a

control so that I may analyze the three presidential years (2004, 2008, and 2012) and the two

non-presidential years (2006 and 2010) in the data series together.

Fixed-Effects Predictor Variables

At the individual level I estimate fixed effects for variables that are known generally

to contribute to political participation among the general U.S. population as well as those

known to contribute to participation specifically among immigrants. These include age,

education level, marital status, employment status, income, years in the U.S., and country of

origin. All of these individual-level variables may be viewed as providing resources known

to support political participation in ways consistent with the resource model of political

participation, which posits that an individual’s position in the economy generates resources

available for political participation (Brady, Verba and Schlozman 1995; Verba et al. 1993). I

expect all of these factors to be positively associated with voting among naturalized

immigrants, although they are included primarily as control variables so that I may focus on

my primary predictor of interest, state immigration policy.

All variables used in the analyses presented here are described in Appendix 4.A,

including their source. The fixed-effects predictors are chosen because of their predictive

value demonstrated in the extant research. A check on correlations among the predictors

reveals correlations among the predictors at <.40 with one exception; the correlation between

years in the U.S. and age is 0.59. Although relatively high I keep both variables in the

analyses in that both are justified by their previous performance in immigrant political

participation literature, both are included as control variables, and neither is my primary

predictor of interest. I describe just two other individual-level variables in detail here.

COUNTRY OF ORIGIN: To account for differing experiences with participatory

democracy in immigrants’ country of origin, I make use of the Polity2 scores contained in the

Polity IV 2012 data series maintained by the Integrated Network for Societal Conflict

Research. The Polity2 scale is -10 (indicating a strongly autocratic government) to +10

(indicating a strongly democratic government). I match Polity2 scores to each respondent’s

country-of-origin identified in the CPS dataset. Based on previous research indicating the

persistence of home country effects (Bueker 2005), I expect that immigrants from home

countries with higher Polity2 scores (indicating a more democratic country of origin) will

demonstrate higher levels of voting.

RACE/ETHNICITY: I make use of the standard race/ethnicity reporting categories in

the Current Population Survey and include sub-analyses for Asian/Pacific Islanders

(nonHispanic), Black (non-Hispanic), Hispanics, and White (non-Hispanic). In reporting

results where I refer to Asian/PI, Black, and White naturalized immigrants, I am referring to

nonHispanic populations who identify as these racial groups. All respondents identifying as

Hispanic are reported under the Hispanic ethnic category.

Findings

Descriptive statistics for all variables included in the analyses are detailed in Table

4.1 (below) by racial/ethnic group; here I call attention to notable difference among the

contexts within which the naturalized immigrant racial/ethnic groups reside. In examining

the focal predictor of interest, immigration policy index, there are some differences in the

state policy environments within which naturalized immigrants reside, with Black and White

immigrants more likely to reside in states with somewhat less favorable immigration policy.

Naturalized immigrants who are Black and White reside in states with mean state policy

indices of 0.53 and 0.58 respectively, somewhat lower than the mean state policy indices

where Asian (0.74) or Hispanic (0.75) naturalized immigrants reside. The other notable

contextual difference evident from the descriptive statistics is in the co-ethnic populations of

the states in which naturalized immigrants reside. Asian/PI and Hispanic naturalized

immigrants tend to live in states with co-ethnic populations higher than the average U.S.

coethnic populations, with Asian/PI naturalized immigrants living in states with average

Asian/PI populations of 10.6% (compared with the national Asian/PI population of 5.8%)

and Hispanic naturalized immigrants living in states with

Table 4.1. Descriptive Statistics

Naturalized Immigrants

Variable

Mean

Min

Max

Asian/PI (non-Hispanic)

Vote in general election

4796

0.4871

0.4999

State policy index

4796

0.7353

1.3275

-3.919

5.370

St. co-ethnic concentration

4796

10.57

11.963

.5196

49.710

Presidential Year

4796

0.4061

0.5000

Age

4796

49.083

15.023

Educ: <High school diploma

4796

0.1083

Educ: High school diploma

4796

0.2031

0.4023

Educ: Some college, no BA

4796

0.2047

0.4036

Educ: BA or higher

4796

0.4839

0.4998

Married

4796

0.7133

0.4523

Employed

4796

0.6503

0.4769

Income: <$30K/yr

4796

0.1883

Income: $30K-$59,999/yr

4796

0.2706

0.4443

Income: >=$60,000/yr

4796

0.5411

0.4984

Years in U.S.

4796

22.126

10.290

Country of Origin Polity Score

4796

-1.3445

7.4200

-10

Black (non-Hispanic)

Vote in general election

1061

0.6136

0.4872

State policy index

1061

0.5279

0.9001

-3.919

5.370

St. co-ethnic concentration

1061

14.428

7.389

.6129

36.379

Presidential Year

1061

0.4846

0.4999

Age

1061

46.055

14.317

Educ: <High school diploma

1061

0.1254

Educ: High school diploma

1061

0.2592

0.4383

Educ: Some college, no BA

1061

0.2582

.04379

Educ: BA or higher

1061

0.3572

.04794

Married

1061

0.5627

0.4963

Employed

1061

0.7191

0.4496

Income: <$30K/yr

1061

0.2658

Income: $30K-$59,999/yr

1061

0.3355

0.4724

Income: >=$60,000/yr

1061

0.3987

0.4899

Years in U.S.

1061

20.031

10.199

Country of Origin Polity Score

1061

1.8162

7.4486

-10

Hispanic

Vote in general election

5349

0.5165

0.4998

State policy index

5349

0.7483

1.2029

-3.771

5.370

St. co-ethnic concentration

5349

23.776

11.882

.8773

46.854

Presidential Year

5349

0.4893

0.4999

Age

5349

47.575

15.109

Educ: <High school diploma

5349

0.3492

Educ: High school diploma

5349

0.2784

0.4482

Educ: Some college, no BA

5349

0.2152

0.4110

Educ: BA or higher

5349

0.1572

0.3640

Married

5349

0.6650

0.4720

Employed

5349

0.6611

0.4734

Income: <$30K/yr

5349

0.3505

Income: $30K-$59,999/yr

5349

0.3584

0.4796

Income: >=$60,000/yr

5349

0.2911

0.4543

Years in U.S.

5349

24.811

11.898

Table 4.1. Descriptive Statistics (Cont’d)

Naturalized Immigrants

Variable

Mean

Min

Max

Country of Origin Polity Score

5349

-0.1974

5.6352

-9

White (non-Hispanic)

Vote in general election

4583

0.6192

0.4826

State policy index

4583

0.5831

1.0693

-3.1817

5.370

St. co-ethnic concentration

4583

68.358

14.981

40.639

97.683

Presidential Year

4583

0.5090

0.4999

Age

4583

55.989

16.755

Educ: <High school diploma

4583

0.1174

Educ: High school diploma

4583

0.2566

0.4368

Educ: Some college, no BA

4583

0.2160

0.4116

Educ: BA or higher

4583

0.4100

0.492

Married

4583

0.6788

0.4679

Employed

4583

0.5479

0.4978

Income: <$30K/yr

4583

0.2302

Income: $30K-$59,999/yr

4583

0.2727

0.4454

Income: >=$60,000/yr

4583

0.4971

0.5000

Years in U.S.

4583

30.578

15.763

Country of Origin Polity Score

4583

3.202

7.501

-10

average Hispanic populations of 23.8% (compared with the U.S. Hispanic population of

16.3%) (Humes, Jones and Ramirez 2011). In contrast, Black naturalized immigrants reside

in states with average Black population of 14.4%, about the same as the overall U.S. Black

population of 14.6%. Also in contrast, White naturalized immigrants reside in states with

average White populations of 68.4%, lower than the U.S. White non-Hispanic population of

76.2% (Humes, Jones and Ramirez 2010).

To test my main hypotheses, that immigration policy will influence voting among

naturalized immigrants and that this influence will be greater among non-White immigrants,

this study employs multi-level mixed-effects with crossed-random effects logit analyses

examining state-level predictors of voting among naturalized immigrants across five national

election cycles. Separate regression models are utilized to test the hypothesis, one for each of

the four major racial/ethnic groups, Asian (non-Hispanic), Black (non-Hispanic), Hispanic,

and White (non-Hispanic). The multi-level mixed-effects with crossed-random effects logit

models regress the binary dependent variable, self-reported voting, on an index of state

immigration policy, state co-ethnic concentration, and a vector of individual-level resource

variables known to be associated with voting behavior. Because coefficients from logit

regressions cannot be interpreted directly, I convert logit coefficients to odds ratios wherever

possible to aid in interpretation. The results of the four logit models are detailed in Table 4.2

below.

Table 4.2. Predictors of Voting among Immigrants from Four Racial/Ethnic Groups, 2004-2012

Multi-level Mixed-Effects with Crossed Random-Effects Logistic Regression

Variables Asian/PI Black Hispanic White

(non-Hispanic) (non-Hispanic) (non-Hispanic)

Random Effects Grouping N=4,796 N=1,061 N=5,349 N=4,583

Variables: Coeff. SE Coeff. SE Coeff. SE Coeff. SE

State policy index

4.23e-13

5.63e-08

3.43e-19

1.54e-10

0.004

.237

1.63e-06

.0021

State co-ethnic concentration

1.33e-13

7.04e-09

3.28e-17

5.39e-10

8.60e-08

.00001

0.003835

6.87e06

Presidential year

0.156

0.1581

0.364

0.3742

0.235

.2366

0.186

.1883

State-level intercept

0.128

0.0517

0.081

0.1335

0.0906

.0414

2.76e-11

3.61e06

Fixed Effects Individual

Level Variables:

Age

1.013***

0.003

1.022***

0.006

1.031***

0.003

1.018***

0.003

Educ: High school diploma

1.855***

0.234

0.808

0.191

1.411***

0.111

1.533***

0.179

Educ: Some college, no BA

2.975***

0.384

1.038

0.255

2.693***

0.239

2.381***

0.298

Educ: BA or higher

4.316***

0.523

1.325

0.323

3.140***

0.320

3.218***

0.386

Married

1.317***

0.095

1.108

0.162

1.465***

0.096

1.780***

0.132

Employed

1.108

0.081

1.818***

0.303

1.401***

0.102

1.258**

0.103

Income: $30,000-$59,999/yr

0.917

0.089

0.885

0.161

0.921

0.068

1.114

0.109

Income: >$60,000/yr

1.105

0.103

1.240

0.246

1.330**

0.112

1.332**

0.129

Years in US

1.017***

0.004

1.023**

0.008

1.010**

0.004

1.031***

0.003

Country of origin

1.009*

0.004

1.026**

0.010

1.000

0.007

1.025***

0.005

Individual-level intercept

0.087***

0.030

0.218***

0.125

0.059***

0.023

0.057***

0.021

LR Test Chi2 191.34*** 69.29*** 317.64*** 183.03***

*=p<.05; **p<.01; ***p<.001

Education: excluded comparison category less than high school diploma, excluded category includes GED

Income: excluded comparison category less than $30,000/yr

The likelihood ratio tests of the final models for each of the four racial/ethnic group

are all significant at p<.0001, indicating that the multilevel mixed-effects logistic regression

is significantly different from the same model run using ordinary (single-level fixed-effects)

logistic regression, and thus confirming that multi-level mixed effects models are warranted.

The results indicate that the random effects of state immigration policy, as measured by an

index combining punitive and beneficial immigration policy enacted by states, are on average

significantly and modestly positively associated with increases in voting among

Hispanic and White naturalized immigrants. This effect is not significant among Asian/PI or

Black immigrants. On average, for each one unit increase in the state immigration policy

index, the odds of voting increase by 0.004% among Hispanic immigrants and increase by

0.0002% among White immigrants, holding all other variables constant.

Co-ethnic concentration, as measured by percentage of the state population sharing

the same racial/ethnic identification as the individual voter, is also on average statistically

significantly and positively associated with increases in voting among Hispanic and White

immigrants, although the effect size is extremely small. For each percentage point increase in

the co-ethnic population at the state level, the odds of voting increase on average by

0.00000009% among Hispanic immigrants24 and 0.384% among White immigrants, holding

all other variables constant. Again, this effect is not significant among Asian/PI or Black

immigrants.

As expected, presidential year is significantly and positively associated with voting

among naturalized immigrants in all four major racial/ethnic groups, with the observed

average effects strongest among Black naturalized immigrants and weakest among Asian/PI

naturalized immigrants. The odds of voting in presidential election years as compared with

non-presidential election years are on average 16.9% higher among Asian/PI immigrants,

43.9% higher among Black immigrants, 26.5% higher among Hispanic immigrants, and

20.4% higher among White immigrants, holding all else constant.

As my primary interest is in the state-level variables, individual-level variables are

included in the analyses primarily as controls. Each of the individual-level variables

represents a potential resource shown in previous research to support higher levels of voting

behavior, thus I expected to observe positive associations with voting. Although the

individual-level resource variables perform largely as expected, there are some notable

differences among the racial/ethnic subgroups. Age and years in the U.S. are consistently

positively associated with increased odds of voting among all four major racial/ethnic groups.

Education is significant and strongly positively associated with voting among Asian,

Hispanic, and White naturalized immigrants, but education is not significantly predictive of

voting among Black immigrants. Being married is significant and strongly positively

associated with voting among Asian/PI, Hispanic, and White immigrants, but not among

Black immigrants. Employment is significant and strongly positively associated with voting

among Black, Hispanic, and White immigrants, but not among Asian/PI immigrants. Country

of origin Polity2 scores are significant and modestly positively associated with voting among

Asian/PI, Black, and White immigrants, but not among Hispanic immigrants. Overall, I

find partial support for Hypothesis 1; among Hispanic and White naturalized immigrants

state immigration policy indices are positively associated with voting. The average effect

size, however, is modest in magnitude when controlling for individuallevel resources and the

size of state co-ethnic population. In contrast, I do not find support for Hypothesis 2. While I

find that the modest effect size of the state immigration policy index is greatest among

Hispanic naturalized immigrants, it is still significant among White immigrants and

unexpectedly not significant at all among Asian and Black immigrants. It is not a surprise

to find that individual-level predictors of voting behavior have much greater effect sizes, as

measured by odds ratios, than do the state-level predictors examined here. This is consistent

with most research utilizing multi-level mixed effects models that finds most of the variation

in individual-level outcome accounted for by individual-level predictors (Hox 2010).

However, the multi-level mixed-effects approach is the most appropriate to examine the

significance and magnitude of the state-level predictors of interest here.

Because I have estimated random coefficients for the state-level impact of

immigration policy index on probability of voting for Hispanics, and because that impact is

significant, I may examine the variation in the random coefficients among states. As

illustrated in Figure 4.1, the (mean across years) coefficients for state policy slopes vary

substantially from -0.0101 to 0.0637. Variation in the coefficient for state policy indicates

that in states with relatively large coefficients for the state policy slope, state policy has a

relatively large impact on voting behavior. While in states with relatively small coefficients

for the state policy slope, state policy has a relatively small impact on voting behavior. It also

indicates that while the mean state policy coefficient for all states in all years reported in

Table 4.1 in the Hispanic model is 0.004, some states have a strongly positive coefficient

while others have negative coefficients indicating opposite directions of impact on the

outcome.

For example, we see that California has the largest (mean) random coefficient at

0.0637, which indicates that each standard deviation increase in the state policy index in

California is associated with a marginal increase in voting among Hispanic naturalized

immigrants of approximately 6.6%, while in the state with the lowest (mean) coefficient,

Washington State, such an increase in the state policy index is associated with a marginal

decrease in voting of about 1.0%. California is clearly an outlier on both axes presented in

Figure 4.1 – with a standardized state policy index score almost two standard deviations

above zero, it enacted during the period of this examination the most beneficial (inclusive)

state immigration policy among all U.S. states. With the highest coefficient for the state

policy slope, California is also an outlier in terms of the strength of impact of that policy on

voting outcomes among Hispanic naturalized immigrants.

Also of note in Figure 4.1 is that the states with relatively large (mean) coefficients

for state immigration policy, those above the horizontal line marking the mean of these

coefficients, tend to be one full standard deviation and more above and below the

standardized policy index mid-point. So while states with standardized policy indices around

zero are demonstrate strength of coefficients around zero (state immigration policy is not

having an impact on voting), those with standardized policy indices about one standard

deviation above (enacting more beneficial/inclusive policy) and one standard deviation below

(enacting more punitive/exclusionary policy) also tend to be those states where policy

impacts are being translated into voting outcomes among Hispanic naturalized immigrants. It

may seem intuitive to think that more extreme policy would likely be associated with more

voter activism and mobilization, but this analysis empirically confirms this is happening in a

number of states.

Finally, in Figure 4.2 below I compare the predicted probabilities of voting among

immigrants by racial/ethnic group. Here we see that the predicted probability of voting

among is 48.71% among Asian/PI immigrants, 61.36% among Black immigrants, 51.65%

among Hispanic immigrants, and 61.93% among White immigrants. Of note, even after

controlling for individual and contextual covariates included in these models, the probability

of voting among Black and White immigrants remains significantly higher than among

Latinos and Asians, and the probability of voting among Latinos remains significantly higher

than among Asians.

Figure

4.2

Predicted

Probability

Voting

among

Immigrants

Race/Ethnicity,

2004

‐

2012

48.71

61.36

51.65

61.93

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

Asian

Black

Hispanic

White

Conclusion

In this chapter I test the effect of state immigration policy on voting among

naturalized immigrants who self-identify as Asian/PI, Black, Hispanic or White. I provide for

a robust test of the effects of state policy by employing a multi-level mixed effects model

with a comprehensive set of individual-level covariates. Overall, this study confirms that for

some groups of naturalized immigrants state immigration policy is impacting political

behavior in the direction predicted by social construction/policy feedback theories, with more

positive policy having a mobilizing impact; from this positive relationship we may infer the

opposite, that more punitive policy is having a de-mobilizing impact. This is remarkable

given that the rhetorical context within the states during consideration of policy reveals that

the purported target of punitive policy is generally not thought to be naturalized immigrants.

Naturalized immigrants by definition are not impacted by the burdens and benefits that the

state policies in question seek to allocate to either undocumented or legal but not-

yetnaturalized immigrants. This suggests that for those populations where public policy is

having an impact (on Hispanic and non-Hispanic White immigrants particularly) the impact

may well occur via interpretive mechanisms; for instance the allocation of interpretive

benefits and burdens that accompany the naturalized immigrant’s psychological identification

with the intended target population of the public policy in question. This area of study

is consequential to the future of U.S. democracy because immigrants generally have low

political participation rates relative to the majority population of U.S.-born whites. In

statewide surveys in California in 2002, Ramakrishnan (2005) identified that 59% of native-

born whites who are third generation or later in the U.S. vote regularly, compared to 38% of

immigrant Latinos and 39% of immigrant Asians.

Additionally, as these groups become a larger share of the U.S. citizen voting age population

(CVAP), a continued disparity in participation could have consequential impacts on a variety

of political processes – from questions of representation and partisan alignment to issues of

public policy.

The variation in the coefficients of state policy (observed in Figure 4.1) attests to the

variation in the size of the effects of state policy on voting impacts. This variation merits

further attention as well, especially in light of previous findings that under some

circumstances anti-immigration policy in the states have resulted in increased mobilization of

some immigrants (Pantoja et al. 2001; Barreto et al. 2005; Rim 2009). It may well be,

consistent with suggestions by Barreto et al. (2009), that the variation observed in Figure 4.1

is attributable in part to the presence of mobilizing institutions in the states, an influence not

tested in this study but worthy of further examination. At the same time caution is warranted

in that the studies cited here all examined mobilization of immigrants in California, clearly

seen in Figure 4.1 as an outlier in terms of the strength of impact of state immigration policy

on voting among Hispanic naturalized citizens.

CHAPTER 5

Second Generation Effects: The Impact of State Immigration Policy and Organizational

Involvement on Political Engagement of U.S.-born Children of Immigrants

This paper is one chapter in a dissertation examining the impact of contentious

immigration policy within individual U.S. states on political engagement among immigrants

and their children. There are an estimated 23.2 million children in the U.S. who have at least

one immigrant parent, and an estimated eighty-four percent of these children are U.S.-born

(Passel 2011). When we consider the exposure of these children to the incorporation and

socialization experiences of their immigrant parents, and take into account that the vast

majority are U.S.-born and thus citizens by birth, the question of the impact of such policy on

their engagement in the U.S. political system becomes salient.

Immigrants and their children currently residing in the U.S. have been exposed during

their political socialization to varying levels of anti-immigrant public sentiment, elite

rhetoric, and restrictionist policymaking at the local, state, and federal levels. The 2000s

witnessed increases in the passage of state and local immigration legislation attempting to

restrict access to employment, housing, education, and other benefits for undocumented

immigrants (Monogon 2013; Chavez and Provine 2009), as well as an unprecedented

increase in detention and deportations of undocumented immigrants and elite rhetoric

conflating undocumented immigration with the threat of terrorism (Pope and Garrett 2012;

Hagan et al. 2008).

My interest with this chapter is in exploring how non-electoral political engagement

among U.S.-born children of immigrants may be impacted by state immigration policies

through the lens of policy design-social construction theory. The research question for this

chapter is “how have the social constructions of immigrants, borne of public sentiment and

elite rhetoric and institutionalized in public policy, impacted the non-electoral political

engagement of U.S.-born children of immigrants?” While there is some early evidence that

state immigration policy is impacting the political engagement of some immigrants (see

Chapters 3 and 4 in this dissertation), less understood is how exposure to contested

immigration policies in the states may affect the political engagement of children of

immigrants. I focus here specifically on the U.S.-born children of immigrants both because

they represent the majority of children of immigrants in the U.S. and because their automatic

citizenship status carries the greatest potential for political (electoral) impact. I hypothesize

that immigration policy affects the political engagement and incorporation of the children of

immigrants via social constructions, and furthermore, that this effect occurs even after

controlling for other known predictors of non-electoral political engagement. Secondarily, I

explore the potential moderating role of community organizations by asking whether there is

evidence that such organizations may be translating salient state immigration policy into

political engagement among U.S.-born children of immigrants. I disaggregate all analyses by

racial/ethnic group in order to assess differences between the four largest racial/ethnic groups

of U.S.-born children of immigrants within the impacts examined here.

Non-Voting Political Participation

Political engagement includes activities intended to influence government action

“either directly by affecting the making or implementation of public policy or indirectly by

influencing the selection of people who make those policies” (Verba et al. 1995, 38).

Nonvoting political engagement then includes activities intended to influence government

action apart from voting - such as volunteering for a political candidate, contacting an elected

official, contributing money to a candidate, or participating in a rally or protest aimed at a

specific policy.

While non-voting political participation remains understudied relative to voting

behavior in the U.S., the factors that drive non-voting political participation are generally

found to be similar to those that drive voting. These include socioeconomic resources (Verba

and Nie 1972), occupation, education, and social connectedness (Leighley and Vedlitz 1999).

In addition to these individual-level factors, scholars studying immigrant communities and

communities of color find that additional contextual factors contribute to non-voting political

participation and civic engagement. For example, Ramakrishnan (2006) finds that Asian and

Latino immigrants have higher rates of non-voting civic volunteerism in the presence of

greater co-ethnic concentrations. Other scholars find that factors related to the country of

origin, such as democratic versus non-democratic and political engagement pre-immigration,

significantly influence a range of immigrant non-voting political activities in the U.S.

(Bueker 2005; Wals 2011).

Children of Immigrants

In this chapter I build upon two veins of previous research findings related to children

of immigrants in order to extend potential policy design-social construction feed-forward

effects of immigration policy to U.S-born children of immigrants. The first vein finds that

immigration policy designs contain social constructions targeting children of immigrants,

both U.S.-born and foreign-born children. I contend that these constructions may confer both

material and interpretive benefits and burdens sufficient to produce measurable outcomes in

children of immigrants. Newton (2008) finds that two major pieces of federal immigration

policy, the Immigration Reform and Control Act of 1986 (IRCA) and the Illegal Immigration

Reform and Immigrant Responsibility Act of 1996 (IIRIRA), contain references to the

children of immigrants and demonstrates that these children have been negatively constructed

in federal immigration policy. For example, the proposed Gallegly Amendment to IIRIRA

1996 contained provisions similar to those found in California’s 1994 Proposition 187 –

preventing children of undocumented immigrants from attending public schools, regardless

of their own citizenship or immigration status.

The second vein of research describes differential outcomes among children of

immigrants, primarily in the areas of education and child development, based on their

parents’ immigration status. Although this second vein does not rely explicitly on policy

design-social construction theory as the explanation for the variations observed, I contend

that such an explanation is sufficiently plausible to inform this study. For example, Bean et

al. (2011) conceptualize parental immigration status as a measurement of membership in the

polity (what they call a ‘political-entry incorporation class’) that ultimately influences the

educational attainment of U.S.-born children of immigrants. They find that Mexican-origin

second generation youth experience negative educational attainment relative to Asian-origin

second generation youth primarily due to delayed maternal ‘membership’ which delays

political and economic incorporation of immigrant families (Bean et al. 2011). In the field of

child development, scholars have found that parental undocumented status has direct effects

on the developmental outcomes of young children, particularly in terms of cognitive

development and educational progress (Yoshikawa and Kholoptseva 2013; Suarez-Orozco et

al. 2011). These authors speculate as to the material burdens of parental removal, parent child

separation due to transnational practices, lower access to means-tested programs, work

conditions, and parental stress as potential mechanisms (Yoshikawa et al. 2013).

Additionally, Filindra et al. (2011) find that state-level policies related to immigrant access to

social welfare are significantly associated with educational outcomes among children of

immigrants in the American states, and speculate that inclusion of immigrants in social

welfare programs may engender “feelings of belongingness” (429) among the children of

immigrants that may enhance their incorporation into school systems.

Finally, descriptive studies demonstrate substantial variation in the household

circumstances within which children of immigrants from different countries grow up. For

example, Chaudry and Fortuny (2011) make use of the American Community Survey to

illustrate differences in rates of poverty among immigrant families (made up most often of

immigrant parents and U.S.-born children), finding that children in households with

immigrant parents from Mexico have relatively high rates of poverty (69%) compared with

children with parents from East Asia (23%) and Europe (21%). And among children in

immigrant households, those in low-income households with roots in Southeast Asia are

much more likely (15%) to rely on public welfare benefits than are those from Mexico (6%).

Although this study is descriptive, I would argue that it provides some insight into the

potential material benefits that may be conferred by immigration policy to immigrant parents

from different countries.

The questions of whether, how, or to what extent immigration policy design may

produce feed-forward effects among children of immigrants in terms of their political

participation remain unanswered. Yet these examples of the presence of social constructions

of children of immigrants contained in federal immigration policy and of variation in

outcomes in the areas of development and education of children of immigrants and

experiences of poverty suggest that policy feed-forward effects feasibly have the capacity to

produce other types of outcomes among this same population. Thus, in this chapter I extend

theories of policy design-social construction to posit that, in addition to directly impacting

the political behavior of immigrants themselves, immigration policy has the potential to

affect the political engagement of children of immigrants.

Community Organizations

The political participation literature identifies two distinct roles that community

organizations play in influencing political participation among racial/ethnic, immigrant, and

other marginalized communities. The first is by directly influencing the capacity of

individual community members to participate politically and civically. This role is perhaps

most completely described by Verba et al. (1995) who find that affiliation with non-political

voluntary associations and religious organizations is significantly, positively, and strongly

associated with voting and non-voting political activities. Similarly, Rosenstone and Hansen

(1993) find that involvement in associations or groups is the single largest predictor of

nonelectoral political participation such as writing to public officials and signing petitions.

These authors posit that involvement with community organizations such as voluntary

association and religious organizations serves to develop civic skills in the individuals that

are subsequently transferrable to political engagement activities (Verba et al. 1995). This

direct effect of participation in community organizations on political participation has been

found among immigrant populations as well (DeSipio 2002).

The second politically relevant role of community organizations is that of direct

political or policy mobilization. Community organizations play an important role in

translating public opinion and policy preferences into non-electoral political mobilization in

ethnic and immigrant communities. Wong (2006) contends that in addition to a general

political socialization role, community organizations in Asian and Latino immigrant

communities effectively organize around issues and are especially effective at mobilizing

community to non-electoral political participation such as petition drives and demonstrations.

There is evidence of this mobilizing role in immigrant communities particularly in the face of

threatening immigration policy. Community-based and national advocacy organizations

played an important role in translating the policy threat environment to political action in the

2006 rallies protesting threatening federal immigration legislation (Rim 2009; Pantoja et al.

2008). This effort was particularly effective in urban areas with rich networks of

communitybased organizations serving ethnic and immigrant communities (Cordero-Guzman

et al. 2008). Research on Latino mobilization following the California anti-immigration

initiative in 1994 also reveals the key role played by immigrant and ethnic mobilizing

organizations in those geographic areas where immigrant mobilization (voting and non-

voting) occurred

(Ramakrishnan 2005; Barreto et al. 2009).

Hypotheses

This study is placed in the tradition of policy design-social construction theory. I

propose that the relationship between state immigration policy and children of immigrants

100

residing in states is such that the children of immigrants receive interpretive messages from

the enactment of state immigration policy and the political rhetoric surrounding such passage,

whether positive or negative. I posit that such interpretive messages have the ability to

influence non-voting political participation among adult U.S.-born children of immigrants.

H1: State immigration policy will be positively associated with non-voting political

participation among U.S.-born children of immigrants.

I also posit that involvement in community organizations by U.S.-born children of

immigrants will result in greater non-voting political participation both through direct

skillbuilding as well as through mobilization stimulated by threatening state immigration

policy.

H2: Organizational involvement will positively impact non-voting political

participation among U.S.-born children of immigrants both directly and indirectly through

its interaction with state immigration policy.

Finally, I posit that because state immigration policy is often racialized as Asian and

Hispanic, U.S.-born Asians and Hispanic/Latinos with immigrant parents may experience the

consequences of the benefits and burdens contained in immigration policy to a greater degree

than U.S.-born Black/African-Americans and Whites with immigrant parents.

H3: The relationship between state immigration policy and non-voting political

participation will be stronger among U.S.-born Asian and Hispanic/Latino children of

immigrants than among U.S.-born Black/African-American and White/European American

children of immigrants.

Data and Methods

I make use of the Current Population Survey (CPS) basic survey and the CPS Civic

Engagement Supplement for 2008-2011. The CPS basic survey is a nationwide survey of the

U.S. population age 15 and over that is administered monthly by the Census Bureau with a

focus on labor force participation. The CPS Civic Engagement Supplement is a supplemental

survey that was administered in November of 2008 through 2011, and contains questions

101

about general civic orientation, community engagement, and non-voting political

participation among a subset of respondents to the basic CPS survey who are at least 15 years

of age. The total number of respondents to the Civic Engagement supplement during the

years of interest is over 224,000. Although the vast majority of respondents are U.S.-born

with U.S.-born parents, each year the survey contains a subset of U.S.-born respondents with

immigrant parents. To the pooled individual-level response from the CPS Civic Engagement

supplement for the years 2008-2011 I have merged in a state- and year-specific policy index

variable (described below) to construct the complete analyzable dataset. After accounting for

missing data on any of the model variables, I am left with a complete dataset of 17,244

U.S.born respondents with at least one foreign-born parent within the four largest

racial/ethnic groups with which to conduct my main analyses.

In order to test my hypotheses, I estimate two separate models within each of the four

largest racial/ethnic groups - one without and one with organizational predictors - for a total

of eight models. For all statistical modeling I make use of Stata/IC 13.1, using the poisson

command for dependent count variables and clustering on state using the vce(cluster)

command to specify robust standard errors. Since respondents are clustered by state

according to CPS sampling methodology, clustering in this manner accounts for intragroup

correlation within states, allowing me to maintain the usual assumption of independence

across respondents while relaxing the assumption within states (Long and Freese 2006). In

postestimation I use Stata’s suest (seemingly unrelated estimation) command to test for

differences in coefficients of interest across separate racial/ethnic group models. Seemingly

unrelated regression techniques were developed for just such a purpose, to test for

differences across independent panel-data equations (Blackwell 2005). Finally, I make use of

Stata’s margins and marginsplot commands to produce and display predicted estimates for

102

postestimation. Brief descriptions of all variables and data sources are included in Table 5.1

below. Here I provide additional detail for the dependent variable and four focal predictors of

interest.

Table 5.1 Descriptive Statistics

U.S.-Born Children of Immigrant Parents

Variable N Mean SD Min Max

Asian (non-Hispanic)

Non-Voting Political Participation

1,293

0.442

0.679

State Immigration Policy Index

1,293

5.401

11.175

-24

Organizational Involvement

1,293

0.587

0.938

Policy Index X Org. Involvement

1,293

3.722

13.833

-63

115

Educ: <High school diploma

1,293

0.087

Educ: High school diploma

1,293

0.190

0.393

Educ: Some college, no BA

1,293

0.569

0.495

Educ: BA or higher

1,293

0.154

0.361

Income: <$25K/yr

1,293

0.149

Income: $25K-$49,999/yr

1,293

0.224

0.417

Income $50K-$99,999/yr

1,293

0.312

0.464

Income: >=$100K/yr

1,293

0.314

0.464

Age

1,293

39.490

19.697

Sex (Female)

1,293

0.502

0.500

Married

1,293

0.402

0.491

Year: 2008

1,293

0.281

Year: 2009

1,293

0.082

0.274

Year: 2010

1,293

0.371

0.483

Year: 2011

1,293

0.266

0.442

Black/African-American (non-Hispanic)

Non-Voting Political Participation

484

0.496

0.746

State Immigration Policy Index

484

3.256

8.942

-24

Organizational Involvement

484

0.688

1.039

Policy Index X Org. Involvement

484

2.229

12.732

-63

Educ: <High school diploma

484

0.097

Educ: High school diploma

484

0.244

0.430

Educ: Some college, no BA

484

0.581

0.494

Educ: BA or higher

484

0.079

0.269

Income: <$25K/yr

484

0.260

Income: $25K-$49,999/yr

484

0.293

0.456

Income $50K-$99,999/yr

484

0.310

0.463

Income: >=$100K/yr

484

0.136

0.344

Age

484

35.558

16.168

Sex (Female)

484

0.543

0.499

Married

484

0.252

0.435

103

Year: 2008

484

0.25

Year: 2009

484

0.074

0.263

Year: 2010

484

0.384

0.487

Year: 2011

484

0.291

0.455

Table 5.1 Descriptive Statistics (Cont’d)

U.S.-Born Children of Immigrant Parents

Variable

Mean

Min

Max

Hispanic/Latino

Non-Voting Political Participation

3,801

0.428

0.667

State Immigration Policy Index

3,801

7.927

10.234

-24

Organizational Involvement

3,801

0.524

0.913

Policy Index X Org. Involvement

3,801

3.810

12.469

-105

115

Educ: <High school diploma

3,801

0.212

Educ: High school diploma

3,801

0.288

0.453

Educ: Some college, no BA

3,801

0.452

0.498

Educ: BA or higher

3,801

0.048

0.215

Income: <$25K/yr

3,801

0.265

Income: $25K-$49,999/yr

3,801

0.322

0.467

Income $50K-$99,999/yr

3,801

0.290

0.454

Income: >=$100K/yr

3,801

0.122

0.328

Age

3,801

34.618

16.318

Sex (Female)

3,801

0.504

0.500

Married

3,801

0.406

0.491

Year: 2008

3,801

0.285

Year: 2009

3,801

0.091

0.287

Year: 2010

3,801

0.373

0.484

Year: 2011

3,801

0.251

0.434

White/European-American (non-Hispanic)

Non-Voting Political Participation

9,661

0.735

0.863

State Immigration Policy Index

9,661

3.455

8.151

-24

Organizational Involvement

9,661

0.882

1.165

Policy Index X Org. Involvement

9,661

2.809

12.464

-84

138

Educ: <High school diploma

9,661

0.835

Educ: High school diploma

9,661

0.274

0.446

Educ: Some college, no BA

9,661

0.501

0.500

Educ: BA or higher

9,661

0.142

0.349

Income: <$25K/yr

9,661

0.215

104

Income: $25K-$49,999/yr

9,661

0.260

0.439

Income $50K-$99,999/yr

9,661

0.311

0.463

Income: >=$100K/yr

9,661

0.224

0.417

Age

9,661

57.010

19.604

Sex (Female)

9,661

0.526

0.499

Married

9,661

0.556

0.497

Year: 2008

9,661

0.289

Year: 2009

9,661

0.082

0.274

Year: 2010

9,661

0.343

0.475

Year: 2011

9,661

0.286

0.452

Unweighted summary statistics using estat sum

Standard deviations not adjusted for clustering

Dependent Variable

I model the dependent variable as self-reported non-voting political participation over

the previous year. The dependent variable is a count variable constructed from three

questions on non-voting political participation common across the CPS Civic Engagement

Supplement surveys from all four years in the dataset (2008-2011). The first question

assesses the frequency with which the respondent discussed politics over the previous year:

“During a TYPICAL MONTH in the past year, when communicating with family and friends,

how often were politics discussed... basically every day, a few times a week, a few times a

month, once a month, or not at all?” I code these responses 1=the first two response options,

indicating frequent discussion of politics, 0=all other response options, and exclude

“Refused” and “Don’t know” responses. The second and third questions assess respondent

involvement in specific non-voting political actions over the previous year: “I am going to

read a list of things some people have done to express their views. Please tell me whether or

not you have done any of the following in the last 12 months, that is between November 20XX

and now…” “Contacted or visited a public official – at any level of government – to express

your opinion…”, “Bought or boycotted a certain product or service because of the

social/political values of the company that provides it…”. I code responses from these two

105

questions 1=yes, 0=no, and exclude “Refused” and “Don’t know” responses. Finally, I

construct the final count variable by simply summing the three coded responses, resulting in

the dependent count variable having a range of 0-3 indicating a range of non-voting political

activities among respondents. Among my population of interest, U.S.-born respondents with

immigrant parents, 55.1% report zero non-voting political activities, 31.0% report one

activity, 10.2% report two activities and 3.7% report three activities (unweighted

mean=0.625, sd=0.813).

Predictor Variables

The main predictor variables of interest are those related to state immigration policies

as well as organizational involvement.

STATE IMMIGRATION POLICY: For the focal predictor, state immigration

policies, I have created an index of state immigration policies for each state. Details of

construction of this variable is contained in Chapter 1 of this dissertation. Variables for each

state 2008-2011 represent the cumulative index of enacted policies from 2003 through the

year in question. I chose a cumulative policy index rather than a discrete year-to-year policy

index because of the political socialization processes which are believed to precede

politicization of both young people and adults (McIntosh and Youniss 2010; Sapiro 1996). A

cumulative measure represents an advantage over year-to-year measure in that it contains in

one year the social constructions, both positive and negative, expressed in state level

immigration policies and experienced by the target population for the previous 5-8 years

combined. The U.S.-born respondents with immigrant parents included in these analyses live

in states with policy indices ranging from -24 to +23 (unweighted mean=4.806, sd=0.527).

106

Consistent with other scholarship of contemporary state immigration policy (Chavez

and Provine 2009; Monogon 2013; Marquez and Schraufnagel 2013; Zingher 2014), I make

use of the National Conference of State Legislature (NCSL) dataset on state immigration

policy to identify enacted state immigration legislation from 2005-2011. As detailed in

Chapter 1, each piece of enacted legislation in the NCLS dataset 2005-2011 has been coded

as to direction –beneficial to immigrants, punitive to immigrants, or neutral– and omnibus

legislation has been disaggregated into separate policies based on substantive content before

coding as to direction. Since NCSL data are available only from 2005 forward, and there is

some evidence that the trend in state immigration policymaking actually began before 2005, I

make use of a report from the Progressive States Network (2008) to identify additional state

immigration policies enacted 2003 and 2004. Coding for 2003 and 2004 policies as to

direction is taken from the publication itself. This variable is limited to enacted policies

passed by the legislature, so it excludes bills introduced and not passed and policy action

taken independently by the executive or judicial branch.

Based on my theoretical considerations, I expect that if U.S.-born respondents with

immigrant parents internalize the constructions of immigrants contained in the state

immigration policies there will be a positive relationship between state immigration policy

index and non-voting political engagement. If, in contrast, U.S.-born respondents with

immigrant parents are resisting the constructions of immigrants contained in state

immigration policies, I expect there will be a negative relationship between the state

immigration policy index and non-voting political engagement. In the case of a positive

relationship more beneficial (integrative) state immigration policy would result in higher

political engagement; while in contrast, in the case of a negative relationship higher political

107

engagement would be associated with punitive (exclusionary) state immigration policy.

ORGANIZATIONAL INVOLVEMENT: In order to test my hypotheses regarding

the potential organizational involvement mechanism by which children of immigrants may

become politicized and pursue non-voting political behavior, I construct an organizational

involvement variable from questions in the CPS Civic Engagement Supplement. This

variable is taken from six questions on organizational involvement common across the

surveys from all four years in the dataset. The first five questions assess organizational

participation in different types of groups in the previous year: “The next questions are about

the groups or organizations in which people sometimes participate. I will read a list of types

of groups and organizations. Please tell me whether or not you participated in any of these

groups during the last 12 months, that is between November 20XX and now…” “A school

group, neighborhood, or community association such as PTA or neighborhood watch

groups?” “A service or civic organization such as American Legion or Lions Club?” “A

sports or recreation organization such as a soccer club or tennis club?” “A church,

synagogue, mosque or other religious institutions or organizations, NOT COUNTING your

attendance at religious services?” “Any other type of organization that I have not

mentioned?” The sixth question assesses leadership experience with an organization in the

previous year: “In the last 12 months, between November 20XX and now, have you been an

officer or served on a committee of any group or organization?” For all six questions I code

these responses 1=yes, indicating organizational involvement or leadership, 0=no, and

exclude “Refused” and “Don’t know” responses. I then sum the responses from each question

to obtain a continuous variable that ranges from 0-6 (unweighted mean=0.722, sd=1.075).

Based on theory previously reviewed, I expect that increased organizational involvement will

be positively associated with the outcome of non-voting political behavior.

108

I intentionally do not include in the construction of this variable questions from the

CPS Civic Engagement Supplement that ask about expressly political organizational

involvement. Thus while the dependent variable contains a list of potential activities clearly

in the political realm, this predictor variable does not. My intent here is to capture

nonpolitical civic, community and religious organizational involvement that, as described in

Verba et al. (1995) have the capacity to confer civic skills and orientations that in turn may

result in increased political behavior.

In addition to testing the direct effect of organizational involvement, I create an

interaction term between organizational involvement and state immigration policy index to

test the indirect effect of organizational involvement on political participation specified in

Hypothesis 2. I expect that, if organizations are instrumental in working with communities to

translate the threat environment created by negative state immigration policies into political

action, this interaction term will be positively associated with the outcome of non-voting

political engagement.

CONTROL VARIABLES: I include a vector of control variables that are known to

contribute to political participation among the general U.S. population. These include

education, income, age, sex, and marital status. All of these individual-level variables may be

viewed as providing resources known to support political participation, consistent with the

resource model of political participation, which predicts that an individual’s position in the

economy generates resources available for political participation (Brady, Verba and

Schlozman 1995; Verba et al. 1995). I expect all of these factors to be positively associated

with non-voting political participation across the models, although they are included

primarily as control variables so that I may focus on my primary predictors of interest.

Finally, I include control variables for each year included in the analysis so as to account for

109

potential election-cycle effects conferred by different years represented in the dataset. Just as

presidential election years typically produce higher rates of voting participation (Burden

2000), I expect that non-voting political participation will be higher in presidential voting

years and lower in non-presidential years. Since 2008 is the only presidential election year

included in the analysis, I expect that the mid-term status of all other years will exert negative

effects on non-voting political participation compared with 2008.

RACE/ETHNICITY: In order to adequately test my hypotheses related to effects of

the main predictors on non-voting political participation among children of immigrants of

different racial/ethnic groups, I run separate models for each of the four largest racial/ethnic

groups represented in the data, Asian, Black/African-American, Hispanic/Latino and

White/European-American. I make use of the standard race/ethnicity self-reporting

categories in the Current Population Survey, and I exclude respondents who check more than

one race. In reporting results where I refer to Asian, Black/African-American, or

White/European-American children of immigrants I am referring to non-Hispanic

populations who identify as these racial groups. All respondents identifying as

Hispanic/Latino are reported under the Hispanic/Latino ethnic category.

Findings

To confirm the distinctiveness of the state immigration policy index on non-voting

political engagement among my population of interest, U.S.-born respondents with

immigrant parents, I first compare the effect of state policy index with just the vector of

control variables among my population of interest compared with U.S.-born respondents with

U.S.-born parents and immigrants with foreign-born parents.

110

Table 5.2. Impact of State Immigration Policy on Non-Voting Political Engagement

By Nativity of Respondent & Parents

By Racial/Ethnic Group

Poisson Coefficients#

US-born with US-born with Immigrants with

Immigrant Parents US-born Foreign-Born

Parents Parents

Policy Index Coef

(robust se)

Policy Index Coef

(robust se)

Policy Index

Coef (robust se)

Asian

0.004 (.005)

N=1,297

-0.003 (.004)

N=852

-0.004(.003)

N=5,600

Black/African-American

0.020 (.012)

N=487

0.003 (.002)

N=14,011

0.0003 (.004)

N=1,660

Hispanic/Latino

-0.005* (.002)

N=3,813

0.003 (0.003)

N=4,956

0.002 (.002)

N=10,057

White/EuropeanAmerican

0.002 (.001)

N=9,694

0.003** (.001)

N=139,015

-0.004 (.002)

N=5,705

# Vector of Control Variables not Shown

*p<0.05, **p<0.01, ***p<0.001

As detailed in Table 5.2, state policy index is exerting a distinctive impact among all

of the racial/ethnic subgroups of U.S.-born respondents with immigrant parents. Although in

this initial testing state immigration policy index is not significantly predictive of non-voting

political engagement among Asian respondents, the coefficient among U.S.-born Asian

respondents with immigrant parents is positive, while it is negative among U.S.-born

respondents with U.S.-born parents and among immigrant respondents. Among

Black/African-American respondents again state immigration policy index is not

significantly predictive of non-voting political engagement; however, the positive coefficient

is substantially larger among U.S.-born respondents with immigrant parents than in the other

111

two categories of respondents. Among Hispanic/Latino respondents we see that the

coefficient for state policy index is negative and significant among U.S.-born respondents

with immigrant parents, in contrast to positive and non-significant coefficients among

U.S.born respondents with U.S.-born parents and among immigrant respondents. And finally

among White/European-American respondents the effect of state immigration policy index is

positive and non-significant among U.S.-born respondents with immigrant parents in contrast

to positive and significant among U.S.-born respondents with U.S.-born parents and negative

and non-significant among immigrant respondents. I conclude that U.S.-born respondents

with immigrant parents constitute a sufficiently distinct subpopulation in terms of the impact

of state immigration policy index on non-voting political engagement to warrant a more

detailed analysis.

Table 5.3 details the results of my more detailed analyses of the impact of state

immigration policy index on non-voting political engagement among U.S.-born respondents

with immigrant parents from each of the four largest racial/ethnic subgroups. In addition to

the basic model with a vector of controls, here I also model the direct and indirect effect of

organizational involvement on the outcome of non-voting political engagement. For the most

part, the vector of control variables perform as expected, although they are not significantly

predictive in all models. Education beyond high school is positively and significantly

predictive of non-voting political engagement among Asians, Blacks/African-Americans,

Hispanic/Latinos, and Whites/European-Americans. Higher family income is positively and

significantly predictive of non-voting political engagement among

Whites/EuropeanAmericans only. Age is significantly and positively associated with non-

voting political engagement only among Asians and Hispanic/Latinos, sex (female) is

significant and positive only among Whites/European-Americans in the second model, and

112

married is significant and positive only among Hispanic/Latinos in the second model. Finally,

as expected years 2009, 2010, and 2011 are negatively and significantly associated with

nonvoting political engagement as compared with 2008 in most of the models tested.

In the model pairs for Asian, African-American/Black, and Whites/European

American U.S.-born respondents with immigrant parents, state immigration policy index is

not significantly associated with non-voting political engagement either before or after the

introduction of organizational involvement variables. Also, in all three of these racial/ethnic

groups, organizational involvement has a significant and substantial direct effect on

nonvoting political engagement, but no significant indirect effect as tested by the interaction

term. Among U.S.-born Asian respondents with immigrant parents, the log of the expected

number of non-voting political activities increases by 0.254 for each unit increase in

organizational involvement score. Among U.S.-born Black/African-American respondents

with immigrant parents the same increase in organizational involvement score yields a

significant increase in the log of expected non-voting political activities by 0.366. And

among U.S.-born White/European-American respondents with immigrant parents a one-unit

increase in organizational involvement produces a significant increase in the log of the

expected number of non-voting political activities of 0.209. This direct effect size is greatest

among U.S.-born Black/African-American respondents with immigrant parents, and a

comparison of the organizational involvement coefficients using

Table 5.3. Policy and Organizational Predictors of Non-Voting Political Engagement Among

U.S.-Born Respondents with Immigrant Parents

Asian

Black/African-American

Without

Organizational

Predictors

With

Organization

al Predictors

Without

Organizational

Predictors

With

Organizational

Predictors

Coef (robust se)

Coef (robust

se)

Coef (robust se)

113

State Immigration

Policy Index

0.004

(.005)

-0.002 (.005)

0.198

(.012)

0.015

(.021)

Organizational

Involvement

____

0.254***

(.069)

____

0.366***

(.075)

Policy x Org

Involvement

____

0.006

(.004)

____

0.001

(.007)

Controls:

Highest Ed - HS

Diploma

-0.084

(.145)

-0.114

(.126)

0.361

(.284)

0.375

(.272)

Highest Ed – Some

College

0.562** (.195)

0.453** (.165)

0.449* (.204)

0.410

(.230)

Highest Ed – BA or

Higher

0.875*** (.245)

0.687** (.230)

1.121***

(.262)

0.807*

(.265)

Fam Income

$25K$49,999/yr

0.130

(.178)

0.184

(.157)

0.070

(.212)

-0.020

(.207)

Fam Income

$50K$09,999/yr

0.057

(.145)

0.082

(.157)

0.384

(.223)

0.228

(.211)

Fam Income

$100K+

0.244

(.145)

0.217

(.131)

0.044

(.198)

-0.098

(.196)

Age

0.005** (.002)

0.005

(.002)

0.002

(.004)

0.002

(.005)

Sex (Female)

0.125

(.078)

0.086

(.077)

0.047

(.095)

0.023

(.098)

Married

-0.054 (.100)

-0.140 (.097)

0.104

(.196)

0.033

(.184)

Year: 2009

-0.517** (.174)

-0.476**

(.183)

-0.957*

(.397)

-0.993*

(.430)

Year: 2010

-0.540** (.183)

-0.498**

(.169)

-0.667**

(.234)

-0.626*

(.233)

Year: 2011

-0.216** (.102)

-0.311**

(.112)

-0.162 (.231)

-0.113

(.213)

Constant

-1.405 (.189)

-1.187***

(.187)

-1.200 (.310)

-1.374

(.361)

N=1,297

N=1,293

N=487

N=484

*p<0.05, **p<0.01, ***p<0.001

Table 5.3. Policy and Organizational Predictors of Non-Voting Political Engagement Among

U.S.-Born Respondents with Immigrant Parents (Cont’d)

Hispanic/Latino

White/European-American

Without

Organizational

Predictors

With

Organizational

Predictors

Without

Organizational

Predictors

With

Organizational

Predictors

114

Coef (robust se)

State Immigration

Policy Index

-0.005* (.002)

-0.008***

(0.002)

0.002

(.001)

0.002

(.002)

Organizational

Involvement

____

0.260***

(.025)

____

0.209***

(.012)

Policy x Org

Involvement

____

0.004*

(0.002)

____

0.0003

(.0007)

Controls:

Highest Ed - HS

Diploma

0.263**

(.097)

0.269**

(.102)

0.290***

(.072)

0.293***

(.066)

Highest Ed – Some

College

0.694*** (.123)

0.618***

(.128)

0.723***

(.074)

0.625***

(.073)

Highest Ed – BA or

Higher

1.250*** (.114)

1.048***

(0.115)

0.936***

(.079)

0.737***

(.078)

Fam Income $25K

$49,999/yr

0.095

(.102)

0.076

(.093)

0.103*

(.051)

0.081

(.046)

Fam Income $50K-

$09,999/yr

0.065

(.112)

0.124

(.110)

0.160**

(.054)

0.125*

(.053)

Fam Income

$100K+

0.311

(.116)

0.204

(.110)

0.250***

(.044)

0.198***

(.043)

Age

0.008***

(.001)

0.007***

(.001)

0.002

(.0008)

0.001

(.0008)

Sex (Female)

-0.011 (.049)

-0.045 (.049)

-0.044 (.029)

-0.070**

(.025)

Married

-0.040 (.039)

-0.095* (.036)

0.074

(.034)

0.024

(.035)

Year: 2009

-0.517** (.183)

-0.450* (.194)

-0.031 (.053)

-0.006

(.050)

Year: 2010

-0.552*** (.089)

-0.504***

(.068)

-0.214***

(.030)

-0.197***

(.025)

Year: 2011

-0.180** (.060)

-0.172**

(.066)

0.094**

(.035)

0.098**

(.033)

Constant

-1.486*** (.073)

-1.510***

(.059)

-1.193***

(.077)

-1.210***

(.079)

N=3,813

N=3,801

N=9,694

N=9,661

*p<0.05, **p<0.01, ***p<0.001

seemingly unrelated estimation comparison procedure indicates that the direct effect of

organizational involvement on non-voting political engagement is significantly higher among

115

Black/African-American respondents with immigrant parents (coeff=0.366) than it is among

White/European-American respondents with immigrant parents (coeff=0.209; chi2=4.15,

p=0.0416).

The pattern is different among U.S.-born Hispanic/Latino respondents with immigrant

parents than among the other three racial/ethnic groups examined here. Here we see that state

immigration policy index has a significant and negative effect on non-voting political

engagement and that the introduction of organizational involvement reveals additional direct

and indirect positive effects. In the Hispanic/Latino base model, the significant impact of

state immigration policy index is such that for each one-unit increase in value on the state

policy index scale the log of the expected number of non-voting political activities decreases

by .005. In the full Hispanic/Latino model we see a direct and positive effect of

organizational involvement indicating that for each unit increase in organizational

involvement the log of the expected number of non-voting political activities increases by

0.260.

The Hispanic/Latino case is distinct from the other three racial/ethnic groups

examined here in that U.S.-born Hispanic/Latino respondents with immigrant parents also

demonstrate a significantly positive indirect effect of organizational involvement through its

interaction with state immigration policy index. Because the main effect of state policy index

remains largely the same after the introduction of the interaction term I conclude that

organizational involvement is exerting a moderating effect on the relationship between state

immigration policy index and non-voting political engagement (Baron and Kenny 1986). In

addition to the direct and negative influence of state policy index on the outcome of

nonvoting political, the presence of organizational involvement moderates the relationship

between state policy and non-voting political engagement for this group in a positive

116

direction such that higher organizational involvement in states with more negative

immigration policies results in greater mobilization to non-voting political engagement than

it does in states with more positive immigration policies.

This moderating impact is illustrated in Figure 5.1 where we see the relationship

between the predicted number of non-voting political activities and the state immigration

policy index by three levels of organizational involvement among U.S.-born Hispanic/Latino

respondents with immigrant parents. The predicted number of non-voting political activities

is highest among those respondents reporting the highest levels of organizational

involvement and lowest among those reporting the lowest levels of organizational

involvement. Among Hispanic/Latino respondents with the highest level of organizational

involvement, the predicted number of non-voting political activities ranges from a high of

2.08 in states with the most negative state immigration policy index to a low of 1.49 in states

with the most positive policy index. This means, for example, that Hispanics/Latinos with

117

high organizational involvement in Alabama in 2011 (the state-year combination with the

most negative cumulative state immigration policy index of -24) demonstrated the highest

level of predicted non-voting political engagement; while those in California in 2011 (the

state-year combination with the most positive cumulative state immigration policy index of

23) demonstrated the lowest level of predicted non-voting political engagement. Among

Hispanics/Latinos with the lowest level of organizational involvement the predicted number

of non-voting political activities ranges from a high of 0.44 in states with the most negative

state immigration policy index to a low of 0.314 in states with the most positive policy index.

Conclusion

The purpose of this chapter was to test the feed-forward effects that heightened state

immigration policy activities may be having on the non-electoral political engagement of

U.S.-born children of immigrants 2008-2011. I also explored the potential moderating role of

community organizations. The findings reported here point to three key conclusions. First,

state immigration policymaking is having the greatest impact on political engagement among

U.S.-born Hispanic/Latino children of immigrants. This study finds that among U.S.-born

children of immigrants, the impact of state policy on non-voting political engagement is seen

only among Hispanics/Latinos and it is a negative impact such that higher

(beneficial/inclusive) state policy indices are associated with the lowest levels of non-voting

political engagement and lower (punitive/exclusionary) state immigration policy indices are

associated with the highest levels of non-voting political engagement. This finding is

consistent with previous literature on increased mobilization among Latinos in California

following the anti-immigrant state initiatives of the 1990s (Barreto et al. 2005; Pantoja and

Segura 2003; Pantoja et al. 2001; Segura et al. 1996) and the threat of federal anti-immigrant

legislation in 2006 (Ramirez 2013; Pantoja et al. 2008; Rim 2009). It is also consistent with

118

the prediction of policy design-social construction theory that, although unusual,

negativelyconstructed groups may be able to use the very policies in which they find

themselves characterized as undeserving as a point of contention in mobilizing to resist and

challenge the negative constructions (Ingram and Schneider 2005). I would speculate that

because the respondents in this study are U.S.-born children of immigrants, they are at once

the target of negative policymaking and at the same time privileged in comparison to their

immigrant parents. Thus the opportunity to resist the negative constructions through

increased mobilization may be greater than for their immigrant parents.

Second, in addition to the important and direct impact local organizational

involvement is having on political engagement among U.S.-born children of immigrants from

all four racial/ethnic groups examined here, there is an additional important moderating effect

that organizational involvement is having in translating state immigration policy into

mobilization. The relationship between state immigration policy and organizational

involvement is such that organizational involvement is accelerating this relationship between

state immigration policy and non-voting political engagement among U.S.-born

Hispanics/Latinos who have immigrant parents. Mobilizing organizations play an important

role in translating perceived grievances into political action (Bowler and Segura 2012). The

finding indicates that community organizations in Hispanic/Latino communities are active in

translating the threat of anti-immigrant policymaking into political engagement, consistent

with observations about the translational role played by ethnic and immigrant organizations

in the 2006 immigrant rallies. Ramirez (2013) finds that it was the interaction of the

threatening policy environment and the presence of ethnic/immigrant mobilizing

organizations that resulted in the political activation that played out as the immigrant protests

of 2006.

119

Finally, these findings confirm the racialized nature of immigration policy in the

contemporary period. State immigration policies are inducing non-voting political

engagement among U.S.-born Hispanic/Latinos with immigrant parents, but not among

similarly-situated U.S.-born children of immigrant parents who are Asian,

Black/AfricanAmerican, or White/European-American. These results appear to confirm

previous scholarship indicating that immigration policymaking in the contemporary period is

racialized as a policy directed primarily at Hispanic/Latino residents (Sanchez 1997;

Huntington 2004; Rumbaut 2009). This study demonstrates that the racialized effects of state

immigration policy accrue not just to the immigrant targets of the policies themselves, but

also to the U.S.-born citizen children of Hispanic/Latino immigrants.

This study has implications for the further study of how policy design-social

construction theory elucidate the ways in which immigration policy may structure political

engagement of immigrants and their children, especially in the socialization processes that

may lead to the results observed here among U.S.-born children growing up in immigrant

homes.

CHAPTER 6

Toward a Structural Explanation for Erosion of the Immigrant Health Advantage:

State Immigration Policy and Adverse Experiences in Immigrant Families

This chapter examines the impact of immigration policy in the U.S. states on the

health and well-being of children being raised by immigrant parents. There are an estimated

23.2 million children in the U.S. who have at least one immigrant parent, and an estimated

eighty-four percent of these children are U.S.-born (Passel 2011). The 2000s witnessed

increased passage of state and local immigration legislation attempting to restrict access to

120

employment, housing, education, and other benefits for undocumented immigrants

(Monogon 2013; Chavez and Provine 2009), as well as an unprecedented increase in

detention and deportations of undocumented immigrants accompanied by elite rhetoric

conflating undocumented immigration with the threat of terrorism (Pope and Garrett 2012;

Hagan et al. 2008).

My interest with this chapter is in exploring the health and well-being of children of

immigrants through the lens of the immigrant health advantage, using a life-course

perspective on the development of adult health. The research question for this chapter is

“how are state immigration policies, both punitive and beneficial, impacting the health and

well-being of children with immigrant parents?” While there is some evidence from other

scholarship that state policies dealing with the allocation of social welfare benefits to

immigrants has impacted access to services (Skinner 2012; O’Leary and Sanchez 2011) and

the educational attainment of children of immigrants (Filindra et al. 2011), less well

understood is the impact that state immigration policies themselves may be having on the

health and well-being of these same children. I contend that immigration policy is a social

structure that actively allocates both material and interpretive benefits and burdens sufficient

to cause observable impacts and hypothesize that punitive (or exclusionary) state

immigration policy will result in worse and beneficial (or inclusionary) policies will result in

better health/well-being outcomes among children of immigrants.

In this chapter I first review the immigrant health advantage, adverse childhood

experiences, and how immigration policy may be impacting the health and well-being of the

children of immigrants. I then conduct bivariate and multivariate analyses to test my specific

hypotheses using a nationally-representative, 50-state survey of households with children

under 18 years of age. I find that state immigration policy has a significant and substantive

121

impact on the health and well-being of some groups of children of immigrants. Finally I

conclude with potential policy implications of the findings.

Immigrant Health Advantage

Public health scholars have long described an immigrant health advantage; in many

immigrant groups the first generation experiences better health status than their U.S.-born

counterparts. Further, the health status of immigrants often erodes with time in the U.S. and

erodes even more among the second generation. The immigrant health advantage has perhaps

been most studied concerning perinatal outcomes among Mexican-origin women, in which

Mexican immigrant women are observed to have better birth outcomes in terms of infant

mortality and low birth weight than U.S.-born Mexican-origin women (de la Rosa 2002). The

immigrant health advantage is also evident in other health outcomes and among other groups.

For example, Mexican-born women and children experience lower rates of asthma

prevalence than their U.S.-born counterparts (Subramanian et al. 2009). All-cause mortality is

lower among immigrant men and immigrant women 25 years and older in the U.S. than

among their U.S.-born counterparts, with notable advantages especially among younger

immigrants and among both Black and Hispanic immigrants (Singh and Siahpush 2001).

Less-acculturated Asian Americans have lower rates of obesity than their more acculturated

counterparts (Wang et al. 2011). African-born Blacks have been found to have better health in

terms of self-rated health status, activity limitations, and limitations due to hypertension than

U.S.-born Blacks (Read et al. 2005). A similar immigrant health advantage as been found

among a variety of immigrant groups in Canada as well (De Maio 2010).

Health scholars have explored a number of possible explanations for the observed and

pervasive erosion of the immigrant health advantage, many of which focus on the unhealthy

behavior changes that accompany the acculturation process as immigrants adjust to life in the

122

U.S. that may account for differences in health outcomes. For example, using nationally

representative data a number of scholars have found significantly higher rates of smoking and

substance abuse among U.S-born women as compared with their immigrant counterparts

across racial/ethnic groups (Lopez-Gonzalez et al. 2005) and among Hispanic/Latinos

specifically (Ojeda et al. 2008). The impact of stress associated with the acculturation process

is an especially popular line of inquiry in examining health status of immigrants themselves

(Johnson and Marchi 2009; Castro et al. 2010; Farley et al. 2005; Finch and Vega 2003) and

of children of immigrants (Smokowski et al. 2009; Unger et al. 2009; Zamboanga et al.

2009).

These acculturation-focused explanations illustrate the importance of viewing the

erosion of the immigrant health advantage from a life-course perspective, one which draws

attention to the manner in which people of color and other disadvantaged groups are

repeatedly exposed to health risk factors throughout childhood and adulthood in a way that

may cause accumulated health disadvantage over time (Gee et al. 2012; Colen 2011).

However, at the same time the individual-level focus and cultural explanations that

characterize much of the scholarship on the erosion of the immigrant health advantage

obscure the possible influence of structural and contextual factors (De Maio 2010). In her

qualitative work on the experiences of first and second-generation Mexican immigrant

women, Viruell-Fuentes (2007) suggests that U.S. social structures that socially exclude

immigrants may be of greater consequence in examining the erosion of the immigrant health

advantage than individual-focused acculturation-related behavioral explanations.

Adverse Childhood Experiences

With the Adverse Childhood Experiences Study (ACES) conducted from 1995-1997,

researchers at Kaiser Permanente in Southern California surveyed over 17,000 adult health

123

maintenance program enrollees (mean age 57 years) about their experiences with trauma,

abuse, and family dysfunction in childhood and found a persistent dose-response relationship

between adverse experiences of childhood and later adult health outcomes (Felitti et al.

1998). In dozens of peer-reviewed publications since that time, investigators have confirmed

the association between early adverse experiences and increased adult health risk behaviors

such as smoking (Anda et al. 1999), alcohol abuse (Dube et al. 2002), drug abuse (Dube et al.

2003a), sexual risk behavior (Hillis et al. 2001), and lack of physical activity (Felitti et al.

1998); poor mental health outcomes such as depressive disorders (Chapman et al. 2004),

anxiety (Edwards et al. 2003), and suicide attempts (Dube et al 2003b); and greater incidence

of chronic illnesses in adulthood such as emphysema, heart disease, stroke, diabetes,

hepatitis, and cancer (Felitti et al. 1998; Brown et al. 2013). Current theories of mechanisms

at work in these associations point to biologic and physiologic changes in the nervous,

endocrine, and immune systems that result from adverse childhood experiences and in turn

contribute to the health risk behaviors noted above (Danese and McEwen 2012). Since

tobacco use, diet and activity, and alcohol use are the top three contributors to actual causes

of death in the U.S. (McGinnis and Foege 1993), the findings of these associations were and

still are treated as a breakthrough in the public health and medical communities (Palusci

2012; Foege 1998).

Recommendations that have come out of this line of research to address the identified

problem of adverse childhood experiences have been primarily around increases in

availability of primary prevention programs aimed at preventing child abuse and neglect,

including early home visitation (Felitti et al. 1998; Anda et al. 1999; Hillis et al. 2001; Dube

et al. 2003b; Hillis et al. 2004) and expansion of early learning and public health programs

(Shonkoff et al. 2009). Additionally, scholars have called for changes in the healthcare

124

delivery system to address the issue, including improved coordination between pediatric and

adult health care on the one hand and social/legal services on the other (Dong et al. 2003);

pediatrician screening for parental drug use, child abuse, and other household dysfunction

(Dube et al. 2003a; Dong et al. 2004); screening adults for adverse childhood experiences

(Brown et al. 2013); and increased availability of early mental health treatment (Shonkoff et

al. 2009).

As might be expected, adverse experiences of childhood are not evenly distributed

across the population; rather such experiences are much more common in low-income and

other marginalized populations. Based on the 2011-2012 National Survey of Children’s

Health, the Centers for Disease Control (CDC) estimate that nationally 22.6% of all children

have experienced 2 or more adverse child or family experiences, but this risk is highest in

low-income populations where 34.8% of all children living in households with incomes less

than 100% Federal Poverty Level have experienced 2 or more adverse experiences (Data

Resource Center 2013). Conspicuously absent from research on adverse childhood

experiences are recommendations to address structural determinants that may contribute to

the unequal distribution of these experiences across society.

While no reports of ACES that this author could locate report specifically on the

health of immigrants or children of immigrants, adverse childhood experiences may provide

insight into the mechanisms at play in the observed erosion of health status from immigrant

generation to second generation among some groups. If we accept that adverse experiences

of childhood set the stage for future adult health, then examining adverse experiences among

children being raised in the U.S. by immigrant parents could provide insight into the

childhood experiences that may contribute to their later poorer health outcomes as adults.

125

Health, Well-Being and Immigration Policy

The World Health Organization’s (WHO) Social Determinants of Health framework

posits that social conditions and social structures, including public policy, are themselves key

determinants of health in any society (WHO 2008). The Social Determinants of Health

framework recognizes the important role that public policy writ large (not limited to what we

might traditionally think of as ‘health policy’ in the U.S.) plays in structuring opportunities

for families and individuals to pursue and achieve health and well-being:

“Economic processes and political decisions condition the private resources available

to individuals and shape the nature of public infrastructure – education, health

services, transportation, environmental controls, availability of food, quality of

housing, occupational health regulations – that form the ‘neo-material’ matrix of

contemporary life” (Solar and Irwin 2007, 10).

An even more active role is theorized by public policy scholars who contend that public

policy is active in distributing not only material benefits and burdens but also interpretive and

symbolic benefits and burdens (Schneider and Ingram 1997). Immigration policy itself is an

important social determinant of health for immigrants and their children in that such policies

define belongingness and racialize new immigrants (Nevins 2002; Viruell-Fuentes et al.

2012). In this way immigration policy may serve to structure the opportunities to which

immigrants, their families, and co-ethnics have access to support their health and well-being.

In the absence of passage of any major federal immigration policy by Congress, the

2000s witnessed the start of an active period of punitive immigration policymaking by the

U.S. states. While Arizona’s SB1070 (enacted in 2010) is perhaps the most well-known

among anti-immigrant state laws, Alabama, Georgia, South Carolina, and Utah have enacted

similarly broad restrictive legislation (Sinema 2012; Wallace 2014). In addition, the National

Conference of State Legislatures (NCSL) Immigration Policy Project has documented

immigration-related policymaking in all 50-states since 2005 (NCSL 2015). Scholars

126

examining this trend in state immigration policymaking have found various contributing

factors including state legislative partisanship (Zingher 2014), citizen ideology (Monogon

2013; Chavez and Provine 2009), Hispanic population (Marquez and Schraufnagel 2013),

special interest groups (Nicholson-Crotty and Nicholson-Crotty 2011), foreign-born

population (Boushey and Luedtke 2011), and the interaction between state budget pressures

and growing racialized immigrant populations (Ybarra et al. 2014).

While there is substantial scholarship examining the increase in state immigration

policymaking during this period of time, there is little scholarship examining the impact this

increase in state policymaking may be having on the health and well-being of immigrants and

their families in the states. Two notable exceptions found direct impacts on the health and

well-being of children in immigrant households, including lower enrollment of eligible

citizen children in the federal Supplemental Nutrition Assistance Program (SNAP) in states

with more restrictive immigration policy in 2008 (Skinner 2012) and immigrant families

avoiding accessing healthcare because of fear related to passage of Arizona’s SB1070 in 2010

(O’Leary and Sanchez 2011). Additionally, findings from previous periods of state

immigration-related policymaking demonstrate impacts on children of immigrants and

inform this chapter. Children whose parents benefited from legalization allowing them to

move from undocumented to legalized status in the 1980s experienced improved educational

outcomes (Bean et al. 2006). Children of immigrants demonstrated higher high school

graduation rates in states with welfare policies that are more inclusive of immigrants

following the immigrant eligibility changes of the 1990s (Filindra et al. 2011).

The relative lack of scholarship examining the impact of the increase in state

immigration policymaking beginning in the 2000s represents an important gap, especially

given the relative lack of scholarship examining the contribution that public policy and other

127

structural factors make to the erosion of the immigrant health advantage noted above. Viruell-

Fuentes et al. (2012) call attention to this gap in their contention that immigration policy is

health policy, in that in addition to materially restricting access to public services, anti-

immigrant policies send unwelcoming messages to immigrants. “The effects of antiimmigrant

policies can, thus, be far reaching in their ability to undermine the health and wellbeing of

undocumented immigrants, their families, and communities” (5).

Hypotheses

I place this chapter in the context of the immigrant health advantage and am interested

in the role that a structural factor (public policy) potentially plays in the erosion of the

immigrant health advantage from the immigrant generation to children of immigrants.

Specifically, with this chapter I seek to understand the role of state-level immigration policy

on child and family-level outcomes among families with immigrant parents from the four

largest racial/ethnic groups in the U.S. I propose that the relationship between state

immigration policy and the health/well-being of children in immigrant families is such that

more punitive (exclusionary) state immigration policy will lead to more negative outcomes

for children of immigrants, and more beneficial (inclusionary) state immigration policy will

lead to more positive outcomes for children of immigrants.

H1: State immigration policy will be positively associated with adverse child/family

experiences among children with immigrant parents.

I also posit that because state immigration policy is often racialized as Asian and Hispanic

(Massey 2013; Junn 2007; Ngai 2004), Asian and Hispanic/Latino children with immigrant

parents may experiences the impact of state immigration policies to a greater degree than

U.S.-born Black/African-American and Whites with immigrant parents.

128

H2: The relationship between state immigration policy and adverse child/family

experiences will be stronger among Asian and Hispanic/Latino children of immigrants

than among Black/African-American and White/European-American children of

immigrants.

Data and Methods

For this chapter I make use of the National Survey of Children’s Health (NSCH) from

2011-2012. The 2011-2012 survey is the third administration of the NSCH since 2003. The

NSCH is a national cross-sectional survey of parents/guardians in households with children

age 0-17 that includes questions regarding health and illness status, development, access to

healthcare, family environment, and perceptions of neighborhoods for a single child sampled

in each household. The survey is designed to produce national and state-level estimates and

utilizes a complex sampling design with states as the primary sampling unit (Blumberg et al.

2012). The 2011-2012 NSCH was administered from February 2011 through June 2012 and

resulted in 95,677 completed child-level interviews, with 1,800-2,200 completed interviews

from each of the 50 states (CDC 2013). Tabulations of the public use file for the 2011-2012

NSCH indicate that the respondents include 1,731 first generation households and another

14,095 second generation households. It is this subsample of households with immigrant

parents that forms the basis of the dataset constructed for this research. To the pooled

individual-level responses from the 2011-2012 NSCH first and second generation household

surveys I have merged in an additional state-level immigration policy variable (described

below) to construct the complete analyzable dataset. After excluding responses in which the

children are not identified with one of the four largest racial/ethnic groups and accounting for

missing data on predictor variables, I am left with a complete dataset of 6,071 interviews

representing children living in households with at least one immigrant parent with which to

conduct my analyses.

129

With this chapter I seek to understand the role of state-level immigration policy

enacted from 2003 to 2012 on child and family-level outcomes among families with

immigrant parents from the four largest racial/ethnic groups in the U.S., while controlling for

individual and family-level factors. Thus following the examination of bivariate comparisons

among the four racial/ethnic groups I move to estimate a multivariate model. Because the

dependent variable is a count of adverse experiences and the primary predictor is a state-level

variable, I estimate a count model with clustering on states. This method allows for clustering

of children and families within states, accounts for unobserved differences between states,

and appropriately adjusts standard errors producing robust standard errors. In order to test my

specific hypotheses, I make use of four such models, one for each of the four largest

racial/ethnic groups present in the NSCH 2011-2012. For all statistical modeling I make use

of Stata/IC 13.1 using the negative binomial regression command, nbreg and the margins

command for post-estimation. I use sampling weights in all estimations.

Dependent Variable

I model the dependent variable as a count of adverse child and family experiences

reported by the parent/guardian respondent in the NSCH 2011-2012 (Data Resource Center

2013). The dependent variable is a simple count variable, coded from 0-10 based on

responses to the series of 10 questions asked in the NSCH 2011-2012 adverse child and

family experiences battery of questions. Eight of the questions call for a simple yes/no

response, which I coded as 0/1. Two of the questions provide a four-point response option

[ACE1 and ACE11] including “very often,” “somewhat often,” “rarely,” and “never.” I coded

the first two responses as 1 and the second two responses as 0. I exclude all respondents

providing all other response options (“don’t know” and “refused”) from the analyses. The

possible range of the additive count dependent variable is 0-10, and 8 is the highest number

130

of adverse child/family experiences observed in the data used for these analyses, with 58% of

respondents reporting a score of zero on the dependent variable.

The NSCH 2011-2012 adverse child/family experiences battery consisted of 10

questions asked of the parent/guardian survey respondent about the child’s experiences. The

questions are detailed in Appendix C alongside the original questions from the Kaiser ACES

1995. The NSCH 2011-2012 questions were modified somewhat, apparently in part to

comport with the limitations of asking a parent/guardian. For example, the NSCH version

does not ask about sexual abuse (included in the original ACES 1995) and does ask about

racial/ethnic discrimination (not asked in the original ACES 1995).

Predictor Variables

To operationalize state immigration policies I have created index of state immigration

policies that includes both punitive (exclusionary) and beneficial (inclusionary) policies.

Consistent with other scholarship of contemporary state immigration policy (Chavez and

Provine 2009; Monogon 2013; Marquez and Schraufnagel 2013; Zingher 2014), I make use

of the National Conference of State Legislature (NCSL) dataset on state immigration policy

to identify enacted state immigration legislation from 2005-2012. Along with a team of

scholars (as described in Ybarra, Sanchez and Sanchez 2014), we have coded each piece of

enacted legislation in the NCLS dataset 2005-2012 as to direction – neutral, beneficial to

immigrants, or punitive to immigrants - and we disaggregated omnibus legislation into

separate policies. This variable is limited to enacted policies passed by the legislature, so it

excludes bills introduced and not passed and policy action taken independently by the

executive or judicial branch. Since NCSL reports no data on enacted state-level immigration

policies prior to 2005, for 2003 and 2004 I make use of a report from the Progressive States

Network (2008) that identifies integrative and punitive policies enacted by each of the 50

131

U.S. states from 1997-2008. For these policies I rely on the policy direction identified in the

publication itself.

I agree with other authors using NCSL data (for example see Nicholson-Crotty and

Nicholson-Crotty 2011) that combining punitive and beneficial state immigration policy into

a single index makes sense both theoretically and empirically. Theoretically, research on state

welfare and economic development policies demonstrates that in the absence of strict control

over who may enter and leave their local jurisdictions, state and local policymakers make use

of a mix of policy burdens and benefits to either attract or repel certain populations (Bailey

and Rom 2004; Eisinger 2000). Empirically, many states seem to be passing both punitive

and beneficial policies in the same years. As detailed in Figure 1.1 in Chapter 1, between

2003-2012 states enacted 1,179 either punitive or beneficial immigration-related policies

(author tabulation), and of these all states enacted both punitive and beneficial policies. Even

the state enacting the least immigration legislation, Wisconsin, enacted one policy coded as

punitive and one coded as beneficial during this period.

To represent the cumulative burden/benefit created by immigration policies enacted

by state legislatures 2003-2012 I construct an index that includes all state immigration

policies enacted the year prior to and the final year of the NSCH 2011-2012 data collection.

Policies are individually counted -1 if punitive, +1 if beneficial and 0 if neutral. I sum all

policies enacted 2003-2012 to create a single cumulative additive index of immigration

policy climate for each state. The resulting unstandardized scores range from -27 to +36.

Among the 50 states represented in this dataset, 6.0% (n=3) states have a value of zero,

132

52.0% (n=26) have negative values and 42.0% (n=21) have positive values. I then convert the

additive index to a z-score to standardize the variable across all states with the resulting

distribution of index scores across states having a mean of zero, a standard deviation of 1.0,

and a range of -2.232 to 3.289. The distribution of the resulting Standardized Cumulative

State Immigration Policy Index is illustrated by the kernel density chart in Figure 6.1. Given

that state policy may contribute to healthy/unhealthy contexts for children in immigrant

families, I expect that more positive immigration policy indices at the state level will be

associated with fewer adverse child/family experiences and more negative indices will be

associated with more child/family adverse experiences.

The NSCH 2011-2012 provides an exceptionally rich source of data for a variety of

covariates reported by the child’s parent/guardian. For purposes of this chapter I have chosen

to estimate effects for covariates known from other studies to contribute to child health and

well-being as well as a number that have either been theorized or found to contribute to the

133

immigrant health advantage. I describe the covariates here in terms of neighborhood factors,

family factors, and individual-level child factors. Although conceptually these factors may be

thought of as occurring at different levels of the child and family ecology, it is important to

recognize that they have all been operationalized at the level of the individual child and

family as all are measured as self-reported by the child’s parent/guardian. Neighborhood

factors include parent/guardian perception of neighborhood helpfulness and child safety and

urban residence (family lives in MSA). Family factors include English spoken in the home,

household poverty, two parents present in the household, total children in the home, parental

mental health, and parental coping.

134

Table 6.1. Descriptive Statistics

Children with Immigrant Parents

Variable N Mean SD Min Max

Asian (non-Hispanic)

Adverse Child/Family Exp. Total

489

0.205

0.530

Immigration Policy Index

489

1.935

1.433

-0.741

3.289

Neighborhood-People Help

489

0.946

0.225

Neighborhood-Child is Safe

489

0.851

0.356

Family Lives in MSA

489

0.873

0.333

English Language Spoken @Home

489

0.546

0.498

HH Poverty <100% FPL

489

0.257

0.438

HH Two Parents

489

0.999

0.008

Total Children in Home

489

2.040

0.773

Mental Health Poor - Mother

489

0.054

0.225

Mental Health Poor - Father

489

0.020

0.140

Parent Not Coping Well

489

0.067

0.250

Child Residential Mobility (moved)

489

1.450

1.660

Child Immigrant

489

0.188

0.391

Child 0 to 5

489

0.323

0.468

Child 6 to 12

489

0.381

0.486

Child 13 to 17

489

0.297

0.457

Black/African-American (non-Hispanic)

Adverse Child/Family Exp. Total

487

0.628

1.057

Immigration Policy Index

487

0..449

1.260

-2.231

3.289

Neighborhood-People Help

487

0.791

0.407

Neighborhood-Child is Safe

487

0.802

0.399

Family Lives in MSA

487

0.818

0.386

English Language Spoken @Home

487

0.851

0.356

HH Poverty <100% FPL

487

0.432

0.496

HH Two Parents

487

omitted

Total Children in Home

487

2.358

0.996

Mental Health Poor - Mother

487

0.033

0.179

Mental Health Poor - Father

487

0.026

0.160

Parent Not Coping Well

487

0.031

0.173

Child Residential Mobility (moved)

487

1.617

1.391

Child Immigrant

487

0.132

0.339

Child 0 to 5

487

0.312

0.464

Child 6 to 12

487

0.373

0.484

Child 13 to 17

487

0.315

0.465

135

Table 6.1. Descriptive Statistics (Cont’d)

Children with Immigrant Parents

Variable N Mean SD Min Max

Hispanic/Latino

Adverse Child/Family Exp. Total

3171

0.608

0.965

Immigration Policy Index

3171

1.118

1.624

-2.231

3.289

Neighborhood-People Help

3171

0.756

0.429

Neighborhood-Child is Safe

3171

0.759

0.428

Family Lives in MSA

3171

0.863

0.344

English Language Spoken @Home

3171

0.217

0.412

HH Poverty <100% FPL

3171

0.772

0.420

HH Two Parents

3171

0.998

0.046

Total Children in Home

3171

2.419

0.951

Mental Health Poor - Mother

3171

0.087

0.282

Mental Health Poor - Father

3171

0.055

0.228

Parent Not Coping Well

3171

0.091

0.287

Child Residential Mobility (moved)

3171

1.944

Child Immigrant

3171

0.118

0.323

Child 0 to 5

3171

0.396

0.489

Child 6 to 12

3171

0.386

0.487

Child 13 to 17

3171

0.219

0.414

White/European-American (non-Hispanic)

Adverse Child/Family Exp. Total

1924

0.381

0.826

Immigration Policy Index

1924

0.584

1.320

-2.231

3.289

Neighborhood-People Help

1924

0.948

0.222

Neighborhood-Child is Safe

1924

0.925

0.263

Family Lives in MSA

1924

0.795

0.404

English Language Spoken @Home

1924

0.855

0.952

HH Poverty <100% FPL

1924

0.249

0.433

HH Two Parents

1924

0.999

0.011

Total Children in Home

1924

2.221

0.914

Mental Health Poor - Mother

1924

0.012

0.108

Mental Health Poor - Father

1924

0.011

0.104

Parent Not Coping Well

1924

0.013

0.112

Child Residential Mobility (moved)

1924

1.538

1.829

Child Immigrant

1924

0.052

0.222

Child 0 to 5

1924

0.401

0.490

136

Child 6 to 12

1924

0.336

0.473

Child 13 to 17

1924

0.263

0.440

Weighted summary statistics using estat sum

Standard deviations not adjusted for clustering

Child factors include child residential mobility, child immigrant, and child age. Details of

NSCH questions and coding for each of these items are listed in Table 6.1 above.

Race/Ethnicity

In order to test hypothesis two, I choose to run separate models for the children of

immigrant parents in the four largest racial/ethnic groups. I make use of the standard

race/ethnicity reporting categories used in the NSCH 2011-2012 to identify the sample child

and include sub-analyses for Asian (non-Hispanic), Black/African American (non-Hispanic),

Hispanics/Latino, and White/European American (non-Hispanic) children. In reporting

results where I refer to Asian, Black, and White I am referring to non-Hispanic children; all

children identified as Hispanic/Latino are reported under the Hispanic/Latino ethnic category.

One important limitation in the dataset is related to how the survey assessed Asian race – the

NSCH reports the Asian race option for respondents from only ten states, and in other states

Asian respondents appear in the “other” race category. Thus the Asian model here includes

respondents from only the 10 states – California, Hawaii, Maryland,

Massachusetts, Minnesota, Nevada, New Jersey, New York, Virginia, and Washington.

Findings

The bivariate comparisons detailed in Table 6.2 reveal important differences in the

occurrence of adverse child/family experiences among families with immigrant and U.S.born

parents and among the racial/ethnic groups examined here. Across all four racial/ethnic

groups, children in families with immigrant parents experience a significantly lower mean

137

occurrence of adverse child/family experiences than do their counterparts with U.S.-born

parents. For example, among Asian children, those with immigrant parents experience a

mean of 0.2543 adverse child/family experiences while those with U.S.-born parents

experience a significantly higher mean occurrence of 0.6815. Similarly among Black

children, those with immigrant parents experience a mean of 0.7941 adverse child/family

Table 6.2. Adverse Child/Family Experiences

Among Children with Immigrant and U.S.-Born Parents

by Race/Ethnicity of Child

National Survey of Children’s Health 2011-2012

Immigrant U.S.-Born Immigrant U.S.-Born

Parents Parents Parents Parents

Mean AC/FE Mean AC/FE %2 or more %2 or more

(95% CI) (95% CI) AC/FE AC/FE

(95% CI) (95% CI)

Asian, non

Hispanic

Black, non

Hispanic

Hispanic/Latino

White/European

American, non-

0.2543

(0.0941, 0.4146)

0.7941

(0.6588, 0.9294)

0.7723

(0.7347, 0.8100)

0.5086

(0.5075, 0.6537)

0.6815

(0.4656, 0.8973)

1.2378

(1.1268, 1.3488)

1.1622

(0.9649, 1.3596)

0.8511

(0.8076, 0.8946)

4.14

(-0.78, 9.06)

18.26

(13.97, 22.56)

16.31

(14.98, 17.63)

12.75

(11.02, 14.49)

16.86

(8.65, 25.07)

31.64

(28.34, 34.95)

30.00

(25.21, 34.77)

20.63

(19.30, 21.96)

Hispanic

weighted using NSCH sampling weights experiences while those with U.S.-born parents

experience a significantly higher 1.2378 mean occurrences. This finding appears consistent

with previous findings of an immigrant advantage in other health and well-being outcomes.

In bivariate comparisons of the proportion of children experiencing two or more

adverse experiences we observe similar differences, with Asian children with immigrant

parents experiencing the lowest rate of two or more adverse child/family experiences

(mean=0.2543; 2 or more=4.14%) and Black children with immigrant parents experiencing

the highest rate (mean=0.7941; 2 or more=18.26%). The occurrence among White and

138

Hispanic children with immigrant parents is between these two extremes, with rates of two or

more adverse experiences among White children with immigrant parents lower

(mean=0.50860; 2 or more=12.75%) and rates among Hispanic children with immigrant

children higher (mean=0.7723; 2 or more 16.31%). In examining the confidence intervals of

these bivariate comparisons we see that the occurrence of adverse child/family experiences,

whether measuring the mean occurrence or the portion of children with two or more, the

higher rates among Black and Hispanic/Latino children with immigrant parents are not

significantly different from one another but are significantly higher than the rates for both

White and Asian children with immigrant parents. Finally, the occurrence for Asian children

with immigrant parents is significantly lower on both measures than the rate for White

children with immigrant parents.

Moving to the multivariate analyses, results are reported in terms of incidence-rate

ratios (IRR), interpreted here as predicting either an increase (IRR over 1.0) or a decrease

(IRRs under 1.0) in the expected number of adverse child/family experiences. The four

regression analyses presented in Table 6.3 below reveal some similarities among predictors

of adverse child/family experiences among children with immigrant parents from the four

racial/ethnic groups and some differences. Most relevant for this analysis the primary

predictor variable, state immigration

139

Table 6.3. Adverse Child/Family Experiences

Among Children with Immigrant Parents By Race/Ethnicity

Asian

Black

State Immigration Policy Index

IRR (Robust SE)

0.7198***

(0.0654)

1.0704

(0.1415)

Neighborhood-People Help

0.3180*

(0.1850)

0.5111* (0.1686)

Neighborhood-Child is Safe

1.354

(0.4979)

0.9922 (0

.0650)

Family Lives in MSA

1.5308

(0.4648)

0.8320

(0.2351)

English Language Spoken @Home

3.2508***

(0.8038)

0.9681

(0.3953)

HH Poverty <100% FPL

1.7798**

(0.3654)

2.130**

(0.4792)

HH Two Parents

0.4638**

(0.1391)

omitted due to collinearity

Total Children in Home

0.7585*

(0.1010)

0.9952

(0.1271)

Mental Health Poor - Mother

5.8000***

(1.4553)

1.2600

(0.3654)

Mental Health Poor - Father

0.7432

(0.5655)

1.8380

(0.8479)

Parent Not Coping Well

0.8884

(0.4598)

2.5022**

(0.8444)

Child Residential Mobility (moved)

1.1647*

(0.0831)

0.9923

(0.0650)

Child Immigrant

1.5254

(0.3605)

0.4909*

(0.1635)

Child 6 to 12

1.1292

(0.2416)

2.8959**

(0.9396)

Child 13 to 17

2.9363**

(0.9995)

2.8048**

(0.9952)

Constant

0.2626*

(0.1688)

0.3630

(0.2384)

489

487

Clusters

140

*p<.05; **p<.01; ***p<.001

Weighted data pw=nschwt

Table 6.3. Adverse Child/Family Experiences (Cont’d)

Among Children with Immigrant Parents By Race/Ethnicity

Hispanic/Latino

White

State Immigration Policy Index

IRR (Robust SE)

0.9419***

(0.0140)

0.8905**

(0.0435)

Neighborhood-People Help

0.8978

(0.0808)

0.6034

(0.1579)

Neighborhood-Child is Safe

0.8712

(0.0773)

0.8217

(0.2488)

Family Lives in MSA

0.9319

(0.0936)

1.1388

(0.1689)

English Language Spoken

@Home

1.4024***

(0.1064)

1.1791

(0.3913)

HH Poverty <100% FPL

1.475**

(0.1749)

2.0518***

(0.2984)

HH Two Parents

1.0692

(0.1487)

0.1759***

(0.0416)

Total Children in Home

0.9730

(0.0426)

0.8248

(0.1304)

Mental Health Poor - Mother

1.664

(0.4876)

1.2248

(0.3603)

Mental Health Poor - Father

1.021

(0.2339)

4.189***

(0.1.3078)

Parent Not Coping Well

1.7791**

(0.3687)

2.3859**

(0.7645)

Child Residential Mobility (moved)

1.160***

(0.0169)

1.0800

(0.0427)

Child Immigrant

1.5461***

(0.1265)

0.4215

(0.2153)

Child 6 to 12

1.2283**

(0.0915)

1.3774

(0.40501)

Child 13 to 17

1.402***

(0.1336)

1.5300

(0.3921)

Constant

0.2689***

(0.0459)

5.7651

(1.7501)

3171

1924

Clusters

141

*p<.05; **p<.01; ***p<.001

Weighted data pw=nschwt

policy index, is significantly and negatively associated with adverse child/family experiences

among Asian, Hispanic/Latino, and White children. Among Asian children with immigrant

parents, for every standard deviation increase in the standardized state immigration policy

index (towards more beneficial/inclusive policy), the number of expected adverse

child/family experiences decreases by 28%, controlling for all other factors. Among

Hispanic/Latino children with immigrant parents, for every standard deviation increase in the

standardized state immigration policy index the number of expected adverse child/family

experiences decrease by 5.8%, controlling for all other factors. Among White children with

immigrant parents the same increase of one standard deviation in standardized state

immigration policy index produces an expected 10.95% decrease in the number of adverse

child/family experiences.

Figures 6.2, 6.3 and 6.4 illustrate the changes in the predicted number of adverse

experiences among Asian, Hispanic/Latino, and White children with immigrant parents.

Figure 3 illustrates the relationship between state immigration policy and adverse experiences

among Asian children with immigrant parents. Of note, the lowest observed state policy

index value of -0.741 for Asian children is almost one and a half standard deviations higher

than the -2.231 observed in the other three racial/ethnic groups, indicating that the

Asian children in this 10-state sample are not residing in the states with the most punitive

(exclusionary) immigrant policy indices. Among Asian children with immigrant parents

living in states with the lowest observed value for state policy index, -0.741, the predicted

number of adverse experiences is 0.261; while for Asian children with immigrant parents

living in states with the highest (most beneficial/inclusionary) immigration policy index of

142

3.27 the predicted number of adverse experiences is 0.070. In Figure 4 we see that for

Hispanic/Latino children residing in those states with the lowest (most punitive/exclusionary)

state immigration policy index of -2.23, the

predicted number of adverse child/family experiences is 0.630; while for those residing in the

states with the highest (most beneficial/inclusionary) state immigration policy index of 3.27,

the

Figure

6.2.

Marginal

Effects

Cumulative

State

Immigration

Policy

Adverse

Experiences

–Asian

Children

with

Immigrant

Parents

-.74

-.24

.26

.76

1.26

1.76

2.26

2.76

3.26

State Policy Index

Figure

6.3.

Marginal

Effects

Cumulative

State

Immigration

Policy

Adverse

Experiences-Hispanic

Children

with

Immigrant

Parents

Predicted

Number

Adverse

Child

Experiences

1 T T

T11 T1

T T

-2.23

-1.73 -1.23

-.73

-23

.27

1.27 1.77

2.27

2.77

3.27

State

Policy

Index

Figure

6.4.

Marginal

Effects

Cumulative

State

Immigration

Policy

Adverse

Experiences

-White

Children

with

Immigrant

Parents

Predicted

Number

Adverse

Child

Experiences

TT11111

1 1 1 1

-2.23 -1.73 -1.23

-.73

-23

1.27 1.77

2.27

2.77

3.27

State

Policy

Index

143

144

predicted number of adverse experiences is 0.453. Figure 5 illustrates the same relationship

among White children with immigrant parents, although the range of predicted adverse

experiences is lower. Among White children with immigrant parents living in states with the

lowest (most punitive/exclusionary) state policy index of -2.23, the predicted number of

adverse experiences is 0.447; while for those in states with the highest (most

beneficial/inclusionary) immigration policy index of 3.27 the predicted number of adverse

experiences is 0.236.

In terms of community-related predictors, parental perceptions that their

neighborhood is one in which people help one another, or social capital, is significantly

associated with decreases in the expected number of adverse child/family experiences among

Asian (68.2%) and Black children (48.9% marginal decrease) with immigrant parents

compared with parents who do not perceive their neighborhood to have high social capital.

Living in a neighborhood that parents perceive as usually or always safe for their children

and living in a metropolitan statistical area are not significantly associated with the number

of adverse child/family experiences for children in any of the four racial/ethnic groups

examined here.

As concerns family factors, English language spoken at home is strongly and

positively associated with an expected increase in adverse child/family experiences among

Asian and Hispanic/Latino children with immigrant parents. Among Asian children with

immigrant parents, those who speak English at home experience a 3.25-fold expected

increase in the number of adverse child/family experiences compared with those who do not

speak English at home. Among Hispanic/Latino children with immigrant parents the

expected increase is 40.2% compared with those families who do not speak English at home.

145

This finding for Asian and Hispanic/Latino children appears consistent with previous

findings related to the immigrant health advantage that suggest erosion of the advantage over

time with family acculturation, given especially that English-language usage is often used as

a primary indicator of acculturation (Hunt et al. 2004). Household poverty is the one

predictor that is strongly and consistently associated with adverse child/family experiences

across all four racial/ethnic groups examined here. Among children of immigrant parents in

poverty the expected increase in the number of adverse child/family experiences is 78.0% for

Asian children, 130% for Black children, 47.5% among Hispanic/Latino children, and 105%

among White children over those children who live in households with incomes above the

100% FPL threshold. Living in a two-parent household is significantly protective in that it is

negatively associated with adverse child/family experiences among Asian and White children

with immigrant parents; such that for children with two parents in the household the expected

decrease in the number of adverse child/family experiences is 24.2% for Asian children and

93.4% for White children.

Poor mental health among mother is strongly and positively associated with an

expected 5.8-fold increase in adverse child/family experiences among Asian children with

immigrant parents. Poor mental health among father is significantly and positively associated

with an expected 4.2-fold increase in adverse experiences among White children with

immigrant parents. Poor parental coping is significantly and positively associated with

expected increases in adverse child/family experiences among Black children (2.5-fold

increase), Hispanic/Latino children (78% increase), and White children (2.4-fold increase)

with immigrant parents.

In terms of child factors, residential mobility is positively and significantly associated

with expected increases in adverse child/family experiences among Asian children (16.5%

146

increase for each addition move) and Hispanic/Latino children (16% increase for each

additional move) with immigrant parents. Child age is strongly and significantly positively

associated with increases in adverse child/family experiences among Asian, Black, and

Hispanic/Latino children with immigrant parents, but not among White children. The effect

of age is likely an exposure effect; with greater age comes greater opportunity to be exposed

to the adverse experiences. For example, Asian adolescents experience a 2.9-fold increase in

the number of expected adverse childhood experiences compared to Asian children 5 and

under, while Black adolescents experience a 2.8-fold expected increase, and Hispanic/Latino

adolescents experience a 40% expected increase compared to their younger counterparts.

Finally, Black child immigrants in a household with immigrant parents experience a

significant 50.9% decrease in the expected number of adverse child/family experiences

compared with U.S.-born Black children with immigrant parents. This is in contrast to

Hispanic/Latino child immigrants who experience a significant 54.6% increase in the

expected number of adverse experiences compared to their U.S.-born counterparts who are

also the children of immigrants. This finding is puzzling, and appears at face value to be

inconsistent with theories of erosion of the immigrant health advantage, at least among

Hispanic/Latino children with immigrant parents. This finding indicates that while Black

immigrant children in immigrant families experience a marginal advantage over their

U.S.born siblings, Hispanic/Latino children experience a marginal disadvantage. Figure 6

details the predicted number of adverse child/family experiences among immigrant and U.S.-

born children of immigrant parents by race/ethnicity for all four racial/ethnic groups

examined here. Although the predicted number of adverse experiences is not significantly

different among immigrant and U.S.-born children of immigrants for Asian and White

147

children, their results are different than those of Black and Hispanic/Latino children, and thus

they are included for comparison.

Conclusion

This chapter examines the impact of state immigration policy on the outcome of

adverse child/family experiences among children being raised by immigrant parents using a

nationally-representative 50-state survey of households with children under 18 years of age.

Further, this chapter explores the differences and similarities of results for children in

immigrant families representing the four largest U.S. racial/ethnic groups – Asian,

Black/African-American, Hispanic/Latino, and White/European-American. Adverse

childhood experiences are an important outcome in that they have been found to be

persistently associated with adult health, including mental health outcomes, health risk

behavior, and chronic illness in adulthood (Felitti et al. 1998; Brown et al. 2013; Dube et al.

Figure

6.5.

Predicted

Number

Adverse

Experiences

among

Children

Immigrant

Parents

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Asian

Black

Hispanic

White

Immigrant

‐

Born

148

2002; Chapman et al. 2004). Because adverse childhood experiences provide a potential

window into the life course development of adult health and illness, they afford an important

opportunity to gain insight into the erosion of the immigrant health advantage that occurs

over the course of multiple generations (de la Rosa 2002; Subramanian et al. 2009; Wang et

al. 2011; Read et al. 2005). This chapter fills important gaps in the existing literature both by

examining adverse childhood experiences among children in immigrant families through the

lens of the immigrant health advantage, and by extending the work of Viruell-Fuentes (2007)

in examining the role of social structures in the erosion of the immigrant health advantage

through social exclusion.

In the bivariate comparisons I find evidence that an immigrant health advantage

occurs with the outcome of adverse child/family experiences in that in each of the four

racial/ethnic groups examined the occurrence of adverse experiences is greater among

children with immigrant parents than among children with U.S.-born parents. Further, there

are significant differences in the observed number of adverse experiences among the children

in the four racial/ethnic groups examined here with Black and Hispanic/Latino children

experiencing the highest occurrence, Asian children experiencing the lowest, and White

children experiencing occurrence in between the two. Given the strong relationship between

adverse childhood experiences and adult health, the roots of racial/ethnic health disparities

are clearly evident in the differences in adverse childhood experiences observed here.

The multivariate analyses reveal that state immigration policy has a significant impact

among Asian, Hispanic/Latino, and White children of immigrants, even after controlling for

relevant community, family, and individual-level factors. The impact is in the expected

direction in that children of immigrants from these three racial/ethnic groups living in states

with more beneficial immigration policies have fewer expected adverse child/family

149

experiences, and the impact is greatest among Asian children. This finding is consistent with

Viruell-Fuentes’ (2007) contention that social structures impact immigrant health and the

erosion of the immigrant health advantage in important ways in addition to individual- and

family-level acculturative factors. Even among other significant predictors included in the

multivariate analyses as community, family, and individual-level factors are those with

structural components that suggest potentially fruitful future research. For example, poverty,

neighborhood social capital, and residential mobility suggest labor market, housing

availability, and residential isolation/segregation as related structural factors impacting the

health and well-being of children in immigrant households and the erosion of the immigrant

health advantage. However, while clearly state immigration policy is helping to explain the

erosion of the immigrant health advantage within Asian, Hispanic, and White immigrant

groups, it is not fully explaining persistent between-group differences.

I hypothesized that because immigration policy is often racialized as Asian and

Hispanic (Massey 2013; Junn 2007; Ngai 2004), the relationship between state immigration

policy and adverse child/family experiences would be strongest among these groups. The

finding that White children of immigrants are impacted as well, and in fact that the effect size

is stronger among White children than Hispanic/Latino children (IRR 0.8905 vs. 0.9419) was

unexpected. This finding suggests that the mechanisms by which state immigration policy are

impacting adverse child/family experiences among children of immigrants is less related to

racialization processes and more related to social exclusion processes that are similarly

impacting Asian, Hispanic/Latino, and White immigrant families. The lack of significant

impact of state immigration policies on adverse child/family experiences in Black immigrant

families, especially in light of the observation that Black immigrant families experience the

highest rate of adverse child/family experiences, suggests other structural factors that

150

systematically disadvantage Black immigrant families unrelated to state immigration policy

are at work. The observation that children in U.S.-born Black families experience the highest

rate of adverse child/family experiences suggests that the structural factors that result in

disadvantage for Black immigrant families may well be racialized structures unmeasured in

this study.

These findings further suggest that public policy related to immigrants is an important

tool for building healthy communities, and that the extent to which immigrant families are

actively included or excluded in community life through public policy has potential

consequences for the health and well-being of immigrant communities and their children for

multiple generations. Although the findings of this chapter are limited to state-level

immigration policymaking in the contemporary period, taken together with findings of other

scholars (Yoshikawa and Kalil 2011; Bean et al. 2011; Suarez-Orozco et al. 2011; Filindra et

al. 2011; Skinner 2012; O’Leary and Sanchez 2011), they suggest that

inclusionary/exclusionary contexts created by immigration policymaking at multiple levels -

national, state, and local - may be effecting the health and well-being of immigrant children

and families broadly with far-reaching consequences.

The English-language finding that maintenance of home-country language is

protective against adverse child/family experiences among Asian and Latino children in

immigrant households has potential practical application in public policy.

Accountabilitybased reform in public education has driven a decrease in dual language

programs and an increase in English-immersion for children who enter the public education

system with limited English abilities (Menken and Solorza 2014). In the absence of access to

dual language programs in public education that would support the maintenance of children’s

first language and the connection language provides to their parents, outside supports to

151

encourage such maintenance may help protect against adverse child/family experiences

among Asian and Hispanic/Latino children with immigrant parents.

The counter-intuitive finding here among Hispanic/Latino children that being an

immigrant is a marginal disadvantage while an advantage among Black children in terms of

adverse experiences may be due to the larger presence of undocumented immigrants in

Hispanic/Latino immigrant families (Passel 2011). The dataset used for these analyses did not

assess the legal status of immigrant children. Thus the disadvantage observed here among

Hispanic/Latino children may be pointing to the disadvantage of being a young person who is

foreign-born and also undocumented, rather than simply the effect of being foreign-born

alone. Although the vast majority of children being raised by immigrant parents in this

dataset are U.S.-born, this finding is potentially consequential and should be explored further.

Finally, this chapter contributes to the already voluminous literature on adverse childhood

experiences by examining their occurrence among children of immigrants, a heretofore

neglected group in this line of research. More importantly however, is the contribution this

chapter makes to the role of social structures in creating contexts that allocate adverse child

experiences. Just as Viruell-Fuentes (2007) called for redirecting our attention away from an

exclusive focus on individual- and family-level acculturative factors and toward social

structures in explaining the erosion of the immigrant health advantage, this chapter calls for

further examination of the social structures that set the stage for adverse childhood

experiences. Until now this literature has been dominated by recommendations of individual-

and family-level interventions designed to interrupt the progression from adverse childhood

experiences to adverse adult health. While these types of interventions may be necessary,

they will always be ameliorative unless and until the social structures that contribute to the

unequal distribution of such risks are discovered and addressed.

152

CHAPTER 7

Conclusion

“Furthermore, what distinguishes political interactions from all other kinds

of social interactions is that they are predominantly oriented toward the

authoritative allocation of values for a society.” – Easton, 1965

This dissertation sought to answer questions related to the impact of contemporary

immigration policymaking in the U.S. states on immigrant communities nationally,

specifically in terms of the political participation of adults and the well-being of children. I

have focused on the impact of immigration policy on the outcomes of interest among legal

immigrants, naturalized immigrants, and the U.S.-born children of immigrants. This

153

dissertation is one of the first studies to examine the relationship between state immigration

policymaking and the outcomes of political engagement and well-being.

This research is particularly relevant during this dynamic time of both increased

subnational immigration policymaking activity and increases in immigrant and second

generation populations. This area of study is consequential to the future of U.S. democracy

because immigrants generally have low political participation rates relative to the majority

population of U.S.-born Whites (Ramakrishnan 2005). As these groups become a larger share

of the U.S. citizen voting age population, a continued disparity in participation could have

consequential impacts on a variety of political processes – from questions of representation

and partisan alignment to issues of public policy. Given this contemporary social context, this

dissertation asked the following research questions: How do contemporary state immigration

policies influence immigrant political incorporation in the U.S.? Are the effects of such

policies observable across a range of immigrant settlement - including new legal immigrants,

naturalized citizens, and U.S.-born second generation citizens? How are such policies

influencing child well-being in immigrant families? And how do the effects of such policies

differ across immigrants identifying with the four largest U.S. racial/ethnic groups?

Gaps in the Literature

In establishing the theory and literature foundation for this dissertation I drew from

three main bodies of literature from three subfields in political science: immigrant political

incorporation literature (found mainly in comparative politics); immigrant political behavior

literature (found mainly in American politics); and policy design-social construction theory

(found in policy studies). I identified gaps and opportunities to extend empirical knowledge

in all three literatures that I proposed to address with this dissertation. In the comparative

immigrant political incorporation literature I found gaps primarily around empirical

154

applications – lack of use of representative data, lack of rigorous quantitative methods, and

lack of subnational U.S. comparisons. I identified gaps in the immigrant political behavior

literature around the need to more adequately operationalize and test the impact of public

policy on the political behavior of immigrants. Additionally, I identified the opportunity to

extend knowledge in the literature on policy design-social construction theory through

empirical application of the feed-forward proposition to immigrants as target populations of

public policy. Finally, as concerns child well-being, I proposed to extend knowledge by

examining adverse childhood experiences among children in immigrant families through the

lens of the immigrant health advantage and by examining the role of social structures in the

erosion of the immigrant health advantage.

I also proposed to extend knowledge of theory in important ways. Neither the

immigrant political incorporation literature nor the immigrant political behavior literature

contain strong theorizing of public policy as a causal factor. Both are generally limited to

viewing public policy as one component of the overall receiving country institutional context

in exclusively instrumental terms, included in empirical studies for its capacity to structure

the material contexts of reception and to confer material benefits and burdens. I proposed that

applying policy design-social construction theory to immigration policy would enrich

theorizing of immigrant political incorporation and immigrant political behavior - with

explicit acknowledgement that values are embodied in the social constructions contained in

public policy and that policy in turn actively communicates meaning and interpretive

messages of belongingness to its target population. Also related to the extension of

knowledge of theory, I proposed to test the feed-forward proposition of policy design-social

construction theory in immigration policy as it applies to adjacent populations – legal

immigrants, naturalized citizens, and U.S.-born children of immigrants – who, while not

155

always the explicit target of state policy aimed at undocumented immigrants, often occupy a

social space close to the target population.

Finally, I proposed to extend existing knowledge with this dissertation through

examination of immigration policy and immigrant political incorporation beyond Latino and

Asian immigrants to include consideration of effects across the four largest U.S. racial/ethnic

subgroups in the immigrant community. This allows for building knowledge about Black and

White immigrant communities often missing in contemporary scholarship on immigrant

political incorporation in the United States. Also and importantly, consistently examining

effects across all four of the largest U.S. racial/ethnic immigrant groups allows for

comparisons that help reveal nuances in this racialized policy area.

Review of Main Findings

The decade under study for this dissertation, 2003-2012, witnessed high levels of

policymaking around immigration in the U.S. states. Both this general increase as well as the

variation in policymaking among the states make state immigration policy during this period

of time an opportune application of the feed-forward proposition theorized by policy

designsocial construction theory. Although I conceptualize political incorporation as

occurring over multiple generations, in the absence of multi-generational longitudinal data I

have measured engagement/behavior at the level of the individual. I have made use of a

variety of nationally-representative survey datasets to seek answers to my larger research

questions by examining subpopulations – first newly-legalized immigrants, then naturalized

immigrants, and finally U.S.-born children of immigrants. I contend that examination of the

role of immigration policy in influencing the political engagement of each of these sub-

156

populations approximates a view to the larger multi-generational process that is political

incorporation of immigrants and their descendants in the contemporary U.S. context.

In Chapter 3 I examined the impact of state immigration policy on political

knowledge and intent to naturalize among new legal immigrants in the six largest

immigrantreceiving states in the nation covering the period 2003-2007. I found that state

immigration policy is strongly and positively associated with political knowledge among

Asian and Black immigrants, suggesting that beneficial (inclusive) policy is politically

mobilizing among these two groups. The same policy is strongly and negatively associated

with intent to naturalize among Asian and Hispanic immigrants, suggesting that punitive

(exclusive) policy is impacting intent to naturalize among these two groups. I also found that

political knowledge appears to be a reasonable approximation of political behavior among

this group of new legal immigrants not yet eligible to vote but that the determinants of intent

to naturalize are different. Thus intent to naturalize may not necessarily be an indicator of

political incorporation among new legal immigrants residing in these six states. Chapter 4

contains my examination of the impact of state immigration policy on voting among

naturalized immigrants nationally, 2004-2012. My findings here indicate that the random

effects of state immigration policy, as measured by an index combining punitive and

beneficial immigration policy enacted by states, were on average significantly and modestly

positively associated with increases in voting among Hispanic and White naturalized

immigrants. This effect was not significant among Asian/PI or Black immigrants.

Additionally, I found that co-ethnic concentration, as measured by percentage of the state

population sharing the same racial/ethnic identification as the individual voter, was also on

average statistically significantly and positively associated with increases in voting among

Hispanic and White immigrants, although the effect size was very small.

157

In Chapter 5 I examined the impact of state immigration policy and organizational

involvement on the non-voting political behavior of U.S.-born children of immigrants

nationally, 2008-2011. Here I found that state immigration policy has a direct negative

impact on non-voting political behavior as well as an indirect positive impact through an

interaction with organizational involvement. Both effects were significant among Hispanic

U.S.-born children of immigrants only, with no significant effects of state policy among

Asian, Black, or White children of immigrants. The moderating relationship between state

immigration policy and organizational involvement was such that organizational involvement

was found to moderate and accelerate the relationship between state immigration policy and

non-voting political engagement among U.S.-born Hispanic/Latinos who have immigrant

parents.

Finally, in Chapter 6 I took up the question of the impact of state immigration policy

on the wellbeing of children of immigrants using a nationally-representative child health

dataset from 2011-2012. I operationalized child wellbeing through the use of a scale of

adverse child/family experiences. The multivariate analyses in this chapter revealed that state

immigration policy had a significant impact among Asian, Hispanic/Latino, and White

children of immigrants, even after controlling for relevant community, family, and

individual-level factors. The impact was in the expected direction in that children of

immigrants from these three racial/ethnic groups living in states with more beneficial

immigration policies had fewer expected adverse child/family experiences, and the impact

was greatest among Asian children.

Multigenerational View of the Findings

The empirical chapters of this dissertation confirm that for some subpopulations

within the immigrant communities public policy is an active social structure conferring

158

benefits and burdens that impact political engagement and child wellbeing, with effects

persisting even after statistically controlling for other known individual-level predictors of

the outcomes examined here. Table 7.1 below summarizes the findings from each of the

empirical chapters of this dissertation in terms of the main effects of state immigration policy

on the sub-population and outcomes examined in each chapter. Viewing the six statistical

tests contained in the four empirical chapters across each of the four largest U.S. racial/ethnic

groups reveals important between-group differences. State immigration policy 2003-2012 has

had significant impact on outcomes most often among Hispanic immigrants and their

children, with the predictor having statistically significant effects in five out of six tests for

Hispanics. Significant impacts among Asians follow, with the predictor having significant

effects in three of the six tests among Asian immigrants and their children. State immigration

policies showed the fewest significant impacts among the Black immigrant community, with

significant effects in just one out of the six empirical tests. The tests also show significant

results in two of the six tests among White immigrants and their children, but recall that

Whites were the omitted reference category in the statistical models used in Chapter 3.

Table 7.1. Summary Impacts of State Immigration Policy on Outcomes of Interest

Chapter 3

New Legal

Immigrants

Chapter 4

Naturalized

Immigrants

Chapter 5

Second

Generation

Chapter 6

Second

Generation

Political

Knowl.

Intent to

Natural.

Voting

Non-voting Political

Behavior

Adverse

Child/Family

Direct

Indirect

Experiences

Asian

—

Black/African

American

159

Hispanic/Latino

—

White/European

American

omitted

—

ns=non significant

That the results, taken together, show the most consistent effects of state immigration

policy among Hispanic and Asian immigrant communities appears consistent with previous

literature that asserts the racialized nature of U.S. immigration policy (Sanchez 1997; Golash-

Boza and Darity 2008; Hopkins 2010). The finding of a positive association between state

immigration policy and voting among naturalized White immigrants was unexpected.

Although the effect size is very small, the finding of significant results is important; political

behavior among White immigrant communities is in some cases impacted by the immigration

policy environment, although not as consistently as among Hispanic and Asian immigrant

communities.

Within the Hispanic immigrant community the pattern across generations seems clear

– policy has little effect on political behavior among new legal immigrants, a modest effect

among naturalized immigrants, and its strongest effects among the children of immigrants.

This pattern suggests that length of exposure to the policy environment, both with time in the

U.S. and in successive generations, plays a role in producing outcomes for the Hispanic

immigrant community. The trends in the impact of state immigration policy along the

political incorporation trajectory of Asian immigrants appears less strong, with the strongest

impacts demonstrated among new legal immigrants, and no significant impact among

naturalized immigrants and the U.S.-born children of immigrants. This trend over the course

of time suggests that exposure to the policy environment, both over time for immigrants and

160

with successive generations, may not necessarily influence political incorporation in the

Asian immigrant community.

Finally, we see the strongest impacts of state immigration policy on the wellbeing

outcome of children, and this impact demonstrates a more consistent pattern than observed

with any of the political incorporation outcomes - it is in the same direction across three

racial/ethnic groups. Given that the children of immigrants under consideration in Chapter 6

are under 18 years of age, while those in Chapter 5 are adults, this finding suggests the

capacity of state immigration policy to impact child/family experiences to a greater extent

and earlier in the developmental process than it impacts political outcomes. This is consistent

with the view of scholars of comparative political incorporation who contend that political

incorporation of immigrants is one type of incorporation and that political incorporation often

occurs after other forms of social and economic incorporation (Mollenkopf and Hoschschild

2009). If the earlier effects of policy on social (in this case family) outcomes is what we are

in fact observing, then we may well expect to find evidence of state immigration policy

producing outcomes in other areas of immigrant life that reflect the social and economic

incorporation of immigrants.

In interpreting these results it is important to remember the concept of adjacency

introduced in Chapter 2. The target population for much of the punitive (exclusionary) state

immigration policymaking during the period of this examination has been undocumented

immigrants. This explicit targeting is perhaps most visible in the large omnibus legislation

passed by numerous states. For example, in 2011 Alabama Governor Bentley said in defense

of Alabama’s HB 56 (The Alabama Taxpayer and Citizen Protection Act), “I campaigned on

the issue of illegal immigration and its cost to the taxpayers of our state…this law was never

designed to hurt fellow human beings…but as Governor of this state it is my sworn duty to

161

uphold our laws, and that’s what I intend to do.” In the empirical chapters of this dissertation

I have tested the impacts of state immigration policy on people in the immigrant communities

– new legal immigrants, naturalized citizens, and U.S.-born children of immigrants – who

themselves were not generally the explicit targets of the punitive

(exclusionary) policymaking but who occupy adjacent social spaces that make them

vulnerable to experiencing spillover effects. That the impacts of state immigration policy are

statistically significant in predicting measures of political incorporation in any of the

statistical tests presented in this dissertation may be taken as evidence of the presence of such

spillover effects.

Additional Contributions

In addition to the main findings specific to each of the empirical chapters and the

multi-generational findings reviewed above, the research represented in this dissertation

makes additional important contributions to the knowledge base of the political incorporation

of immigrant communities and of policy design-social construction theory that will inform

future research in these fields. I review three additional contributions here; these include

elucidating the variation among states in the strength of the impact state immigration policy

is exerting on political participation, extending our knowledge of target group contestations

of social constructions contained in public policy, and deepening our understanding of the

important role that values play in the repeating recursive cycles of political participation and

policymaking.

The use of multi-level modeling in Chapter 4 allowed me to estimate random

coefficients for each state, which in turn provided insight into the variation in states in terms

of the strength of the effect of state immigration policy on voting among Hispanic naturalized

citizens. By displaying the (mean) random coefficients by state along with the (mean)

162

standardized state policy index score (see Figure 4.1) I illustrated that the strongest effects of

state policy on voting among Hispanic naturalized citizens tend to occur in the states with

standardized policy index scores about one standard deviation and more above and below the

mean. While it may seem intuitive to conclude that more extreme policy would likely be

associated with more voter activism and mobilization, this analysis empirically confirms this

is happening in a number of states. Importantly, examining random coefficients of state

immigration policy in this way also elucidated the extent to which California is an outlier

among states, both in terms of its enactment of beneficial (inclusive) state immigration policy

and in the strength of the impact of that policy on voting outcomes among Hispanic

naturalized immigrants.

Another important contribution is made by the findings in Chapter 5, related to the

negative effect of state immigration policy on non-voting political behavior, suggesting

contestation of social constructions contained in public policy by Hispanic U.S.-born children

of immigrants. Policy design-social construction theory predicts that, although unusual,

negatively-constructed groups may be capable of using the very policies in which they find

themselves characterized as undeserving as a point of contention in mobilizing to resist and

challenge the negative constructions (Ingram and Schneider 2005). Although the children of

immigrants examined in Chapter 5 were exclusively U.S.-born, we see examples of other

children of immigrants, undocumented youth, contesting negative constructions of them and

their families as an opportunity for mobilization. Early scholarly work finds that in the wake

of the Deferred Action for Childhood Arrivals (DACA) executive order in 2012 the

DREAMers and their allies have successfully contested a negative construction of them in

the popular media (Garcia Rodriguez 2014). Schneider and Ingram (2005) posit that public

policy works to effect political outcomes in large part by way of a social identity mechanism

163

- social identity is important in political engagement and mobilization, and at the same time

public policy contains messages about the social identity of target populations. While the

mechanisms at work that (in some cases) enable negatively constructed immigrant target

populations to contest the construction is not entirely clear, related scholarship suggests a

possible group consciousness mechanism by which anti-immigrant policymaking may be

triggering pan-ethnic solidarity and increased political organizing among target groups

(Schmidt et al. 2010). That we observe these contestation effects among Hispanic U.S.-born

children of immigrants in Chapter 5 but not among naturalized immigrants in Chapter 4

(along with the scholarship cited above on the DREAMers) suggests that this contestation is

more likely among children of immigrants than among immigrants themselves. That we

observe these contestation effects among Hispanic U.S.-born children of immigrants but not

among U.S.-born children in the other three racial/ethnic groups examined here suggests that

there is a racialized component to state immigration policy, to the process of contestation, or

both.

Applying policy design-social construction to the value-laden immigration policy

arena permitted a deepening of our understanding of the important role that values play in the

repeating cycles of political participation and policymaking described in Chapter 2. Social

constructions of target populations contained in public policy transmit values in the form of

meanings about those populations and their place in the polity (Schneider and Ingram 1997).

Policy design contains “implicit ideas, values, and broader meaning within society” that

bring about “patterns of political voice, power and democratic responsiveness” (Schneider

and Sidney 2009, 112). “The net result is a powerful influence on the behavior and

understanding of self by target populations” (Pierce et al. 2014, 6). Similarly, the process of

immigrant political incorporation is defined by the extent to which immigrants come to be

164

included in the polity. Value-laden discourse is institutionalized in laws, and laws related to

immigration construct categories of belongingness and membership that often become a part

of the social identities of those residing within U.S. borders (Nevins 2002). Political

incorporation encompasses both “membership, connoting recognition and belonging, felt by

the immigrant ‘outsiders’ and also host country insiders” and “capacity for influence, that is,

to successfully make claims in the polity” (Briggs 2013, 323).

The conceptual parallels between the work of social constructions contained in public

policy on the one hand and the work of immigrant political incorporation on the other is

striking. Both contain tension between legitimate inclusion (welcomeness, belonging) in the

polity and exclusion from the polity, and both are long-term processes defined at any point in

time by their location in the struggle between inclusion and exclusion. Given these striking

parallels, what have the findings of this dissertation applying policy design-social

construction theory to immigration policy and immigrant political incorporation revealed?

That the values embedded in the social constructions of immigration policy, because they

define and structure inclusion and exclusion, are consequential for both our democracy and

for the real lived experiences of immigrants in America. The value-embedded social

constructions contained in immigration policy matter for our democracy in that they are

structuring political behavior for large and growing subsets of the immigrant communities

whose descendants will in the future make up a substantial portion of our electorate. And

they matter for the lived experiences of immigrants in America because inclusion is

important for the healthy development of communities, of families, and of children.

Limitations of Dissertation and Future Research

The findings of the studies contained in this dissertation were intended to be

generalizable to the context of U.S. states nationally, and to move beyond the early and

165

relatively narrow studies represented in the literature thus far that apply the feed-forward

proposition of the policy design-social construction theory to immigration policy (Coutin

1998). The one exception here is the study contained in Chapter 3, where data limitations

meant that the empirical examination was limited to just the six largest immigrant-receiving

states. Although those six states contained almost two-thirds of new legal immigrants in the

national sample (64.5%), the presence of only six states limited the variation in state policy

available for the statistical test. Thus the findings from the study in Chapter 3 specifically

cannot be generalized beyond those six states.

The other main limitation of this research concerns the populations to which the

findings are applicable. As stated earlier, in these empirical chapters I test the impact of state

immigration policy on populations that occupy social spaces adjacent to the target population

of much of the punitive (exclusionary) state immigration policy enacted 2003-2012, not to

undocumented immigrants who themselves are often the targets. Thus while these findings

may be generalized to many groups in immigrant communities, the findings may not be

generalizable to undocumented immigrants. This observation speaks to the need for more

research to elucidate the impacts of the feed-forward effects of state immigration policy on

the political engagement of undocumented immigrant target populations.

Additionally, these findings point to the need for future research to extend our

knowledge on the contestation of social constructions in immigration policy and political

mobilization among Hispanic immigrant community beyond California, especially among the

children of immigrants. Most of the limited research to date on this has been done in

California (Ramakrishnan 2005; Barreto et al. 2009; Pantoja et al. 2008). While instructive,

since I found California to be an extreme outlier in the extent to which immigration policy

results in political engagement among immigrants, more research should be done to explore

166

how these processes might be the same or different in other states that have also experienced

heightened levels of state immigration policymaking. One potentially fruitful line of research

in this regard would be exploring similarities and differences between U.S.-born children of

immigrants and the DREAMers.

As concerns the outcome of child wellbeing, the findings contained in Chapter 6 point

to the need for further research into how structural factors are impacting the erosion of the

immigrant health advantage in Asian, Hispanic, and White immigrant communities. While I

have tested the impact of state immigration policy as one important structural factor, the

findings suggest that other structural factors such as neighborhood poverty, social capital,

housing, labor markets, and residential segregation may also be contributing to the erosion of

the immigrant health advantage.

Final Remarks

Finally, in terms of applied policy recommendations, the lesson contained in these

findings is that sub-national immigration policy is an important tool for social inclusion

among immigrants and their families. While much of the scholarship on sub-national

immigration policymaking to date has focused on the anti-immigrant side of the equation,

because I have made use of positive-to-negative indices to operationalize state immigration

policy in the quantitative models contained herein we are able to witness inclusive

immigration policy resulting in positive outcomes. We observe inclusive immigration policy

contributing to increased political engagement in the form of increased political knowledge

among Asian and Black new immigrants, increased voting among Hispanic and White

naturalized immigrants, and increased non-voting political participation among Hispanic

U.S.-born children of immigrants (indirect). We see even more clearly where inclusive

immigration policy is contributing to child wellbeing in Asian, Hispanic, and White

167

immigrant families. Taken together these findings suggest that inclusive sub-national

immigration policy, both state and local, could be a powerful tool for generating social

inclusion in immigrant communities – contributing to a more vibrant polity and healthier

families.

LIST OF APPENDICES

Appendix 1.A. Coding Detail for State Immigration Policy Variable

Appendix 4.A. Chapter 4 Variable Descriptions and Sources

168

Appendix 5.A. Chapter 5 Variable Descriptions and Sources

Appendix 6.A. Chapter 6 Variable Descriptions and Sources

Appendix 6.B. Adverse Childhood Experiences Questions

Appendix 1.A. Coding Detail for State Immigration Policy Variable

The raw data for our dependent variable comes from the National Conference of State Legislatures

(NCSL), which compiles a list of all state immigration laws passed by year. Our research team

drew data from NCSL between 2005 and 2012. Because we are only interested in laws that have

the potential to tangibly effect immigrant populations, we removed from our analysis several types

of bills including: bills vetoed by the governor, bills not enacted, proclamations, and resolutions.

169

Therefore, our dependent variable only captures enacted immigration policies. In general, our

dependent variable is simply a basic count of immigration policies enacted per state per year.

There are three parts to our coding process: (1) NCSL policy type, (2) ideological direction, and

(3) immigrant target population. Based on these aspects of coding and our interest in substantive

policy area, what began as a single raw total of immigration laws passed per year by state became

35 possible permutations of our initial dependent variable. These include raw totals of immigration

laws passed by state per year, totals per NSCL policy type, and totals aggregated by ideological

direction. What follows is the basic process by which all permutations of our dependent variable

were created.

Categorization of laws based on policy type was provided to us by NSCL in their raw data. NCSL

staffers placed each law into one of ten categories which include: Health and Welfare, Education,

Employment, Identification, Human Trafficking, Law Enforcement Services, Language, Property,

Elections and Miscellaneous. We retained these categories and their coding process in its entirety.

Omnibus bills were divided into their individual clauses and counted as unique policies based on

sub-topic/ideological direction combinations. This procedure allows us to include comprehensive

immigration laws in our analysis, while still allowing us to code for ideological direction. If we

had not done so, there would have been cases where legislation included individual policies with

opposite ideological directions. Therefore, without coding each policy these laws would have been

excluded from analysis. Omnibus bills are important pieces of immigration legislation because

they tend to signal a comprehensive reform; therefore, they remain in our analysis. It is important

to note that, omnibus coding procedures were used infrequently in our coding process. In total,

there were 17 omnibus bills passed between 2005 and 2012. Hereafter we use the term “policies”

to refer to enacted laws or portions of omnibus laws that were coded as separate policies.

Our major contribution to policy studies lies in the second portion of our coding protocol:

classification of policies based on their ideological direction. We created a three-category typology

to account for policy direction- (1) punitive, (2) beneficial, and (3) neutral. Using the bill summary

provided by NSCL, we coded each policy in our data set to reflect its place on our three category

ideological scale. In dealing with ideological coding, it is important to note that the majority of

policies displayed a clear ideological direction. Where there was any doubt about the ideological

direction of a policy our team examined the complete bill text for further analysis using the same

protocol describe above. All policy directions were resolved following further analysis.

Punitive immigration policies are those that are unfavorable to or adversely affect the immigrant

population residing in a state. We deem policies “unfavorable” when they put forth limitations,

exclusions, restrictions, and/or prohibitions on immigrant communities. In addition, policies

which seek to detain, deport, or lay bare the legal status of a state’s resident are also considered

unfavorable to immigrant communities. Punitive policies tend to use language such as “prohibit,”

“exclude,” “illegal,” or “limit.” The following are examples of the punitive laws we encountered.

Appendix 1.A. Coding Detail for State Immigration Policy Variable (Cont’d)

Tennessee H.B. 111, 2006: “Prohibits contractors from contracting with state Agencies

within one year of the discovery that the contractor employs illegal immigrants.”

170

Georgia S.B. 350, 2008: “This law requires that a reasonable effort be made to determine

the nationality of persons convicted of driving without a license, in addition to any person

charged with a felony or with driving under the influence who is confined to jail.”

Conversely, beneficial immigration policies are those that are favorable or advantageous to a

states’ immigrant population. We deem policies “favorable” when they enlarge qualifications for

access to government goods and services, establish supportive mechanisms for immigrant

populations, or espouse legal “blindness” to immigration status. Favorable laws tend to use

language such as “qualify,” “acceptable,” “unauthorized,” “receive,” or “assistance.” The

following are a few examples of the beneficial laws we encountered.

Virginia S.B. 821, 2005: “Allows anyone age 19 or older to waive the learner’s permit and driver’s

education requirements for a driver’s license if that person has a foreign license.”

Nebraska L.B. 239, 2006: “Allows unauthorized immigrant students to qualify for in-state

tuition.”

Finally, we classify neutral policies as those laws that, while dealing with immigration policy,

made no new impact, positive or negative, on the state’s immigrant population. Neutral policies

simply shift funding sources while maintaining extant funding levels. These policies were not

designed to newly affect immigration policy, but rather fulfill state budgetary needs. There are

only four neutral laws in the entire data set.

The Final portion of our coding protocol deals with discerning the immigrant target population that

each piece of legislation was designed to affect. We only undertake this protocol when a policy

takes on opposite ideological directions dependent on the immigrant population type. In, nearly all

cases, the distinction did not affect our coding. When there was a discrepancy, we divided the law

into distinct policies and coded them separately just as we did for omnibus legislation. In essence,

to undertake this protocol, the law must clearly call out differing provisions based on immigration

status. In our data, this situation was only encountered when the bill text treated undocumented

immigrants in a punitive manner, but legal resident aliens in a beneficial manner. The example

below is indicative of this point.

Georgia S.B. 492, 2008: “This law states that non-citizen students shall not be classified as

instate for tuition purposes unless the student is legally in the state and the board of regents

determines their in-state classification. Lawful permanent residents, refugees and asylees can

receive equal consideration for in-state tuition as U.S. citizens. International students who

reside in the United States under nonimmigrant status who do not abandon foreign domicile

shall not be eligible for in-state classification.”

Once all three coding steps where complete, we created totals for the number of policies per state

and year by ideological direction and policy type. Finally, we total the number of punitive policies

across all policy types to arrive at the dependent variable in the present analysis. The dependent

variable is the total number of punitive policies passed by each state each year from 2005-2010.

Reproduced from Ybarra, Sanchez and Sanchez (2014)

Appendix 4.A. Chapter 4 Variable Descriptions and Sources

Variable

Description

Source

171

Voted General

Election

From PES1 in CPS, 1=Yes, 0=No. Excluded all

other response options.

CPS Voting and Registration

supplements 2004-2012.

Naturalized

Immigrants

Identified naturalized immigrants from

PRCTISHP from CPS.

CPS basic survey,

evennumbered years, 2004-

2012.

Race/Ethnicity

Identified race/ethnicity from CPS

PTDTRACE and PEHSPNON. Recoded to

four mutually exclusive categories: nonHispanic

Asian/PI, Black, and White; and Hispanic.

Excluded all others.

CPS basic survey,

evennumbered years, 2004-

2012.

State policy

index

Constructed from enacted state policies for each

two-year period ending with evennumbers

years; example 2003-2004 policies aggregated

for 2004. Index=total beneficial policies

enacted – total punitive policies enacted for

each state. Converted to z-scores.

2005-2012 NCSL reporting

from www.ncsl.org , coding

per Ybarra, Sanchez &

Sanchez (2014). 2003-2004

data from Progressive States

Networks (2008).

State co-ethnic

concentration

State-level population percentage of each of the

four major racial/ethnic groups – Asian/PI

(non-Hispanic), Black (non-Hispanic),

Hispanic, and White (non-Hispanic), averaged

over each two-year period ending with

evennumbered years. Percentage racial/ethnic

concentration matched to each respondent

race/ethnicity for each year.

2005-2012 data from

American Communities

Survey (ACS) 1-year data.

Values for 2003 and 2004

estimated by interpolating

between 2000 Census and

2005 ACS using Stata ipolate

command.

Presidential year

Presidential years=2004, 2008 and 2012. Non-

presidential years=2006 and 2010.

CPS Voting and Registration

supplements, 2004-2012.

Age Continuous 18-79 age in years, 80=80-84 years CPS basic survey,

evenold, 85=85+ years old. PEAGE in CPS. numbered years, 2004-

2012.

Education

Highest level of school completed or degree

received, from PEEDUCA in CPS. Recoded so

that 1=less than HS including GED, 2=HS

diploma, 3=some college, no BA, 4=BA+.

Married

Marital status from PEMARITL in CPS.

Recoded so 1=married with spouse either

present or absent, 0=widowed, divorced,

separated or never married.

Employed

Labor force participation, from PEMLR in

CPS. Recoded so that 1=employed either at

work or absent, 0=all other possible responses.

Income

Annual family income, from HUFAMINC in

CPS. Recoded so that 1=less than $30K/yr.,

2=$30K-$59,999K/yr., 3=$60K+=/yr.

Years in U.S. From CPS PRINUSYR. Top coded to highest

year in each CPS range, then subtracted from

year of survey to estimate years in U.S.

172

Country of Polity2 scores matched to country of birth from Integration Network for

Origin CPS PENATVTY foreign country codes. Societal Conflict Research

Measures level of democracy on scale from -10

http://www.systemicpeace. (most oppressive) to +10 (most democratic).

org/inscr/inscr.htm

Appendix 5.A. Chapter 5 Variable Descriptions and Sources

Variable Description Source

Respondent U.S.-born respondents identified from PRCITSHP in CPS basic survey,

2008Nativity CPS. Recoded so that 1=Native, born in U.S. or Puerto 2011. Rico or other U.S.

Island area, or born abroad of American parents; 0= all else.

Parental Foreign-born parents identified from PEMNTVT

Nativity (mother’s nativity) and PEFNTVTY (father’s nativity) in

CPS. Recoded so 1=foreign-born to identify respondents

with foreign-born parent(s).

Race/Ethnicity Identified from PTDTRACE and PEHSPNON from CPS.

Recoded to four mutually exclusive categories – non-

Hispanic Asian, Black/African-American, and

White/European-American, and Hispanic/Latino.

Excluded all other responses.

Non-Voting

Political

Participation

Constructed from recodes of political activities in last

12 months. PEQ2, PEQ4A, and PEQ4C. Recoded to

additive scale, values 0-3. Excluding refused/ DK.

CPS Civic Engagement

Supplement, 2008-2011.

Policy Index

Constructed from all enacted state policies 2003-2011.

Variables for each state 2008-2011 represent

cumulative index of enacted policies from 2003

through that year. Index created by subtracting punitive

policies per state from beneficial policies. Each

enacted policy coded +1 if beneficial for immigrants, -

1 if punitive, and 0 if neutral; omnibus policies split

into substantive parts and counted as multiple policies.

2005-2011 NCSL

reporting from

www.ncsl.org, coding

per Ybarra, Sanchez &

Sanchez (2014). 2003-

2004 data from

Progressive States

Network (2008).

Organizational

Involvement

Organizational involvement last 12 months from

PEQ5A, PEQ5B, PEQ5C, PEQ5D, PEQ5E, PEQ6,

1=yes, 0=all else for each question. Excluding

refused/DK. Responses to each question summed to

create a summary variable with values 0-6.

CPS Civic Engagement

Supplement, 2008-2011.

Education Highest level of school completed or degree received, CPS basic survey,

2008from PEEDUCA in CPS. Recoded so that 1=less than 2011. HS including

GED, 2=HS diploma, 3=some college, no BA, 4=BA or higher.

Income

Annual family income, from HUFAMINC in CPS.

Recoded so that 1=less than $25K/yr.,

2=$25K$49,999K/yr., 3=$50K-$99,000/yr,

4=$100K+/yr.

173

Age

From PEAGE in CPS, continuous 18-79 age in years,

80=80-84 years old, 85=85+ years old.

Sex

From PESEX in CPS, coded 1=female, 0=male.

Married

Marital status from PEMARITL in CPS. Recoded so

1=married with spouse either present or absent,

0=widowed, divorced, separated or never married.

Year Dummy variables for year of survey 2008, 2009, 2010 and

2011. 2008 excluded category.

Appendix 6.A. Variable Descriptions and Sources

Variable Description Source

Policy Index

Cumulative index constructed from all enacted state

policies 2003-2012. Index created by subtracting

punitive policies per state from beneficial policies.

Each enacted policy coded +1 if beneficial for

immigrants, -1 if punitive, and 0 if neutral; omnibus

policies split into substantive parts and counted as

multiple policies. Converted to z-scores.

2005-2012 NCSL

reporting from

www.ncsl.org, coding per

Ybarra, Sanchez &

Sanchez (2014). 2003-

2004 data from

Progressive States

Network (2008).

Neighborhood- [K10Q30] “People in this neighborhood help each National Survey of

People Help other out.” 1=Yes; 0=No. Children’s Health 2011-

174

2012

Child

Age

[AGE_X] “Please tell me the [age/ages] of the [child/children] less than

18 years old living in this household.”

Appendix 6.B. Adverse Childhood Experiences Questions

Original Kaiser ACES Questions 1995 NSCH 2011-2012 ACFE Questions

While you were growing up, during your first

18 years of life:

Neighborhood-

Child is Safe

[K10Q40] “How often do you feel <child> is safe in

your community or neighborhood?” 1=Always and

Usually; 0=Sometimes and Never.

Family lives in

MSA

Zip code [C11Q22] and state [LOC_STATE] assessed

by NSCH but not reported, instead used to create

Metropolitan Statistical Area (MSA) variable that is

reported. 1=Yes; 0=No.

English

Language

[K1Q03] “What is the primary language spoken in

your home?” 1=English; 0=all other languages.

HH Poverty

<100% FPL

Household income assessed in survey questions

[K11Q52-K11Q58], combined with household size and

reported in dataset as % FPL.

HH Two

Parents

Parental membership in household assessed in

[C0Q02A].

Total Children

in Home

[S_NUMB] “How many people less than 18 years old

live in this household?” Top-coded by NSCH to 4.

Mental Health

Poor-Mother

[K9Q23] “Would you say that, in general, <child’s

mother type/your> mental and emotional health is

excellent, very good, good, fair or poor?” 1=poor and

fair; 0=excellent, very good, and good.

Mental Health

Poor-Father

[K9Q24] “Would you say that, in general, <child’s

father type/your> mental and emotional health is

excellent, very good, good, fair or poor?” 1=poor and

fair; 0=excellent, very good, and good.

Parent not

Coping Well

[K8Q30] “In general, how well do you feel you are

coping with the day-to-day demands of

[parenthood/raising children?] 1=not very well and not

very well at all; 0=somewhat well and very well.

Child

Residential

Mobility

[K11Q43] “How many times has <child> ever moved

to a new address?”

Child

Immigrant

[K11Q33] “Was <child> born in the United States?”

1=Yes; 0=No.

175

Did a parent or other adult in the household

often or very often…

Swear at you, insult you, put you down, or

humiliate you? OR

Act in a way that made you afraid that you might

be physically hurt?

No questions regarding verbal abuse or

fear of physical abuse.

Did a parent or other adult in the household

often or very often… Push, grab, slap or

throw something at you? Ever hit you so

hard that you had marks or were injured?

[ACE7] Was <child> ever the victim of

violence or witness any violence in his/her

neighborhood?

Did an adult or person at least 5 years older than

you ever…

Touch or fondle you our have you touch their

body in a sexual way? OR

Try to or actually have oral, anal, or vaginal sex

with you?

No question regarding sexual abuse.

Did you often or very often feel that…

No one in your family loved you or thought you

were important or special?

Your family didn’t look out for each other, feel

close to each other, or support each other?

No question about supportive family.

Did you often or very often feel that…

You didn’t have enough to eat, had to wear dirty

clothes, and had no one to protect you?

Your parents were too drunk or high to take care

of you or take you to the doctor if you needed

it?

[ACE1] Since <child> was born, how often

has it been very hard to get by on your

family’s income, for example, it was hard

to cover the basics like food or housing?

Were your parents ever separated or divorced?

[ACE3] Did <child> ever live with a parent

or guardian who got divorced or separated

after <child> was born?

No question regarding parental death.

[ACE4] Did <child> ever live with a parent

or guardian who died?

Was your mother or stepmother:

Often or very often pushed, grabbed, slapped, or

had something thrown at her? OR

Sometimes, often or very often kicked,

bitten, hit with a fist, or hit with something hard?

Ever repeatedly hit over at least a few minutes or

threatened with a gun or knife?

[ACE6] Did <child> ever see or hear any

parents, guardians, or any other adults in

his/her home slap, hit, kick, punch, or beat

each other up?

Appendix 6.B. Adverse Childhood Experiences Questions (Cont’d)

Original Kaiser ACES Questions 1995 NSCH 2011-2012 ACFE Questions

176

8 Did you live with anyone who was a problem

drinker or alcoholic or who used street

drugs?

[ACE9] Did <child> ever live with anyone

who had a problem with alcohol or drugs?

9 Was a household member depressed or

mentally ill or did a household member attempt

suicide?

[ACE8] Did <child> ever live with anyone

who was mentally ill or suicidal, or severely

depressed for more than a couple of weeks?

10 Did a household member go to prison?

[ACE5] Did <child> ever live with a parent

or guardian who served time in jail or

prison after <child> was born?

No questions about discrimination.

[ACE10] Was <child> ever treated or

judged unfairly because of his/her race or

ethnic group?

[ACE11] During the past year, how often

was <child> treated or judged unfairly?

ACES questionnaire from original study from: http://acestudy.org/ace_score

Students also viewed